QM Lecture Notes

Eyal Buks
Quantum Mechanics - Lecture

Notes
March 4, 2013
Technion
Preface
The dynamics of a quantum system is governed by the celebrated Schrdinger
equation
i
d
dt
[ = H[ , (0.1)
where i =

1 and = 1.05457266 10
34
J s is Plancks h-bar constant.
However, what is the meaning of the symbols [ and H? The answers will
be given in the rst part of the course (chapters 1-4), which reviews several
physical and mathematical concepts that are needed to formulate the theory
of quantum mechanics. We will learn that [ in Eq. (0.1) represents the
ket-vector state of the system and H represents the Hamiltonian operator.
The operator H is directly related to the Hamiltonian function in classical
physics, which will be dened in the rst chapter. The ket-vector state and
its physical meaning will be introduced in the second chapter. Chapter 3
reviews the position and momentum operators, whereas chapter 4 discusses
dynamics of quantum systems. The second part of the course (chapters 5-7)
is devoted to some relatively simple quantum systems including a harmonic
oscillator, spin, hydrogen atom and more. In chapter 8 we will study quantum
systems in thermal equilibrium. The third part of the course (chapters 9-13)
is devoted to approximation methods such as perturbation theory, semiclas-
sical and adiabatic approximations. Light and its interaction with matter are
the subjects of chapter 14-15. Finally, systems of identical particles will be
discussed in chapter 16 and open quantum systems in chapter 17. Most of
the material in these lecture notes is based on the textbooks [1, 2, 3, 4, 6, 7].
Contents
1. Hamiltons Formalism of Classical Physics . . . . . . . . . . . . . . . . 1
1.1 Action and Lagrangian . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.2 Principle of Least Action . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.3 Hamiltonian . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
1.4 Poissons Brackets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
1.5 Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
1.6 Solutions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
2. State Vectors and Operators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
2.1 Linear Vector Space . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
2.2 Operators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
2.3 Diracs notation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
2.4 Dual Correspondence . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
2.5 Matrix Representation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
2.6 Observables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
2.6.1 Hermitian Adjoint . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
2.6.2 Eigenvalues and Eigenvectors . . . . . . . . . . . . . . . . . . . . . . 23
2.7 Quantum Measurement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
2.8 Example - Spin 1/2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
2.9 Unitary Operators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
2.10 Trace . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
2.11 Commutation Relation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
2.12 Simultaneous Diagonalization of Commuting Operators . . . . . 35
2.13 Uncertainty Principle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
2.14 Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
2.15 Solutions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
3. The Position and Momentum Observables . . . . . . . . . . . . . . . . 49
3.1 The One Dimensional Case . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
3.1.1 Position Representation . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
3.1.2 Momentum Representation . . . . . . . . . . . . . . . . . . . . . . . . 54
3.2 Transformation Function . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
3.3 Generalization for 3D . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57
3.4 Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58
Contents
3.5 Solutions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60
4. Quantum Dynamics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69
4.1 Time Evolution Operator . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69
4.2 Time Independent Hamiltonian . . . . . . . . . . . . . . . . . . . . . . . . . . 70
4.3 Example - Spin 1/2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71
4.4 Connection to Classical Dynamics . . . . . . . . . . . . . . . . . . . . . . . . 73
4.5 Symmetric Ordering . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74
4.6 Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76
4.7 Solutions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80
5. The Harmonic Oscillator . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97
5.1 Eigenstates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 98
5.2 Coherent States . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100
5.3 Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 103
5.4 Solutions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 110
6. Angular Momentum . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 139
6.1 Angular Momentum and Rotation . . . . . . . . . . . . . . . . . . . . . . . . 140
6.2 General Angular Momentum . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 142
6.3 Simultaneous Diagonalization of J
2
and J
z
. . . . . . . . . . . . . . . . 142
6.4 Example - Spin 1/2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 147
6.5 Orbital Angular Momentum . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 147
6.6 Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 153
6.7 Solutions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 160
7. Central Potential . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 187
7.1 Simultaneous Diagonalization of the Operators H, L
2
and L
z
188
7.2 The Radial Equation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 190
7.3 Hydrogen Atom . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 192
7.4 Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 197
7.5 Solutions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 199
8. Density Operator . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 207
8.1 Time Evolution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 211
8.2 Quantum Statistical Mechanics . . . . . . . . . . . . . . . . . . . . . . . . . . . 211
8.3 Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 212
8.4 Solutions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 220
9. Time Independent Perturbation Theory . . . . . . . . . . . . . . . . . . 259
9.1 The Level E
n
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 260
9.1.1 Nondegenerate Case . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 261
9.1.2 Degenerate Case . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 263
9.2 Example . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 263
9.3 Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 266
Eyal Buks Quantum Mechanics - Lecture Notes 6
Contents
9.4 Solutions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 271
10. Time-Dependent Perturbation Theory . . . . . . . . . . . . . . . . . . . . 293
10.1 Time Evolution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 293
10.2 Perturbation Expansion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 294
10.3 The Operator O(t) = u
0
(t, t
0
) u(t, t
0
) . . . . . . . . . . . . . . . . . . . . . 295
10.4 Transition Probability. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 296
10.4.1 The Stationary Case . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 298
10.4.2 The Near-Resonance Case . . . . . . . . . . . . . . . . . . . . . . . . . 299
10.4.3 H
1
is Separable . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 300
10.5 Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 300
10.6 Solutions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 302
11. WKB Approximation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 305
11.1 WKB Wavefunction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 305
11.2 Turning Point . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 308
11.3 Bohr-Sommerfeld Quantization Rule . . . . . . . . . . . . . . . . . . . . . . 312
11.4 Tunneling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 314
11.5 Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 315
11.6 Solutions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 315
12. Path Integration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 323
12.1 Charged Particle in Electromagnetic Field . . . . . . . . . . . . . . . . . 323
12.2 Classical Limit . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 327
12.3 Aharonov-Bohm Eect . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 328
12.3.1 Two-slit Interference . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 330
12.3.2 Gauge Invariance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 331
12.4 One Dimensional Path Integrals . . . . . . . . . . . . . . . . . . . . . . . . . . 333
12.4.1 One Dimensional Free Particle . . . . . . . . . . . . . . . . . . . . . 334
12.4.2 Expansion Around the Classical Path . . . . . . . . . . . . . . . 335
12.4.3 One Dimensional Harmonic Oscillator . . . . . . . . . . . . . . . 337
12.5 Semiclassical Limit . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 341
12.6 Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 342
12.7 Solutions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 343
13. Adiabatic Approximation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 351
13.1 Momentary Diagonalization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 351
13.2 Gauge Transformation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 353
13.3 Adiabatic Limit . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 353
13.4 The Case of Two Dimensional Hilbert Space . . . . . . . . . . . . . . . 354
13.5 Transition Probability. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 356
13.5.1 The Case of Two Dimensional Hilbert Space . . . . . . . . . 357
13.6 Slow and Fast Coordinates. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 360
13.7 Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 363
13.8 Solutions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 363
Contents
14. The Quantized Electromagnetic Field . . . . . . . . . . . . . . . . . . . . . 365
14.1 Classical Electromagnetic Cavity . . . . . . . . . . . . . . . . . . . . . . . . . 365
14.2 Quantum Electromagnetic Cavity . . . . . . . . . . . . . . . . . . . . . . . . 370
14.3 Periodic Boundary Conditions . . . . . . . . . . . . . . . . . . . . . . . . . . . 372
14.4 Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 373
14.5 Solutions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 373
15. Light Matter Interaction. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 381
15.1 Hamiltonian . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 381
15.2 Transition Rates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 382
15.2.1 Spontaneous Emission . . . . . . . . . . . . . . . . . . . . . . . . . . . . 382
15.2.2 Stimulated Emission and Absorption. . . . . . . . . . . . . . . . 383
15.2.3 Selection Rules . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 384
15.3 Semiclassical Case . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 386
15.4 Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 388
15.5 Solutions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 388
16. Identical Particles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 393
16.1 Basis for the Hilbert Space . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 393
16.2 Bosons . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 396
16.3 Fermions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 397
16.4 Changing the Basis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 399
16.5 Many Particle Observables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 401
16.5.1 One-Particle Observables . . . . . . . . . . . . . . . . . . . . . . . . . . 401
16.5.2 Two-Particle Observables . . . . . . . . . . . . . . . . . . . . . . . . . . 402
16.6 Hamiltonian . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 403
16.7 Momentum Representation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 406
16.8 Spin . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 408
16.9 The Electron Gas . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 409
16.10Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 411
16.11Solutions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 412
17. Superconductivity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 423
17.1 Macroscopic Wavefunction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 423
17.1.1 Single Particle in Electromagnetic Field . . . . . . . . . . . . . 423
17.1.2 Drude Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 425
17.1.3 The Macroscopic Quantum Model . . . . . . . . . . . . . . . . . . 427
17.1.4 London Equations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 428
17.2 The Josephson Eect . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 432
17.2.1 The First Josephson Relation . . . . . . . . . . . . . . . . . . . . . . 432
17.2.2 The Second Josephson Relation . . . . . . . . . . . . . . . . . . . . 433
17.2.3 The Energy of a Josephson Junction . . . . . . . . . . . . . . . . 434
17.3 RF SQUID . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 434
17.3.1 Lagrangian . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 435
17.3.2 Flux Quantum Bit. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 438
Contents
17.3.3 Qubit Readout . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 438
17.4 BCS Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 447
17.4.1 Phonon Mediated Electron-Electron Interaction . . . . . . 447
17.4.2 The Hamiltonian . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 451
17.4.3 Bogoliubov Transformation . . . . . . . . . . . . . . . . . . . . . . . . 451
17.4.4 The Energy Gap . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 454
17.4.5 The Ground State . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 456
17.4.6 Pairing Wavefunction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 458
17.5 The Josephson Eect . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 459
17.5.1 The Second Josephson Relation . . . . . . . . . . . . . . . . . . . . 461
17.5.2 The Energy of a Josephson Junction . . . . . . . . . . . . . . . . 462
17.5.3 The First Josephson Relation . . . . . . . . . . . . . . . . . . . . . . 465
17.6 Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 466
17.7 Solutions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 467
18. Open Quantum Systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 471
18.1 Classical Resonator . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 471
18.2 Quantum Resonator Coupled to Thermal Bath . . . . . . . . . . . . . 472
18.2.1 The closed System. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 472
18.2.2 Coupling to Thermal Bath . . . . . . . . . . . . . . . . . . . . . . . . 473
18.2.3 Thermal Equilibrium . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 476
18.3 Two Level System Coupled to Thermal Bath. . . . . . . . . . . . . . . 479
18.3.1 The Closed System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 479
18.3.2 Coupling to Thermal Baths . . . . . . . . . . . . . . . . . . . . . . . . 480
18.3.3 Thermal Equilibrium . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 484
18.3.4 Correlation Functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 485
18.3.5 The Bloch Equations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 487
18.4 Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 487
18.5 Solutions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 488
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 495
Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 497
1. Hamiltons Formalism of Classical Physics
In this chapter the Hamiltons formalism of classical physics is introduced,
with a special emphasis on the concepts that are needed for quantum me-
chanics.
1.1 Action and Lagrangian
Consider a classical physical system having N degrees of freedom. The clas-
sical state of the system can be described by N independent coordinates q
n
,
where n = 1, 2, , N. The vector of coordinates is denoted by
Q = (q
1
, q
2
, , q
N
) . (1.1)
Consider the case where the vector of coordinates takes the value Q
1
at time
t
1
and the value Q
2
at a later time t
2
> t
1
, namely
Q(t
1
) = Q
1
, (1.2)
Q(t
2
) = Q
2
. (1.3)
The action S associated with the evolution of the system from time t
1
to
time t
2
is dened by
S =
t2
_
t
1
dt / , (1.4)
where / is the Lagrangian function of the system. In general, the Lagrangian
is a function of the coordinates Q, the velocities

Q and time t, namely
/ = /
_
Q,

Q; t
_
, (1.5)
where
Q = ( q
1
, q
2
, , q
N
) , (1.6)
and where overdot denotes time derivative. The time evolution of Q, in turn,
depends of the trajectory taken by the system from point Q
1
at time t
1
Chapter 1. Hamiltons Formalism of Classical Physics
t
Q
t
1
t
2
Q
2
Q
1
t
Q
t
1
t
2
Q
2
Q
1
Fig. 1.1. A trajectory taken by the system from point Q1 at time t1 to point Q2
at time t2.
to point Q
2
at time t
2
(see Fig. 1.1). For a given trajectory the time
dependency is denoted as
Q(t) = Q
(t) . (1.7)
1.2 Principle of Least Action
For any given trajectory Q(t) the action can be evaluated using Eq. (1.4).
Consider a classical system evolving in time from point Q
1
at time t
1
to point
Q
2
at time t
2
along the trajectory Q
(t). The trajectory Q
(t), which is
obtained from the laws of classical physics, has the following unique property
known as the principle of least action:
Proposition 1.2.1 (principle of least action). Among all possible trajec-
tories from point Q
1
at time t
1
to point Q
2
at time t
2
the action obtains its
minimal value by the classical trajectory Q
(t).
In a weaker version of this principle, the action obtains a local minimum
for the trajectory Q
(t). As the following theorem shows, the principle of

least action leads to a set of equations of motion, known as Euler-Lagrange
equations.
Theorem 1.2.1. The classical trajectory Q
(t), for which the action obtains

its minimum value, obeys the Euler-Lagrange equations of motion, which are
given by
1.2. Principle of Least Action
d
dt
/
q
n
=
/
q
n
, (1.8)
where n = 1, 2, , N.
Proof. Consider another trajectory Q
(t) from point Q

1
at time t
1
to point
Q
2
at time t
2
(see Fig. 1.2). The dierence
Q = Q
(t) Q
(t) = (q
1
, q
2
, , q
N
) (1.9)
is assumed to be innitesimally small. To lowest order in Q the change in
the action S is given by
S =
t2
_
t1
dt /
=
t2
_
t1
dt
_
N
n=1
/
q
n
q
n
+
N
n=1
/
q
n
q
n
_
=
t2
_
t
1
dt
_
N
n=1
/
q
n
q
n
+
N
n=1
/
q
n
d
dt
q
n
_
.
(1.10)
Integrating the second term by parts leads to
S =
t2
_
t1
dt
N
n=1
_
/
q
n

d
dt
/
q
n
_
q
n
+
N
n=1
_
/
q
n
q
n
t
2
t
1
.
(1.11)
The last term vanishes since
Q(t
1
) = Q(t
2
) = 0 . (1.12)
The principle of least action implies that
S = 0 . (1.13)
This has to be satised for any Q, therefore the following must hold
d
dt
/
q
n
=
/
q
n
. (1.14)
t
Q
t
1
t
2
Q
2
Q
1
t
Q
t
1
t
2
Q
2
Q
1
Fig. 1.2. The classical trajectory Q (t) and the trajectory Q
(t).
In what follows we will assume for simplicity that the kinetic energy T of
the system can be expressed as a function of the velocities

Q only (namely,
it does not explicitly depend on the coordinates Q). The components of the
generalized force F
n
, where n = 1, 2, , N, are derived from the potential
energy U of the system as follows
F
n
=
U
q
n
+
d
dt
U
q
n
. (1.15)
When the potential energy can be expressed as a function of the coordinates
Q only (namely, when it is independent on the velocities

Q), the system is
said to be conservative. For that case, the Lagrangian can be expressed in
terms of T and U as
/ = T U . (1.16)
Example 1.2.1. Consider a point particle having mass m moving in a one-
dimensional potential U (x). The Lagrangian is given by
/ = T U =
m x
2
2
U (x) . (1.17)
From the Euler-Lagrange equation
d
dt
/
x
=
/
x
, (1.18)
one nds that
m x =
U
x
. (1.19)
1.3. Hamiltonian
1.3 Hamiltonian
The set of Euler-Lagrange equations contains N second order dierential
equations. In this section we derive an alternative and equivalent set of equa-
tions of motion, known as Hamilton-Jacobi equations, that contains twice the
number of equations, namely 2N, however, of rst, instead of second, order.
Denition 1.3.1. The variable canonically conjugate to q
n
is dened by
p
n
=
/
q
n
. (1.20)
Denition 1.3.2. The Hamiltonian of a physical system is a function of
the vector of coordinates Q, the vector of canonical conjugate variables P =
(p
1
, p
2
, , p
N
) and time, namely
H = H(Q, P; t) , (1.21)
is dened by
H =
N
n=1
p
n
q
n
/ , (1.22)
where / is the Lagrangian.
Theorem 1.3.1. The classical trajectory satises the Hamilton-Jacobi equa-
tions of motion, which are given by
q
n
=
H
p
n
, (1.23)
p
n
=
H
q
n
, (1.24)
where n = 1, 2, , N.
Proof. The dierential of H is given by
dH = d
N
n=1
p
n
q
n
d/
=
N
n=1
_
_
_
_
_
_
q
n
dp
n
+p
n
d q
n
/
q
n
..
d
dt
L
q
n
dq
n
/
q
n
..
p
n
d q
n
_
_
_
_
_
_
/
t
dt
=
N
n=1
( q
n
dp
n
p
n
dq
n
)
/
t
dt .
(1.25)
Thus the following holds
q
n
=
H
p
n
, (1.26)
p
n
=
H
q
n
, (1.27)
/
t
=
H
t
. (1.28)
Corollary 1.3.1. The following holds
dH
dt
=
H
t
. (1.29)
Proof. Using Eqs. (1.23) and (1.24) one nds that
dH
dt
=
N
n=1
_
H
q
n
q
n
+
H
p
n
p
n
_
. .
=0
+
H
t
=
H
t
. (1.30)
The last corollary implies that H is time independent provided that H
does not depend on time explicitly, namely, provided that H/t = 0. This
property is referred to as the law of energy conservation. The theorem below
further emphasizes the relation between the Hamiltonian and the total energy
of the system.
Theorem 1.3.2. Assume that the kinetic energy of a conservative system is
given by
T =

n,m
nm
q
n
q
m
, (1.31)
where
nm
are constants. Then,the Hamiltonian of the system is given by
H = T +U , (1.32)
where T is the kinetic energy of the system and where U is the potential
energy.
Proof. For a conservative system the potential energy is independent on ve-
locities, thus
p
l
=
/
q
l
=
T
q
l
, (1.33)
where / = T U is the Lagrangian. The Hamiltonian is thus given by
1.4. Poissons Brackets
H =
N
l=1
p
l
q
l
/
=

l
T
q
l
q
l
(T U)
=

l,n,m
nm
_
_
_
_
_
q
m
q
n
q
l
..
nl
+ q
n
q
m
q
l
..
ml
_
_
_
_
_
q
l
T +U
= 2
n,m
nm
q
n
q
m
. .
T
T +U
= T +U .
(1.34)
1.4 Poissons Brackets
Consider two physical quantities F and G that can be expressed as a function
of the vector of coordinates Q, the vector of canonical conjugate variables P
and time t, namely
F = F (Q, P; t) , (1.35)
G = G(Q, P; t) , (1.36)
The Poissons brackets are dened by
F, G =
N
n=1
_
F
q
n
G
p
n

F
p
n
G
q
n
_
, (1.37)
The Poissons brackets are employed for writing an equation of motion for a
general physical quantity of interest, as the following theorem shows.
Theorem 1.4.1. Let F be a physical quantity that can be expressed as a
function of the vector of coordinates Q, the vector of canonical conjugate
variables P and time t, and let H be the Hamiltonian. Then, the following
holds
dF
dt
= F, H +
F
t
. (1.38)
Proof. Using Eqs. (1.23) and (1.24) one nds that the time derivative of F
is given by
dF
dt
=
N
n=1
_
F
q
n
q
n
+
F
p
n
p
n
_
+
F
t
=
N
n=1
_
F
q
n
H
p
n

F
p
n
H
q
n
_
+
F
t
= F, H +
F
t
.
(1.39)
Corollary 1.4.1. If F does not explicitly depend on time, namely if F/t =
0, and if F, H = 0, then F is a constant of the motion, namely
dF
dt
= 0 . (1.40)
1.5 Problems
1. Consider a particle having charge q and mass m in electromagnetic eld
characterized by the scalar potential and the vector potential A. The
electric eld E and the magnetic eld B are given by
E =
1
c
A
t
, (1.41)
and
B = A . (1.42)
Let r = (x, y, z) be the Cartesian coordinates of the particle.
a) Verify that the Lagrangian of the system can be chosen to be given
by
/ =
1
2
m r
2
q +
q
c
A r , (1.43)
by showing that the corresponding Euler-Lagrange equations are
equivalent to Newtons 2nd law (i.e., F = mr).
b) Show that the Hamilton-Jacobi equations are equivalent to Newtons
2nd law.
c) Gauge transformation The electromagnetic eld is invariant un-
der the gauge transformation of the scalar and vector potentials
A A+ , (1.44)

1
c

t
(1.45)
where = (r, t) is an arbitrary smooth and continuous function
of r and t. What eect does this gauge transformation have on the
Lagrangian and Hamiltonian? Is the motion aected?
1.6. Solutions
L C L C
Fig. 1.3. LC resonator.
2. Consider an LC resonator made of a capacitor having capacitance C in
parallel with an inductor having inductance L (see Fig. 1.3). The state
of the system is characterized by the coordinate q , which is the charge
stored by the capacitor.
a) Find the Euler-Lagrange equation of the system.
b) Find the Hamilton-Jacobi equations of the system.
c) Show that q, p = 1 .
3. Show that Poisson brackets satisfy the following relations
q
j
, q
k
= 0 , (1.46)
p
j
, p
k
= 0 , (1.47)
q
j
, p
k
=
jk
, (1.48)
F, G = G, F , (1.49)
F, F = 0 , (1.50)
F, K = 0 if K constant or F depends only on t , (1.51)
E +F, G = E, G +F, G , (1.52)
E, FG = E, F G+F E, G . (1.53)
4. Show that the Lagrange equations are coordinate invariant.
5. Consider a point particle having mass m moving in a 3D central po-
tential, namely a potential V (r) that depends only on the distance
r =
_
x
2
+y
2
+z
2
from the origin. Show that the angular momentum
L = r p is a constant of the motion.
1.6 Solutions
1. The Lagrangian of the system (in Gaussian units) is taken to be given
by
/ =
1
2
m r
2
q +
q
c
A r . (1.54)
a) The Euler-Lagrange equation for the coordinate x is given by
d
dt
/
x
=
/
x
, (1.55)
where
d
dt
/
x
= m x +
q
c
_
A
x
t
+ x
A
x
x
+ y
A
x
y
+ z
A
x
z
_
, (1.56)
and
/
x
= q
x
+
q
c
_
x
A
x
x
+ y
A
y
x
+ z
A
z
x
_
, (1.57)
thus
m x = q
x

q
c
A
x
t
. .
qEx
+
q
c
_
_
y
_
A
y
x

A
x
y
_
. .
(A)
z
z
_
A
x
z

A
z
x
_
. .
(A)
y
. .
( r(A))
x
_
_
,
(1.58)
or
m x = qE
x
+
q
c
( r B)
x
. (1.59)
Similar equations are obtained for y and z in the same way. These 3
equations can be written in a vector form as
mr = q
_
E+
1
c
r B
_
. (1.60)
b) The variable vector canonically conjugate to the coordinates vector
r is given by
p =
/
r
= m r+
q
c
A . (1.61)
The Hamiltonian is thus given by
1.6. Solutions
H = p r /
= r
_
p
1
2
m r
q
c
A
_
+q
=
1
2
m r
2
+q
=
_
p
q
c
A
_
2
2m
+q .
(1.62)
The Hamilton-Jacobi equation for the coordinate x is given by
x =
H
p
x
, (1.63)
thus
x =
p
x
q
c
A
x
m
, (1.64)
or
p
x
= m x+
q
c
A
x
. (1.65)
The Hamilton-Jacobi equation for the canonically conjugate variable
p
x
is given by
p
x
=
H
x
, (1.66)
where
p
x
= m x+
q
c
_
x
A
x
x
+ y
A
x
y
+ z
A
x
z
_
+
q
c
A
x
t
, (1.67)
and
H
x
=
q
c
_
p
x
q
c
A
x
m
A
x
x
+
p
y
q
c
A
y
m
A
y
x
+
p
z
q
c
A
z
m
A
z
x
_
q
x
=
q
c
_
x
A
x
x
+ y
A
y
x
+ z
A
z
x
_
q
x
,
(1.68)
thus
m x = q
x

q
c
A
x
t
+
q
c
_
y
_
A
y
x

A
x
y
_
z
_
A
x
z

A
z
x
__
.
(1.69)
The last result is identical to Eq. (1.59).
c) Clearly, the elds E and B, which are given by Eqs. (1.41) and (1.42)
respectively, are unchanged since
t
_
()
t
= 0 , (1.70)
and
() = 0 . (1.71)
Thus, even though both / and H are modied, the motion, which
depends on E and B only, is unaected.
2. The kinetic energy in this case T = L q
2
/2 is the energy stored in the
inductor, and the potential energy U = q
2
/2C is the energy stored in the
capacitor.
a) The Lagrangian is given by
/ = T U =
L q
2
2

q
2
2C
. (1.72)
The Euler-Lagrange equation for the coordinate q is given by
d
dt
/
q
=
/
q
, (1.73)
thus
L q +
q
C
= 0 . (1.74)
This equation expresses the requirement that the voltage across the
capacitor is the same as the one across the inductor.
b) The canonical conjugate momentum is given by
p =
/
q
= L q , (1.75)
and the Hamiltonian is given by
H = p q / =
p
2
2L
+
q
2
2C
. (1.76)
Hamilton-Jacobi equations read
q =
p
L
(1.77)
p =
q
C
, (1.78)
thus
L q +
q
C
= 0 . (1.79)
1.6. Solutions
c) Using the denition (1.37) one has
q, p =
q
q
p
p

q
p
p
q
= 1 . (1.80)
3. All these relations are easily proven using the denition (1.37).
4. Let / = /
_
Q,

Q; t
_
be a Lagrangian of a system, where Q = (q
1
, q
2
, )
is the vector of coordinates,

Q = ( q
1
, q
2
, ) is the vector of veloci-
ties, and where overdot denotes time derivative. Consider the coordinates
transformation
x
a
= x
a
(q
1
, q
2
, ..., t) , (1.81)
where a = 1, 2, . The following holds
x
a
=
x
a
q
b
q
b
+
x
a
t
, (1.82)
where the summation convention is being used, namely, repeated indices
are summed over. Moreover
/
q
a
=
/
x
b
x
b
q
a
+
/
x
b
x
b
q
a
, (1.83)
and
d
dt
_
/
q
a
_
=
d
dt
_
/
x
b
x
b
q
a
_
. (1.84)
As can be seen from Eq. (1.82), one has
x
b
q
a
=
x
b
q
a
. (1.85)
Thus, using Eqs. (1.83) and (1.84) one nds
d
dt
_
/
q
a
_
/
q
a
=
d
dt
_
/
x
b
x
b
q
a
_
/
x
b
x
b
q
a

/
x
b
x
b
q
a
=
_
d
dt
_
/
x
b
_
/
x
b
_
x
b
q
a
+
_
d
dt
_
x
b
q
a
_
x
b
q
a
_
/
x
b
.
(1.86)
As can be seen from Eq. (1.82), the second term vanishes since
x
b
q
a
=

2
x
b
q
a
q
c
q
c
+

2
x
b
tq
a
=
d
dt
_
x
b
q
a
_
,
thus
d
dt
_
/
q
a
_
/
q
a
=
_
d
dt
_
/
x
b
_
/
x
b
_
x
b
q
a
. (1.87)
The last result shows that if the coordinate transformation is reversible,
namely if det (x
b
/q
a
) ,= 0 then Lagrange equations are coordinate
invariant.
5. The angular momentum L is given by
L = r p = det
_
_
x y z
x y z
p
x
p
y
p
z
_
_
, (1.88)
where r = (x, y, z) is the position vector and where p = (p
x
, p
y
, p
z
) is the
momentum vector. The Hamiltonian is given by
H =
p
2
2m
+V (r) . (1.89)
Using
x
i
, p
j
=
ij
, (1.90)
L
z
= xp
y
yp
x
, (1.91)
one nds that
_
p
2
, L
z
_
=
_
p
2
x
, L
z
_
+
_
p
2
y
, L
z
_
+
_
p
2
z
, L
z
_
=
_
p
2
x
, xp
y
_
_
p
2
y
, yp
x
_
= 2p
x
p
y
+ 2p
y
p
x
= 0 ,
(1.92)
and
_
r
2
, L
z
_
=
_
x
2
, L
z
_
+
_
y
2
, L
z
_
+
_
z
2
, L
z
_
= y
_
x
2
, p
x
_
+
_
y
2
, p
y
_
x
= 0 .
(1.93)
Thus
_
f
_
r
2
_
, L
z
_
= 0 for arbitrary smooth function f
_
r
2
_
, and con-
sequently H, L
z
= 0. In a similar way one can show that H, L
x
=
H, L
y
= 0, and therefore
_
H, L
2
_
= 0.
2. State Vectors and Operators
In quantum mechanics the state of a physical system is described by a state
vector [, which is a vector in a vector space T, namely
[ T . (2.1)
Here, we have employed the Diracs ket-vector notation [ for the state vec-
tor, which contains all information about the state of the physical system
under study. The dimensionality of T is nite in some specic cases (no-
tably, spin systems), however, it can also be innite in many other cases
of interest. The basic mathematical theory dealing with vector spaces hav-
ing innite dimensionality was mainly developed by David Hilbert. Under
some conditions, vector spaces having innite dimensionality have properties
similar to those of their nite dimensionality counterparts. A mathematically
rigorous treatment of such vector spaces having innite dimensionality, which
are commonly called Hilbert spaces, can be found in textbooks that are de-
voted to this subject. In this chapter, however, we will only review the main
properties that are useful for quantum mechanics. In some cases, when the
generalization from the case of nite dimensionality to the case of arbitrary
dimensionality is nontrivial, results will be presented without providing a
rigorous proof and even without accurately specifying what are the validity
conditions for these results.
2.1 Linear Vector Space
A linear vector space T is a set [ of mathematical objects called vectors.
The space is assumed to be closed under vector addition and scalar multipli-
cation. Both, operations (i.e., vector addition and scalar multiplication) are
commutative. That is:
1. [ +[ = [ +[ T for every [ T and [ T
2. c [ = [ c T for every [ T and c ( (where ( is the set of
complex numbers)
A vector space with an inner product is called an inner product space.
An inner product of the ordered pair [ , [ T is denoted as [. The
inner product is a function T
2
( that satises the following properties:
Chapter 2. State Vectors and Operators
[ ( , (2.2)
[ = [
, (2.3)
(c
1
[
1
+c
2
[
2
) = c
1
[
1
+c
2
[
2
, where c
1
, c
2
( , (2.4)
[ and [ 0. Equality holds i [ = 0 .
(2.5)
Note that the asterisk in Eq. (2.3) denotes complex conjugate. Below we list
some important denitions and comments regarding inner product:
The real number
_
[ is called the norm of the vector [ T.
A normalized vector has a unity norm, namely [ = 1.
Every nonzero vector 0 ,= [ T can be normalized using the transfor-
mation
[
[
_
[
. (2.6)
The vectors [ T and [ T are said to be orthogonal if [ = 0.
A set of vectors [
n
n
, where [
n
T is called a complete orthonormal
basis if
The vectors are all normalized and orthogonal to each other, namely
m
[
n
=
nm
. (2.7)
Every [ T can be written as a superposition of the basis vectors,
namely
[ =
n
c
n
[
n
, (2.8)
where c
n
(.
By evaluating the inner product
m
[, where [ is given by Eq. (2.8)
one nds with the help of Eq. (2.7) and property (2.4) of inner products
that
m
[ =
m
_
n
c
n
[
n
_
=

n
c
n
m
[
n
. .
=nm
= c
m
. (2.9)
The last result allows rewriting Eq. (2.8) as
[ =
n
c
n
[
n
=

n
[
n
c
n
=
n
[
n

n
[ . (2.10)
2.3. Diracs notation
2.2 Operators
Operators, as the denition below states, are function from T to T:
Denition 2.2.1. An operator A : T T on a vector space maps vectors
onto vectors, namely A[ T for every [ T.
Some important denitions and comments are listed below:
The operators X : T T and Y : T T are said to be equal, namely
X = Y , if for every [ T the following holds
X[ = Y [ . (2.11)
Operators can be added, and the addition is both, commutative and asso-
ciative, namely
X +Y = Y +X , (2.12)
X + (Y +Z) = (X +Y ) +Z . (2.13)
An operator A : T T is said to be linear if
A(c
1
[
1
+c
2
[
2
) = c
1
A[
1
+c
2
A[
2
(2.14)
for every [
1
, [
2
T and c
1
, c
2
(.
The operators X : T T and Y : T T can be multiplied, where
XY [ = X (Y [) (2.15)
for any [ T.
Operator multiplication is associative
X (Y Z) = (XY ) Z = XY Z . (2.16)
However, in general operator multiplication needs not be commutative
XY ,= Y X . (2.17)
2.3 Diracs notation
In Diracs notation the inner product is considered as a multiplication of two
mathematical objects called bra and ket
[ = [
..
bra
[
..
ket
. (2.18)
While the ket-vector [ is a vector in T, the bra-vector [ represents a
functional that maps any ket-vector [ T to the complex number [.
While the multiplication of a bra-vector on the left and a ket-vector on the
right represents inner product, the outer product is obtained by reversing the
order
A
= [ [ . (2.19)
The outer product A
is clearly an operator since for any [ T the object

A
[ is a ket-vector
A
[ = ([ [) [ = [ [
. .
C
T . (2.20)
Moreover, according to property (2.4), A
is linear since for every [

1
, [
2

T and c
1
, c
2
( the following holds
A
(c
1
[
1
+c
2
[
2
) = [ [ (c
1
[
1
+c
2
[
2
)
= [ (c
1
[
1
+c
2
[
2
)
= c
1
A
[
1
+c
2
A
[
2
.
(2.21)
With Diracs notation Eq. (2.10) can be rewritten as
[ =
_
n
[
n

n
[
_
[ . (2.22)
Since the above identity holds for any [ T one concludes that the quantity
in brackets is the identity operator, which is denoted as 1, namely
1 =
n
[
n

n
[ . (2.23)
This result, which is called the closure relation, implies that any complete
orthonormal basis can be used to express the identity operator.
2.4 Dual Correspondence
As we have mentioned above, the bra-vector [ represents a functional map-
ping any ket-vector [ T to the complex number [. Moreover, since
the inner product is linear [see property (2.4) above], such a mapping is linear,
namely for every [
1
, [
2
T and c
1
, c
2
( the following holds
[ (c
1
[
1
+c
2
[
2
) = c
1
[
1
+c
2
[
2
. (2.24)
The set of linear functionals from T to (, namely, the set of functionals F : T
( that satisfy
2.4. Dual Correspondence
F (c
1
[
1
+c
2
[
2
) = c
1
F ([
1
) +c
2
F ([
2
) (2.25)
for every [
1
, [
2
T and c
1
, c
2
(, is called the dual space T
. As
the name suggests, there is a dual correspondence (DC) between T and T
,
namely a one to one mapping between these two sets, which are both linear
vector spaces. The duality relation is presented using the notation
[ [ , (2.26)
where [ T and [ T
. What is the dual of the ket-vector [ =

c
1
[
1
+c
2
[
2
, where [
1
, [
2
T and c
1
, c
2
(? To answer this question
we employ the above mentioned general properties (2.3) and (2.4) of inner
products and consider the quantity [ for an arbitrary ket-vector [ T
[ = [
= (c
1
[
1
+c
2
[
2
)
= c
1
[ +c
2
2
[
= (c
1
[ +c
2
[) [ .
(2.27)
From this result we conclude that the duality relation takes the form
c
1
[ +c
2
[ c
1
[
1
+c
2
[
2
. (2.28)
The last relation describes how to map any given ket-vector [ T
to its dual F = [ : T (, where F T
is a linear functional that

maps any ket-vector [ T to the complex number [. What is the
inverse mapping? The answer can take a relatively simple form provided that
a complete orthonormal basis exists, and consequently the identity operator
can be expressed as in Eq. (2.23). In that case the dual of a given linear
functional F : T ( is the ket-vector [F
D
T, which is given by
[F
D
=

n
(F ([
n
))
[
n
. (2.29)
The duality is demonstrated by proving the two claims below:
Claim. [
DD
= [ for any [ T, where [
DD
is the dual of the dual of
[.
Proof. The dual of [ is the bra-vector [, whereas the dual of [ is found
using Eqs. (2.29) and (2.23), thus
[
DD
=

n
[
n
. .
=
n
|
[
n
=

n
[
n

n
[
=

n
[
n

n
[
. .
=1
[
= [ .
(2.30)
Claim. F
DD
= F for any F T
, where F
DD
is the dual of the dual of F.
Proof. The dual [F
D
T of the functional F T
is given by Eq. (2.29).

Thus with the help of the duality relation (2.28) one nds that dual F
DD
T
of [F
D
is given by
F
DD
=
n
F ([
n
)
n
[ . (2.31)
Consider an arbitrary ket-vector [ T that is written as a superposition
of the complete orthonormal basis vectors, namely
[ =
m
c
m
[
m
. (2.32)
Using the above expression for F
DD
and the linearity property one nds that
F
DD
[ =

n,m
c
m
F ([
n
)
n
[
m
. .
mn
=

n
c
n
F ([
n
)
= F
_
n
c
n
[
n
_
= F ([) ,
(2.33)
therefore, F
DD
= F.
2.5 Matrix Representation
Given a complete orthonormal basis, ket-vectors, bra-vectors and linear op-
erators can be represented using matrices. Such representations are easily
obtained using the closure relation (2.23).
2.5. Matrix Representation
The inner product between the bra-vector [ and the ket-vector [ can
be written as
[ = [ 1 [
=

n
[
n

n
[
=
_
[
1
[
2

_
_
_
_
1
[
2
[
.
.
.
_
_
_ .
(2.34)
Thus, the inner product can be viewed as a product between the row vector
[ =
_
[
1
[
2

_
, (2.35)
which is the matrix representation of the bra-vector [, and the column
vector
[ =
_
_
_
1
[
2
[
.
.
.
_
_
_ , (2.36)
which is the matrix representation of the ket-vector [. Obviously, both
representations are basis dependent.
Multiplying the relation [ = X[ from the right by the basis bra-vector
m
[ and employing again the closure relation (2.23) yields
m
[ =
m
[ X[ =
m
[ X1 [ =

n
m
[ X[
n

n
[ , (2.37)
or in matrix form
_
_
_
1
[
2
[
.
.
.
_
_
_ =
_
_
_
1
[ X[
1

1
[ X[
2

2
[ X[
1

2
[ X[
2

.
.
.
.
.
.
_
_
_
_
_
_
1
[
2
[
.
.
.
_
_
_ . (2.38)
In view of this expression, the matrix representation of the linear operator
X is given by
X =
_
_
_
1
[ X[
1

1
[ X[
2

2
[ X[
1

2
[ X[
2

.
.
.
.
.
.
_
_
_ . (2.39)
Alternatively, the last result can be written as
X
nm
=
n
[ X[
m
, (2.40)
where X
nm
is the element in row n and column m of the matrix represen-
tation of the operator X.
Such matrix representation of linear operators can be useful also for mul-
tiplying linear operators. The matrix elements of the product Z = XY are
given by
m
[ Z [
n
=
m
[ XY [
n
=
m
[ X1Y [
n
=

l
m
[ X[
l

l
[ Y [
n
.
(2.41)
Similarly, the matrix representation of the outer product [ [ is given
by
[ [ =
_
_
_
1
[
2
[
.
.
.
_
_
_
_
[
1
[
2

_
=
_
_
_
1
[ [
1

1
[ [
2

2
[ [
1

2
[ [
2

.
.
.
.
.
.
_
_
_ .
(2.42)
2.6 Observables
Measurable physical variables are represented in quantum mechanics by Her-
mitian operators.
2.6.1 Hermitian Adjoint
Denition 2.6.1. The Hermitian adjoint of an operator X is denoted as X
and is dened by the following duality relation

[ X
X[ . (2.43)
Namely, for any ket-vector [ T, the dual to the ket-vector X[ is the
bra-vector [ X
.
Denition 2.6.2. An operator is said to be Hermitian if X = X
.
Below we prove some simple relations:
Claim. [ X[ = [ X
Proof. Using the general property (2.3) of inner products one has
[ X[ = [ (X[) =
__
[ X
_
[
_
= [ X
. (2.44)
Note that this result implies that if X = X
then [ X[ = [ X[
.
2.6. Observables
Claim.
_
X
= X
Proof. For any [ , [ T the following holds
[ X[ =
_
[ X[
= [ X
= [
_
X
[ , (2.45)
thus
_
X
= X.
Claim. (XY )
= Y
Proof. Applying XY on an arbitrary ket-vector [ T and employing the

duality correspondence yield
XY [ = X (Y [)
_
[ Y
_
X
= [ Y
, (2.46)
thus
(XY )
= Y
. (2.47)
Claim. If X = [ [ then X
= [ [
Proof. By applying X on an arbitrary ket-vector [ T and employing the
duality correspondence one nds that
X[ = ([ [) [ = [ ( [) ( [)
[ = [ [ = [ X
,
(2.48)
where X
= [ [.
2.6.2 Eigenvalues and Eigenvectors
Each operator is characterized by its set of eigenvalues, which is dened
below:
Denition 2.6.3. A number a
n
( is said to be an eigenvalue of an op-
erator A : T T if for some nonzero ket-vector [a
n
T the following
holds
A[a
n
= a
n
[a
n
. (2.49)
The ket-vector [a
n
is then said to be an eigenvector of the operator A with
an eigenvalue a
n
.
The set of eigenvectors associated with a given eigenvalue of an operator
A is called eigensubspace and is denoted as
T
n
= [a
n
T such that A[a
n
= a
n
[a
n
. (2.50)
Clearly, T
n
is closed under vector addition and scalar multiplication, namely
c
1
[
1
+c
2
[
2
T
n
for every [
1
, [
2
T
n
and for every c
1
, c
2
(. Thus,
the set T
n
is a subspace of T. The dimensionality of T
n
(i.e., the minimum
number of vectors that are needed to span T
n
) is called the level of degeneracy
g
n
of the eigenvalue a
n
, namely
g
n
= dimT
n
. (2.51)
As the theorem below shows, the eigenvalues and eigenvectors of a Her-
mitian operator have some unique properties.
Theorem 2.6.1. The eigenvalues of a Hermitian operator A are real. The
eigenvectors of A corresponding to dierent eigenvalues are orthogonal.
Proof. Let a
1
and a
2
be two eigenvalues of A with corresponding eigen vectors
[a
1
and [a
2
A[a
1
= a
1
[a
1
, (2.52)
A[a
2
= a
2
[a
2
. (2.53)
Multiplying Eq. (2.52) from the left by the bra-vector a
2
[, and multiplying
the dual of Eq. (2.53), which since A = A
is given by
a
2
[ A = a
2
a
2
[ , (2.54)
from the right by the ket-vector [a
1
yield
a
2
[ A[a
1
= a
1
a
2
[a
1
, (2.55)
a
2
[ A[a
1
= a
2
a
2
[a
1
. (2.56)
Thus, we have found that
(a
1
a
2
) a
2
[a
1
= 0 . (2.57)
The rst part of the theorem is proven by employing the last result (2.57)
for the case where [a
1
= [a
2
. Since [a
1
is assumed to be a nonzero ket-
vector one concludes that a
1
= a
1
, namely a
1
is real. Since this is true for
any eigenvalue of A, one can rewrite Eq. (2.57) as
(a
1
a
2
) a
2
[a
1
= 0 . (2.58)
The second part of the theorem is proven by considering the case where
a
1
,= a
2
, for which the above result (2.58) can hold only if a
2
[a
1
= 0.
Namely eigenvectors corresponding to dierent eigenvalues are orthogonal.
Consider a Hermitian operator A having a set of eigenvalues a
n
n
. Let
g
n
be the degree of degeneracy of eigenvalue a
n
, namely g
n
is the dimension
of the corresponding eigensubspace, which is denoted by T
n
. For simplic-
ity, assume that g
n
is nite for every n. Let [a
n,1
, [a
n,2
, , [a
n,g
n
be
2.6. Observables
an orthonormal basis of the eigensubspace T
n
, namely a
n,i
[a
n,i
=
ii
.
Constructing such an orthonormal basis for T
n
can be done by the so-called
Gram-Schmidt process. Moreover, since eigenvectors of A corresponding to
dierent eigenvalues are orthogonal, the following holds
a
n
,i
[a
n,i
=
nn

ii
, (2.59)
In addition, all the ket-vectors [a
n,i
are eigenvectors of A
A[a
n,i
= a
n
[a
n,i
. (2.60)
Projectors. Projector operators are useful for expressing the properties of
an observable.
Denition 2.6.4. An Hermitian operator P is called a projector if P
2
= P.
Claim. The only possible eigenvalues of a projector are 0 and 1.
Proof. Assume that [p is an eigenvector of P with an eigenvalue p, namely
P [p = p [p. Applying the operator P on both sides and using the fact that
P
2
= P yield P [p = p
2
[p, thus p (1 p) [p = 0, therefore since [p is
assumed to be nonzero, either p = 0 or p = 1.
A projector is said to project any given vector onto the eigensubspace
corresponding to the eigenvalue p = 1.
Let [a
n,1
, [a
n,2
, , [a
n,gn
be an orthonormal basis of an eigensub-
space T
n
corresponding to an eigenvalue of an observable A. Such an ortho-
normal basis can be used to construct a projection P
n
onto T
n
, which is given
by
P
n
=
gn
i=1
[a
n,i
a
n,i
[ . (2.61)
Clearly, P
n
is a projector since P
n
= P
n
and since
P
2
n
=
gn
i,i
=1
[a
n,i
a
n,i
[a
n,i

. .
ii
a
n,i
[ =
gn
i=1
[a
n,i
a
n,i
[ = P
n
. (2.62)
Moreover, it is easy to show using the orthonormality relation (2.59) that the
following holds
P
n
P
m
= P
m
P
n
= P
n
nm
. (2.63)
For linear vector spaces of nite dimensionality, it can be shown that the
set [a
n,i
n,i
forms a complete orthonormal basis of eigenvectors of a given
Hermitian operator A. The generalization of this result for the case of ar-
bitrary dimensionality is nontrivial, since generally such a set needs not be
complete. On the other hand, it can be shown that if a given Hermitian oper-
ator A satises some conditions (e.g., A needs to be completely continuous)
then completeness is guarantied. For all Hermitian operators of interest for
this course we will assume that all such conditions are satised. That is, for
any such Hermitian operator A there exists a set of ket vectors [a
n,i
, such
that:
1. The set is orthonormal, namely
a
n
,i
[a
n,i
=
nn

ii
, (2.64)
2. The ket-vectors [a
n,i
are eigenvectors, namely
A[a
n,i
= a
n
[a
n,i
, (2.65)
where a
n
.
3. The set is complete, namely closure relation [see also Eq. (2.23)] is satis-
ed
1 =

n
gn
i=1
[a
n,i
a
n,i
[ =

n
P
n
, (2.66)
where
P
n
=
gn
i=1
[a
n,i
a
n,i
[ (2.67)
is the projector onto eigen subspace T
n
with the corresponding eigenvalue
a
n
.
The closure relation (2.66) can be used to express the operator A in terms
of the projectors P
n
A = A1 =
n
gn
i=1
A[a
n,i
a
n,i
[ =

n
a
n
gn
i=1
[a
n,i
a
n,i
[ , (2.68)
that is
A =

n
a
n
P
n
. (2.69)
The last result is very useful when dealing with a function f (A) of the
operator A. The meaning of a function of an operator can be understood in
terms of the Taylor expansion of the function
f (x) =

m
f
m
x
m
, (2.70)
2.6. Observables
where
f
m
=
1
m!
d
m
f
dx
m
. (2.71)
With the help of Eqs. (2.63) and (2.69) one nds that
f (A) =

m
f
m
A
m
=

m
f
m
_
n
a
n
P
n
_
m
=

m
f
m
n
a
m
n
P
n
=

n
m
f
m
a
m
n
. .
f(an)
P
n
,
(2.72)
thus
f (A) =

n
f (a
n
) P
n
. (2.73)
Exercise 2.6.1. Express the projector P
n
in terms of the operator A and
its set of eigenvalues.
Solution 2.6.1. We seek a function f such that f (A) = P
n
. Multiplying
from the right by a basis ket-vector [a
m,i
yields
f (A) [a
m,i
=
mn
[a
m,i
. (2.74)
On the other hand
f (A) [a
m,i
= f (a
m
) [a
m,i
. (2.75)
Thus we seek a function that satisfy
f (a
m
) =
mn
. (2.76)
The polynomial function
f (a) = K

m=n
(a a
m
) , (2.77)
where K is a constant, satises the requirement that f (a
m
) = 0 for every
m ,= n. The constant K is chosen such that f (a
n
) = 1, that is
f (a) =

m=n
a a
m
a
n
a
m
, (2.78)
Thus, the desired expression is given by
P
n
=

m=n
Aa
m
a
n
a
m
. (2.79)
2.7 Quantum Measurement
Consider a measurement of a physical variable denoted as A
(c)
performed on
a quantum system. The standard textbook description of such a process is
described below. The physical variable A
(c)
is represented in quantum me-
chanics by an observable, namely by a Hermitian operator, which is denoted
as A. The correspondence between the variable A
(c)
and the operator A will
be discussed below in chapter 4. As we have seen above, it is possible to con-
struct a complete orthonormal basis made of eigenvectors of the Hermitian
operator A having the properties given by Eqs. (2.64), (2.65) and (2.66). In
that basis, the vector state [ of the system can be expressed as
[ = 1 [ =
n
gn
i=1
a
n,i
[ [a
n,i
. (2.80)
Even when the state vector [ is given, quantum mechanics does not gener-
ally provide a deterministic answer to the question: what will be the outcome
of the measurement. Instead it predicts that:
1. The possible results of the measurement are the eigenvalues a
n
of the
operator A.
2. The probability p
n
to measure the eigen value a
n
is given by
p
n
= [ P
n
[ =
g
n
i=1
[a
n,i
[[
2
. (2.81)
Note that the state vector [ is assumed to be normalized.
3. After a measurement of A with an outcome a
n
the state vector collapses
onto the corresponding eigensubspace and becomes
[
P
n
[
_
[ P
n
[
. (2.82)
It is easy to show that the probability to measure something is unity
provided that [ is normalized:
n
p
n
=
n
[ P
n
[ = [
_
n
P
n
_
[ = 1 . (2.83)
2.8. Example - Spin 1/2
We also note that a direct consequence of the collapse postulate is that two
subsequent measurements of the same observable performed one immediately
after the other will yield the same result. It is also important to note that the
above standard textbook description of the measurement process is highly
controversial, especially, the collapse postulate. However, a thorough discus-
sion of this issue is beyond the scope of this course.
Quantum mechanics cannot generally predict the outcome of a specic
measurement of an observable A, however it can predict the average, namely
the expectation value, which is denoted as A. The expectation value is easily
calculated with the help of Eq. (2.69)
A =
n
a
n
p
n
=

n
a
n
[ P
n
[ = [ A[ . (2.84)
2.8 Example - Spin 1/2
Spin is an internal degree of freedom of elementary particles. Electrons, for
example, have spin 1/2. This means, as we will see in chapter 6, that the
state of a spin 1/2 can be described by a state vector [ in a vector space
of dimensionality 2. In other words, spin 1/2 is said to be a two-level system.
The spin was rst discovered in 1921 by Stern and Gerlach in an experiment
in which the magnetic moment of neutral silver atoms was measured. Silver
atoms have 47 electrons, 46 out of which ll closed shells. It can be shown
that only the electron in the outer shell contributes to the total magnetic
moment of the atom. The force F acting on a magnetic moment moving in
a magnetic eld B is given by F = ( B). Thus by applying a nonuniform
magnetic eld B and by monitoring the atoms trajectories one can measure
the magnetic moment.
It is important to keep in mind that generally in addition to the spin
contribution to the magnetic moment of an electron, also the orbital motion
of the electron can contribute. For both cases, the magnetic moment is related
to angular momentum by the gyromagnetic ratio. However this ratio takes
dierent values for these two cases. The orbital gyromagnetic ratio can be
evaluated by considering a simple example of an electron of charge e moving
in a circular orbit or radius r with velocity v. The magnetic moment is given
by
orbital
=
AI
c
, (2.85)
where A = r
2
is the area enclosed by the circular orbit and I = ev/ (2r)
is the electrical current carried by the electron, thus
orbital
=
erv
2c
. (2.86)
This result can be also written as
orbital
=

B
L , (2.87)
where L = m
e
vr is the orbital angular momentum, and where
B
=
e
2m
e
c
(2.88)
is the Bohrs magneton constant. The proportionality factor
B
/ is the
orbital gyromagnetic ratio. In vector form and for a more general case of
orbital motion (not necessarily circular) the orbital gyromagnetic relation is
given by
orbital
=

B
L . (2.89)
On the other hand, as was rst shown by Dirac, the gyromagnetic ratio
for the case of spin angular momentum takes twice this value
spin
=
2
B
S . (2.90)
Note that we follow here the convention of using the letter L for orbital
angular momentum and the letter S for spin angular momentum.
The Stern-Gerlach apparatus allows measuring any component of the
magnetic moment vector. Alternatively, in view of relation (2.90), it can be
said that any component of the spin angular momentum S can be measured.
The experiment shows that the only two possible results of such a measure-
ment are +/2 and /2. As we have seen above, one can construct a com-
plete orthonormal basis to the vector space made of eigenvectors of any given
observable. Choosing the observable S
z
= S z for this purpose we construct
a basis made of two vectors [+; z , [; z. Both vectors are eigenvectors of
S
z
S
z
[+; z =

2
[+; z , (2.91)
S
z
[; z =
2
[; z . (2.92)
In what follow we will use the more compact notation
[+ = [+; z , (2.93)
[ = [; z . (2.94)
The orthonormality property implied that
+ [+ = [ = 1 , (2.95)
[+ = 0 . (2.96)
The closure relation in the present case is expressed as
[+ +[ +[ [ = 1 . (2.97)
In this basis any ket-vector [ can be written as
[ = [+ + [ +[ [ . (2.98)
The closure relation (2.97) and Eqs. (2.91) and (2.92) yield
S
z
=

2
([+ +[ [ [) (2.99)
It is useful to dene also the operators S
+
and S
S
+
= [+ [ , (2.100)
S
= [ +[ . (2.101)
In chapter 6 we will see that the x and y components of S are given by
S
x
=

2
([+ [ +[ +[) , (2.102)
S
y
=

2
(i [+ [ +i [ +[) . (2.103)
All these ket-vectors and operators have matrix representation, which for the
basis [+; z , [; z is given by
[+ =
_
1
0
_
, (2.104)
[ =
_
0
1
_
, (2.105)
S
x
=

2
_
0 1
1 0
_
, (2.106)
S
y
=

2
_
0 i
i 0
_
, (2.107)
S
z
=

2
_
1 0
0 1
_
, (2.108)
S
+
=
_
0 1
0 0
_
, (2.109)
S
=
_
0 0
1 0
_
. (2.110)
Exercise 2.8.1. Given that the state vector of a spin 1/2 is [+; z calculate
(a) the expectation values S
x
and S
z
(b) the probability to obtain a value
of +/2 in a measurement of S
x
.
Solution 2.8.1. (a) Using the matrix representation one has
S
x
= +[ S
x
[+ =

2
_
1 0
_
_
0 1
1 0
__
1
0
_
= 0 , (2.111)
S
z
= +[ S
z
[+ =

2
_
1 0
_
_
1 0
0 1
__
1
0
_
=

2
. (2.112)
(b) First, the eigenvectors of the operator S
x
are found by solving the equa-
tion S
x
[ = [, which is done by diagonalization of the matrix represen-
tation of S
x
. The relation S
x
[ = [ for the two eigenvectors is written
in a matrix form as
2
_
0 1
1 0
_
_
1
2
1
2
_
=

2
_
1
2
1
2
_
, (2.113)
2
_
0 1
1 0
_
_
1
2
_
=
2
_
1
2
_
. (2.114)
That is, in ket notation
S
x
[; x =
2
[; x , (2.115)
where the eigenvectors of S
x
are given by
[; x =
1
2
([+ [) . (2.116)
Using this result the probability p
+
is easily calculated
p
+
= [+ [+; x[
2
=
+[
1
2
([+ +[)
2
=
1
2
. (2.117)
2.9 Unitary Operators
Unitary operators are useful for transforming from one orthonormal basis to
another.
Denition 2.9.1. An operator U is said to be unitary if U
= U
1
, namely
if UU
= U
U = 1.
Consider two observables A and B, and two corresponding complete and
orthonormal bases of eigenvectors
A[a
n
= a
n
[a
n
, a
m
[a
n
=
nm
,

n
[a
n
a
n
[ = 1 , (2.118)
B[b
n
= b
n
[b
n
, b
m
[b
n
=
nm
,

n
[b
n
b
n
[ = 1 . (2.119)
The operator U, which is given by
2.10. Trace
U =
n
[b
n
a
n
[ , (2.120)
transforms each of the basis vector [a
n
to the corresponding basis vector [b
n
U [a
n
= [b
n
. (2.121)
It is easy to show that the operator U is unitary
U
U =

n,m
[a
n
b
n
[b
m
. .
nm
a
m
[ =

n
[a
n
a
n
[ = 1 . (2.122)
The matrix elements of U in the basis [a
n
are given by
a
n
[ U [a
m
= a
n
[b
m
, (2.123)
and those of U
by
a
n
[ U
[a
m
= b
n
[a
m
.
Consider a ket vector
[ =
n
[a
n
a
n
[ , (2.124)
which can be represented as a column vector in the basis [a
n
. The nth
element of such a column vector is a
n
[. The operator U can be employed
for nding the corresponding column vector representation of the same ket-
vector [ in the other basis [b
n
b
n
[ =
m
b
n
[a
m
a
m
[ =

m
a
n
[ U
[a
m
a
m
[ . (2.125)
Similarly, Given an operator X the relation between the matrix elements
a
n
[ X[a
m
in the basis [a
n
to the matrix elements b
n
[ X[b
m
in the
basis [b
n
is given by
b
n
[ X[b
m
=

k,l
b
n
[a
k
a
k
[ X[a
l
a
l
[b
m
=

k,l
a
n
[ U
[a
k
a
k
[ X[a
l
a
l
[ U [a
m
.
(2.126)
2.10 Trace
Given an operator X and an orthonormal and complete basis [a
n
, the
trace of X is given by
Tr (X) =

n
a
n
[ X[a
n
. (2.127)
It is easy to show that Tr (X) is independent on basis, as is shown below:
Tr (X) =

n
a
n
[ X[a
n
=

n,k,l
a
n
[b
k
b
k
[ X[b
l
b
l
[a
n
=

n,k,l
b
l
[a
n
a
n
[b
k
b
k
[ X[b
l
=

k,l
b
l
[b
k
. .
kl
b
k
[ X[b
l
=

k
b
k
[ X[b
k
.
(2.128)
The proof of the following relation
Tr (XY ) = Tr (Y X) , (2.129)
is left as an exercise.
2.11 Commutation Relation
The commutation relation of the operators A and B is dened as
[A, B] = AB BA . (2.130)
As an example, the components S
x
, S
y
and S
z
of the spin angular momentum
operator, satisfy the following commutation relations
[S
i
, S
j
] = i
ijk
S
k
, (2.131)
where
ijk
=
_
_
_
0 i, j, k are not all dierent
1 i, j, k is an even permutation of x, y, z
1 i, j, k is an odd permutation of x, y, z
(2.132)
is the Levi-Civita symbol. Equation (2.131) employs the Einsteins conven-
tion, according to which if an index symbol appears twice in an expression,
it is to be summed over all its allowed values. Namely, the repeated index k
should be summed over the values x, y and z:
ijk
S
k
=
ijx
S
x
+
ijy
S
y
+
ijz
S
z
. (2.133)
2.12. Simultaneous Diagonalization of Commuting Operators
Moreover, the following relations hold
S
2
x
= S
2
y
= S
2
z
=
1
4
2
, (2.134)
S
2
= S
2
x
+S
2
y
+S
2
z
=
3
4
2
. (2.135)
The relations below, which are easy to prove using the above denition,
are very useful for evaluating commutation relations
[F, G] = [G, F] , (2.136)
[F, F] = 0 , (2.137)
[E +F, G] = [E, G] + [F, G] , (2.138)
[E, FG] = [E, F] G+F [E, G] . (2.139)
2.12 Simultaneous Diagonalization of Commuting
Operators
Consider an observable A having a set of eigenvalues a
n
. Let g
n
be the
degree of degeneracy of eigenvalue a
n
, namely g
n
is the dimension of the
corresponding eigensubspace, which is denoted by T
n
. Thus the following
holds
A[a
n,i
= a
n
[a
n,i
, (2.140)
where i = 1, 2, , g
n
, and
a
n
,i
[a
n,i
=
nn

ii
. (2.141)
The set of vectors [a
n,1
, [a
n,2
, , [a
n,gn
forms an orthonormal basis for
the eigensubspace T
n
. The closure relation can be written as
1 =
n
gn
i=1
[a
n,i
a
n,i
[ =
n
P
n
, (2.142)
where
P
n
=
g
n
i=1
[a
n,i
a
n,i
[ . (2.143)
Now consider another observable B, which is assumed to commute with
A, namely [A, B] = 0.
Claim. The operator B has a block diagonal matrix in the basis [a
n,i
,
namely a
m,j
[ B[a
n,i
= 0 for n ,= m.
Proof. Multiplying Eq. (2.140) from the left by a
m,j
[ B yields
a
m,j
[ BA[a
n,i
= a
n
a
m,j
[ B[a
n,i
. (2.144)
On the other hand, since [A, B] = 0 one has
a
m,j
[ BA[a
n,i
= a
m,j
[ AB[a
n,i
= a
m
a
m,j
[ B[a
n,i
, (2.145)
thus
(a
n
a
m
) a
m,j
[ B[a
n,i
= 0 . (2.146)
For a given n, the g
n
g
n
matrix a
n,i
[ B[a
n,i
is Hermitian, namely
a
n,i
[ B[a
n,i
= a
n,i
[ B[a
n,i

. Thus, there exists a unitary transformation

U
n
, which maps T
n
onto T
n
, and which diagonalizes the block of B in the
subspace T
n
. Since T
n
is an eigensubspace of A, the block matrix of A in the
new basis remains diagonal (with the eigenvalue a
n
). Thus, we conclude that
a complete and orthonormal basis of common eigenvectors of both operators
A and B exists. For such a basis, which is denoted as [n, m, the following
holds
A[n, m = a
n
[n, m , (2.147)
B[n, m = b
m
[n, m . (2.148)
2.13 Uncertainty Principle
Consider a quantum system in a state [n, m, which is a common eigenvector
of the commuting observables A and B. The outcome of a measurement of
the observable A is expected to be a
n
with unity probability, and similarly,
the outcome of a measurement of the observable B is expected to be b
m
with unity probability. In this case it is said that there is no uncertainty
corresponding to both of these measurements.
Denition 2.13.1. The variance in a measurement of a given observable A
of a quantum system in a state [ is given by
_
(A)
2
_
, where A = AA,
namely
_
(A)
2
_
=
_
A
2
2AA +A
2
_
=

A
2
_
A
2
, (2.149)
where
A = [ A[ , (2.150)
A
2
_
= [ A
2
[ . (2.151)
2.13. Uncertainty Principle
Example 2.13.1. Consider a spin 1/2 system in a state [ = [+; z. Using
Eqs. (2.99), (2.102) and (2.134) one nds that
_
(S
z
)
2
_
=

S
2
z
_
S
z
2
= 0 , (2.152)
_
(S
x
)
2
_
=

S
2
x
_
S
x
2
=
1
4
2
. (2.153)
The last example raises the question: can one nd a state [ for which
the variance in the measurements of both S
z
and S
x
vanishes? According to
the uncertainty principle the answer is no.
Theorem 2.13.1. The uncertainty principle - Let A and B be two observ-
ables. For any ket-vector [ the following holds
_
(A)
2
__
(B)
2
_
1
4
[[A, B][
2
. (2.154)
Proof. Applying the Schwartz inequality [see Eq. (2.166)], which is given by
u [u v [v [u [v[
2
, (2.155)
for the ket-vectors
[u = A[ , (2.156)
[v = B[ , (2.157)
and exploiting the fact that (A)
= A and (B)
= B yield
_
(A)
2
__
(B)
2
_
[AB[
2
. (2.158)
The term AB can be written as
AB =
1
2
[A, B] +
1
2
[A, B]
+
, (2.159)
where
[A, B] = AB BA , (2.160)
[A, B]
+
= AB +BA . (2.161)
While the term [A, B] is anti-Hermitian, whereas the term [A, B]
+
is
Hermitian, namely
([A, B])
= (AB BA)
= BAAB = [A, B] ,
_
[A, B]
+
_
= (AB +BA)
= BA+AB = [A, B]
+
.
In general, the following holds
[ X[ = [ X
=
_
[ X[
if X is Hermitian
[ X[
if X is anti-Hermitian
, (2.162)
thus
AB =
1
2
[A, B]
. .
I
+
1
2
[A, B]
+
_
. .
R
, (2.163)
and consequently
[AB[
2
=
1
4
[[A, B][
2
+
1
4
[A, B]
+
_
2
. (2.164)
Finally, with the help of the identity [A, B] = [A, B] one nds that
_
(A)
2
__
(B)
2
_
1
4
[[A, B][
2
. (2.165)
2.14 Problems
1. Derive the Schwartz inequality
[u [v[
_
u [u
_
v [v , (2.166)
where [u and [v are any two vectors of a vector space T.
2. Derive the triangle inequality:
_
(u[ +v[) ([u +[v)
_
u [u +
_
v [v . (2.167)
3. Show that if a unitary operator U can be written in the form U = 1+iF,
where is a real innitesimally small number, then the operator F is
Hermitian.
4. A Hermitian operator A is said to be positive-denite if, for any vector
[u, u[ A[u 0. Show that the operator A = [a a[ is Hermitian and
positive-denite.
5. Show that if A is a Hermitian positive-denite operator then the following
hold
[u[ A[v[
_
u[ A[u
_
v[ A[v . (2.168)
6. Find the expansion of the operator (AB)
1
in a power series in ,
assuming that the inverse A
1
of A exists.
7. The derivative of an operator A() which depends explicitly on a para-
meter is dened to be
dA()
d
= lim
0
A( +) A()
. (2.169)
Show that
d
d
(AB) =
dA
d
B +A
dB
d
. (2.170)
2.14. Problems
8. Show that
d
d
_
A
1
_
= A
1
dA
d
A
1
. (2.171)
9. Let [u and [v be two vectors of nite norm. Show that
Tr ([u v[) = v [u . (2.172)
10. If A is any linear operator, show that A
A is a positive-denite Her-
mitian operator whose trace is equal to the sum of the square moduli of
the matrix elements of A in any arbitrary representation. Deduce that
Tr
_
A
A
_
= 0 is true if and only if A = 0.
11. Show that if Aand B are two positive-denite observables, then Tr (AB)
0.
12. Show that for any two operators A and L
e
L
Ae
L
= A+[L, A] +
1
2!
[L, [L, A]] +
1
3!
[L, [L, [L, A]]] + . (2.173)
13. Show that if Aand B are two operators satisfying the relation [[A, B] , A] =
0 , then the relation
[A
m
, B] = mA
m1
[A, B] (2.174)
holds for all positive integers m .
14. Show that
e
A
e
B
= e
A+B
e
(1/2)[A,B]
, (2.175)
provided that [[A, B] , A] = 0 and [[A, B] , B] = 0.
15. Proof Kondos identity
_
A, e
H
= e
H
_
0
e
H
[H, A] e
H
d , (2.176)
where A and H are any two operators and is real.
16. Show that Tr (XY ) = Tr (Y X).
17. Consider the two normalized spin 1/2 states [ and [. The operator
A is dened as
A = [ [ [ [ . (2.177)
Find the eigenvalues of the operator A.
18. A molecule is composed of six identical atoms A
1
, A
2
, . . . , A
6
which
form a regular hexagon. Consider an electron, which can be localized on
each of the atoms. Let [
n
be the state in which it is localized on the
nth atom (n = 1, 2, , 6). The electron states will be conned to the
space spanned by the states [
n
, which is assumed to be orthonormal.
The Hamiltonian of the system is given by
H = H
0
+ . (2.178)
The eigenstates of H
0
are the six states [
n
, with the same eigenvalue
E
0
. The operator is described by
[
1
= a [
2
a [
6
,
[
2
= a [
3
a [
1
,
.
.
.
[
6
= a [
1
a [
5
.
(2.179)
Find the eigenvalues and eigen vectors of H. Clue: Consider a solution of
the form
[k =
6
n=1
e
ikn
[
n
. (2.180)
2.15 Solutions
1. Let
[ = [u +[v , (2.181)
where (. The requirement [ 0 leads to
u [u +u [v +
v [u +[[
2
v [v 0 . (2.182)
By choosing
=
v [u
v [v
, (2.183)
one has
u [u
v [u
v [v
u [v
u [v
v [v
v [u +
v [u
v [v
2
v [v 0 , (2.184)
thus
[u [v[
_
u [u
_
v [v . (2.185)
2.15. Solutions
2. The following holds
(u[ +v[) ([u +[v) = u [u +v [v + 2 Re (u [v)
u [u +v [v + 2 [u [v[ .
(2.186)
Thus, using Schwartz inequality one has
(u[ +v[) ([u +[v) u [u +v [v + 2
_
u [u
_
v [v
=
_
_
u [u +
_
v [v
_
2
.
(2.187)
3. Since
1 = U
U =
_
1 iF
_
(1 +iF) = 1 +i
_
F F
_
+O
_
2
_
, (2.188)
one has F = F
.
4. In general, ([ [)
= [ [, thus clearly the operator A is Hermitian.

Moreover it is positive-denite since for every [u the following holds
u[ A[u = u [a a [u = [a [u[
2
0 . (2.189)
5. Let
[ = [u
v[ A[u
v[ A[v
[v .
Since A is Hermitian and positive-denite the following holds
0 [ A[
=
_
u[
u[ A[v
v[ A[v
v[
_
A
_
[u
v[ A[u
v[ A[v
[v
_
= u[ A[u
[u[ A[v[
2
v[ A[v

[u[ A[v[
2
v[ A[v
+
[u[ A[v[
2
v[ A[v
,
(2.190)
thus
[u[ A[v[
_
u[ A[u
_
v[ A[v . (2.191)
Note that this result allows easy proof of the following: Under the same
conditions (namely, A is a Hermitian positive-denite operator) Tr (A) =
0 if and only if A = 0.
6. The expansion is given by
(AB)
1
=
_
A
_
1 A
1
B
__
1
=
_
1 A
1
B
_
1
A
1
=
_
1 +A
1
B +
_
A
1
B
_
2
+
_
A
1
B
_
3
+
_
A
1
.
(2.192)
7. By denition:
d
d
(AB) = lim
0
A( +) B( +) A() B()
= lim
0
(A( +) A()) B()
+ lim
0
A( +) (B( +) B())
=
dA
d
B +A
dB
d
.
(2.193)
8. Taking the derivative of both sides of the identity 1 = AA
1
on has
0 =
dA
d
A
1
+A
dA
1
d
, (2.194)
thus
d
d
_
A
1
_
= A
1
dA
d
A
1
. (2.195)
9. Let [n be a complete orthonormal basis, namely
n
[n n[ = 1 . (2.196)
In this basis
Tr ([u v[) =

n
n [u v [n = v[
_
n
[n n[
_
[u = v [u . (2.197)
10. The operator A
A is Hermitian since
_
A
A
_
= A
A, and positive-
denite since the norm of A[u is nonnegative for every [u, thus one
has u[ A
A[u 0. Moreover, using a complete orthonormal basis [n

one has
Tr
_
A
A
_
=

n
n[ A
A[n
=

n,m
n[ A
[m m[ A[n
=

n,m
[m[ A[n[
2
.
(2.198)
11. Let [b
be a complete orthonormal basis made of eigenvectors of B

(i.e., B[b
= b
[b
). Using this basis for evaluating the trace one has

Tr (AB) =

b
[ AB[b
=

b
..
0
b
[ A[b
. .
0
0 . (2.199)
2.15. Solutions
12. Let f (s) = e
sL
Ae
sL
, where s is real. Using Taylor expansion one has
f (1) = f (0) +
1
1!
df
ds
s=0
+
1
2!
d
2
f
ds
2
s=0
+ , (2.200)
thus
e
L
Ae
L
= A+
1
1!
df
ds
s=0
+
1
2!
d
2
f
ds
2
s=0
+ , (2.201)
where
df
ds
= Le
sL
Ae
sL
e
sL
Ae
sL
L = [L, f (s)] , (2.202)
d
2
f
ds
2
=
_
L,
df
ds
_
= [L, [L, f (s)]] , (2.203)
therefore
e
L
Ae
L
= A+[L, A] +
1
2!
[L, [L, A]] +
1
3!
[L, [L, [L, A]]] + . (2.204)
13. The identity clearly holds for the case m = 1. Moreover, assuming it
holds for m, namely assuming that
[A
m
, B] = mA
m1
[A, B] , (2.205)
one has
_
A
m+1
, B
= A[A
m
, B] + [A, B] A
m
= mA
m
[A, B] + [A, B] A
m
.
(2.206)
It is easy to show that if [[A, B] , A] = 0 then [[A, B] , A
m
] = 0, thus one
concludes that
_
A
m+1
, B
= (m+ 1) A
m
[A, B] . (2.207)
14. Dene the function f (s) = e
sA
e
sB
, where s is real. The following holds
df
ds
= Ae
sA
e
sB
+e
sA
Be
sB
=
_
A+e
sA
Be
sA
_
e
sA
e
sB
Using Eq. (2.174) one has
e
sA
B =

m=0
(sA)
m
m!
B
=

m=0
s
m
(BA
m
+ [A
m
, B])
m!
=

m=0
s
m
_
BA
m
+mA
m1
[A, B]
_
m!
= Be
sA
+s

m=1
(sA)
m1
(m1)!
[A, B]
= Be
sA
+se
sA
[A, B] ,
(2.208)
thus
df
ds
= Ae
sA
e
sB
+Be
sA
e
sB
+se
sA
[A, B] e
sB
= (A+B + [A, B] s) f (s) .
(2.209)
The above dierential equation can be easily integrated since [[A, B] , A] =
0 and [[A, B] , B] = 0. Thus
f (s) = e
(A+B)s
e
[A,B]
s
2
2
. (2.210)
For s = 1 one gets
e
A
e
B
= e
A+B
e
(1/2)[A,B]
. (2.211)
15. Dene
f ()
_
A, e
H
, (2.212)
g () e
H
_
0
e
H
[H, A] e
H
d . (2.213)
Clearly, f (0) = g (0) = 0 . Moreover, the following holds
df
d
= AHe
H
+He
H
A = Hf + [H, A] e
H
, (2.214)
dg
d
= Hg + [H, A] e
H
, (2.215)
namely, both functions satisfy the same dierential equation. Therefore
f = g.
16. Using a complete orthonormal basis [n one has
2.15. Solutions
Tr (XY ) =

n
n[ XY [n
=

n,m
n[ X[m m[ Y [n
=

n,m
m[ Y [n n[ X[m
=

m
m[ Y X[m
= Tr (Y X) .
(2.216)
Note that using this result it is easy to show that Tr (U
+
XU) = Tr (X)
, provided that U is a unitary operator.
17. Clearly A is Hermitian, namely A
= A, thus the two eigenvalues

1
and
2
are expected to be real. Since the trace of an operator is basis
independent, the following must hold
Tr (A) =
1
+
2
, (2.217)
and
Tr
_
A
2
_
=
2
1
+
2
2
. (2.218)
On the other hand, with the help of Eq. (2.172) one nds that
Tr (A) = Tr ([ [) Tr ([ [) = 0 , (2.219)
and
Tr
_
A
2
_
= Tr ([ [ [) + Tr ([ [ [) Tr ([ [ [) Tr ([ [ [)
= 2 [ Tr ([ [) [ Tr ([ [)
= 2
_
1 [ [[
2
_
,
(2.220)
thus
=
_
1 [ [[
2
. (2.221)
Alternatively, this problem can also be solved as follows. In general, the
state [ can be decomposed into a parallel to and a perpendicular to [
terms, namely
[ = a [ +c [ , (2.222)
where a, c (, the vector [ is orthogonal to [, namely [ = 0, and
in addition [ is assumed to be normalized, namely [ = 1. Since [
is normalized one has [a[
2
+[c[
2
= 1. The matrix representation of A in
the orthonormal basis [ , [ is given by
A =
_
[ A[ [ A[
[ A[ [ A[
_
=
_
[c[
2
ac
c [c[
2
_

A . (2.223)
Thus,
Tr
_
A
_
= 0 , (2.224)
and
Det
_
A
_
= [c[
2
_
[c[
2
+[a[
2
_
=
_
1 [ [[
2
_
, (2.225)
therefore the eigenvalues are
=
_
1 [ [[
2
. (2.226)
18. Following the clue, we seek a solution to the eigenvalue equation
H[k = E
m
[k , (2.227)
where
[k =
6
n=1
e
ikn
[
n
. (2.228)
thus
H[k = E
0
[k a
6
n=1
e
ikn
_
(n1)
_
+
(n+1)
__
= E[k , (2.229)
where n
is the modulus of n divided by 6 (e.g., 1
= 1, 0
= 6, 7
= 1).
A solution is obtained if
e
6ik
= 1 , (2.230)
or
k
m
=
m
3
, (2.231)
where m = 1, 2, , 6. The corresponding eigen vectors are denoted as
[k
m
=
6
n=1
e
ik
m
n
[
n
, (2.232)
and the following holds
2.15. Solutions
H[k
m
= E
0
[k
m
ae
ikm
6
n=1
e
ikm(n1)
(n1)
_
ae
ikm
6
n=1
e
ikm(n+1)
(n+1)
_
= (E
0
2acos k
m
) [k
m
,
(2.233)
thus
H[k
m
= E
m
[k , (2.234)
where
E
m
= E
0
2a cos k
m
. (2.235)
3. The Position and Momentum Observables
Consider a point particle moving in a 3 dimensional space. We rst treat
the system classically. The position of the particle is described using the
Cartesian coordinates q
x
, q
y
and q
z
. Let
p
j
=
/
q
j
(3.1)
be the canonically conjugate variable to the coordinate q
j
, where j x, y, z
and where / is the Lagrangian. As we have seen in exercise 4 of set 1, the
following Poissons brackets relations hold
q
j
, q
k
= 0 , (3.2)
p
j
, p
k
= 0 , (3.3)
q
j
, p
k
=
jk
. (3.4)
In quantum mechanics, each of the 6 variables q
x
, q
y
, q
z
, p
x
, p
y
and p
z
is
represented by an Hermitian operator, namely by an observable. It is postu-
lated that the commutation relations between each pair of these observables
is related to the corresponding Poissons brackets according to the rule
,
1
i
[, ] . (3.5)
Namely the following is postulated to hold
[q
j
, q
k
] = 0 , (3.6)
[p
j
, p
k
] = 0 , (3.7)
[q
j
, p
k
] = i
jk
. (3.8)
Note that here we use the same notation for a classical variable and its
quantum observable counterpart. In this chapter we will derive some results
that are solely based on Eqs. (3.6), (3.7) and (3.8).
3.1 The One Dimensional Case
In this section, which deals with the relatively simple case of a one dimen-
sional motion of a point particle, we employ the less cumbersome notation
Chapter 3. The Position and Momentum Observables
x and p for the observables q
x
and p
x
. The commutation relation between
these operators is given by [see Eq. (3.8)]
[x, p] = i . (3.9)
The uncertainty principle (2.154) employed for x and p yields
_
(x)
2
__
(p
x
)
2
_

2
4
. (3.10)
3.1.1 Position Representation
Let x
be an eigenvalue of the observable x, and let [x
be the corresponding
eigenvector, namely
x[x
= x
[x
. (3.11)
Note that x
since x is Hermitian. As we will see below transformation

between dierent eigenvectors [x
can be performed using the translation

operator J (
x
).
Denition 3.1.1. The translation operator is given by
J (
x
) = exp
_
i
x
p
_
, (3.12)
where
x
.
Recall that in general the meaning of a function of an operator can be
understood in terms of the Taylor expansion of the function, that is, for the
present case
J (
x
) =

n=0
1
n!
_
i
x
p
_
n
. (3.13)
It is easy to show that J (
x
) is unitary
J
(
x
) = J (
x
) = J
1
(
x
) . (3.14)
Moreover, the following composition property holds
J (
x1
) J (
x2
) = J (
x1
+
x2
) . (3.15)
Theorem 3.1.1. Let x
be an eigenvalue of the observable x, and let [x
be
the corresponding eigenvector. Then the ket-vector J (
x
) [x
is a normalized
eigenvector of x with an eigenvalue x
+
x
.
3.1. The One Dimensional Case
Proof. With the help of Eq. (3.77), which is given by
[x, B(p)] = i
dB
dp
, (3.16)
and which is proven in exercise 1 of set 3, one nds that
[x, J (
x
)] = i
x
i
J (
x
) . (3.17)
Using this result one has
xJ (
x
) [x
= ([x, J (
x
)] +J (
x
) x) [x
= (x
+
x
) J (
x
) [x
, (3.18)
thus the ket-vector J (
x
) [x
is an eigenvector of x with an eigenvalue x
x
. Moreover, J (
x
) [x
is normalized since J is unitary.

In view of the above theorem we will in what follows employ the notation
J (
x
) [x
= [x
+
x
. (3.19)
An important consequence of the last result is that the spectrum of eigenval-
ues of the operator x is continuous and contains all real numbers. This point
will be further discussed below.
The position wavefunction
(x
) of a state vector [ is dened as:
(x
) = x
[ . (3.20)
Given the wavefunction
(x
) of state vector [, what is the wavefunction

of the state O[, where O is an operator? We will answer this question below
for some cases:
1. The operator O = x. In this case
x
[ x[ = x
[ = x
(x
) , (3.21)
namely, the desired wavefunction is obtained by multiplying
(x
) by
x
.
2. The operator O is a function A(x) of the operator x. Let
A(x) =
n
a
n
x
n
. (3.22)
be the Taylor expansion of A(x). Exploiting the fact that x is Hermitian
one nds that
x
[ A(x) [ =

n
a
n
x
[ x
n
. .
x
n
x
|
[ =

n
a
n
x
n
x
[ = A(x
(x
) .
(3.23)
3. The operator O = J (
x
). In this case
x
[ J (
x
) [ = x
[ J
(
x
) [ = x
x
[ =
(x
x
) .
(3.24)
4. The operator O = p. In view of Eq. (3.12), the following holds
J (
x
) = exp
_
ip
x
_
= 1 +
i
x
p +O
_
(
x
)
2
_
, (3.25)
thus
x
[ J (
x
) [ =
(x
) +
i
x
[ p [ +O
_
(
x
)
2
_
. (3.26)
On the other hand, according to Eq. (3.24) also the following holds
x
[ J (
x
) [ =
(x
+
x
) . (3.27)
Equating the above two expressions for x
[ J (
x
) [ yields
x
[ p [ = i
(x
+
x
)
(x
x
+O(
x
) . (3.28)
Thus, in the limit
x
0 one has
x
[ p [ = i
d
dx
. (3.29)
To mathematically understand the last result, consider the dierential
operator
J (
x
) = exp
_
x
d
dx
_
= 1 +
x
d
dx
+
1
2!
_
x
d
dx
_
2
+ .
(3.30)
In view of the Taylor expansion of an arbitrary function f (x)
f (x
0
+
x
) = f (x
0
) +
x
df
dx
+
(
x
)
2
2!
d
2
f
dx
2
+
= exp
_
x
d
dx
_
f
x=x0
=

J (
x
) f
x=x0
,
(3.31)
one can argue that the operator

J (
x
) generates translation.
3.1. The One Dimensional Case
As we have pointed out above, the spectrum (i.e., the set of all eigen-
values) of x is continuous. On the other hand, in the discussion in chapter
2 only the case of an observable having discrete spectrum has been consid-
ered. Rigorous mathematical treatment of the case of continuous spectrum
is nontrivial mainly because typically the eigenvectors in such a case cannot
be normalized. However, under some conditions one can generalize some of
the results given in chapter 2 for the case of an observable having a continu-
ous spectrum. These generalization is demonstrated below for the case of the
position operator x:
1. The closure relation (2.23) is written in terms of the eigenvectors [x
as
dx
[x
[ = 1 , (3.32)
namely, the discrete sum is replaced by an integral.
2. With the help of Eq. (3.32) an arbitrary ket-vector can be written as
[ =
dx
[x
[ =
dx
(x
) [x
, (3.33)
and the inner product between a ket-vector [ and a bra-vector [ as
[ =
dx
[x
[ =
dx
(x
(x
) . (3.34)
3. The normalization condition reads
1 = [ =
dx
(x
)[
2
. (3.35)
4. The orthonormality relation (2.64) is written in the present case as
x
[x
= (x
) . (3.36)
Note that the above orthonormality relation (3.36) is consistent with
the closure relation (3.32). This can be seen by evaluating the operator
1
2
= 1 1 using Eqs. (3.32) and (3.36)
1
2
=
dx
dx
[x
[x
. .
(x
)
x
[ =
dx
[x
[ , (3.37)
thus, as expected 1
2
= 1.
5. In a measurement of the observable x, the quantity
f (x
) = [x
[[
2
= [
(x
)[
2
(3.38)
represents the probability density to nd the particle at the point x = x
.
6. That is, the probability to nd the particle in the interval (x
1
, x
2
) is given
by
p
(x1,x2)
=
_
x2
x1
dx
f (x
) . (3.39)
This can be rewritten as
p
(x
1
,x
2
)
= [ P
(x
1
,x
2
)
[ , (3.40)
where the projection operator P
(x
1
,x
2
)
is given by
P
(x1,x2)
=
_
x
2
x
1
dx
[x
[ . (3.41)
The operator P
(x1,x2)
is considered to be a projection operator since for
every x
0
(x
1
, x
2
) the following holds
P
(x1,x2)
[x
0
=
_
x2
x1
dx
[x
[x
0
. .
(x
x0)
= [x
0
. (3.42)
7. Any realistic measurement of a continuous variable such as position is
subjected to nite resolution. Assuming that a particle has been mea-
sured to be located in the interval (x
x
/2, x
+
x
/2), where
x
is the
resolution of the measuring device, the collapse postulate implies that
the state of the system undergoes the following transformation
[
P
(x
x
/2,x
+
x
/2)
[
_
[ P
(x
x
/2,x
+
x
/2)
[
. (3.43)
8. Some observables have a mixed spectrum containing both a discrete and
continuous subsets. An example of such a mixed spectrum is the set of
eigenvalues of the Hamiltonian operator of a potential well of nite depth.
3.1.2 Momentum Representation
Let p
be an eigenvalue of the observable p, and let [p
be the corresponding
eigenvector, namely
p [p
= p
[p . (3.44)
Note that p
since p is Hermitian. Similarly to the case of the position

observable, the closure relation is written as
3.2. Transformation Function
_
dp
[p
[ = 1 , (3.45)
and the orthonormality relation as
p
[p
= (p
) . (3.46)
The momentum wavefunction
(p
) of a given state [ is dened as
(p
) = p
[ . (3.47)
The probability density to measure a momentum value of p = p
is
[
(p
)[
2
= [p
[[
2
. (3.48)
Any ket-vector can be decomposed into momentum eigenstates as
[ =
dp
[p
[ =
dp
(p
) [p
. (3.49)
The inner product between a ket-vector [ and a bra-vector [ can be
expressed as
[ =
dp
[p
[ =
dp
(p
(p
) . (3.50)
The normalization condition reads
1 = [ =
dp
(p
)[
2
. (3.51)
3.2 Transformation Function
What is the relation between the position wavefunction
(x
) and its mo-

mentum counterpart
(p
)?
Claim. The transformation function x
[p
is given by
x
[p
=
1
2
exp
_
ip
_
. (3.52)
Proof. On one hand, according to Eq. (3.44)
x
[ p [p
= p
[p
, (3.53)
and on the other hand, according to Eq. (3.29)
x
[ p [p
= i

x
[p
, (3.54)
thus
p
[p
= i

x
[p
. (3.55)
The general solution of this dierential equation is
x
[p
= N exp
_
ip
_
, (3.56)
where N is a normalization constant. To determine the constant N we employ
Eqs. (3.36) and (3.45):
(x
)
= x
[x
=
_
dp
[p
[x
dp
[N[
2
exp
_
ip
(x
_
= [N[
2
dke
ik(x
)
. .
2(x
)
.
(3.57)
Thus, by choosing N to be real one nds that
x
[p
=
1
2
exp
_
ip
_
. (3.58)
The last result together with Eqs. (3.32) and (3.45) yield
(x
) = x
[ =
dp
[p
[ =
dp
e
ip
(p
2
, (3.59)
(p
) = p
[ =
dx
[x
[ =
dx
ip
(x
2
. (3.60)
That is, transformations relating
(x
) and
(p
) are the direct and inverse

Fourier transformations.
3.3. Generalization for 3D
3.3 Generalization for 3D
According to Eq. (3.6) the observables q
x
, q
y
and q
z
commute with each
other, hence, a simultaneous diagonalization is possible. Denoting the com-
mon eigenvectors as
[r
x
, q
y
, q
z
_
, (3.61)
one has
q
x
[r
= q
x
, q
y
, q
z
_
, (3.62)
q
y
[r
= q
x
, q
y
, q
z
_
, (3.63)
q
z
[r
= q
x
, q
y
, q
z
_
. (3.64)
The closure relation is written as
1 =
dx
dy
dz
[r
[ , (3.65)
r
[r
= (r
) . (3.66)
Similarly, according to Eq. (3.7) the observables p
x
, p
y
and p
z
commute
with each other, hence, a simultaneous diagonalization is possible. Denoting
the common eigenvectors as
[p
x
, p
y
, p
z
_
, (3.67)
one has
p
x
[p
= p
x
, p
y
, p
z
_
, (3.68)
p
y
[p
= p
x
, p
y
, p
z
_
, (3.69)
p
z
[p
= p
x
, p
y
, p
z
_
. (3.70)
The closure relation is written as
1 =
dp
x
dp
y
dp
z
[p
[ , (3.71)
p
[p
= (p
) . (3.72)
The translation operator in three dimensions can be expressed as
J () = exp
_
i p
_
, (3.73)
where =(
x
,
y
,
z
)
3
, and where
J () [r
= [r
+ . (3.74)
The generalization of Eq. (3.52) for three dimensions is
r
[p
=
1
(2)
3/2
exp
_
ip
_
. (3.75)
3.4 Problems
1. Show that
[p, A(x)] = i
dA
dx
, (3.76)
[x, B(p)] = i
dB
dp
, (3.77)
where A(x) is a dierentiable function of x and B(p) is a dierentiable
function of p.
2. Show that the mean value of x in a state described by the wavefunction
(x), namely
x =
+
_
dx
(x) x (x) , (3.78)

is equal to the value of a for which the expression
F (a)
+
_
dx
(x +a) x
2
(x +a) (3.79)
obtains a minimum, and that this minimum has the value
F
min
= (x)
2
=

x
2
_
x
2
. (3.80)
3. Consider a Gaussian wave packet, whose x space wavefunction is given
by
(x
) =
1
1/4
d
exp
_
ikx
x
2
2d
2
_
. (3.81)
Calculate
a)
_
(x)
2
__
(p)
2
_
b) p
[
3.4. Problems
4. Show that the state [ with wave function
x
[ =
_
1/
2a for [x[ a
0 for [x[ > a
(3.82)
the uncertainty in momentum is innity.
5. Show that
p = i
dx
[x
d
dx
[ . (3.83)
6. Show that
1
(2)
3
_
d
3
p
exp
_
ip
(r
_
= (r
) . (3.84)
7. Find eigenvectors and corresponding eigenvalues of the operator
O = p +Kx , (3.85)
where K is a real constant, p is the momentum operator, which is the
canonically conjugate to the position operator x. Calculate the wavefunc-
tion of the eigenvectors.
8. Let [ be the state vector of a point particle having mass m that moves
in one dimension along the x axis. The operator p
is dened by the
following requirements: (1) p
is Hermitian (i.e. p
= p
) (2) [x, p
] = 0
(i.e. p
commutes with the position operator x) and (3)

[ (p p
)
2
[ = min
O
[ (p O)
2
[ , (3.86)
where p is the momentum operator (i.e. the minimum value of the quan-
tity [ (p O)
2
[ is obtained when the operator O is chosen to be
p
).
a) Calculate the matrix elements x
[ p
[x
of the operator p
in the
position representation.
b) The operator T is the dierence between the true momentum op-
erator and p
T = p p
. (3.87)
Calculate the variance
_
(T)
2
_
with respect to the state [
_
(T)
2
_
= [ T
2
[ [ T [
2
. (3.88)
c) Use your results to prove the uncertainty relation (3.10)
_
(x)
2
__
(p)
2
_

2
4
. (3.89)
where
_
(x)
2
_
= [ x
2
[ [ x[
2
, (3.90)
and where
_
(p)
2
_
= [ p
2
[ [ p [
2
. (3.91)
3.5 Solutions
1. The commutator [x, p] = i is a constant, thus the relation (2.174) can
be employed
[p, x
m
] = imx
m1
= i
dx
m
dx
, (3.92)
[x, p
m
] = imp
m1
= i
dp
m
dp
. (3.93)
This holds for any m, thus, for any dierentiable function A(x) of x and
for any dierentiable function B(p) of p one has
[p, A(x)] = i
dA
dx
, (3.94)
[x, B(p)] = i
dB
dp
. (3.95)
F (a) =
+
_
dx
(x +a) x
2
(x +a)
=
+
_
dx
(x
) (x
a)
2
(x)
=
_
(x a)
2
_
=

x
2
_
2a x +a
2
.
(3.96)
The requirement
dF
da
= 0 (3.97)
3.5. Solutions
leads to 2 x + 2a = 0, or a = x. At that point one has
F
min
=
_
(x x)
2
_
=

x
2
_
x
2
. (3.98)
3. The following hold
x =
+
_
dx
(x
) x
(x
)
=
1
1/2
d
+
_
dx
exp
_
x
2
d
2
_
x
= 0 ,
(3.99)
x
2
_
=
+
_
dx
(x
) x
2
(x
)
=
1
1/2
d
+
_
dx
exp
_
x
2
d
2
_
x
2
=
1
1/2
d
d
3
1/2
2
=
d
2
2
,
(3.100)
p = i
+
_
dx
(x
)
d
dx
=
i
1/2
d
+
_
dx
exp
_
x
2
d
2
__
ik
x
d
2
_
=
i
1/2
d
ikd
1/2
= k ,
(3.101)
p
2
_
= (i)
2
+
_
dx
(x
)
d
2
dx
2
= (i)
2
1
1/2
d
+
_
dx
exp
_
x
2
d
2
_
__
_
ik
x
d
2
_
2
1
d
2
__
= (i)
2
1
1/2
d
_
1
2
_
d
2d
4
k
2
+d
2
d
4
= (k)
2
_
1 +
1
2 (dk)
2
_
,
(3.102)
a) thus
_
(x)
2
__
(p)
2
_
=
d
2
2
_
(k)
2
_
1 +
1
2 (dk)
2
_
(k)
2
_
=

2
4
.
(3.103)
b) Using Eq. (3.60) one has
p
[ =
1
2
_
dx
exp
_
ip
(x
)
=
1
2
1
1/4
dx
exp
__
ik
ip
_
x
x
2
2d
2
_
=
1/4
exp
_
(k p
)
2
d
2
2
2
_
.
(3.104)
4. The momentum wavefunction is found using Eq. (3.60)
(p
) =
1
dx
exp
_
ip
_
x
[
=
1
4a
a
_
a
dx
exp
_
ip
_
=
_
a
sin
ap
ap
.
(3.105)
The momentum wavefunction
(p
) is normalizable, however, the inte-

grals for evaluating both p and

p
2
_
do not converge.
3.5. Solutions
5. Using Eqs. (3.29) and (3.32) one has
p [ =
dx
[x
[ p [
= i
dx
[x
d
dx
[ ,
(3.106)
thus, since [ is an arbitrary ket vector, the following holds
p = i
dx
[x
d
dx
[ . (3.107)
6. With the help of Eqs. (3.66), (3.71) and (3.75) one nds that
(r
) = r
[r
=
_
d
3
p
[p
[r
=
1
(2)
3
_
d
3
p
exp
_
ip
(r
_
.
(3.108)
7. Using the identity (2.173), which is given by
e
L
Ae
L
= A+[L, A] +
1
2!
[L, [L, A]] +
1
3!
[L, [L, [L, A]]] + . (3.109)
and the identity (3.76), which is given by
[g (x) , p] = i
dg
dx
, (3.110)
one nds that
e
g(x)
pe
g(x)
= A+i
dg
dx
+
i
2!
_
g (x) ,
dg
dx
_
+
i
3!
_
g (x) ,
_
g (x) ,
dg
dx
__
+ .
(3.111)
Choosing g (x) to be given by
g (x) =
Kx
2
2i
(3.112)
yields
UpU
= A+Kx = O , (3.113)
where the unitary operator U is given by
U = e
iKx
2
2
.
Thus, the vectors [ (p
), which are dene as

[(p
) = U [p
, (3.114)
where [p
is an eigenvector of p with eigenvalue p
(i.e. p [p
= p
[p
),
are eigenvectors of O, and the following holds
O[ (p
) = p
[(p
) . (3.115)
With the help of Eq. (3.52), which is given by
x
[p
=
1
2
e
ip
, (3.116)
one nds that the wavefunction (x
; p
) = x
[ (p
) of the state
[ (p
) is given by
(x
; p
) = e
iKx
2
2
x
[p
=
1
2
e
i
_
p
Kx
2
2
_
.
(3.117)
8. With the help of Eq. (3.32) one nds that
p
dx
dx
[x
[ p
[x
[ . (3.118)
The requirement [x, p
] = 0 implies that
dx
dx
[x
[ p
[x
(x
) x
[ = 0 , (3.119)
hence x
[ p
[x
= 0 unless x
= x
. Thus by using the notation

x
[ p
[x
(x
) (x
) , (3.120)
the operator p
can be expressed as
p
dx
[x
(x
) x
[ . (3.121)
The requirement that p
is Hermitian implies that
(x
) is real.
3.5. Solutions
a) With the help of Eq. (3.83), which is given by
p = i
dx
[x
d
dx
[ , (3.122)
one nds that
p p
dx
[x
_
i
d
dx
(x
)
_
x
[ , (3.123)
hence in terms of the wavefunction
(x
) = x
[ of [ one has
(p p
) [ =
dx
[x
_
i
d
dx
(x
)
_
(x
)
=
dx
[x
(x
)
_
i
dlog
dx
(x
)
_
.
(3.124)
Similarly
[ (p p
) =
dx
(x
)
_
i
dlog
dx
(x
)
_
x
[ , (3.125)
and thus
[ (p p
)
2
[
=
dx
(x
i
dlog
dx
(x
2
.
(3.126)
where
(x
) = [
(x
)[
2
. (3.127)
The minimum value is obtained when (recall that
(x
) is required
to be real)
(x
) =

2i
_
dlog
dx

dlog
dx
_
=

2i
dlog
dx
,
(3.128)
and thus
x
[ p
[x
=

2i
dlog
dx
(x
) . (3.129)
Note: Comparing this result with the expression for the current den-
sity J associated with the state [ [see Eq. (4.181)] yields the fol-
lowing relation
J =

m
Im
_
dx
_
=
(x
)
m

2i
_
dlog
dx

dlog
dx
_
=
(x
)
m

(x
) .
(3.130)
b) As can be seen from Eqs. (3.123) and (3.128) the following holds
T = i
dx
[x
_
_
d
dx
+
1
2
dlog
dx
_
_
x
[ , (3.131)
hence
[ T [ = i
dx
dx
+
1
2
_
dx
dx
__
=
i
2
dx
dx
+
d
dx
_
=
i
2
dx
d (x
)
dx
= 0 ,
(3.132)
thus [see Eqs. (3.126) and (3.128)]
_
(T)
2
_
= [ T
2
[
=
_
2
_
2

_
dx
(x
dlog
dx
+
dlog
dx
2
=
_
2
_
2

_
dx
(x
)
_
dlog(x
)
dx
_
2
.
(3.133)
Note that the result [ T [ = 0 implies that p
and p have the

same expectation value, i.e. [ p
[ = [ p [. On the other hand,

3.5. Solutions
contrary to p, the operator p
commutes with the position operator

x.
c) Using the relation [ p
[ = [ p [ one nds that

_
(p)
2
_
_
(p
)
2
_
= [ p
2
[ [ p
2
[
= [ (p p
)
2
[ +[
_
pp
+p
p 2p
2
_
[
= [ T
2
[ +[
_
pp
+p
p 2p
2
_
[ .
(3.134)
As can be see from Eq. (3.128), the following holds
p
dx
[x
Im
_
dlog
dx
_
x
[ , (3.135)
thus
[
_
pp
+p
p 2p
2
_
[
=
2
dx

_
_
|p[x
+
x
[p|
2 Im
_
dlog
dx
_
_
_
Im
_
dlog
dx
_
= 0 ,
(3.136)
and therefore _
(p)
2
_
= [ T
2
[ +
_
(p
)
2
_
[ T
2
[
=
_
2
_
2

_
dx
(x
)
_
dlog (x
)
dx
_
2
.
(3.137)
For general real functions f (x
) , g (x
) : the Schwartz in-

equality (2.166) implies that
dx
f (x
) g (x
dx
(f (x
))
2

_
dx
(g (x
))
2
. (3.138)
Implementing this inequality for the functions
f (x
) =
_
(x
) (x
x) , (3.139)
g (x
) =
_
(x
)
dlog (x
)
dx
, (3.140)
where
x =
dx
(x
) x

is the expectation value of x, yields
dx
(x
)
_
dlog (x
)
dx
_
2
dx
(x
) (x
x)
dlog(x
)
dx
2
_
(x)
2
_ ,
(3.141)
where
_
(x)
2
_
=
dx
(x
) (x
x)
2
(3.142)
is the variance of x. By integrating by parts one nds that
dx
(x
) (x
x)
dlog (x
)
dx
dx
(x
x)
d(x
)
dx
dx
(x
)
= 1 .
(3.143)
Combining these results [see Eqs. (3.137) and (3.141)] lead to
_
(x)
2
__
(p)
2
_
2
_
2
. (3.144)
4. Quantum Dynamics
The time evolution of a state vector [ is postulated to be given by the
Schrdinger equation
i
d[
dt
= H[ , (4.1)
where the Hermitian operator H = H
is the Hamiltonian of the system.

The Hamiltonian operator is the observable corresponding to the classical
Hamiltonian function that we have studied in chapter 1. The time evolution
produced by Eq. (4.1) is unitary, as is shown below:
Claim. The norm [ is time independent.
Proof. Since H = H
, the dual of the Schrdinger equation (4.1) is given by

i
d[
dt
= [ H . (4.2)
Using this one has
d [
dt
=
_
d[
dt
_
[+[
d[
dt
=
1
i
([ H[ +[ H[) = 0 . (4.3)
4.1 Time Evolution Operator
The time evolution operator u(t, t
0
) relates the state vector at time [(t
0
)
with its value [(t) at time t:
[(t) = u(t, t
0
) [(t
0
) . (4.4)
Claim. The time evolution operator satises the Schrdinger equation (4.1).
Proof. Expressing the Schrdinger equation (4.1) in terms of Eq. (4.4)
i
d
dt
u(t, t
0
) [(t
0
) = Hu(t, t
0
) [(t
0
) , (4.5)
and noting that [(t
0
) is t independent yield
Chapter 4. Quantum Dynamics
i
_
d
dt
u(t, t
0
)
_
[(t
0
) = Hu(t, t
0
) [(t
0
) . (4.6)
Since this holds for any [(t
0
) one concludes that
i
du(t, t
0
)
dt
= Hu(t, t
0
) . (4.7)
This results leads to the following conclusion:
Claim. The time evolution operator is unitary.
Proof. Using Eq. (4.7) one nds that
d
_
u
u
_
dt
= u
du
dt
+
du
dt
u
=
1
i
_
u
Hu u
Hu
_
= 0 .
(4.8)
Furthermore, for t = t
0
clearly u(t
0
, t
0
) = u
(t
0
, t
0
) = 1. Thus, one concludes
that u
u = 1 for any time, namely u is unitary.

4.2 Time Independent Hamiltonian
A special case of interest is when the Hamiltonian is time independent. In
this case the solution of Eq. (4.7) is given by
u(t, t
0
) = exp
_
iH(t t
0
)
_
. (4.9)
The operator u(t, t
0
) takes a relatively simple form in the basis of eigenvectors
of the Hamiltonian H. Denoting these eigenvectors as [a
n,i
, where the index
i is added to account for possible degeneracy, and denoting the corresponding
eigenenergies as E
n
one has
H[a
n,i
= E
n
[a
n,i
, (4.10)
where
a
n
,i
[a
n,i
=
nn

ii
. (4.11)
By using the closure relation, which is given by
1 =
n
g
n
i=1
[a
n,i
a
n,i
[ , (4.12)
and Eq. (4.9) one nds that
u(t, t
0
) = exp
_
iH(t t
0
)
_
1
=

n
gn
i=1
exp
_
iH(t t
0
)
_
[a
n,i
a
n,i
[
=

n
g
n
i=1
exp
_
iE
n
(t t
0
)
_
[a
n,i
a
n,i
[ .
(4.13)
Using this results the state vector [(t) can be written as
[(t) = u(t, t
0
) [(t
0
)
=

n
gn
i=1
exp
_
iE
n
(t t
0
)
_
a
n,i
[(t
0
) [a
n,i
.
(4.14)
Note that if the system is initially in an eigenvector of the Hamiltonian
with eigenenergy E
n
, then according to Eq. (4.14)
[(t) = exp
_
iE
n
(t t
0
)
_
[(t
0
) . (4.15)
However, the phase factor multiplying [(t
0
) has no eect on any mea-
surable physical quantity of the system, that is, the systems properties are
time independent. This is why the eigenvectors of the Hamiltonian are called
stationary states.
In classical mechanics, the potential energy U of a magnetic moment in a
magnetic eld B is given by
U = B . (4.16)
The magnetic moment of a spin 1/2 is given by [see Eq. (2.90)]
spin
=
2
B
S , (4.17)
where S is the spin angular momentum vector and where
B
=
e
2m
e
c
(4.18)
is the Bohrs magneton (note that the electron charge is taken to be negative
e < 0). Based on these relations we hypothesize that the Hamiltonian of a
spin 1/2 in a magnetic eld B is given by
H =
e
m
e
c
S B . (4.19)
Assume the case where
B = Bz , (4.20)
where B is a constant. For this case the Hamiltonian is given by
H = S
z
, (4.21)
where
=
[e[ B
m
e
c
(4.22)
is the so-called Larmor frequency. In terms of the eigenvectors of the operator
S
z
S
z
[ =
2
[ , (4.23)
where the compact notation [ stands for the states [; z, one has
H[ =
2
[ , (4.24)
namely the states [ are eigenstates of the Hamiltonian. Equation (4.13) for
the present case reads
u(t, 0) = e
it
2
[+ +[ +e
it
2
[ [ . (4.25)
Exercise 4.3.1. Consider spin 1/2 in magnetic eld given by B = Bz, where
B is a constant. Given that [(0) = [+; x at time t = 0 calculate (a) the
probability p
(t) to measure S
x
= /2 at time t; (b) the expectation value
S
x
(t) at time t.
Solution 4.3.1. Recall that [see Eq. (2.102)]
[; x =
1
2
([+ [) (4.26)
(a) Using Eq. (4.25) one nds
p
(t) = [; x[ u(t, 0) [(0)[

2
=
1
2
(+[ [)
_
e
it
2
[+ +[ +e
it
2
[ [
_
([+ +[)
2
=
1
2
_
e
it
2
e
it
2
_
2
,
(4.27)
4.4. Connection to Classical Dynamics
thus
p
+
(t) = cos
2
_
t
2
_
, (4.28)
p
(t) = sin
2
_
t
2
_
. (4.29)
(b)Using the results for p
+
and p
one has
S
x
=

2
(p
+
p
)
=

2
_
cos
2
_
t
2
_
sin
2
_
t
2
__
=

2
cos (t) .
(4.30)
4.4 Connection to Classical Dynamics
In chapter 1 we have found that in classical physics, the dynamics of a variable
A
(c)
is governed by Eq. (1.38), which is given by
dA
(c)
dt
=
_
A
(c)
, H
(c)
_
+
A
(c)
t
. (4.31)
We seek a quantum analogy to this equation. To that end, we derive an
equation of motion for the expectation value A of the observable A that
corresponds to the classical variable A
(c)
. In general, the expectation value
can be expressed as
A = (t)[ A[(t) = (t
0
)[ u
(t, t
0
) Au(t, t
0
) [(t
0
) = (t
0
)[ A
(H)
[(t
0
) ,
(4.32)
where u is the time evolution operator and
A
(H)
= u
(t, t
0
) Au(t, t
0
) . (4.33)
The operator A
(H)
is called the Heisenberg representation of A. We rst
derive an equation of motion for the operator A
(H)
. By using Eq. (4.7) one
nds that the following holds
du
dt
=
1
i
Hu , (4.34)
du
dt
=
1
i
u
H , (4.35)
therefore
dA
(H)
dt
=
du
dt
Au +u
A
du
dt
+u
A
t
u
=
1
i
_
u
HAu +u
AHu
_
+u
A
t
u
=
1
i
_
u
Huu
Au +u
Auu
Hu
_
+u
A
t
u
=
1
i
_
H
(H)
A
(H)
+A
(H)
H
(H)
_
+
A
(H)
t
.
(4.36)
Thus, we have found that
dA
(H)
dt
=
1
i
_
A
(H)
, H
(H)
_
+
A
(H)
t
. (4.37)
Furthermore, the desired equation of motion for A is found using Eqs. (4.32)
and (4.37)
dA
dt
=
1
i
[A, H] +
_
A
t
_
. (4.38)
We see that the Poissons brackets in the classical equation of motion (4.31)
for the classical variable A
(c)
are replaced by a commutation relation in the
quantum counterpart equation of motion (4.38) for the expectation value A
,
1
i
[, ] . (4.39)
Note that for the case where the Hamiltonian is time independent, namely
for the case where the time evolution operator is given by Eq. (4.9), u com-
mutes with H, namely [u, H] = 0, and consequently
H
(H)
= u
Hu = H . (4.40)
4.5 Symmetric Ordering
What is in general the correspondence between a classical variable and its
quantum operator counterpart? Consider for example the system of a point
particle moving in one dimension. Let x
(c)
be the classical coordinate and let
p
(c)
be the canonically conjugate momentum. As we have done in chapter 3,
the quantum observables corresponding to x
(c)
and p
(c)
are the Hermitian
operators x and p. The commutation relation [x, p] is derived from the cor-
responding Poissons brackets
_
x
(c)
, p
(c)
_
according to the rule
,
1
i
[, ] , (4.41)
4.5. Symmetric Ordering
namely
_
x
(c)
, p
(c)
_
= 1 [x, p] = i . (4.42)
However, what is the quantum operator corresponding to a general func-
tion A
_
x
(c)
, p
(c)
_
of x
(c)
and p
(c)
? This question raises the issue of order-
ing. As an example, let A
_
x
(c)
, p
(c)
_
= x
(c)
p
(c)
. Classical variables obviously
commute, therefore x
(c)
p
(c)
= p
(c)
x
(c)
. However, this is not true for quantum
operators xp ,= px. Moreover, it is clear that both operators xp and px cannot
be considered as observables since they are not Hermitian
(xp)
= px ,= xp , (4.43)
(px)
= xp ,= px . (4.44)
A better candidate to serve as the quantum operator corresponding to the
classical variables x
(c)
p
(c)
is the operator (xp +px) /2, which is obtained from
x
(c)
p
(c)
by a procedure called symmetric ordering. A general transformation
that produces a symmetric ordered observable A(x, p) that corresponds to
a given general function A
_
x
(c)
, p
(c)
_
of the classical variable x
(c)
and its
canonical conjugate p
(c)
is given below
A(x, p) =
A
_
x
(c)
, p
(c)
_
dx
(c)
dp
(c)
,
(4.45)
where
=
1
(2)
2
e
i
((x
(c)
x)+(p
(c)
p))
dd . (4.46)
This transformation is called the Weyl transformation. The identity
dke
ik(x
)
= 2 (x
) , (4.47)
implies that
1
2
_
e
i
(x
(c)
x)
d =
_
x
(c)
x
_
, (4.48)
1
2
_
e
i
(p
(c)
p)
d =
_
p
(c)
p
_
. (4.49)
At rst glance these relations may lead to the (wrong) conclusion that the
term equals to
_
x
(c)
x
_
_
p
(c)
p
_
, however, this is incorrect since x
and p are non-commuting operators.
4.6 Problems
1. Consider spin 1/2 in magnetic eld given by B = Bz, where B is a
constant. At time t = 0 the system is in the state [+; x. Calculate S
x
,
S
y
and S
z
as a function of time t.
2. Consider a point particle having mass m moving in one dimension under
the inuence of the potential V (x). Let [
n
be a normalized eigenvector
of the Hamiltonian of the system with eigenvalue E
n
. Show that the cor-
responding wavefunction
n
(x
) in the coordinate representation satises

the following equation
2
2m
d
2
n
(x
)
dx
2
+V (x
)
n
(x
) = E
n
n
(x
) . (4.50)
3. Consider the Hamiltonian operator
H =
p
2
2m
+V (r) , (4.51)
where r = (x, y, z) is the vector of position operators, p = (p
x
, p
y
, p
z
)
is the vector of canonical conjugate operators, and the mass m is a con-
stant. Let [
n
be a normalizable eigenvector of the Hamiltonian H with
eigenvalue E
n
. Show that
n
[ p[
n
= 0 . (4.52)
4. Show that in the p representation the Schrdinger equation
i
d[
dt
= H[ , (4.53)
where H is the Hamiltonian
H =
p
2
2m
+V (r) , (4.54)
can be transformed into the integro-dierential equation
i
d
dt
=
p
2
2m
+
_
dp
U (p p
, (4.55)
where
(p
, t) = p
[ is the momentum wave function and

where
U (p) = (2)
3
_
drV (r) exp
_
p r
_
. (4.56)
5. Consider a particle of mass m in a scalar potential energy V (r). Prove
Ehrenfests theorem
m
d
2
dt
2
r = V (r) . (4.57)
4.6. Problems
6. Show that if the potential energy V (r) can be written as a sum of func-
tions of a single coordinate, V (r) = V
1
(x
1
) +V
2
(x
2
) +V
3
(x
3
), then the
time-independent Schrdinger equation can be decomposed into a set of
one-dimensional equations of the form
d
2
i
(x
i
)
dx
2
i
+
2m
2
[E
i
V
i
(x
i
)]
i
(x
i
) = 0 , (4.58)
where i 1, 2, 3, with (r) =
1
(x
1
)
2
(x
2
)
3
(x
3
) and E = E
1
+
E
2
+E
3
.
7. Show that, in one-dimensional problems, the energy spectrum of the
bound states is always non-degenerate.
8. Let
n
(x) (n = 1, 2, 3, ) be the eigen-wave-functions of a one-
dimensional Schrdinger equation with eigen-energies E
n
placed in order
of increasing magnitude (E
1
< E
2
< . ). Show that between any two
consecutive zeros of
n
(x),
n+1
(x) has at least one zero.
9. What conclusions can be drawn about the parity of the eigen-functions
of the one-dimensional Schrdinger equation
d
2
(x)
dx
2
+
2m
2
(E V (x)) (x) = 0 (4.59)
if the potential energy is an even function of x , namely V (x) = V (x).
10. Show that the rst derivative of the time-independent wavefunction is
continuous even at points where V (x) has a nite discontinuity.
11. A particle having mass mis conned by a one dimensional potential given
by
V
s
(x) =
_
W if [x[ a
0 if [x[ > a
, (4.60)
where a > 0 and W > 0 are real constants. Show that the particle has
at least one bound state (i.e., a state having energy E < 0).
12. Consider a particle having mass m conned in a potential well given by
V (x) =
_
0 if 0 x a
if x < 0 or x > a
. (4.61)
The eigen energies are denoted by E
n
and the corresponding eigen states
are denoted by [
n
, where n = 1, 2, (as usual, the states are num-
bered in increasing order with respect to energy). The state of the system
at time t = 0 is given by
[(0) = a
1
[
1
+a
2
[
2
+a
3
[
3
. (4.62)
(a) The energy E of the system is measured at time t = 0 . What is the
probability to measure a value smaller than 3
2
2
/
_
ma
2
_
? (b) Calculate
the standard deviation E =
_
E
2
E
2
at time t = 0 . (c) the same
as (b), however for any time t > 0 . (d) The energy was measured at
time t and the value of 2
2
2
/
_
ma
2
_
was found. The energy is measured
again at later time t
0
> t . Calculate E and E at time t
0
.
13. Consider a point particle having mass m in a one dimensional potential
given by
V (x) = (x) , (4.63)
where (x) is the delta function. The value of the parameter suddenly
changes from
1
at times t < 0 to the value
2
at times t > 0. Both
1
and
2
are positive real numbers. Given that the particle was in the
ground state at times t < 0, what is the probability p that the particle
will remain bounded at t > 0?
14. Consider a point particle having mass m in a one dimensional potential
given by
V (x) =
_
(x) [x[ < a
[x[ a
, (4.64)
where (x) is the delta function and is a constant. Let E
0
be the energy
of the ground state. Under what conditions E
0
< 0?
15. Thomas-Reiche-Kuhn sum rule - Let
H =
p
2
2m
+V (r) (4.65)
be the Hamiltonian of a particle of mass m moving in a potential V (r).
Show that
k
(E
k
E
l
) [k[ x[l[
2
=

2
2m
, (4.66)
where the sum is taken over all energy eigen-states of the particle (where
H[k = E
k
[k ), and x is the x component of the position vector operator
r (the Thomas-Reiche-Kuhn sum rule).
16. A particle having mass m is conned in a one dimensional potential well
given by
V (x) =
_
0 0 < x < a
else
.
a) At time t = 0 the position was measured and the result was x = a/2.
The resolution of the position measurement is x , where x << a.
After time
1
the energy was measured. Calculate the probability p
n
to measure that the energy of the system is E
n
, where E
n
are the
eigen energies of the particle in the well, and where n = 1, 2, .
4.6. Problems
b) Assume that the result of the measurement in the previous section
was E
2
. At a later time
2
>
1
the momentum p of the particle
was measured. Calculate the expectation value p.
17. A particle having mass m is in the ground state of an innite potential
well of width a, which is given by
V
1
(x) =
_
0 0 < x < a
else
. (4.67)
At time t = 0 the potential suddenly changes and becomes
V
2
(x) =
_
0 0 < x < 2a
else
, (4.68)
namely the width suddenly becomes 2a. (a) Find the probability p to
nd the particle in the ground state of the new well. (b) Calculate the
expectation value of the energy H before and after the change in the
potential.
18. Calculate the uncertainties in position
_
(x)
2
_
and in momentum
_
(p)
2
_
of the energy eigenstates of a particle having mass m, which
is conned in a one dimensional potential well given by
V (x) =
_
0 0 < x < a
else
. (4.69)
19. The continuity equation - Consider a point particle having mass m
and charge q placed in an electromagnetic eld. Show that
d
dt
+J = 0 , (4.70)
where
=
(4.71)
is the probability density, (x
) is the wavefunction,
J =

m
Im(
)
q
mc
A (4.72)
is the current density, and A is the electromagnetic vector potential.
20. A particle having mass m moves in one dimension under the inuence of
the potential V (x
). In the range [x
[ > a the potential V (x
) vanishes.
Consider a solution to the time independent Schrdinger equation, whose
wavefunction (x
) in the range [x
[ > a is taken to be given by

(x
) =
_
A
1
e
ikx
+B
1
e
ikx
< a
A
2
e
ikx
+B
2
e
ikx
> a
, (4.73)
where A
1
, B
1
, A
2
, B
2
and k are all constants. Find a relation that the
constants A
1
, B
1
, A
2
and B
2
must satisfy.
21. Calculate the Weyl transformation A(x, p) of the classical variable A
_
x
(c)
, p
(c)
_
=
p
(c)
x
(c)
.
22. Invert Eq. (4.45), i.e. express the variable A
_
x
(c)
, p
(c)
_
as a function of
the operator A(x, p).
4.7 Solutions
1. The operators S
x
, S
y
and S
z
are given by Eqs. (2.102), (2.103) and (2.99)
respectively. The Hamiltonian is given by Eq. (4.21). Using Eqs. (4.38)
and (2.131) one has
dS
x
dt
=

i
[S
x
, S
z
] = S
y
, (4.74)
dS
y
dt
=

i
[S
y
, S
z
] = S
x
, (4.75)
dS
z
dt
=

i
[S
z
, S
z
] = 0 , (4.76)
where
=
[e[ B
m
e
c
. (4.77)
At time t = 0 the system is in state
[+; x =
1
2
([+ +[) , (4.78)
thus
S
x
(t = 0) =

4
(+[ +[) ([+ [ +[ +[) ([+ +[) =

2
.
S
y
(t = 0) =

4
(+[ +[) (i [+ [ +i [ +[) ([+ +[) = 0 .
S
z
(t = 0) =

4
(+[ +[) ([+ +[ [ [) ([+ +[) = 0 .
The solution is easily found to be given by
S
x
(t) =
_
2
_
cos (t) , (4.79)
S
y
(t) =
_
2
_
sin(t) , (4.80)
S
z
(t) = 0 . (4.81)
2. The Hamiltonian operator H is given by
H =
p
2
2m
+V (x) . (4.82)
4.7. Solutions
Multiplying the relation
H[
n
= E
n
[
n
(4.83)
from the left by x
[ yields [see Eqs. (3.23) and (3.29)]
2
2m
d
2
n
(x
)
dx
2
+V (x
)
n
(x
) = E
n
n
(x
) , (4.84)
where
n
(x
) = x
[
n
(4.85)
is the wavefunction in the coordinate representation.
3. Using [x, p
x
] = [y, p
y
] = [z, p
z
] = i one nds that
[H, r] =
_
p
2
2m
, r
_
=
1
2m
__
p
2
x
, x
,
_
p
2
y
, y
,
_
p
2
z
, z
_
=

im
(p
x
, p
y
, p
z
)
=

im
p .
(4.86)
Thus
n
[ p[
n
=
im

n
[ [H, r] [
n
=
im

n
[ (Hr rH) [
n
=
imE
n

n
[ (r r) [
n
= 0 .
(4.87)
4. Multiplying Eq. (4.53) from the left by the bra p
[ and inserting the

closure relation
1 =
_
dp
[p
[ (4.88)
yields
i
d
(p
)
dt
=
_
dp
[ H[p
(p
) . (4.89)
The following hold
p
[ p
2
[p
= p
2
(p
) , (4.90)
and
p
[ V (r) [p
=
_
dr
_
dr
[r
[ V (r) [r
[p
= (2)
3
_
dr
_
dr
exp
_
ip
_
V (r
) (r
) exp
_
ip
_
= (2)
3
_
dr
exp
_
i (p
) r
_
V (r
)
= U (p
) ,
(4.91)
thus the momentum wave function
(p
) satises the following equation

i
d
dt
=
p
2
2m
+
_
dp
U (p
. (4.92)
5. The Hamiltonian is given by
H =
p
2
2m
+V (r) . (4.93)
Using Eq. (4.38) one has
dx
dt
=
1
i
[x, H] =
1
i2m
_
x, p
2
x
_
=
p
x
m
, (4.94)
and
dp
x
dt
=
1
i
[p
x
, V (r)] , (4.95)
or with the help of Eq. (3.76)
dp
x
dt
=
_
V
x
_
. (4.96)
This together with Eq. (4.94) yield
m
d
2
x
dt
2
=
_
V
x
_
. (4.97)
Similar equations are obtained for y and z, which together yield Eq.
(4.57).
6. Substituting a solution having the form
(r) =
1
(x
1
)
2
(x
2
)
3
(x
3
) (4.98)
4.7. Solutions
into the time-independent Schrdinger equation, which is given by
2
(r) +
2m
2
[E V (r)] (r) = 0 , (4.99)
and dividing by (r) yield
3
i=1
_
1
i
(x
i
)
d
2
i
(x
i
)
dx
2
i

2m
2
V
i
(x
i
)
_
=
2m
2
E . (4.100)
In the sum, the i th term (i 1, 2, 3) depends only on x
i
, thus each
term must be a constant
1
i
(x
i
)
d
2
i
(x
i
)
dx
2
i

2m
2
V
i
(x
i
) =
2m
2
E
i
, (4.101)
where E
1
+E
2
+E
3
= E.
7. Consider two eigen-wave-functions
1
(x) and
2
(x) having the same
eigenenergy E. The following holds
d
2
1
dx
2
+
2m
2
(E V (x))
1
= 0 , (4.102)
d
2
2
dx
2
+
2m
2
(E V (x))
2
= 0 , (4.103)
thus
1
1
d
2
1
dx
2
=
1
2
d
2
2
dx
2
, (4.104)
or
2
d
2
1
dx
2

1
d
2
2
dx
2
=
d
dx
_
2
d
1
dx

1
d
2
dx
_
= 0 , (4.105)
therefore
2
d
1
dx

1
d
2
dx
= C , (4.106)
where C is a constant. However, for bound states
lim
x
(x) = 0 , (4.107)
thus C = 0, and consequently
1
1
d
1
dx
=
1
2
d
2
dx
. (4.108)
Integrating the above equation yields
log
1
= log
2
+ , (4.109)
where is a constant. Therefore
1
= e
2
, (4.110)
and therefore
2
is just proportional to
1
(both represent the same
physical state).
8. Consider two eigen-wave-functions
n
(x) and
n+1
(x) with E
n
< E
n+1
.
As we saw in the previous exercise, the spectrum is non-degenerate. More-
over, the Schrdinger equation
d
2
dx
2
+
2m
2
(E V (x)) = 0 , (4.111)
which the eigen-wave-functions satisfy, is real. Therefore given that (x)
is a solution with a given eigenenergy E, then also
(x) is a solution
with the same E. Therefore, all eigen-wave-functions can be chosen to be
real (i.e., by the transformation (x) ( (x) +
(x)) /2). We have

d
2
n
dx
2
+
2m
2
(E
n
V (x))
n
= 0 , (4.112)
d
2
n+1
dx
2
+
2m
2
(E
n+1
V (x))
n+1
= 0 . (4.113)
By multiplying the rst Eq. by
n+1
, the second one by
n
, and sub-
tracting one has
n+1
d
2
n
dx
2

n
d
2
n+1
dx
2
+
2m
2
(E
n
E
n+1
)
n
n+1
= 0 , (4.114)
or
d
dx
_
n+1
d
n
dx

n
d
n+1
dx
_
+
2m
2
[E
n
E
n+1
]
n
n+1
= 0 . (4.115)
Let x
1
and x
2
be two consecutive zeros of
n
(x) (i.e.,
n
(x
1
) =
n
(x
2
) = 0). Integrating from x
1
to x
2
yields
_
_
n+1
d
n
dx

n
..
=0
d
n+1
dx
x2
x1
=
2m
2
(E
n+1
E
n
)
. .
>0
_
x
2
x
1
dx
n
n+1
.
(4.116)
Without lost of generality, assume that
n
(x) > 0 in the range (x
1
, x
2
).
Since
n
(x) is expected to be continuous, the following must hold
d
n
dx
x=x1
> 0 , (4.117)
d
n
dx
x=x2
< 0 . (4.118)
4.7. Solutions
As can be clearly seen from Eq. (4.116), the assumption that
n+1
(x) > 0
in the entire range (x
1
, x
2
) leads to contradiction. Similarly, the possibil-
ity that
n+1
(x) < 0 in the entire range (x
1
, x
2
) is excluded. Therefore,
n+1
must have at least one zero in this range.
9. Clearly if (x) is an eigen function with energy E, also (x) is an
eigen function with the same energy. Consider two cases: (i) The level E
is non-degenerate. In this case (x) = c (x), where c is a constant.
Normalization requires that [c[
2
= 1. Moreover, since the wavefunctions
can be chosen to be real, the following holds: (x) = (x). (ii) The
level E is degenerate. In this case every superposition of (x) and (x)
can be written as a superposition of an odd eigen function
odd
(x) and
an even one
even
(x), which are dened by
odd
(x) = (x) (x) , (4.119)
even
(x) = (x) + (x) . (4.120)
10. The time-independent Schrdinger equation reads
d
2
(x)
dx
2
+
2m
2
(E V (x)) (x) = 0 . (4.121)
Assume V (x) has a nite discontinuity at x = x
0
. Integrating the
Schrdinger equation in the interval (x
0
, x
0
+) yields
_
d (x)
dx
x0+
x0
=
2m
2
x0+
_
x0
(V (x) E) (x) = 0 . (4.122)
In the limit 0 the right hand side vanishes (assuming (x) is
bounded). Therefore d(x) /dx is continuous at x = x
0
.
11. Since V
s
(x) = V
s
(x) the ground state wavefunction is expected to be
an even function of x. Consider a solution having an energy E and a
wavefunction of the form
(x) =
_
_
_
Ae
x
if x > a
Bcos (kx) if a x a
Ae
x
if x < a
, (4.123)
where
=
2mE
, (4.124)
and
k =
_
2m(W +E)
. (4.125)
Requiring that both (x) and d (x) /dx are continuous at x = a yields
Ae
a
= Bcos (ka) , (4.126)
and
Ae
a
= kBsin(ka) , (4.127)
or in a matrix form
C
_
A
B
_
=
_
0
0
_
, (4.128)
where
C =
_
e
a
cos (ka)
e
a
k sin(ka)
_
. (4.129)
A nontrivial solution exists i Det (C) = 0, namely i
k
= tan(ka) . (4.130)
This condition can be rewritten using Eqs. (4.124) and (4.125) and the
dimensionless parameters
K = ka , (4.131)
K
0
=
2mW
a , (4.132)
as
cos
2
K =
1
1 + tan
2
K
=
1
1 +
_
k
_
2
=
_
K
K
0
_
2
. (4.133)
Note, however, that according to Eq. (4.130) tanK > 0. Thus, Eq. (4.130)
is equivalent to the set of equations
[cos K[ =
K
K
0
, (4.134)
tanK > 0 . (4.135)
This set has at least one solution (this can be seen by plotting the func-
tions [cos K[ and K/K
0
).
12. Final answers: (a) [a
1
[
2
+[a
2
[
2
. (b)
E =

2
2
2ma
2
_
3
n=1
[a
n
[
2
n
4
_
3
n=1
[a
n
[
2
n
2
_
2
. (4.136)
(c) The same as at t = 0. (d) E = 2
2
2
/
_
ma
2
_
, E = 0.
4.7. Solutions
13. The Schrdinger equation for the wavefunction (x) is given by
_
d
2
dx
2
+
2m
2
E
_
(x) = 0 . (4.137)
The boundary conditions at x = 0 are
_
0
+
_
=
_
0
_
, (4.138)
d (0
+
)
dx

d (0
)
dx
=
2
a
0
(0) , (4.139)
where
a
0
=

2
m
. (4.140)
Due to symmetry V (x) = V (x) the solutions are expected to have
denite symmetry (even (x) = (x) or odd (x) = (x)). For
the ground state, which is expected to have even symmetry, we consider
a wavefunction having the form
(x) = Ae
|x|
, (4.141)
where A is a normalization constants and where
=
2mE
. (4.142)
The parameter is real for E < 0. This even wavefunction satises the
Schrdinger equation for x ,= 0 and the boundary condition (4.138). The
condition (4.139) leads to a single solution for the energy of the ground
state
E =
m
2
2
2
. (4.143)
Thus the normalized wavefunction of the ground state
0
(x) is given by
0,
(x) =
_
m
2
exp
_
2
[x[
_
. (4.144)
The probability p that the particle will remain bounded is given by
p =
0,
1
(x)
0,
2
(x) dx
2
=
4m
2
_

0
exp
_
m(
1
+
2
)
2
x
_
dx
2
=
4
1
2
(
1
+
2
)
2
.
(4.145)
14. The Schrdinger equation for the wavefunction (x) is given by
_
d
2
dx
2
+
2m
2
(E V )
_
(x) = 0 . (4.146)
The boundary conditions imposed upon (x) by the potential are [see
Eq. (4.122)]
(a) = 0 , (4.147)
_
0
+
_
=
_
0
_
, (4.148)
d (0
+
)
dx

d (0
)
dx
=
2
a
0
(0) , (4.149)
where
a
0
=

2
m
. (4.150)
Due to symmetry V (x) = V (x) the solutions are expected to have
denite symmetry (even (x) = (x) or odd (x) = (x)). For
the ground state, which is expected to have even symmetry, we consider
a wavefunction having the form
(x) =
_
Asinh((x a)) x > 0
Asinh((x +a)) x < 0
, (4.151)
where A is a normalization constants and where
=
2mE
0
. (4.152)
The parameter is real for E
0
< 0. This even wavefunction satises Eq.
(4.146) for x ,= 0 and the boundary conditions (4.147) and (4.148). The
condition (4.149) reads
a
0
= tanh(a) . (4.153)
Nontrivial ( ,= 0) real solution exists only when a > a
0
, thus E
0
< 0 i
a > a
0
=

2
m
. (4.154)
15. Using Eq. (4.37) one has
dx
(H)
dt
=
1
i
_
x
(H)
, H
_
, (4.155)
therefore
k[
dx
(H)
dt
[l =
1
i
k[ x
(H)
HHx
(H)
[l =
i (E
k
E
l
)
k[ x
(H)
[l .
4.7. Solutions
(4.156)
Integrating yields
k[ x
(H)
(t) [l = k[ x
(H)
(t = 0) [l exp
_
i (E
k
E
l
) t
_
. (4.157)
Using this result one has
k
(E
k
E
l
) [k[ x[l[
2
=

k
(E
k
E
l
)
k[ x
(H)
[l
2
=

k
(E
k
E
l
) k[ x
(H)
[l l[ x
(H)
[k
=

2i
k
_
k[
dx
(H)
dt
[l l[ x
(H)
[k k[ x
(H)
[l l[
dx
(H)
dt
[k
_
=

2i
k
_
l[ x
(H)
[k k[
dx
(H)
dt
[l l[
dx
(H)
dt
[k k[ x
(H)
[l
_
=

2i
l[ x
(H)
dx
(H)
dt

dx
(H)
dt
x
(H)
[l .
(4.158)
Using again Eq. (4.37) one has
dx
(H)
dt
=
1
i
_
x
(H)
, H
_
=
p
(H)
x
m
, (4.159)
therefore
k
(E
k
E
l
) [k[ x[l[
2
=

2im
l[
_
x
(H)
, p
(H)
x
_
[l
=

2im
i
=

2
2m
.
(4.160)
16. The wavefunctions of the normalized eigenstates are given by
n
(x) =
_
2
a
sin
nx
a
, (4.161)
and the corresponding eigenenergies are
E
n
=

2
2
n
2
2ma
2
. (4.162)
a) The wavefunction after the measurement is a normalized wavepacket
centered at x = a/2 and having a width x
(x) =
_
1
x
a
2
x
2
0 else
. (4.163)
Thus in the limit x << a
p
n
=
_
a
0
dx
n
(x) (x)
2
2
x
a
sin
2
n
2
. (4.164)
Namely, p
n
= 0 for all even n, and the probability of all energies with
odd n is equal.
b) Generally, for every bound state in one dimension p = 0 [see Eq.
(4.52)].
17. For a well of width a the wavefunctions of the normalized eigenstates are
given by
(a)
n
(x) =
_
2
a
sin
nx
a
, (4.165)
E
(a)
n
=

2
2
n
2
2ma
2
. (4.166)
(a) The probability is given by
p =
_
a
0
dx
(a)
1
(x)
(2a)
1
(x)
2
=
32
9
2
. (4.167)
(b) For times t < 0 it is given that H = E
(a)
1
. Immediately after the
change (t = 0
+
) the wavefunction remains unchanged. A direct evaluation
of H using the new Hamiltonian yields the same result H = E
(a)
1
as for
t < 0. At later times t > 0 the expectation value H remains unchanged
due to energy conservation.
18. The wavefunctions of the normalized eigenstates are given by [see Eq.
(4.162)]
n
(x) =
_
2
a
sin
nx
a
, (4.168)
and the corresponding eigenenergies are [see Eq. (4.161)]
E
n
=

2
2
n
2
2ma
2
, (4.169)
4.7. Solutions
where n = 1, 2, . By symmetry for all states x = a/2. Furthermore,
for all states p = 0 [see Eq. (4.52)]. For the nth state the following
holds
x
2
_
=
2
a
_
a
0
dx
x
2
sin
2
nx
a
=
a
2
_
2n
2
2
3
_
6n
2
2
, (4.170)
thus
_
(x)
2
_
=
x
2
_
x
2
= a
2
_
1
12

1
2n
2
2
_
and [see Eq. (3.29)]
_
(p)
2
_
=
p
2
_
=
2
a
_
n
a
_
2
_
a
0
dx
sin
2
nx
a
=
_
n
a
_
2
, (4.171)
thus [compare with the uncertainty principle (3.10)]
_
(x)
2
__
(p)
2
_
=

2
4
_
n
2
2
3
2
_
. (4.172)
19. The Schrdinger equation is given by
i
d[
dt
= H[ , (4.173)
where the Hamiltonian is given by [see Eq. (1.62)]
H =
_
p
q
c
A
_
2
2m
+q . (4.174)
Multiplying from the left by x
[ yields
i
d
dt
=
1
2m
_
i
q
c
A
_
2
+q , (4.175)
where
= (x
) = x
[ . (4.176)
Multiplying Eq. (4.175) by
, and subtracting the complex conjugate of

Eq. (4.175) multiplied by yields
i
d
dt
=
1
2m
_
_
i
q
c
A
_
2

_
i
q
c
A
_
2
_
, (4.177)
where
=
(4.178)
is the probability density. Moreover, the following holds
_
i
q
c
A
_
2

_
i
q
c
A
_
2
2
+
_
q
c
_
2
A
2
+
iq
c
A+
iq
c
A
_
2
+
_
q
c
_
2
A
2
iq
c
A
iq
c
A
_
=
2
_
2

2
_
+
iq
c
(
A +
A +A
+A
)
=
2
(
) +
iq
c
(
A +A
) .
(4.179)
Thus, Eq. (4.177) can be written as
d
dt
+J = 0 , (4.180)
where
J =

m
Im(
)
q
mc
A . (4.181)
20. The current density J [see Eq. (4.72)] that is associated with the wave-
function (x
) = Ae
ikx
+Be
ikx
is given by
J =

m
Im
_
_
=

m
Im
_
ik
_
A
e
ikx
+B
e
ikx
__
Ae
ikx
Be
ikx
__
=

m
Im
_
ik
_
[A[
2
[B[
2
+AB
e
2ikx
Be
2ikx
__
=
k
m
_
[A[
2
[B[
2
_
.
(4.182)
Thus for a solution to the time independent Schrdinger equation, for
which the current density =
is time independent, the continuity

equation (4.70) yields the relation
[A
1
[
2
[B
1
[
2
= [A
2
[
2
[B
2
[
2
. (4.183)
A(x, p) =
1
(2)
2
_ _ _ _
p
(c)
x
(c)
e
i
[(x
(c)
x)+(p
(c)
p)]
dddx
(c)
dp
(c)
.
4.7. Solutions
(4.184)
e
A
e
B
= e
A+B
e
(1/2)[A,B]
, (4.185)
one has
e
x
e
p
= e
(x+p)
e
1
2
2
[x,p]
, (4.186)
thus
A(x, p) =
1
(2)
2
_ _ _ _
p
(c)
x
(c)
e
i
(x
(c)
+p
(c)
)
e
i

2
e
x
e
p
dddx
(c)
dp
(c)
=
1
(2)
2
_ _ _ _
p
(c)
x
(c)
e
i
[((x
(c)
+
2
)+p
(c)
)]
e
x
e
p
dddx
(c)
dp
(c)
.
(4.187)
Changing the integration variable
x
(c)
= x
(c)

2
, (4.188)
one has
A(x, p) =
1
(2)
2
_ _ _ _
p
(c)
_
x
(c)

2
_
e
i
(x
(c)
+p
(c)
)
e
x
e
p
dddx
(c)
dp
(c)
=
1
(2)
2
_ _ _ _
p
(c)
_
x
(c)

2
_
e
i
(x
(c)
x)
e
i
(p
(c)
p)
dddx
(c)
dp
(c)
.
(4.189)
Using the identity
dke
ik(x
)
= 2 (x
) , (4.190)
one nds that
1
2
_
e
i
(x
(c)
x)
d =
_
x
(c)
x
_
, (4.191)
1
2
_
e
i
(p
(c)
p)
d =
_
p
(c)
p
_
, (4.192)
thus
A(x, p) =
1
2
_ _ _
p
(c)
_
x
(c)

2
_
e
i
(p
(c)
p)
ddx
(c)
dp
(c)
1
2
_
e
i
(x
(c)
x)
d
=
1
2
_ _ _
p
(c)
_
x
(c)

2
_
e
i
(p
(c)
p)
ddx
(c)
dp
(c)
_
x
(c)
x
_
=
1
2
_ _
p
(c)
_
x

2
_
e
i
(p
(c)
p)
ddp
(c)
=
_
p
(c)
xdp
(c)
1
2
_
e
i
(p
(c)
p)
d
1
2
_ _
p
(c)
2
e
i
(p
(c)
p)
ddp
(c)
= px
1
2
_ _
p
(c)
2
e
i
(p
(c)
p)
ddp
(c)
= px
1
2

2i
_ _
p
(c)
e
i
(p
(c)
p)
p
(c)
ddp
(c)
= px

2i
_
dp
(c)
p
(c)

p
(c)
1
2
_
de
i
(p
(c)
p)
. .
(p
(c)
p)
.
(4.193)
Integration by parts yields
A(x, p) = px

2i
_ _
p
(c)
p
(c)
_
_
p
(c)
p
_
dp
(c)
= px

2i
= px +
[x, p]
2
=
xp +px
2
.
(4.194)
22. Below we derive an expression for the variable A
_
x
(c)
, p
(c)
_
in terms
of the matrix elements of the operator A(x, p) in the basis of position
eigenvectors [x
. To that end we begin by evaluating the matrix element

_
x
A(x, p)
+
x
2
_
using Eqs. (4.187), (3.19) and (4.191)
4.7. Solutions
_
x
A(x, p)
+
x
2
_
=
1
(2)
2
_ _ _ _
A
_
x
(c)
, p
(c)
_
e
i
[((x
(c)
+
2
)+p
(c)
)]
_
x
x
e
+
x
2
_
dddx
(c)
dp
(c)
=
1
(2)
2
_ _ _ _
A
_
x
(c)
, p
(c)
_
e
i
[((x
(c)
+
2
)+p
(c)
)]
e

_
x
2
_
_
x
+
x
2
+
_
dddx
(c)
dp
(c)
=
1
(2)
2
_ _
A
_
x
(c)
, p
(c)
_
e
p
(c)
dx
(c)
dp
(c)
_
e
i
[(x
(c)
x
)]
d
=
1
2
_ _
A
_
x
(c)
, p
(c)
_
e
p
(c)
dx
(c)
dp
(c)
_
x
(c)
x
_
=
1
2
_
A
_
x
, p
(c)
_
e
p
(c)
dp
(c)
.
Applying the inverse Fourier transform, i.e. multiplying by e
i
and
integrating over x
yields
_ _
x
A(x, p)
+
x
2
_
e
i
dx
=
1
2
_
A
_
x
, p
(c)
_
dp
(c)
_
e
i
(p
p
(c)
)
dx
,
(4.195)
thus with the help of Eq. (4.192) one nds the desired inversion of Eq.
(4.45) is given by
A(x
, p
) =
_ _
x
A(x, p)
+
x
2
_
e
i
dx
. (4.196)
Note that A(x
, p
), which appears on the left hand side of the above

equation (4.196) is a classical variable, whereas A(x, p) on the right hand
side is the corresponding quantum operator. A useful relations can be
obtained by integrating A(x
, p
) over p
. With the help of Eq. (4.191)

one nds that
_
A(x
, p
) dp
=
_
dx
_
x
A(x, p)
+
x
2
__
e
i
dp
= 2 x
[ A(x, p) [x
.
(4.197)
Another useful relations can be obtained by integrating A(x
, p
) over
x
.With the help of Eqs. (3.52) and (4.192) one nds that
_
A(x
, p
) dx
=
_ _ _
x
A(x, p)
+
x
2
_
e
i
dx
dx
=
_ _ _ _ _
x
2
[p
[ A(x, p) [p
+
x
2
_
e
i
dx
dx
dp
dp
=
1
2
_ _ _ _
e
i
(p
)
e
i
2
(p
)
p
[ A(x, p) [p
e
i
dx
dx
dp
dp
=
_ _ _
(p
) e
i
2
(p
)
p
[ A(x, p) [p
e
i
dx
dp
dp
=
_ _
p
[ A(x, p) [p
e
i
(p
)
dx
dp
= 2
_
p
[ A(x, p) [p
(p
) dp
= 2 p
[ A(x, p) [p
.
(4.198)
5. The Harmonic Oscillator
Consider a particle of mass m in a parabolic potential well
U (x) =
1
2
m
2
x
2
,
where the angular frequency is a constant. The classical equation of motion
for the coordinate x is given by [see Eq. (1.19)]
m x =
U
x
= m
2
x . (5.1)
It is convenient to introduce the complex variable , which is given by
=
1
x
0
_
x +
i
x
_
, (5.2)
where x
0
is a constant having dimension of length. Using Eq. (5.1) one nds
that
=
1
x
0
_
x +
i
x
_
=
1
x
0
_
x
i
2
x
_
= i . (5.3)
The solution is given by
=
0
e
it
, (5.4)
where
0
= (t = 0). Thus, x and x oscillate in time according to
x = x
0
Re
_
0
e
it
_
, (5.5)
x = x
0
Im
_
0
e
it
_
. (5.6)
The Hamiltonian is given by [see Eq. (1.34)]
H =
p
2
2m
+
m
2
x
2
2
. (5.7)
In quantum mechanics the variables x and p are regarded as operators satis-
fying the following commutation relations [see Eq. (3.9)]
[x, p] = xp px = i . (5.8)
Chapter 5. The Harmonic Oscillator
5.1 Eigenstates
The annihilation and creation operators are dened as
a =
_
m
2
_
x +
ip
m
_
, (5.9)
a
=
_
m
2
_
x
ip
m
_
. (5.10)
The inverse transformation is given by
x =
_

2m
_
a +a
_
, (5.11)
p = i
_
m
2
_
a +a
_
. (5.12)
The following holds
_
a, a
=
i
2
([p, x] [x, p]) = 1 , (5.13)
The number operator, which is dened as
N = a
a, (5.14)
can be expressed in terms of the Hamiltonian
N = a
a
=
m
2
_
x
ip
m
__
x +
ip
m
_
=
m
2
_
p
2
m
2
2
+x
2
+
i [x, p]
m
_
=
1
_
p
2
2m
+
m
2
x
2
2
_
1
2
=
H

1
2
.
(5.15)
Thus, the Hamiltonian can be written as
H =
_
N +
1
2
_
. (5.16)
The operator N is Hermitian, i.e. N = N
, therefore its eigenvalues are

expected to be real. Let [n be the set of eigenvectors of N and let n be
the corresponding set of eigenvalues
5.1. Eigenstates
N [n = n[n . (5.17)
According to Eq. (5.16) the eigenvectors of N are also eigenvectors of H
H[n = E
n
[n , (5.18)
where the eigenenergies E
n
are given by
E
n
=
_
n +
1
2
_
. (5.19)
Theorem 5.1.1. Let [n be a normalized eigenvector of the operator N with
eigenvalue n. Then (i) the vector
[n + 1 = (n + 1)
1/2
a
[n (5.20)
is a normalized eigenvector of the operator N with eigenvalue n + 1; (ii) the
vector
[n 1 = n
1/2
a [n (5.21)
is a normalized eigenvector of the operator N with eigenvalue n 1
Proof. Using the commutation relations
_
N, a
= a
_
a, a
= a
, (5.22)
[N, a] =
_
a
, a
a = a , (5.23)
one nds that
Na
[n =
__
N, a
+a
N
_
[n = (n + 1) a
[n , (5.24)
and
Na[n = ([N, a] +aN) [n = (n 1) a [n . (5.25)
Thus, the vector a
[n, which is proportional to [n + 1, is an eigenvector of

the operator N with eigenvalue n + 1 and the vector a [n, which is propor-
tional to [n 1, is an eigenvector of the operator N with eigenvalue n 1.
Normalization is veried as follows
n + 1 [n + 1 = (n + 1)
1
n[ aa
[n = (n + 1)
1
n[
_
a, a
+a
a [n = 1 ,
(5.26)
and
n 1 [n 1 = n
1
n[ a
a [n = 1 . (5.27)
As we have seen from the above theorem the following hold
a[n =
n[n 1 , (5.28)
a
[n =
n + 1 [n + 1 . (5.29)
Claim. The spectrum (i.e. the set of eigenvalues) of N are the nonnegative
integers 0, 1, 2, .
Proof. First, note that since the operator N is positive-denite the eigenval-
ues are necessarily non negative
n = n[ a
a [n 0 . (5.30)
On the other hand, according to Eq. (5.28), if n is an eigenvalue also n 1
is an eigenvalue, unless n = 0. For the later case according to Eq. (5.28)
a [0 = 0. Therefore, n must be an integer, since otherwise one reaches a
contradiction with the requirement that n 0.
According to exercise 6 of set 4, in one-dimensional problems the energy
spectrum of the bound states is always non-degenerate. Therefore, one con-
cludes that all eigenvalues of N are non-degenerate. Therefore, the closure
relation can be written as
1 =

n=0
[n n[ . (5.31)
Furthermore, using Eq. (5.29) one can express the state [n in terms of the
ground state [0 as
[n =
_
a
_
n
n!
[0 . (5.32)
5.2 Coherent States
As can be easily seen from Eqs. (5.11), (5.12), (5.28) and (5.29), all energy
eigenstates [n have vanishing position and momentum expectation values
n[ x[n = 0 , (5.33)
n[ p [n = 0 . (5.34)
Clearly these states dont oscillate in phase space as classical harmonic os-
cillators do. Can one nd quantum states having dynamics that resembles
classical harmonic oscillators?
5.2. Coherent States
Denition 5.2.1. Consider a harmonic oscillator having ground state [0.
A coherent state [ with a complex parameter is dened by
[ = D() [0 , (5.35)
where
D() = exp
_
a
a
_
, (5.36)
is the displacement operator.
In the set of problems at the end of this chapter the following results are
obtained:
The displacement operator is unitary D
() D() = D() D
() = 1.
The coherent state [ is an eigenvector of the operator a with an eigenvalue
, namely
a [ = [ . (5.37)
For any function f
_
a, a
_
having a power series expansion the following
holds
D
() f
_
a, a
_
D() = f
_
a +, a
_
. (5.38)
The displacement operator satises the following relations
D() = e
||
2
2
e
a
a
= e
||
2
2
e
a
e
a
, (5.39)
D() = e
2
x
e
m
+
2
p
e
2
4
, (5.40)
D() D(
) = exp
_
2
_
D( +
) . (5.41)
Coherent state expansion in the basis of number states
[ = e
||
2
2
n=0
n!
[n . (5.42)
The following expectation values hold
H
= [ H[ =
_
[[
2
+ 1/2
_
, (5.43)
[ H
2
[ =
2
2
_
[[
4
+ 2 [[
2
+ 1/4
_
, (5.44)
H
=
_
[ (H)
2
[ = [[ , (5.45)
x
= [ x[ =
_
2
m
Re () , (5.46)
p
= [ p [ =
2m Im() , (5.47)
x
=
_
[ (x)
2
[ =
_

2m
, (5.48)
p
=
_
[ (p)
2
[ =
_
m
2
, (5.49)
x
=

2
. (5.50)
The wave function of a coherent state is given by
(x
) = x
[
= exp
_
2
4
_
_
m
_
1/4
exp
_
_
x
2x
_
2
+i p
_
.
(5.51)
The following closure relation holds
1 =
1
_ _
[ [ d
2
, (5.52)
where d
2
denotes innitesimal area in the complex plane, namely d
2
=
dRe dIm.
Given that at time t = 0 the oscillator is in a coherent state with para-
meter
0
, namely [(t = 0) = [
0
, the time evolution can be found with
the help of Eqs. (4.14), (5.19) and (5.42)
[ (t) = e
|
0
|
2
2
n=0
exp
_
iE
n
t
_

n
0
n!
[n
= e
it/2
e
|
0
|
2
2
n=0
exp(int)

n
0
n!
[n
= e
it/2
e
|
0
|
2
2
n=0
_
0
e
it
_
n
n!
[n
= e
it/2
=
0
e
it
_
.
(5.53)
5.3. Problems
In view of Eqs. (5.43), (5.45) (5.48) and (5.49), we see from this results that
H
, H
, x
and p
are all time independent. On the other hand, as

can be seen from Eqs. (5.46) and (5.47) the following holds
x
= [ x[ =
_
2
m
Re
_
0
e
it
_
, (5.54)
p
= [ p [ =
2m Im
_
0
e
it
_
. (5.55)
These results show that indeed, x
and p
have oscillatory time depen-

dence identical to the dynamics of the position and momentum of a classical
harmonic oscillator [compare with Eqs. (5.5) and (5.6)].
5.3 Problems
1. Calculate the wave functions
n
(x
) = x
[n of the number states [n

of a harmonic oscillator.
2. Show that
exp
_
2Xt t
2
_
=

n=0
H
n
(X)
t
n
n!
, (5.56)
where H
n
(X) is the Hermite polynomial of order n, which is dened by
H
n
(X) = exp
_
X
2
2
__
X
d
dX
_
n
exp
_
X
2
2
_
. (5.57)
3. Show that for the state [n of a harmonic oscillator
_
(x)
2
__
(p)
2
_
=
_
n +
1
2
_
2
2
. (5.58)
4. Consider a free particle in one dimension having mass m. Express the
Heisenberg operator x
(H)
(t) in terms x
(H)
(0) and p
(H)
(0). At time t = 0
the system in in the state [
0
. Express the variance
_
(x)
2
_
(t) at time
t, where x = x x, in terms of the following expectation values at
time t = 0
x
0
=
0
[ x[
0
, (5.59)
p
0
=
0
[ p [
0
, (5.60)
(xp)
0
=
0
[ xp [
0
, (5.61)
(x)
2
0
=
0
[ (x x
0
)
2
[
0
, (5.62)
(p)
2
0
=
0
[ (p p
0
)
2
[
0
. (5.63)
5. Consider a harmonic oscillator of angular frequency and mass m.
a) Express the Heisenberg picture x
(H)
(t) and p
(H)
(t) in terms x
(H)
(0)
and p
(H)
(0).
b) Calculate the following commutators
_
p
(H)
(t
1
) , x
(H)
(t
2
)
,
_
p
(H)
(t
1
) , p
(H)
(t
2
)
and
_
x
(H)
(t
1
) , x
(H)
(t
2
)
.
6. Consider a particle having mass m conned by a one dimensional poten-
tial V (x), which is given by
V (x) =
_
m
2
2
x
2
x > 0
x 0
, (5.64)
where is a constant.
a) Calculate the eigenenergies of the system.
b) Calculate the expectation values

x
2
_
of all energy eigenstates of the
particle.
7. Calculate the possible energy values of a particle in the potential given
by
V (x) =
m
2
2
x
2
+x . (5.65)
8. A particle is in the ground state of harmonic oscillator with potential
energy
V (x) =
m
2
2
x
2
. (5.66)
Find the probability p to nd the particle in the classically forbidden
region.
9. Consider a particle having mass m in a potential V given by
V (x, y, z) =
_
m
2
z
2
2

a
2
x
a
2
and
a
2
y
a
2
else
, (5.67)
where and a are positive real constants. Find the eigen energies of the
system.
10. Consider a harmonic oscillator having angular resonance frequency
0
.
At time t = 0 the systems state is given by
[(t = 0) =
1
2
([0 +[1) , (5.68)
where the states [0 and [1 are the ground and rst excited states, re-
spectively, of the oscillator. Calculate as a function of time t the following
quantities:
a) x
b) p
c)

x
2
_
5.3. Problems
d) xp
11. Harmonic oscillator having angular resonance frequency is in state
[(t = 0) =
1
2
([0 +[n) (5.69)
at time t = 0, where [0 is the ground state and [n is the eigenstate
with eigenenergy (n + 1/2) (n is a non zero integer). Calculate the
expectation value x for time t 0.
12. Consider a harmonic oscillator having mass m and angular resonance
frequency . At time t = 0 the systems state is given by [(0) = c
0
[0+
c
1
[1 , where [n are the eigenstates with energies E
n
= (n + 1/2).
Given that H = , [(0) is normalized, and x (t = 0) =
1
2
_

m
,
calculate x (t) at times t > 0.
13. Show that
D() = e
||
2
2
e
a
a
= e
||
2
2
e
a
e
a
. (5.70)
14. Show that the displacement operator D() is unitary.
15. Show that
[ = e
||
2
2
n=0
n!
[n . (5.71)
16. Show that the coherent state [ is an eigenvector of the operator a with
an eigenvalue , namely
a[ = [ . (5.72)
17. Show that
D() = exp
__
m
2
x
_
exp
_
m
+
2
p
_
exp
_
2
4
_
.
(5.73)
18. Show that for any function f
_
a, a
_
having a power series expansion the
following holds
D
() f
_
a, a
_
D() = f
_
a +, a
_
. (5.74)
19. Show that the following holds for a coherent state [:
a) [ H[ =
_
[[
2
+ 1/2
_
.
b) [ H
2
[ =
2
2
_
[[
4
+ 2 [[
2
+ 1/4
_
.
c)
_
[ (H)
2
[ = [[.
d) x
= [ x[ =
_
2
m
Re ().
e) p
= [ p [ =
2m Im().
f) x
=
_
[ (x)
2
[ =
_

2m
.
g) p
=
_
[ (p)
2
[ =
_
m
2
.
20. Consider a harmonic oscillator of mass m and angular resonance fre-
quency . The Hamiltonian is given by
H =
p
2
2m
+
1
2
m
2
x
2
. (5.75)
The system at time t is in a normalized state [, which is an eigenvector
of the annihilation operator a, thus
a[ = [ , (5.76)
where the eigenvalue is a complex number. At time t > 0 the energy of
the system is measured. What are the possible results E
n
and what are
the corresponding probabilities p
n
(t)?
21. Show that the wave function of a coherent state is given by
(x
) = x
[
= exp
_
2
4
_
_
m
_
1/4
exp
_
_
x x
2x
_
2
+i p
_
.
(5.77)
22. Show that
D() D(
) = exp
_
2
_
D( +
) . (5.78)
23. Show that the following closure relation holds
1 =
1
_ _
[ [ d
2
, (5.79)
where d
2
denotes innitesimal area in the complex plane, namely
d
2
= dRe dIm.
24. Calculate the inner product between two coherent states [ and [,
where , (.
25. A one dimensional potential acting on a particle having mass m is given
by
V
1
(x) =
1
2
m
2
x
2
+m
2
x . (5.80)
5.3. Problems
a) Calculate the Heisenberg representation of the position operator
x
(H)
(t) and its canonically conjugate operator p
(H)
(t).
b) Given that the particle at time t = 0 is in the state [0, where the
state [0 is the ground state of the potential
V
1
(x) =
1
2
m
2
x
2
. (5.81)
Calculate the expectation value x at later times t > 0.
26. A particle having mass m is in the ground state of the one-dimensional
potential well V
1
(x) = (1/2) m
2
(x
x
)
2
for times t < 0 . At time
t = 0 the potential suddenly changes and becomes V
2
(x) = (1/2) m
2
x
2
.
a) Calculate the expectation value x at times t > 0.
b) Calculate the variance
_
(x)
2
_
at times t > 0 , where x = xx.
c) The energy of the particle is measured at time t > 0 . What are the
possible results and what are the probabilities to obtain any of these
results.
27. Consider a particle having mass m in the ground state of the potential
well V
a
(x) = (1/2) m
2
x
2
for times t < 0 . At time t = 0 the potential
suddenly changes and becomes V
b
(x) = gx . (a) Calculate the expecta-
tion value x at times t > 0 . (b) Calculate the variance
_
(x)
2
_
at
times t > 0 , where x = x x.
28. Consider a particle of mass m in a potential of a harmonic oscillator
having angular frequency . The operator S (r) is dened as
S (r) = exp
_
r
2
_
_
a
_
2
a
2
__
, (5.82)
where r is a real number, and a and a
are the annihilation and creation

operators respectively. The operator T is dened as
T = S (r) aS
(r) . (5.83)
a) Find an expression for the operator T of the form T = Aa + Ba
,
where both A and B are constants.
b) The vector state [r is dened as
[r = S
(r) [0 , (5.84)
where [0 is the ground state of the harmonic oscillator. Calculate
the expectation values r[ x[r of the operator x (displacement) and
the expectation value r[ p [r of the operator p (momentum).
c) Calculate the variance (x)
2
of x and the variance (p)
2
of p.
29. The state [r from the previous exercise, which is called a squeezed state,
can be alternatively dened as a normalized state that satises the rela-
tion
Q(r) [r = 0 , (5.85)
where the operator Q(r) is dened by
Q(r) = a coshr +a
sinhr , (5.86)
r is a real number, and a and a
are the annihilation and creation opera-

tors respectively. Based on the above denition calculate the expectation
values r[ x[r of the position operator x, the expectation value r[ p [r
of the momentum operator p, the variance (x)
2
of x and the variance
(p)
2
of p with respect to the state [r.
30. Consider one dimensional motion of a particle having mass m. The Hamil-
tonian is given by
H =
0
a
a +
1
a
aa , (5.87)
where
a =
_
m
0
2
_
x +
ip
m
0
_
, (5.88)
is the annihilation operator, x is the coordinate and p is its canonical
conjugate momentum. The frequencies
0
and
1
are both positive.
a) Calculate the eigenenergies of the system.
b) Let [0 be the ground state of the system. Calculate
i. 0[x[0
ii. 0[p[0
iii.
_
0[ (x)
2
[0
_
iv.
_
0[ (p)
2
[0
_
31. The Hamiltonian of a system is given by
H = N , (5.89)
where the real non-negative parameter has units of energy, and where
the operator N is given by
N = b
b . (5.90)
The following holds
b
b +bb
= 1 , (5.91)
b
2
= 0 , (5.92)
_
b
_
2
= 0 . (5.93)
a) Find the eigenvalues of H. Clue: show rst that N
2
= N.
5.3. Problems
b) Let [0 be the ground state of the system, which is assumed to be
non-degenerate. Dene the two states
[+ = A
+
_
1 +b
_
[0 , (5.94a)
[ = A
_
1 b
_
[0 , (5.94b)
where the real non-negative numbers A
+
and A
are normalization
constants. Calculate A
+
and A
. Clue: show rst that b
[0 is an
eigenvector of N.
c) At time t = 0 the system is in the state
[(t = 0) = [+ , (5.95)
Calculate the probability p (t) to nd the system in the state [ at
time t > 0.
32. Normal ordering - Let f
_
a, a
_
be a function of the annihilation a and
creation a
operators. The normal ordering of f

_
a, a
_
, which is denoted
by : f
_
a, a
_
: places the a operators on the right and the a
operators
on the left. Some examples are given below
: aa
: = a
a , (5.96)
: a
a : = a
a , (5.97)
:
_
a
a
_
n
: =
_
a
_
n
a
n
. (5.98)
Normal ordering is linear, i.e. : f +g :=: f : + : g :. Show that the projec-
tion operator P
n
= [n n[, where [n is an eigenvector of the Hamiltonian
of a harmonic oscillator, can be expressed as
P
n
=
1
n!
:
_
a
_
n
exp
_
a
a
_
a
n
: . (5.99)
33. Consider a harmonic oscillator of angular frequency and mass m. A
time dependent force is applied f (t). The function f (t) is assumed to
vanish f (t) 0 in the limit t . Given that the oscillator was
initially in its ground state [0 at t calculate the probability p
n
to nd the oscillator in the number state [n in the limit t .
34. The operator T is dened by
T =
dx
[x
[ , (5.100)
where [x
is an eigenvector of the position operator x having eigenvalue

x
, i.e. x[x
= x
[x
. Express the operator T as a function of the number

operator N = a
a.
35. Consider a system having Hamiltonian H given by
H = a
a +
1
_
a
a
_
k
, (5.101)
where a and a
are the annihilation and creation operators, both and
1
are positive, and where k is integer. At initial time t = 0 the state
of the system is an eigenstate of the operator a with eigenvalue , i.e.
[ (t = 0) = [
c
, where a [
c
= [
c
.
a) Find a general expression for the state of the system [ (t) at time
t > 0.
b) Evaluate [ (t) at time t = 2/
1
.
c) Evaluate [ (t) at time t = /
1
.
d) Evaluate [ (t) at time t = /2
1
for the case where k is even.
36. Consider two normalized coherent states [ and [, where , (.
The operator A is dened as
A = [ [ [ [ . (5.102)
Find the eigenvalues of the operator A.
5.4 Solutions
H =
p
2
2m
+
m
2
x
2
2
.
Using Eqs. (3.21), (3.29), (5.9) and (5.10) one has
x
[ a [n =
_
2x
2
0
_
1/2
_
x
n
(x
) +x
2
0
d
n
dx
_
, (5.103)
x
[ a
[n =
_
2x
2
0
_
1/2
_
x
n
(x
) x
2
0
d
n
dx
_
, (5.104)
where
x
0
=
_

m
. (5.105)
For the ground state [0, according to Eq. (5.28), a [0 = 0, thus
x
0
(x
) +x
2
0
d
0
dx
= 0 . (5.106)
0
(x
) = A
0
exp
_
1
2
_
x
x
0
_
2
_
, (5.107)
where the normalization constant A
0
is found from the requirement
5.4. Solutions
_

[
0
(x
)[
2
dx = 1 , (5.108)
thus
[A
0
[
2
_

exp
_
_
x
x
0
_
2
_
dx
. .
x0
= 1 . (5.109)
Choosing A
0
to be real leads to
0
(x
) =
1
1/4
x
1/2
0
exp
_
1
2
_
x
x
0
_
2
_
. (5.110)
All other wavefunctions are found using Eqs. (5.32) and (5.104)
n
(x
) =
1
(2x
0
)
n/2
n!
_
x
x
2
0
d
dx
_
n
0
(x
)
=
1
1/4
2
n
n!
1
x
n+1/2
0
_
x
x
2
0
d
dx
_
n
exp
_
1
2
_
x
x
0
_
2
_
.
(5.111)
Using the notation
H
n
(X) = exp
_
X
2
2
__
X
d
dX
_
n
exp
_
X
2
2
_
, (5.112)
the expression for
n
(x
) can be rewritten as
n
(x
) =
exp
_
x
2
2x
2
0
_
H
n
_
x
x
0
_
1/4
x
1/2
0
2
n
n!
. (5.113)
The term H
n
(X), which is called the Hermite polynomial of order n, is
calculated below for some low values of n
H
0
(X) = 1 , (5.114)
H
1
(X) = 2X , (5.115)
H
2
(X) = 4X
2
2 , (5.116)
H
3
(X) = 8X
3
12X , (5.117)
H
4
(X) = 16X
4
48X
2
+ 12 . (5.118)
2. The relation (5.56), which is a Taylor expansion of the function f (t) =
exp
_
2Xt t
2
_
around the point t = 0, implies that
H
n
(X) =
d
n
dt
n
exp
_
2Xt t
2
_
t=0
. (5.119)
The identity 2Xt t
2
= X
2
(X t)
2
yields
H
n
(X) = exp
_
X
2
_
d
n
dt
n
exp
_
(X t)
2
_
t=0
. (5.120)
Moreover, using the relation
d
dt
exp
_
(X t)
2
_
=
d
dX
exp
_
(X t)
2
_
, (5.121)
one nds that
H
n
(X) = exp
_
X
2
_
(1)
n
d
n
dX
n
exp
_
(X t)
2
_
t=0
= exp
_
X
2
_
(1)
n
d
n
dX
n
exp
_
X
2
_
.
(5.122)
Note that for an arbitrary function g (X) the following holds
exp
_
X
2
_
d
dX
exp
_
X
2
_
g =
_
2X
d
dX
_
g , (5.123)
and
exp
_
X
2
2
__
X
d
dX
_
exp
_
X
2
2
_
g =
_
2X
d
dX
_
g , (5.124)
thus
H
n
(X) = exp
_
X
2
2
__
X
d
dX
_
n
exp
_
X
2
2
_
. (5.125)
3. With the help of Eqs. (5.9), (5.10), (5.11), (5.12) and (5.13) one nds
n[ x[n = 0 , (5.126)
n[ x
2
[n =

2m
n[ aa
+a
a [n =

2m
(2n + 1) , (5.127)
n[ p [n = 0 , (5.128)
n[ p
2
[n =
m
2
n[ aa
+a
a [n =
m
2
(2n + 1) , (5.129)
thus
_
(x)
2
__
(p)
2
_
=
_
n +
1
2
_
2
2
.
H =
p
2
2m
. (5.130)
5.4. Solutions
Using Eqs. (4.37) and (5.8) one nds that
dx
(H)
dt
=
1
i
_
x
(H)
, H
(H)
_
=
p
(H)
im
_
x
(H)
, p
(H)
_
=
p
(H)
m
, (5.131)
and
dp
(H)
dt
=
1
i
_
p
(H)
, H
(H)
_
= 0 . (5.132)
The solution is thus
x
(H)
(t) = x
(H)
(0) +
1
m
p
(H)
(0) t . (5.133)
With the help of Eq. (5.133) one nds that
_
(x)
2
_
(t) =

x
2
_
(t) (x (t))
2
=
0
[
_
x
(H)
(0) +
1
m
p
(H)
(0) t
_
2
[
0
0
[
_
x
(H)
(0) +
1
m
p
(H)
(0) t
_
[
0
_
2
= (x)
2
0
+
t
2
m
2
(p)
2
0
+
2t
m
((xp)
0
x
0
p
0
) .
(5.134)
H =
p
2
2m
+
m
2
x
2
2
. (5.135)
dx
(H)
dt
=
1
i
_
x
(H)
, H
(H)
_
=
p
(H)
m
, (5.136)
and
dp
(H)
dt
=
1
i
_
p
(H)
, H
(H)
_
= m
2
x
(H)
. (5.137)
a) The solutions of the above equations are given by
x
(H)
(t) = x
(H)
(0) cos (t) +
sin(t)
m
p
(H)
(0) , (5.138)
and
p
(H)
(t) = p
(H)
(0) cos (t) m sin(t) x
(H)
(0) . (5.139)
b) Using the expressions for x
(H)
(t) and p
(H)
(t) and Eq. (5.8) one nds
that _
p
(H)
(t
1
) , x
(H)
(t
2
)
_
= (cos (t
1
) cos (t
2
) + sin(t
1
) sin(t
2
))
_
x
(H)
(0) , p
(H)
(0)
_
= icos ( (t
1
t
2
)) ,
(5.140)
_
p
(H)
(t
1
) , p
(H)
(t
2
)
_
= m (cos (t
1
) sin(t
2
) sin(t
1
) cos (t
2
))
_
x
(H)
(0) , p
(H)
(0)
_
= im sin( (t
1
t
2
)) ,
(5.141)
and _
x
(H)
(t
1
) , x
(H)
(t
2
)
_
=
1
m
(cos (t
1
) sin(t
2
) sin(t
1
) cos (t
2
))
_
x
(H)
(0) , p
(H)
(0)
_
=
i
m
sin( (t
1
t
2
)) .
(5.142)
6. Due to the innite barrier for x 0 the wavefunction must vanish at
x = 0. This condition is satised by the wavefunction of all number
states [n with odd value of n (the states [n are eigenstates of the regu-
lar harmonic oscillator with potential V (x) =
_
m
2
/2
_
x
2
). These wave-
functions obviously satisfy the Schrdinger equation for x > 0.
a) Thus the possible energy values are E
k
= (2k + 3/2) where k =
0, 1, 2, .
b) The corresponding normalized wavefunctions are given by
k
(x) =
_
2
2k+1
(x) x > 0
0 x 0
, (5.143)
where
n
(x) is the wavefunction of the number states [n. Thus for
a given k
5.4. Solutions
x
2
_
k
=
_
0
dx

k
(x)
2
x
2
= 2
_
0
dx

2k+1
(x)
2
x
2
=
dx

2k+1
(x)
2
x
2
= 2k + 1[ x
2
[2k + 1 ,
(5.144)
thus with the help of Eq. (5.127) one nds that
x
2
_
k
=

m
_
2k +
3
2
_
. (5.145)
7. The potential can be written as
V (x) =
m
2
2
_
x +

m
2
_
2

2
2m
2
. (5.146)
This describes a harmonic oscillator centered at x
0
= /m
2
having
angular resonance frequency . The last constant term represents energy
shift. Thus, the eigenenergies are given by
E
n
= (n + 1/2)
2
/2m
2
, (5.147)
where n = 0, 1, 2, .
8. In the classically forbidden region V (x) > E
0
= /2, namely [x[ > x
0
where
x
0
=
_

m
. (5.148)
Using Eq. (5.110) one nds
p = 2
_

x0
[
0
(x)[
2
dx
=
2
1/2
x
0
_

x
0
exp
_
_
x
x
0
_
2
_
dx
= 1 erf (1)
= 0.157 .
(5.149)
9. The answer is [see Eqs. (4.162) and (5.19)]
E
nx,ny,nz
=

2
2
_
n
2
x
+n
2
y
_
2ma
2
+
_
n
z
+
1
2
_
, (5.150)
where n
x
and n
y
are positive integers and n
z
is a nonnegative integer.
10. With the help of Eq. (4.14) one has
[(t) =
1
2
e
i
0
t
2
_
[0 +e
i0t
[1
_
. (5.151)
Moreover, the following hold
x =
_

2m
0
_
a +a
_
, (5.152)
p = i
_
m
0
2
_
a +a
_
, (5.153)
a [n =
n[n 1 , (5.154)
a
[n =
n + 1 [n + 1 , (5.155)
_
a, a
= 1 , (5.156)
thus
a)
x =
_

2m
0
(t)[
_
a +a
_
[(t)
=
_

2m
0
1
2
_
0[ +e
i
0
t
1[
_ _
a +a
_ _
[0 +e
i
0
t
[1
_
=
_

2m
0
1
2
_
e
i
0
t
+e
i
0
t
_
=
_

2m
0
cos (
0
t) .
(5.157)
b)
p = i
_
m
0
2
(t)[
_
a +a
_
[(t)
= i
_
m
0
2
1
2
_
0[ +e
i0t
1[
_ _
a +a
_ _
[0 +e
i0t
[1
_
=
_
m
0
2
sin(
0
t) .
(5.158)
c)
5.4. Solutions
x
2
_
=

2m
0
(t)[
_
a +a
_
2
[(t)
=

2m
0
(t)[
_
a
2
+
_
a
_
2
+
_
a, a
+ 2a
a
_
[(t)
=

2m
0
_
1 + 2
1
2
_
=

m
0
.
(5.159)
d) Similarly
p
2
_
=
m
0
2
(t)[
_
a +a
_
2
[(t)
=
m
0
2
(t)[
_
a
2
+
_
a
_
2
_
a, a
2a
a
_
[(t)
= m
0
,
(5.160)
thus
xp =
_
1
cos
2
(
0
t)
2
_
1
sin
2
(
0
t)
2
=

2
_
2 +
1
4
sin
2
(2
0
t) .
(5.161)
11. The state [ (t) is given by
[(t) =
1
2
_
exp
_
iE
0
t
_
[0 + exp
_
iE
n
t
_
[n
_
, (5.162)
where
E
n
=
_
n +
1
2
_
, (5.163)
thus, using
x =
_

2m
_
a +a
_
, (5.164)
and
a [n =
n[n 1 , (5.165)
a
[n =
n + 1 [n + 1 , (5.166)
one nds that x (t) = 0 if n > 1, and for n = 1
x (t) =
_

2m
(t)[
_
a +a
_
[ (t)
=
_

2m
cos (t) .
(5.167)
12. Since H = and [(0) is normalized one has
[c
0
[
2
= [c
1
[
2
=
1
2
, (5.168)
thus [(0) can be written as
[(0) =
_
1
2
_
[0 +e
i
[1
_
, (5.169)
where is real. Given that at time t = 0
x (t = 0) =
1
2
_

m
, (5.170)
one nds using the identities
x =
_

2m
_
a +a
_
, (5.171)
a [n =
n[n 1 , (5.172)
a
[n =
n + 1 [n + 1 , (5.173)
that
cos =
2
2
. (5.174)
Using this result one can evaluate p (t = 0), where
p = i
_
m
2
_
a +a
_
, (5.175)
thus
p (t = 0) =
_
m
2
sin =
_
m
2
2
2
= mx (t = 0) . (5.176)
Using these results together with Eq. (5.138) yields
x (t) =
1
2
_

m
(cos (t) sin(t))
=
_

2m
cos
_
t

4
_
.
(5.177)
5.4. Solutions
13. According to identity (2.175), which states that
e
A+B
= e
A
e
B
e
1
2
[A,B]
= e
B
e
A
e
1
2
[A,B]
, (5.178)
provided that
[A, [A, B]] = [B, [A, B]] = 0 , (5.179)
one nds with the help of Eq. (5.13) that
D() = exp
_
a
a
_
= e
||
2
2
e
a
a
= e
||
2
2
e
a
e
a
.
(5.180)
D
() = e
||
2
2
e
a
a
= e
||
2
2
e
a
e
a
, (5.181)
thus
D
() D() = D() D
() = 1 . (5.182)
15. Using Eqs. (5.35), (5.28) and (5.29) one nds that
[ = e
||
2
2
e
a
a
[0 = e
||
2
2
e
a
[0
= e
||
2
2
n=0
n!
[n .
(5.183)
16. Using Eqs. (5.42) and (5.28) one has
a[ = e
||
2
2
n=0
n!
a [n
= e
||
2
2
n=1
n1
_
(n 1)!
[n 1
= [ .
(5.184)
17. Using Eqs. (5.36), (5.9) and (5.10) one has
D() = exp
_
_
m
2
(
) x i
_
1
2m
( +
) p
_
, (5.185)
thus with the help of Eqs. (2.175) and (5.8) the desired result is obtained
D() = exp
__
m
2
x
_
exp
_
m
+
2
p
_
exp
_
2
4
_
.
(5.186)
18. Using the operator identity (2.173)
e
L
Ae
L
= A+[L, A] +
1
2!
[L, [L, A]] +
1
3!
[L, [L, [L, A]]] + , (5.187)
and the denition (5.36)
D() = exp
_
a
a
_
, (5.188)
one nds that
D
() aD() = a + , (5.189)
D
() a
D() = a
. (5.190)
Exploiting the unitarity of D()
D() D
() = 1
it is straightforward to show that for any function f
_
a, a
_
having a
power series expansion the following holds
D
() f
_
a, a
_
D() = f
_
a +, a
_
(5.191)
(e.g., D
a
2
D = D
aDD
aD = (a +)
2
).
19. Using Eq. (5.74) and the following identities
H =
_
a
a +
1
2
_
, (5.192)
x =
_

2m
_
a +a
_
, (5.193)
p = i
_
m
2
_
a +a
_
, (5.194)
all these relations are easily obtained.
20. Expressing the state [ in the basis of eigenvectors of the Hamiltonian
[n
[ =

n=0
c
n
[n , (5.195)
using
5.4. Solutions
a[ = [ , (5.196)
and
a[n =
n[n 1 , (5.197)
one nds
n=0
c
n
n[n 1 =

n=0
c
n
[n , (5.198)
thus
c
n+1
=

n + 1
c
n
, (5.199)
therefore
[ = A

n=0
n!
[n . (5.200)
The normalization constant A is found by
1 = [A[
2

n=0
_
[[
2
_
n
n!
= [A[
2
e
||
2
. (5.201)
Choosing A to be real yields
A = e
||
2
2
, (5.202)
thus
c
n
= e
||
2
2

n
n!
. (5.203)
Note that this result is identical to Eq. (5.42), thus [ is a coherent
state. The possible results of the measurement are
E
n
=
_
n +
1
2
_
, (5.204)
and the corresponding probabilities, which are time independent, are
given by
p
n
(t) = [c
n
[
2
= e
||
2
_
[[
2
_
n
n!
. (5.205)
21. Using the relations
x
=
_
2
m
Re () , (5.206)
p
2m Im() , (5.207)
Eq. (5.73) can be written as
D() = exp
_
i p
_
exp
_
i x
_
exp
_
2
4
_
. (5.208)
exp
_
i x
_
[x
= [x
+x
,
thus
x
[ = x
[ exp
_
i p
_
exp
_
i x
_
exp
_
2
4
_
[0
= exp
_
2
4
_
exp
_
i p
_
x
[0 .
(5.209)
Using Eq. (5.110) the wavefunction of the ground state is given by
x
[0 =
1
(2)
1/4
1
exp
_
_
x
2x
_
2
_
, (5.210)
where
x
=
_

2m
, (5.211)
thus
x
[ = exp
_
2
4
_
exp
_
i p
_ exp
_
_
x
2x
_
2
_
(2)
1/4
= exp
_
2
4
_
_
m
_
1/4
exp
_
_
x x
2x
_
2
+i p
_
.
(5.212)
22. Using Eqs. (5.36) and (2.175) this relation is easily obtained.
23. With the help of Eq. (5.42) one has
1
_ _
[ [ d
2
=
1
n,m
[n m[
1
n!m!
_ _
e
||
2
m
d
2
.
5.4. Solutions
(5.213)
Employing polar coordinates in the complex plane = e
i
, where is
non-negative real and is real, leads to
1
_ _
[ [ d
2
=
1
n,m
[n m[
1
n!m!
_
0
d
n+m+1
e
2
2
_
0
de
i(nm)
. .
2
nm
=

n
[n n[
2
n!
_
0
d
2n+1
e
2
=

n
[n n[
1
n!
(n + 1)
. .
=n!
=

n
[n n[
= 1 .
(5.214)
24. Using Eqs. (5.35) and (5.41) one nds that
[ = 0[ D
() D() [0
= 0[ D() D() [0
= exp
_
2
_
0[ D( +) [0
= exp
_
2
_
0 [ .
(5.215)
Thus, with the help of Eq. (5.42) one has
[ = exp
_
2
_
e
||
2
2
= exp
_
[[
2
2

[[
2
2
+
_
= exp
_
[ [
2
2
+i Im(
)
_
.
(5.216)
V
1
(x) =
1
2
m
2
(x +)
2
1
2
m
2
2
=
1
2
m
2
x
2
1
2
m
2
2
,
(5.217)
where
x
= x + . (5.218)
a) Thus, using Eqs. (5.138) and (5.139) together with the relations
x
(H)
(t) = x
(H)
(t) + , (5.219)
p
(H)
(t) = p
(H)
(t) , (5.220)
one nds
x
(H)
(t) =
_
x
(H)
(0) +
_
cos (t) +
sin(t)
m
p
(H)
(0) , (5.221)
p
(H)
(t) = p
(H)
(0) cos (t) m sin(t)
_
x
(H)
(0) +
_
. (5.222)
b) For this case at time t = 0
_
x
(H)
(0)
_
= 0 , (5.223)
_
p
(H)
(0)
_
= 0 , (5.224)
thus
_
x
(H)
(t)
_
= (cos (t) 1) . (5.225)
26. The state of the system at time t = 0 is given by
[(t = 0) = exp
_
i
x
p
_
[0 , (5.226)
where [0 is the ground state of the potential V
2
. In general a coherent
state with parameter can be written as
[ = exp
__
m
2
x
_
exp
_
m
+
2
p
_
exp
_
2
4
_
[0 .
(5.227)
a) Thus [ (t = 0) = [
0
, where
0
=
x
_
m
2
. (5.228)
The time evolution of a coherent state is given by
[ (t) = e
it/2
=
0
e
it
_
, (5.229)
5.4. Solutions
x (t) =
_
2
m
Re
_
0
e
it
=
x
cos (t) , (5.230)
b) According to Eq. (5.48)
_
(x)
2
_
(t) =

2m
. (5.231)
c) In general a coherent state can be expanded in the basis of number
states [n
[ = e
||
2
/2
n!
[n , (5.232)
thus the probability to measure energy E
n
= (N + 1/2) at time t
is given by
P
n
= [n[ (t)[
2
=
e
[
2
0
[
2n
0
n!
=
1
n!
exp
_
m
2
x
2
__
m
2
x
2
_
n
.
(5.233)
27. At time t = 0 the following holds
x = 0 , (5.234)
p = 0 , (5.235)
_
(x)
2
_
=

x
2
_
=

2m
, (5.236)
_
(p)
2
_
=

p
2
_
=
m
2
. (5.237)
Moreover, to calculate xp it is convenient to use
x =
_

2m
_
a +a
_
, (5.238)
p = i
_
m
2
_
a +a
_
, (5.239)
_
a, a
= 1 , (5.240)
thus at time t = 0
xp = i
2
0[ aa
a [0 = i
2
. (5.241)
The Hamiltonian for times t > 0 is given by
H =
p
2
2m
+gx . (5.242)
Using the Heisenberg equation of motion for the operators x and x
2
one
nds
dx
(H)
dt
=
1
i
_
x
(H)
, H
, (5.243)
dp
(H)
dt
=
1
i
_
p
(H)
, H
, (5.244)
dx
2
(H)
dt
=
1
i
_
x
2
(H)
, H
_
, (5.245)
or using [x, p] = i
dx
(H)
dt
=
p
(H)
m
, (5.246)
dp
(H)
dt
= g , (5.247)
dx
2
(H)
dt
=
1
m
_
x
(H)
p
(H)
+p
(H)
x
(H)
_
=
1
m
_
2x
(H)
p
(H)
i
_
, (5.248)
thus
p
(H)
(t) = p
(H)
(0) gt , (5.249)
x
(H)
(t) = x
(H)
(0) +
p
(H)
(0) t
m

gt
2
2m
, (5.250)
x
2
(H)
(t) = x
2
(H)
(0)
it
m
+
2
m
_
t
0
x
(H)
(t
) p
(H)
(t
) dt
= x
2
(H)
(0)
it
m
+
2
m
_
t
0
_
x
(H)
(0) +
p
(H)
(0) t
m

gt
2
2m
_
_
p
(H)
(0) gt
dt
= x
2
(H)
(0)
it
m
+
2
m
_
t
0
_
x
(H)
(0) p
(H)
(0) +
p
2
(H)
(0) t
m

gt
2
2m
p
(H)
(0) x
(H)
(0) gt
p
(H)
(0) gt
2
m
+
g
2
t
3
2m
_
dt
= x
2
(H)
(0)
it
m
+
2
m
_
x
(H)
(0) p
(H)
(0) t +
p
2
(H)
(0) t
2
2m

p
(H)
(0) gt
3
6m

x
(H)
(0) gt
2
2

p
(H)
(0) gt
3
3m
+
g
2
t
4
8m
_
.
(5.251)
Using the initial conditions Eqs. (5.234), (5.235), (5.236), (5.237) and
(5.241) one nds
x(t) =
gt
2
2m
, (5.252)
5.4. Solutions
x(t)
2
=
g
2
t
4
4m
2
, (5.253)
p (t) = gt , (5.254)
x
2
(t)
_
=

2m

it
m
+
2
m
_
it
2
+
t
2
4
+
g
2
t
4
8m
_
, (5.255)
and
_
(x)
2
(t)
_
=
x
2
(t)
_
x(t)
2
=

2m
+
t
2
2m
=

2m
_
1 +
2
t
2
_
.
(5.256)
28. Using the operator identity (2.173), which is given by
e
L
Oe
L
= O+[L, O] +
1
2!
[L, [L, O]] +
1
3!
[L, [L, [L, O]]] + , (5.257)
for the operators
O = a , (5.258)
L =
r
2
_
_
a
_
2
a
2
_
, (5.259)
and the relations
_
a, a
= 1 , (5.260)
[L, O] = ra
, (5.261)
[L, [L, O]] = r
2
a , (5.262)
[L, [L, [L, O]]] = r
3
a
, (5.263)
[L, [L, [L, [L, O]]]] = r
4
a , (5.264)
etc., one nds
T =
_
1 +
r
2
2!
+
r
4
4!
+
_
a
_
r +
r
3
3!
+
_
a
+ , (5.265)
a) Thus
T = Aa +Ba
, (5.266)
where
A = coshr , (5.267)
B = sinhr . (5.268)
b) Using the relations
x =
_

2m
_
a +a
_
, (5.269)
p = i
_
m
2
_
a +a
_
. (5.270)
one nds
r[ x[r =
_

2m
0[ S (r)
_
a +a
_
S
(r) [0
=
_

2m
_
0[ T [0 +0[ T
[0
_
= 0 ,
(5.271)
r[ p [r = i
_
m
2
0[ S (r)
_
a +a
_
S
(r) [0
=
_

2m
_
0[ T [0 +0[ T
[0
_
= 0 .
(5.272)
c) Note that S (r) is unitary, namely S
(r) S (r) = 1, since the operator

a
2
_
a
_
2
is anti Hermitian. Thus
r[ x
2
[r =

2m
0[ S (r)
_
a +a
_ _
a +a
_
S
(r) [0
=

2m
0[ S (r)
_
a +a
_
S
(r) S (r)
_
a +a
_
S
(r) [0
=

2m
0[
_
T +T
_
2
[0
=
(A+B)
2
2m
0[
_
a +a
_
2
[0
=
(coshr sinhr)
2
2m
=
e
2r
2m
,
(5.273)
and
r[ p
2
[r =
m
2
0[ S (r)
_
a a
_
2
S
(r) [0
=
m
2
0[
_
T T
_
2
[0
=
m (AB)
2
2
0[
_
a a
_
2
[0
=
m (coshr + sinhr)
2
2
=
me
2r
2
.
(5.274)
Thus
5.4. Solutions
(x)
2
=
e
2r
2m
, (5.275)
(p)
2
=
me
2r
2
, (5.276)
(x) (p) =

2
. (5.277)
29. With the help of Eqs. (5.9) and (5.10) one nds that
x = e
r
_

2m
_
Q(r) +Q
(r)
_
, (5.278)
p = ie
r
_
m
2
_
Q(r) Q
(r)
_
, (5.279)
r[ x[r = 0 , (5.280)
r[ p [r = 0 . (5.281)
Using the commutation relation
_
Q(r) , Q
(r)
=
_
cosh
2
r sinh
2
r
_ _
a, a
= 1 ,
one obtains
r[ x
2
[r =
e
2r
2m
r[
_
Q(r) +Q
(r)
_
2
[r
=
e
2r
2m
r[ Q(r) Q
(r) [r
=
e
2r
2m
,
(5.282)
and similarly
r[ p
2
[r =
me
2r
2
r[
_
Q(r) Q
(r)
_
2
[r
=
me
2r
2
r[ Q(r) Q
(r) [r
=
me
2r
2
,
(5.283)
thus
(x)
2
=
e
2r
2m
, (5.284)
(p)
2
=
me
2r
2
. (5.285)
30. Using the commutation relation
_
a, a
= 1 , (5.286)
one nds
H =
0
N +
1
_
N
2
N
_
, (5.287)
where
N = a
a (5.288)
is the number operator.
a) The eigenvectors of N
N [n = n[n , (5.289)
(where n = 0, 1, ) are also eigenvectors of H and the following
holds
H[n = E
n
[n , (5.290)
where
E
n
=
_
0
n +
1
_
n
2
n
_
. (5.291)
Note that
E
n+1
E
n
=
0
+ 2
1
n , (5.292)
thus E
n+1
> E
n
.
b) Using the relations
x =
_

2m
0
_
a
+a
_
, (5.293)
p = i
_
m
0
2
_
a
a
_
, (5.294)
x
2
=

2m
0
_
a
+aa + 2N + 1
_
, (5.295)
p
2
=
m
0
2
_
a
aa + 2N + 1
_
, (5.296)
a[n =
n[n 1 , (5.297)
a
[n =
n + 1 [n + 1 , (5.298)
one nds
i. 0[x[0 = 0
ii. 0[p[0 = 0
5.4. Solutions
iii.
_
0[ (x)
2
[0
_
=

2m
0
iv.
_
0[ (p)
2
[0
_
=
m
0
2
31. The proof of the clue is:
N
2
= b
bb
b = b
_
1 b
b
_
b = N . (5.299)
Moreover, N is Hermitian, thus N is a projector.
a) Let [n be the eigenvectors of N and n the corresponding real eigen-
values (N is Hermitian)
N [n = n[n . (5.300)
Using the clue one nds that n
2
= n, thus the possible values of n
are 0 (ground state) and 1 (excited state). Thus, the eigenvalues of
H are 0 and .
b) To verify the statement in the clue we calculate
Nb
[0 = b
bb
[0 = b
(1 N) [0 = b
[0 , (5.301)
thus the state b
[0 is indeed an eigenvector of N with eigenvalue 1

(excited state). In what follows we use the notation
[1 = b
[0 . (5.302)
Note that [1 is normalized since
1[1 = 0[ bb
[0 = 0[ (1 N) [0 = 0[0 = 1 . (5.303)
Moreover, since [0 and [1 are eigenvectors of an Hermitian operator
with dierent eigenvalues they must be orthogonal to each other
0[1 = 0 . (5.304)
Using Eqs. (5.302), (5.303) and (5.304) one nds
+[+ = 2 [A
+
[
2
, (5.305)
[ = 2 [A
[
2
. (5.306)
choosing the normalization constants to be non-negative real num-
bers lead to
A
+
= A
=
1
2
. (5.307)
c) Using N
2
= N one nds
exp
_
iHt
_
= 1 +

n=1
1
n!
_
iHt
_
n
= 1 +N

n=1
1
n!
_
it
_
n
= 1 +N
_
1 +

n=0
1
n!
_
it
_
n
_
= 1 +N
_
1 + exp
_
it
__
.
(5.308)
Thus
p
0
(t) =
[ exp
_
iHt
_
[+
2
=
1
4
(0[ 1[)
_
1 +N
_
1 + exp
_
it
___
([0 +[1)
2
=
1
4
1 exp
_
it
2
= sin
2
_
t
2
_
.
(5.309)
32. The closure relation (5.31) can be written as
1 =

n,m=0
[n m[
n,m
. (5.310)
With the help of Eq. (5.32) together with the relation
1
n!
_
d
d
_
n
=0
=
n,m
, (5.311)
which is obtained using the Taylor power expansion series of the function
m
, one nds that
5.4. Solutions
1 =

n,m=0
[n m[
n,m
=

n,m=0
1
n!
m!
[n m[
_
d
d
_
n
=0
=

n,m=0
_
a
_
n
n!
[0 0[
a
m
m!
_
d
d
_
n
=0
=
_

n=0
_
a
_
n
_
d
d
_
n
n!
_
[0 0[
_

m=0
a
m
m
m!
_
=0
= exp
_
a
d
d
_
[0 0[ exp(a)
=0
.
(5.312)
Denote the normal ordering representation of the operator [0 0[ by Z,
i.e.
[0 0[ =: Z: . (5.313)
For general functions f, g and h of the operators a and a
it is easy to
show that the following holds
: fg : =: gf : , (5.314)
: fgh: =: fhg : , (5.315)
and
: f (: g : ) : =: fg : . (5.316)
Thus
1 = exp
_
a
d
d
_
: Z: exp(a)
=0
= : exp
_
a
d
d
_
Z exp(a)
=0
:
= : exp
_
a
d
d
_
exp(a) Z
=0
:
= :

n,m
_
a
d
d
_
n
n!
(a)
m
m!
Z
=0
:
= :

n,m
_
a
_
n
_
d
d
_
n
n!

m
=0
. .
n,m
a
m
m!
Z:
= : exp
_
a
a
_
Z:
= : exp
_
a
a
_
(: Z: ) : ,
(5.317)
and therefore
[0 0[ =: exp
_
a
a
_
: . (5.318)
Using again Eq. (5.32) one nds that
P
n
= [n n[ =
1
n!
:
_
a
_
n
exp
_
a
a
_ _
a
_
n
: . (5.319)
33. The Hamiltonian H, which is given by
H =
p
2
2m
+
m
2
x
2
2
+xf (t) , (5.320)
can be expressed in terms of the annihilation a and creation a
operators
[see Eqs. (5.11) and (5.12)] as
H =
_
a
a +
1
2
_
+f (t)
_

2m
_
a +a
_
. (5.321)
The Heisenberg equation of motion for the operator a is given by [see
Eq. (4.37)]
da
dt
= ia i
_
1
2m
f (t) . (5.322)
The solution of this rst order dierential equation is given by
a(t) = e
i(tt0)
a (t
0
) i
_
1
2m
_
t
t0
dt
e
i(tt
)
f (t
) , (5.323)
5.4. Solutions
where the initial time t
0
will be taken below to be . The Heisenberg
operator a
(t) is found from the Hermitian conjugate of the last result.

Let P
n
(t) be the Heisenberg representation of the projector [n n[. The
probability p
n
(t) to nd the oscillator in the number state [n at time t
is given by
p
n
(t) = 0[ P
n
(t) [0 . (5.324)
To evaluate p
n
(t) it is convenient to employ the normal ordering repre-
sentation of the operator P
n
(5.99). In normal ordering the rst term of
Eq. (5.323), which is proportional to a (t
0
) does not contribute to p
n
(t)
since a (t
0
) [0 = 0 and also 0[ a
(t
0
) = 0. To evaluate p
n
= p
n
(t )
the integral in the second term of Eq. (5.323) is evaluate from t
0
=
to t = +. Thus one nds that
p
n
=
e
n
n!
, (5.325)
where
=
1
2m
dt
e
it
f (t
2
. (5.326)
34. As can be seen from the denition of T, the following holds
x
[ T[ =
dx
[x
[
= x
[ ,
(5.327)
thus the wave function of T[ is (x
) given that the wave function

of [ is (x
). For the wavefunctions

n
(x
) = x
[n of the number
states [n, which satisfy N [n = n[n, the following holds
n
(x
) =
_
n
(x
) n odd
n
(x
) n even
, (5.328)
thus
T[n =
_
[n n odd
[n n even
, (5.329)
or T[n = (1)
n
[n ,thus, the operator T can be expressed as a function
of N
T = e
iN
. (5.330)
35. Initially, the system is in a coherent state given by Eq. (5.42)
[(t = 0) = [
c
= e
||
2
2
n=0
n!
[n . (5.331)
The notation [
c
is used to label coherent states satisfying a [
c
=
[
c
.
a) Since a
a commutes with
_
a
a
_
k
, the time evolution operator is given
by [see Eq. (4.9)]
u(t) = exp
_
iHt
_
= e
i
1(a
a)
k
t
e
ia
at
, (5.332)
thus
[ (t) = u(t) [ (t = 0)
= e
i1(a
a)
k
t
e
ia
at
e
||
2
2
n=0
n!
[n
= e
i
1(a
a)
k
t
e
||
2
2
n=0
_
e
it
_
n
n!
[n
= e
||
2
2
n=0
_
e
it
_
n
n!
e
i
n
[n ,
(5.333)
where
n
=
1
tn
k
. (5.334)
b) At time t = 2/
1
the phase factor
n
is given by
n
= 2n
k
, thus
e
i
n
= 1 , (5.335)
and therefore
_
2
1
__
=

2i
1
_
c
. (5.336)
c) At time t = /
1
the phase factor
n
is given by
n
= n
k
. Using
the fact that
mod
_
n
k
, 2
_
=
_
0 n is even
1 n is odd
, (5.337)
one has
e
i
n
= (1)
n
, (5.338)
5.4. Solutions
and therefore
1
__
= e
||
2
2
n=0
_
e
1
_
n
n!
(1)
n
[n
=

1
_
c
.
(5.339)
d) At time t = /2
1
the phase factor
n
is given by
n
= (/2) n
k
.
For the case where k is even one has
mod
_
n
k
, 4
_
=
_
0 n is even
1 n is odd
, (5.340)
thus
e
i
n
=
_
1 n is even
i n is odd
, (5.341)
and therefore
_

2
1
__
= e
||
2
2
n=0
_
e
i
2
1
_
n
n!
e
i
n
[n . (5.342)
This state can be expressed as a superposition of two coherent states
_

2
1
__
=
1
2
_
e
i
4
i
2
1
_
c
+e
i
4
i
2
1
_
c
_
. (5.343)
36. Let
n
be the set of eigenvalues of A. Clearly A is Hermitian, namely
A
= A, thus the eigenvalues

n
are expected to be real. Since the trace
of an operator is basis independent, the following must hold
Tr (A) =
n
, (5.344)
and
Tr
_
A
2
_
=
2
n
. (5.345)
On the other hand, with the help of Eq. (2.172) one nds that
Tr (A) = Tr ([ [) Tr ([ [) = 0 , (5.346)
and
Tr
_
A
2
_
= Tr ([ [ [) + Tr ([ [ [)
Tr ([ [ [) Tr ([ [ [)
= 2 [ Tr ([ [) [ Tr ([ [)
= 2
_
1 [ [[
2
_
.
(5.347)
Clearly, A cannot have more than two nonzero eigenvalues, since the
dimensionality of the subspace spanned by the vectors [ , [ is at
most 2, and therefore A has three eigenvalues 0,
+
and
, where [see
Eq. (5.216)]
=
_
1 [ [[
2
=
_
1 e
||
2
. (5.348)
6. Angular Momentum
Consider a point particle moving in three dimensional space. The orbital
angular momentum L is given by
L = r p = det
_
_
x y z
x y z
p
x
p
y
p
z
_
_
,
where r = (x, y, z) is the position vector and where p = (p
x
, p
y
, p
z
) is the
momentum vector. In classical physics the following holds:
Claim.
L
i
, L
j
=
ijk
L
k
, (6.1)
where
ijk
=
_
_
_
0 i, j, k are not all dierent
1 i, j, k is an even permutation of x, y, z
1 i, j, k is an odd permutation of x, y, z
. (6.2)
Proof. Clearly, Eq. (6.1) holds for the case i = j. Using Eq. (1.48), which
reads
x
i
, p
j
=
ij
, (6.3)
one has
L
x
, L
y
= yp
z
zp
y
, zp
x
xp
z
= yp
z
, zp
x
+zp
y
, xp
z
= y p
z
, z p
x
+xz, p
z
p
y
= yp
x
+xp
y
= L
z
.
(6.4)
In a similar way one nds that L
y
, L
z
= L
x
and L
z
, L
x
= L
y
. These
results together with Eq. (1.49) complete the proof.
Using the rule (4.41) , (1/i) [, ] one concludes that in quantum
mechanics the following holds:
[L
i
, L
j
] = i
ijk
L
k
. (6.5)
Chapter 6. Angular Momentum
6.1 Angular Momentum and Rotation
We have seen before that the unitary operator u(t, t
0
) is the generator of time
evolution [see Eq. (4.4)]. Similarly, we have seen that the unitary operator
J () = exp
_
i p
_
(6.6)
[see Eq. (3.73)] is the generator of linear translations:
J () [r
= [r
+ . (6.7)
Below we will see that one can dene a unitary operator that generates ro-
tations.
Exercise 6.1.1. Show that
D
z
()
_
_
x
y
z
_
_
D
z
() = R
z
_
_
x
y
z
_
_
, (6.8)
where
D
z
() = exp
_
iL
z
_
, (6.9)
and where
R
z
=
_
_
cos sin 0
sin cos 0
0 0 1
_
_
. (6.10)
Solution 6.1.1. Equation (6.8) is made of 3 identities:
D
z
() xD
z
() = xcos y sin , (6.11)
D
z
() yD
z
() = xsin +y cos , (6.12)
D
z
() zD
z
() = z . (6.13)
As an example, we prove below the rst one. Using the identity (2.173), which
is given by
e
L
Ae
L
= A+ [L, A] +
1
2!
[L, [L, A]] +
1
3!
[L, [L, [L, A]]] + , (6.14)
one has
D
z
() xD
z
()
= x +
i
[L
z
, x] +
1
2!
_
i
_
2
[L
z
, [L
z
, x]] +
1
3!
_
i
_
3
[L
z
, [L
z
, [L
z
, x]]] + .
(6.15)
6.1. Angular Momentum and Rotation
Furthermore with the help of
L
z
= xp
y
yp
x
, (6.16)
[x
i
, p
j
] = i
ij
, (6.17)
one nds that
[L
z
, x] = y [p
x
, x] = iy ,
[L
z
, [L
z
, x]] = ix[p
y
, y] = (i)
2
x ,
[L
z
, [L
z
, [L
z
, x]]] = (i)
2
[L
z
, x] = (i)
3
y ,
[L
z
, [L
z
, [L
z
, [L
z
, x]]]] = (i)
4
x ,
.
.
. (6.18)
thus
D
z
() xD
z
() = x
_
1

2
2!
+

4
4!
+
_
y
_

3
3!
+
_
= xcos y sin .
(6.19)
The other 2 identities in Eq. (6.8) can be proven in a similar way.
The matrix R
z
[see Eq. (6.10)] represents a geometrical rotation around
the z axis with angle . Therefore, in view of the above result, we refer to the
operator D
z
() as the generator of rotation around the z axis with angle .
It is straightforward to generalize the above results and to show that rotation
around an arbitrary unit vector n axis with angle is given by
D
n
() = exp
_
iL n
_
. (6.20)
In view of Eq. (3.73), it can be said that linear momentum p generates
translations. Similarly, in view of the above equation (6.20), angular momen-
tumL generates rotation. However, there is an important distinction between
these two types of geometrical transformations. On one hand, according to
Eq. (3.7) the observables p
x
, p
y
and p
z
commute with each other, and con-
sequently translation operators with dierent translation vectors commute
[J (
1
) , J (
2
)] = 0 . (6.21)
On the other hand, as can be seen from Eq. (6.5), dierent components of L do
not commute and therefore rotation operators D
n
() with dierent rotations
axes n need not commute. Both the above results, which were obtained from
commutation relations between quantum operators, demonstrate two well
known geometrical facts: (i) dierent linear translations commute, whereas
(ii) generally, dierent rotations do not commute.
6.2 General Angular Momentum
Elementary particles carry angular momentum in two dierent forms. The
rst one is the above discussed orbital angular momentum, which is com-
monly labeled as L. This contribution L = r p has a classical analogue,
which was employed above to derive the commutation relations (6.5) from
the corresponding Poissons brackets relations. The other form of angular
momentum is spin, which is commonly labeled as S. Contrary to the orbital
angular momentum, the spin does not have any classical analogue. In a gen-
eral discussion on angular momentum in quantum mechanics the label J is
commonly employed.
L - orbital angular momentum
S - spin angular momentum
J - general angular momentum
In the discussion below we derive some properties of angular momentum
in quantum mechanics, where our only assumption is that the components of
the angular momentum vector of operators J = (J
x
, J
y
, J
z
) obey the following
commutation relations
[J
i
, J
j
] = i
ijk
J
k
. (6.22)
Namely, we assume that Eq. (6.5), which was obtained from the corresponding
Poissons brackets relations for the case of orbital angular momentum holds
for general angular momentum.
6.3 Simultaneous Diagonalization of J
2
and J
z
As we have seen in chapter 2, commuting operators can be simultaneously
diagonalized. In this section we seek such simultaneous diagonalization of the
operators J
2
and J
z
, where
J
2
= J
2
x
+J
2
y
+J
2
z
. (6.23)
As is shown by the claim below, these operators commute.
Claim. The following holds
_
J
2
, J
x
=
_
J
2
, J
y
=
_
J
2
, J
z
= 0 . (6.24)
Proof. Using Eq. (6.22) one nds that
_
J
2
, J
z
=
_
J
2
x
, J
z
+
_
J
2
y
, J
z
= i (J
x
J
y
J
y
J
x
+J
y
J
x
+J
x
J
y
) = 0 .
(6.25)
In a similar way one can show that
_
J
2
, J
x
=
_
J
2
, J
y
= 0.
6.3. Simultaneous Diagonalization of J
2
and J
z
The common eigenvectors of the operators J
2
and J
z
are labeled as [a, b,
and the corresponding eigenvalues are labeled as a
2
and b respectively
J
2
[a, b = a
2
[a, b , (6.26)
J
z
[a, b = b [a, b . (6.27)
Recall that we have shown in chapter 5 for the case of harmonic oscillator
that the ket-vectors a [n and a
[n are eigenvectors of the number operator

N provided that [n is an eigenvector of N. Somewhat similar claim can
be made regrading the current problem under consideration of simultaneous
diagonalization of J
2
and J
z
:
Theorem 6.3.1. Let [a, b be a normalized simultaneous eigenvector of the
operators J
2
and J
z
with eigenvalues
2
a and b respectively, i.e.
J
2
[a, b = a
2
[a, b , (6.28)
J
z
[a, b = b [a, b , (6.29)
a, b [a, b = 1 . (6.30)
Then (i) the vector
[a, b + 1
1
[a b (b + 1)]
1/2
J
+
[a, b (6.31)
where
J
+
= J
x
+iJ
y
, (6.32)
is a normalized simultaneous eigenvector of the operators J
2
and J
z
with
eigenvalues
2
a and (b + 1) respectively, namely
J
2
[a, b + 1 = a
2
[a, b + 1 , (6.33)
J
z
[a, b + 1 = (b + 1) [a, b + 1 . (6.34)
(ii) The vector
[a, b 1
1
[a b (b 1)]
1/2
J
[a, b (6.35)
where
J
= J
x
iJ
y
, (6.36)
is a normalized simultaneous eigenvector of the operators J
2
and J
z
with
eigenvalues
2
a and (b 1) respectively, namely
J
2
[a, b 1 = a
2
[a, b 1 , (6.37)
J
z
[a, b 1 = (b 1) [a, b 1 . (6.38)
Proof. The following holds
J
2
(J
[a, b) =
_
_
_
_
J
2
, J
. .
0
+J
J
2
_
_
_[a, b = a
2
(J
[a, b) . (6.39)
Similarly
J
z
(J
[a, b) = ([J
z
, J
] +J
J
z
) [a, b , (6.40)
where
[J
z
, J
] = [J
z
, J
x
iJ
y
] = (iJ
y
J
x
) = J
, (6.41)
thus
J
z
(J
[a, b) = (b 1) (J
[a, b) . (6.42)
Using the following relations
J
+
J
+
= J
J
+
= (J
x
iJ
y
) (J
x
+iJ
y
)
= J
2
x
+J
2
y
+i [J
x
, J
y
]
= J
2
J
2
z
J
z
,
(6.43)
J
= J
+
J
= (J
x
+iJ
y
) (J
x
iJ
y
)
= J
2
x
+J
2
y
+i [J
y
, J
x
]
= J
2
J
2
z
+J
z
,
(6.44)
one nds that
a, b[ J
+
J
+
[a, b = a, b[ J
2
[a, b a, b[ J
z
(J
z
+) [a, b
=
2
[a b (b + 1)] ,
(6.45)
and
a, b[ J
[a, b = a, b[ J
2
[a, b a, b[ J
z
(J
z
) [a, b
=
2
[a b (b 1)] .
(6.46)
Thus the states [a, b + 1 and [a, b 1 are both normalized.
6.3. Simultaneous Diagonalization of J
2
and J
z
What are the possible values of b? Recall that we have shown in chapter 5
for the case of harmonic oscillator that the eigenvalues of the number operator
N must be nonnegative since the operator N is positive-denite. Below we
employ a similar approach to show that:
Claim. b
2
a
Proof. Both J
2
x
and J
2
y
are positive-denite, therefore
a, b[ J
2
x
+J
2
y
[a, b 0 . (6.47)
On the other hand, J
2
x
+J
2
y
= J
2
J
2
z
, therefore a b
2
0.
As we did in chapter 5 for the case of the possible eigenvalues n of the
number operator N, also in the present case the requirement b
2
a restricts
the possible values that b can take:
Claim. For a given value of a the possible values of b are b
max
, b
max
+ 1, , b
max
1, b
max
where a = b
max
(b
max
+ 1). Moreover, the possible values of b
max
are 0, 1/2, 1, 3/2, 2, .
Proof. There must be a maximum value b
max
for which
J
+
[a, b
max
= 0 . (6.48)
Thus, also
J
+
J
+
[a, b
max
= 0 (6.49)
holds. With the help of Eq. (6.43) this can be written as
_
J
2
J
2
z
J
z
_
[a, b
max
= [a b
max
(b
max
+ 1)]
2
[a, b
max
= 0 . (6.50)
Since [a, b
max
,= 0 one has
a b
max
(b
max
+ 1) = 0 , (6.51)
or
a = b
max
(b
max
+ 1) . (6.52)
In a similar way with the help of Eq. (6.44) one can show that there exists a
minimum value b
min
for which
a = b
min
(b
min
1) . (6.53)
From the last two equations one nds that
b
max
(b
max
+ 1) = b
min
(b
min
1) , (6.54)
or
(b
max
+b
min
) (b
max
b
min
+ 1) = 0 . (6.55)
Thus, since b
max
b
min
+ 1 > 0 one nds that
b
min
= b
max
. (6.56)
The formal solutions of Eqs. (6.52) and (6.53) are given by
b
max
=
1
2

1
2
1 + 4a , (6.57)
and
b
min
=
1
2

1
2
1 + 4a . (6.58)
Furthermore, a is an eigenvalue of a positive-denite operator J
2
, therefore
a 0. Consequently, the only possible solutions for which b
max
b
min
are
b
max
=
1
2
+
1
2
1 + 4a 0 , (6.59)
and
b
min
=
1
2

1
2
1 + 4a = b
max
0 . (6.60)
That is, for a given value of a, both b
max
and b
min
are uniquely de-
termined. The value b
min
is obtained by successively applying the oper-
ator J
to the state [a, b

max
an integer number of times, and therefore
b
max
b
min
= 2b
max
must be an integer. Consequently, the possible values of
b
max
are 0, 1/2, 1, 3/2, .
We now change the notation [a, b for the simultaneous eigenvectors to
the more common notation [j, m, where
j = b
max
, (6.61)
m = b . (6.62)
Our results can be summarized by the following relations
J
2
[j, m = j (j + 1)
2
[j, m , (6.63)
J
z
[j, m = m[j, m , (6.64)
J
+
[j, m =
_
j (j + 1) m(m+ 1)[j, m+ 1 , (6.65)
J
[j, m =
_
j (j + 1) m(m1)[j, m1 , (6.66)
where the possible values j can take are
j = 0,
1
2
, 1,
3
2
, , (6.67)
and for each given j, the quantum number m can take any of the 2j + 1
possible values
m = j, j + 1, , j 1, j . (6.68)
6.5. Orbital Angular Momentum
The vector space of a spin 1/2 system is the subspace spanned by the ket-
vectors [j = 1/2, m = 1/2 and [j = 1/2, m = 1/2. In this subspace the
spin angular momentum is labeled using the letter S, as we have discussed
above. The matrix representation of some operators of interest in this basis
can be easily found with the help of Eqs. (6.63), (6.64), (6.65) and (6.66):
S
2
=
3
2
4
_
1 0
0 1
_
, (6.69)
S
z
=
2
_
1 0
0 1
_
z
, (6.70)
S
+
=
_
0 1
0 0
_
, (6.71)
S
=
_
0 0
1 0
_
. (6.72)
The above results for S
+
and S
together with the identities

S
x
=
S
+
+S
2
, (6.73)
S
y
=
S
+
S
2i
, (6.74)
can be used to nd the matrix representation of S
x
and S
y
S
x
=

2
_
0 1
1 0
_
x
, (6.75)
S
y
=

2
_
0 i
i 0
_
y
. (6.76)
The matrices
x
,
y
and
z
are called Paulis matrices, and are related to
the corresponding spin angular momentum operators by the relation
S
k
=
k
. (6.77)
6.5 Orbital Angular Momentum
As we have discussed above, orbital angular momentum L = r p refers to
spatial motion. For this case the states [l, m (here, the letter l is used instead
of j since we are dealing with orbital angular momentum) can be described
using wave functions. In this section we calculate these wave functions. For
this purpose it is convenient to employ the transformation from Cartesian to
spherical coordinates
x = r sin cos , (6.78)
y = r sin sin , (6.79)
z = r cos , (6.80)
where
r 0 , (6.81)
0 , (6.82)
0 2 . (6.83)
Exercise 6.5.1. Show that:
1.
r
[ L
z
[ = i

[ . (6.84)
2.
r
[ L
[ = i exp(i)
_
i

cot

_
r
[ . (6.85)
3.
r
[ L
2
[ =
2
_
1
sin
2
2
+
1
sin

_
sin

__
r
[ . (6.86)
Solution 6.5.1. Using the relations
L = r p = det
_
_
x y z
x y z
p
x
p
y
p
z
_
_
, (6.87)
r
[ r [ = r
[ , (6.88)
r
[ p[ =

i
r
[ , (6.89)
[see Eqs. (3.21) and (3.29)] one nds that
r
[ L
x
[ =

i
_
y

z
z

y
_
(r
) , (6.90)
r
[ L
y
[ =

i
_
z

x
x

z
_
(r
) , (6.91)
r
[ L
z
[ =

i
_
x

y
y

x
_
(r
) , (6.92)
where
(r
) = r
[ . (6.93)
r =
_
x
2
+y
2
+z
2
, (6.94)
cos =
z
_
x
2
+y
2
+z
2
, (6.95)
cot =
x
y
. (6.96)
=
x

x
+
y

y
+
z

z
= r sin sin

x
+r sin cos

y
= y

x
+x

y
,
(6.97)
thus using Eq. (6.92) one has
r
[ L
z
[ = i

(r
) . (6.98)
2. Using Eqs. (6.90) and (6.91) together with the relation L
+
= L
x
+ iL
y
one has
i
[ L
+
[ =
i
[ L
x
+iL
y
[
=
_
y

z
z

y
+iz

x
ix

z
_
(r
)
=
_
z
_
i

x

y
_
i (x +iy)

z
_
(r
)
=
_
z
_
i

x

y
_
ir sine
i

z
_
(r
) .
(6.99)
Thus, by using the identity
=
x

x
+
y

y
+
z

z
= r cos
_
cos

x
+ sin

y
_
r sin

z
,
(6.100)
or
r sin

z
= r cos
_
cos

x
+ sin

y
_
, (6.101)
one nds that
i
[ L
+
[ =
_
z
_
i

x

y
_
ie
i
_
cot
_
x

x
+y

y
_
__
(r
)
=
_
i
_
z e
i
xcot
_

x

_
z +ie
i
y cot
_

y
+ie
i

(r
)
= e
i
_
_
i cot
_
_
ze
i
tan
. .
xiy
x
_
_

x
cot
_
_
ze
i
tan
. .
xiy
+iy
_
_

y
+i

_
_
(r
)
= e
i
_
cot
_
y

x
x

y
_
+i

(r
)
= e
i
_
i

cot

(r
) .
(6.102)
In a similar way one evaluates r
[ L
[. Both results can be expressed

as
r
[ L
[ = i exp(i)
_
i

cot

(r
) . (6.103)
3. Using the result of the previous section one has
r
[ L
x
[ =
1
2
r
[ (L
+
+L
) [
=
i
2
_
e
i
_
cot

_
+e
i
_
cot

+i

__
(r
)
= i
_
cos cot

+ sin

(r
) .
(6.104)
Similarly
r
[ L
y
[ = i
_
sincot

cos

(r
) , (6.105)
thus
r
[ L
2
[ = r
[ L
2
x
+L
2
y
+L
2
z
[
=
2
_
_
cos cot

+ sin

_
2
+
_
sincot

cos

_
2
+

2
2
_
(r
)
=
2
_
_
1 + cot
2
_

2
2
+ cot

+

2
2
_
(r
)
=
2
_
1
sin
2
2
+
1
sin

_
sin

__
(r
) .
(6.106)
Spherical Harmonics. The above exercise allows translating the relations
(6.63) and (6.64), which are given by
L
2
[l, m = l (l + 1)
2
[l, m , (6.107)
L
z
[l, m = m [l, m , (6.108)
into dierential equations for the corresponding wavefunctions
_
1
sin
2
2
+
1
sin

_
sin

__
(r
) = l (l + 1)
(r
) , (6.109)
i

(r
) = m
(r
) , (6.110)
where
m = l, l + 1, , l 1, l . (6.111)
We seek solutions having the form
(r
) = f (r) Y
m
l
(, ) . (6.112)
We require that both f (r) and Y
m
l
(, ) are normalized
1 =
_
0
drr
2
[f (r)[
2
, (6.113)
1 =
_
0
d sin
2
_
0
d[Y
m
l
(, )[
2
. (6.114)
These normalization requirements guarantee that the total wavefunction is
normalized
1 =
dx
dy
dz [
(r
)[
2
. (6.115)
Substituting into Eqs. (6.109) and (6.110) yields
_
1
sin
2
2
+
1
sin

_
sin

_
Y
m
l
= l (l + 1) Y
m
l
, (6.116)
i

Y
m
l
= mY
m
l
. (6.117)
The functions Y
m
l
(, ) are called spherical harmonics
In the previous section, which discusses the case of general angular mo-
mentum, we have seen that the quantum number m can take any half integer
value 0, 1/2, 1, 3/2, [see Eq. (6.67)]. Recall that the only assumption em-
ployed in order to obtain this result was the commutation relations (6.22).
However, as is shown by the claim below, only integer values are allowed for
the case of orbital angular momentum. In view of this result, one may argue
that the existence of spin, which corresponds to half integer values, is in fact
predicted by the commutation relations (6.22) only.
Claim. The variable m must be an integer.
Proof. Consider a solution having the form
Y
m
l
(, ) = F
m
l
() e
im
. (6.118)
Clearly, Eq. (6.117) is satised. The requirement
Y
m
l
(, ) = Y
m
l
(, + 2) , (6.119)
namely, the requirement that Y
m
l
(, ) is continuos, leads to
e
2im
= 1 , (6.120)
thus m must be an integer.
The spherical harmonics Y
m
l
(, ) can be obtained by solving Eqs. (6.116)
and (6.117). However, we will employ an alternative approach, in which in
the rst step we nd the spherical harmonics Y
l
l
(, ) by solving the equation
L
+
[l, l = 0 , (6.121)
which is of rst order [contrary to Eq. (6.116), which is of the second order].
Using the identity (6.85), which is given by
r
[ L
+
[ = ie
i
_
i

cot

_
r
[ , (6.122)
one has
_

l cot
_
F
l
l
() = 0 . (6.123)
F
l
l
() = C
l
(sin)
l
, (6.124)
where C
l
is a normalization constant. Thus, Y
l
l
is given by
Y
l
l
(, ) = C
l
(sin)
l
e
il
. (6.125)
In the second step we employ the identity (6.66), which is given by
J
[j, m =
_
j (j + 1) m(m1) [j, m1 , (6.126)
6.6. Problems
and Eq. (6.85), which is given by
r
[ L
[ = iexp(i)
_
i

cot

_
r
[ , (6.127)
to derive the following recursive relation
e
i
_
mcot
_
Y
m
l
(, ) =
_
l (l + 1) m(m1)Y
m1
l
(, ) ,
(6.128)
which allows nding Y
m
l
(, ) for all possible values of m provided that
Y
l
l
(, ) is given. The normalized spherical harmonics are found using this
method to be given by
Y
m
l
(, ) =
(1)
l
2
l
l!
_
2l + 1
4
(l +m)!
(l m)!
e
im
(sin)
m
d
lm
d(cos )
lm
(sin)
2l
.
(6.129)
As an example, closed form expressions for the cases l = 0 and l = 1 are
given below
Y
0
0
(, ) =
1
4
, (6.130)
Y
1
1
(, ) =
_
3
8
sine
i
, (6.131)
Y
0
1
(, ) =
_
3
4
cos . (6.132)
6.6 Problems
1. Let R
(where i x, y, z ) be the 33 rotation matrices (as dened in

the lecture). Show that for innitesimal angle the following holds
[R
x
() , R
y
()] = 1 R
z
_
2
_
, (6.133)
where
[R
x
() , R
y
()] = R
x
() R
y
() R
y
() R
x
() . (6.134)
2. Show that
exp
_
iJ
z
_
J
x
exp
_
iJ
z
_
= J
x
cos J
y
sin . (6.135)
3. The components of the Pauli matrix vector are given by:
x
=
_
0 1
1 0
_
,
y
=
_
0 i
i 0
_
,
z
=
_
1 0
0 1
_
. (6.136)
a) Show that
( a) ( b) = a b +i (a b) , (6.137)
where a and b are vector operators which commute with , but not
necessarily commute with each other.
b) Show that
exp
_
i n
2
_
= 1cos

2
i nsin

2
, (6.138)
where n is a unit vector and where 1 is the 2 2 identity matrix.
4. Find the eigenvectors and eigenvalues of the matrix n ( n is a unit
vector).
5. Consider an electron in a state in which the component of its spin along
the z axis is +/2 . What is the probability that the component of the spin
along an axis z
, which makes an angle with the z axis, will measured

to be +/2 or /2 . What is the average value of the component of the
spin along this axis?
6. The 2 2 matrix U is given by
U =
1 +i( n)
1 i( n)
, (6.139)
where
=
x
x +
y
y +
z
z (6.140)
is the Pauli vector matrix,
n = n
x
x +n
y
y +n
z
z (6.141)
is a unit vector, i.e. n n = 1, and n
x
, n
y
, n
z
and are all real parameters.
Note that generally for a matrix or an operator
1
A
A
1
.
a) show that U is unitary.
b) Show that
dU
d
=
2i ( n)
1 +
2
U . (6.142)
c) Calculate U by solving the dierential equation in the previous sec-
tion.
6.6. Problems
7. A particle is located in a box, which is divided into a left and right
sections. The corresponding vector states are denoted as [L and [R
respectively. The Hamiltonian of the system is given by
H = E
L
[L L[ +E
R
[R R[ +([L R[ +[R L[) . (6.143)
The particle at time t = 0 is in the left section
[(t = 0) = [L . (6.144)
Calculate the probability p
R
(t) to nd the particle in the state [R at
time t.
8. A magnetic eld given by
B(t) = B
0
z +B
1
( xcos (
1
t) + ysin(
1
t)) (6.145)
is applied to a spin 1/2 particle. At time t = 0 the state is given by
[ (t = 0) = [+; z . (6.146)
Calculate the probability P
+
(t) to nd the system in the state [; z
at time t > 0.
9. Find the time evolution of the state vector of a spin 1/2 particle in
a magnetic eld along the z direction with time dependent magnitude
B(t) = B(t) z.
10. A magnetic eld given by B = Bcos (t) z, where B is a constant, is
applied to a spin 1/2. At time t = 0 the spin is in state [ (t), which
satises
S
x
[ (t = 0) =

2
[ (t = 0) , (6.147)
Calculate the expectation value S
z
at time t 0.
11. Consider a spin 1/2 particle. The time dependent Hamiltonian is given
by
H =
4S
z
1 + (t)
2
, (6.148)
where is a real non-negative constant and S
z
is the z component of the
angular momentum operator. Calculate the time evolution operator u of
the system.
12. Consider a spin 1/2 particle in an eigenstate of the operator S n with
eigenvalue +/2 , where S is the vector operator of angular momentum
and where n is a unit vector. The angle between the unit vector n and
the z axis is . Calculate the expectation values
a) S
z

b)
_
(S
z
)
2
_
13. An ensemble of spin 1/2 particles are in a normalized state
[ = [+ + [ ,
where the states [+ and [ are the eigenstates of S
z
(the z component
of the angular momentum operator). At what direction the magnetic eld
should be aligned in a Stern-Gerlach experiment in order for the beam
not to split.
14. Consider a spin 1/2 particle having gyromagnetic ratio in a magnetic
eld given by B(t) u . The unit vector is given by
u =(sin cos , sin sin, cos ) , (6.149)
where , are angles in spherical coordinates. The eld intensity is given
by
B(t) =
_
_
_
0 t < 0
B
0
0 < t <
0 t >
. (6.150)
At times t < 0 the spin was in state [+ , namely in eigenstate of S
z
with
positive eigenvalue. Calculate the probability P
(t) to nd the spin in

state [ at time t , where t > .
15. Consider a spin 1/2 particle. The Hamiltonian is given by
H = S
x
, (6.151)
where is a Larmor frequency and where S
x
is the x component of the
angular momentum operator. The z component of the angular momen-
tum is measured at the times t
n
= nT/N where n = 0, 1, 2, , N, N is
integer and T is the time of the last measurement.
a) Find the matrix representation of the time evolution operator u(t)
in the basis of [; z states.
b) What is the probability p
same
to get the same result in all N + 1
measurements. Note that the initial state of the particle is unknown.
c) For a xed T calculate the limit lim
N
p
same
.
16. Consider a spin 1/2 particle. No external magnetic eld is applied. Three
measurements are done one after the other. In the rst one the z com-
ponent of the angular momentum is measured, in the second one the
component along the direction u is measured and in the third measure-
ment, again the z component is measured. The unit vector u is described
using the angles and
u =(sin cos , sin sin, cos ) . (6.152)
same
to have the same result in the 1st and
3rd measurements.
6.6. Problems
17. Let (t) be the expectation value of the magnetic moment associated
with spin 1/2 particle ( = S , where S is the angular momentum and
is the gyromagnetic ratio). Show that in the presence of a time varying
magnetic eld B(t) the following holds
d
dt
(t) = (t) B(t) . (6.153)
18. The Hamiltonian of an electron of mass m, charge q, spin 1/2, placed in
electromagnetic eld described by the vector potential A(r, t) and the
scalar potential (r, t), can be written as [see Eq. (1.62)]
H =
_
p
q
c
A
_
2
2m
+q
q
2mc
B , (6.154)
where B =A. Show that this Hamiltonian can also be written as
H =
1
2m
_

_
p
q
c
A
__
2
+q . (6.155)
19. Show that
j, m[
_
(J
x
)
2
+ (J
y
)
2
_
[j, m =
2
_
j
2
+j m
2
_
. (6.156)
20. Find the condition under which the Hamiltonian of a charged particle in
a magnetic eld
H =
1
2m
_
p
q
c
A
_
2
. (6.157)
can be written as
H =
1
2m
p
2
q
mc
p A+
q
2
2mc
2
A
2
. (6.158)
21. Consider a point particle having mass m and charge q moving under the
inuence of electric eld E and magnetic eld B, which are related to
the scalar potential and to the vector potential A by
E =
1
c
A
t
, (6.159)
and
B = A . (6.160)
Find the coordinates representation of the time-independent Schrdinger
equation H[ = E[.
22. A particle of mass m and charge e interacts with a vector potential
A
x
= A
z
= 0 , (6.161)
A
y
= Bx . (6.162)
Calculate the ground state energy. Clue: Consider a wave function of the
form
(x, y, z) = (x) exp(ik
y
y) exp(ik
z
z) . (6.163)
23. Find the energy spectrum of a charged particle having mass mand charge
q moving in uniform and time-independent magnetic eld B = Bz and
electric eld E = E x.
24. Consider a particle having mass m and charge e moving in xy plane under
the inuence of the potential U (y) =
1
2
m
2
0
y
2
. A uniform and time-
independent magnetic eld given by B = Bz is applied perpendicularly
to the xy plane. Calculate the eigenenergies of the particle.
25. Consider a particle with charge q and mass conned to move on a circle
of radius a in the xy plane, but is otherwise free. A uniform and time
independent magnetic eld B is applied in the z direction.
a) Find the eigenenergies.
b) Calculate the current J
m
for each of the eigenstates of the system.
26. The Hamiltonian of a non isotropic rigid rotator is given by
H =
L
2
x
2I
xy
+
L
2
y
2I
xy
+
L
2
z
2I
z
, (6.164)
where L is the vector angular momentum operator. At time t = 0 the
state of the system is described by the wavefunction
(, ) = Asin cos , (6.165)
where , are angles in spherical coordinates and A is a normalization
constant. Calculate the expectation value L
z
at time t > 0 .
27. The eigenstates of the angular momentum operators L
2
and L
z
with
l = 1 and m = 1, 0, 1 are denoted as [1, 1, [1, 0 and [1, 1.
a) Write the 3 3 matrix of the operator L
x
in this l = 1 subspace.
b) Calculate the expectation value L
x
for the state
1
2
_
[1, 1 +
2 [1, 0 +[1, 1
.
c) The same as the previous section for the state
1
2
[[1, 1 [1, 1].
d) Write the 3 3 matrix representation in this basis of the rotation
operator at angle around the z axis.
e) The same as in the previous section for an innitesimal rotation with
angle d around the x axis.
28. Consider a particle of mass m in a 3D harmonic potential
V (x, y, z, ) =
1
2
m
2
_
x
2
+y
2
+z
2
_
. (6.166)
6.6. Problems
The state vector [ of the particle satisfy
a
x
[ =
x
[ , (6.167)
a
y
[ =
y
[ , (6.168)
a
z
[ =
z
[ , (6.169)
where
x
,
y
and
z
are complex and a
x
, a
y
and a
z
are annihilation
operators
a
x
=
_
m
2
_
x +
ip
x
m
_
, (6.170)
a
y
=
_
m
2
_
y +
ip
y
m
_
, (6.171)
a
z
=
_
m
2
_
z +
ip
z
m
_
, (6.172)
Let L be the vector operator of the orbital angular momentum.
a) Calculate L
z
.
b) Calculate L
z
.
29. A rigid rotator is prepared in a state
[ = A([1, 1 [1, 1) , (6.173)
where A is a normalization constant, and where the symbol [l, m denotes
an angular momentum state with quantum numbers l and m. Calculate
a) L
x
.
b)
_
(L
x
)
2
_
.
30. The Hamiltonian of a top is given by
H =
L
2
x
+L
2
y
2I
1
+
L
2
z
2I
2
, (6.174)
where L is the angular momentum vector operator. Let [
0
be the
ground state of the system.
a) Calculate the quantity A
z
(), which is dened as
A
z
() =
0
[ exp
_
iL
z
_
Hexp
_
iL
z
_
[
0
. (6.175)
b) Calculate the quantity A
x
(), which is dened as
A
x
() =
0
[ exp
_
iL
x
_
Hexp
_
iL
x
_
[
0
. (6.176)
31. The wavefunction of a point particle is given by
(r) = (x +y + 2z) f (r) , (6.177)
where f (r) is a function of the radial coordinate r =
_
x
2
+y
2
+z
2
.
a) In a measurement of L
2
what are the possible outcomes and the
corresponding probabilities.
b) The same for a measurement of L
z
.
32. Consider a system comprising of two spin 1/2 particles.
a) Show that
_
S
2
, S
z
= 0 , (6.178)
where S = S
1
+S
2
, S
z
= S
1z
+S
2z
and where S
1
and S
2
are the an-
gular momentum vector operators of the rst and second spin repet-
itively, i.e. S
1
= (S
1x
, S
1y
, S
1z
) and S
2
= (S
2x
, S
2y
, S
2z
).
b) Find an orthonormal basis of common eigenvectors of S
2
and S
z
[recall that the existence of such a basis is guaranteed by the result
of the previous section
_
S
2
, S
z
= 0, see Eqs. (2.147) and (2.148)].

6.7 Solutions
1. By cyclic permutation of
R
z
=
_
_
cos sin 0
sin cos 0
0 0 1
_
_
, (6.179)
one has
R
x
=
_
_
1 0 0
0 cos sin
0 sin cos
_
_
, (6.180)
R
y
=
_
_
cos 0 sin
0 1 0
sin 0 cos
_
_
. (6.181)
On one hand
1 [R
x
() , R
y
()]
=
_
_
1 1 + cos
2
sin sincos
1 cos
2
1 sincos sin
sin sincos sincos sin 1
_
_
=
_
_
1
2
0
2
1 0
0 0 1
_
_
+O
_
3
_
.
(6.182)
On the other hand
6.7. Solutions
R
z
_
2
_
=
_
_
cos
2
sin
2
0
sin
2
cos
2
0
0 0 1
_
_
=
_
_
1
2
0
2
1 0
0 0 1
_
_
+O
_
3
_
, (6.183)
thus
1 [R
x
() , R
y
()] = R
z
_
2
_
+O
_
3
_
. (6.184)
2. Using the identity (2.173), which is given by
e
L
Ae
L
= A+[L, A] +
1
2!
[L, [L, A]] +
1
3!
[L, [L, [L, A]]] + , (6.185)
and the commutation relations (6.22), which are given by
[J
i
, J
j
] = i
ijk
J
k
, (6.186)
one has
exp
_
iJ
z
_
J
x
exp
_
iJ
z
_
= J
x
+
i
[J
z
, J
x
] +
1
2!
_
i
_
2
[J
z
, [J
z
, J
x
]]
+
1
3!
_
i
_
3
[J
z
, [J
z
, [J
z
, J
x
]]] +
= J
x
_
1
1
2!
2
+
_
J
y
_

1
3!
3
+
_
J
x
cos J
y
sin .
(6.187)
3. The components of the Pauli matrix vector are given by:
x
=
_
0 1
1 0
_
,
y
=
_
0 i
i 0
_
,
z
=
_
1 0
0 1
_
. (6.188)
a) The following holds
a =
_
a
z
a
x
ia
y
a
x
+ia
y
a
z
_
, (6.189)
b =
_
b
z
b
x
ib
y
b
x
+ib
y
b
z
_
, (6.190)
thus
( a) ( b) =
_
a
z
b
z
+ (a
x
ia
y
) (b
x
+ib
y
) a
z
(b
x
ib
y
) (a
x
ia
y
) b
z
(a
x
+ia
y
) b
z
a
z
(b
x
+ib
y
) a
z
b
z
+ (a
x
+ia
y
) (b
x
ib
y
)
_
= a b
_
1 0
0 1
_
+i (a
y
b
z
a
z
b
y
)
_
0 1
1 0
_
+i (a
z
b
x
a
x
b
z
)
_
0 i
i 0
_
+i (a
x
b
y
a
y
b
x
)
_
1 0
0 1
_
= a b +i (a b) .
(6.191)
b) Using (a) one has
( n)
2
= 1 , (6.192)
thus with the help of the Taylor expansion of the functions cos (x)
and sin(x) one nds
exp
_
i n
2
_
= cos
_
n
2
_
i sin
_
n
2
_
= 1cos

2
i nsin

2
.
(6.193)
4. In spherical coordinates the unit vectors n is expressed as
n = (cos sin, sinsin, cos ) , (6.194)
thus
n =
_
cos sine
i
sine
i
cos
_
. (6.195)
The eigen values
+
and
are found solving
+
+
= Tr ( n) = 0 , (6.196)
and
= Det ( n) = 1 , (6.197)
thus
= 1 . (6.198)
The normalized eigenvectors can be chosen to be given by
6.7. Solutions
[+ =
_
cos

2
exp
_
i
2
_
sin

2
exp
_
i
2
_
_
, (6.199)
[ =
_
sin

2
exp
_
i
2
_
cos

2
exp
_
i
2
_
_
. (6.200)
5. Using Eq. (6.199) one nds the probability p
+
to measure +/2 is given
by
p
+
=
_
1 0
_
_
cos

2
exp
_
i
2
_
sin

2
exp
_
i
2
_
_
2
= cos
2

2
, (6.201)
and the probability p
to measure /2 is
p
= 1 p
+
= sin
2

2
. (6.202)
The average value of the component of the spin along z
axis is thus
2
_
cos
2

2
sin
2

2
_
=

2
cos . (6.203)
6. In general, note that all smooth functions of the matrix ( n) commute,
a fact that greatly simplies the calculations.
1
1 i( n)
= 1 +i( n) + [i( n)]
2
+ , (6.204)
thus
_
1
1 i( n)
_
= 1 i( n) + [(i) ( n)]
2
+
=
1
1 +i( n)
,
(6.205)
therefore
UU
=
1 +i( n)
1 i( n)
1 i( n)
1 +i( n)
= 1 , (6.206)
and similarly U
U = 1.
b) Exploiting again the fact that all smooth functions of the matrix
( n) commute and using Eq. (6.192) one has
dU
d
= i
[1 i( n)] ( n) + [1 +i( n)] ( n)
[1 i( n)]
2
= i
2 ( n)
[1 i( n)]
2
= i
2 ( n)
[1 i( n)] [1 +i( n)]
1 +i( n)
1 i( n)
=
2i ( n)
1 +
2
U .
(6.207)
c) By integration one has
U = U
0
exp
_
2i ( n)
_

0
d
1 +
2
_
U
0
exp
_
2i ( n) tan
1
_
,
(6.208)
where U
0
is a the matrix U at = 0. With the help of Eq. (6.138)
one thus nds that
U = U
0
_
1 cos
_
2 tan
1
_
+i nsin
_
2 tan
1
_
, (6.209)
Using the identities
cos
_
2 tan
1
_
=
1
2
1 +
2
, (6.210)
sin
_
2 tan
1
_
=
2
1 +
2
, (6.211)
and assuming U
0
= 1 one nds that
U =
1
2
1 +
2
+i n
2
1 +
2
. (6.212)
7. In terms of Pauli matrices
H =E
a
0
+
x
+E
d
z
, (6.213)
where
E
a
=
E
L
+E
R
2
, E
d
=
E
L
E
R
2
, (6.214)
and
0
=
_
1 0
0 1
_
,
x
=
_
0 1
1 0
_
,
z
=
_
1 0
0 1
_
. (6.215)
Using Eq. (6.138), which is given by
exp
_
i n
2
_
= cos

2
i nsin

2
, (6.216)
6.7. Solutions
the time evolution operator u(t) can be calculated
u(t) = exp
_
iHt
_
= exp
_
iE
a
0
t
_
exp
_
i (
x
+E
d
z
) t
_
= exp
_
iE
a
t
_
exp
_
i n
_
2
+E
2
d
t
_
,
(6.217)
where
n =
(, 0, E
d
)
_
2
+E
2
d
, (6.218)
thus
u(t) = exp
_
iE
a
t
_
_
cos
_
2
+E
2
d
t
x
+E
d
z
_
2
+E
2
d
sin
t
_
2
+E
2
d
_
.
(6.219)
The probability p
R
(t) is thus given by
p
R
(t) = [R[ u(t) [ (t = 0)[
2
= [R[ u(t) [L[
2
=

2
2
+
_
ELER
2
_
2
sin
2
t
_
2
+
_
ELER
2
_
2
.
(6.220)
H =
0
S
z
+
1
(cos (t) S
x
+ sin(t) S
y
) , (6.221)
where
0
=
[e[ B
0
m
e
c
, (6.222)
1
=
[e[ B
1
m
e
c
. (6.223)
The matrix representation in the basis [+ , [ (where [+ = [+; z
and [ = [; z) is found using Eqs. (6.70), (6.75) and (6.76)
H =
2
_

0

1
exp(it)
1
exp(it)
0
_
. (6.224)
The Schrdinger equation is given by
i
d
dt
[ = H[ . (6.225)
It is convenient to express the general solution as
[ (t) = b
+
(t) exp
_
it
2
_
[+ +b
(t) exp
_
it
2
_
[ . (6.226)
Substituting into the Schrdinger equation yields
i
d
dt
__
e
it
2
0
0 e
it
2
__
b
+
b
__
=
1
2
_

0

1
e
it
1
e
it
0
__
e
it
2
0
0 e
it
2
__
b
+
b
_
,
(6.227)
or
i
2
_
e
it
2
0
0 e
it
2
__
b
+
b
_
+
_
e
it
2
0
0 e
it
2
__
b
+
_
=
i
2
_

0

1
e
it
1
e
it
0
__
e
it
2
0
0 e
it
2
__
b
+
b
_
.
(6.228)
By multiplying from the left by
_
e
it
2
0
0 e
it
2
_
one has
i
2
_
1 0
0 1
__
b
+
b
_
+
_
b
+
_
=
i
2
_
0

1
1

0
__
b
+
b
_
, (6.229)
or
i
_
b
+
_
=

2
_
b
+
b
_
, (6.230)
where
=
_

1
1

_
=
z
+
1
x
, (6.231)
and
=
0
. (6.232)
6.7. Solutions
At time t = 0
_
b
+
(0)
b
(0)
_
=
_
1
0
_
. (6.233)
The time evolution is found using Eq. (6.138)
_
b
+
(t)
b
(t)
_
= exp
_
it
2
__
b
+
(0)
b
(0)
_
=
_
_
cos i
sin
2
1
+()
2
i
1 sin
2
1
+()
2
i
1 sin
2
1
+()
2
cos +i
sin
2
1
+()
2
_
_
_
1
0
_
,
(6.234)
where
=
_
2
1
+ ()
2
t
2
. (6.235)
The probability is thus given by
P
+
(t) =

2
1
2
1
+ ()
2
sin
2
_
2
1
+ ()
2
t
2
. (6.236)
9. The Schrdinger equation is given by
i
d[
dt
= H[ , (6.237)
where
H = S
z
, (6.238)
and
(t) =
[e[ B(t)
m
e
c
. (6.239)
In the basis of the eigenvectors of S
z
one has
[ = c
+
[+ +c
[ , (6.240)
and
i ( c
+
[+ + c
[) = S
z
(c
+
[+ +c
[) , (6.241)
where
S
z
[ =
2
[ , (6.242)
thus one gets 2 decoupled equations
c
+
=
i
2
c
+
, (6.243)
c
=
i
2
c
. (6.244)
c
(t) = c
(0) exp
_
i
2
_
t
0
(t
) dt
_
= c
(0) exp
_
i [e[
2m
e
c
_
t
0
B(t
) dt
_
.
(6.245)
10. At time t = 0
[(t = 0) =
1
2
([+ +[) . (6.246)
Using the result of the previous problem and the notation
c
=
eB
mc
, (6.247)
one nds
[ (t) =
1
2
_
_
exp
_
_
i
c
2
t
_
0
cos (t
) dt
_
_
[+ + exp
_
_
i
c
2
t
_
0
cos (t
) dt
_
_
[
_
_
=
1
2
_
exp
_
i
c
sint
2
_
[+ + exp
_
i
c
sint
2
_
[
_
,
(6.248)
thus
S
z
(t) = (t)[ S
z
[ (t) = 0 . (6.249)
11. The Schrdinger equation for u is given by
i
du
dt
= Hu , (6.250)
thus
du
dt
=
4iS
z
1
1 + (t)
2
u . (6.251)
By integration one nds
6.7. Solutions
u(t) = u(0) exp
_
_
4iS
z
t
_
0
dt
1 + (t
)
2
_
_
= u(0) exp
_
4iS
z
tan
1
(t)
_
.
(6.252)
Setting an initial condition u(t = 0) = 1 yields
u(t) = exp
_
4iS
z
tan
1
(t)
_
. (6.253)
The matrix elements of u(t) in the basis of the eigenstates [ of S
z
are
given by
+[ u(t) [+ = exp
_
2i tan
1
(t)
_
=
1 +it
1 it
, (6.254)
[ u(t) [ = exp
_
2i tan
1
(t)
_
=
1 it
1 +it
, (6.255)
+[ u(t) [ = [ u(t) [+ = 0 . (6.256)
12. The eigenvector of S n, where n =(sin cos , sin sin, cos ) with
eigenvalue +/2 is given by [see Eq. (6.199)]
[+; S n = cos

2
e
i
2
[+ + sin

2
e
i
2
[ . (6.257)
The operator S
z
is written as
S
z
=

2
([+ +[ [ [) . (6.258)
a) Thus
+; S n[ S
z
[+; S n =

2
_
cos
2

2
sin
2
2
_
=

2
cos . (6.259)
b) Since S
2
z
is the identity operator times
2
/4 one has
_
(S
z
)
2
_
=
S
2
z
_
S
z
2
=

2
4
_
1 cos
2
_
=

2
4
sin
2
. (6.260)
13. We seek a unit vector n such that
[ = [+; S n , (6.261)
where [+; S n is given by Eq. (6.199)
[+; S n = cos

+
2
exp
_
i
+
2
_
[++sin

+
2
exp
_
i
+
2
_
[ , (6.262)
thus the following hold
ctg

+
2
=
, (6.263)
and
+
= arg () arg () . (6.264)
Similarly, by requiring that
[ = [; S n , (6.265)
where
[; S n = sin

2
exp
_
2
_
[++cos

2
exp
_
i
2
_
[ , (6.266)
one nds
tan

2
=
, (6.267)
= arg () arg () + . (6.268)

14. The Hamiltonian at the time interval 0 < t < is given by
H = B
0
(S u) , (6.269)
where is the gyromagnetic ratio and S is the angular momentum op-
erator. The eigenvectors of S u with eigenvalue /2 are given by [see
Eqs. (6.199) and (6.200)]
[+; S u = cos

2
e
i
2
[+ + sin

2
e
i
2
[ , (6.270)
[; S u = sin

2
e
i
2
[+ + cos

2
e
i
2
[ , (6.271)
Thus in the time interval 0 < t < the state vector is given by
6.7. Solutions
[ = [+; S u +; S u [+ exp
_
iB
0
t
2
_
+[; S u ; S u [+ exp
_
iB
0
t
2
_
= [+; S u cos

2
e
i
2
exp
_
iB
0
t
2
_
[; S u sin

2
e
i
2
exp
_
iB
0
t
2
_
= e
i
_
cos
2

2
exp
_
iB
0
t
2
_
+ sin
2

2
exp
_
iB
0
t
2
__
[+
+sin

2
cos

2
_
exp
_
iB
0
t
2
_
exp
_
iB
0
t
2
__
[
= e
i
_
1 + cos
2
exp
_
iB
0
t
2
_
+
1 cos
2
exp
_
iB
0
t
2
__
[+
+i sin sin
_
B
0
t
2
_
[
= e
i
_
cos
_
B
0
t
2
_
+i cos sin
_
B
0
t
2
__
[+ +i sin sin
_
B
0
t
2
_
[ .
(6.272)
Thus for t >
P
(t) = sin
2
sin
2
_
B
0
2
_
. (6.273)
An alternative solution - The Hamiltonian in the basis of [ states is
given by
H =
B
0
2
( u) , (6.274)
where is the Pauli matrix vector
1
=
_
0 1
1 0
_
,
2
=
_
0 i
i 0
_
,
3
=
_
1 0
0 1
_
. (6.275)
The time evolution operator is given by
u(t) = exp
_
iHt
_
= exp
_
iB
0
t
2
( u)
_
. (6.276)
Using the identity (6.138) one nds
u(t) = I cos
_
B
0
t
2
_
+i u sin
_
B
0
t
2
_
=
_
_
cos
_
B0t
2
_
+i cos sin
_
B0t
2
_
i sine
i
sin
_
B0t
2
_
i sine
i
sin
_
B
0
t
2
_
cos
_
B
0
t
2
_
i cos sin
_
B
0
t
2
_
_
_
,
(6.277)
thus for t >
P
(t) =
_
0 1
_
u(t)
_
1
0
_
2
= sin
2
sin
2
_
B
0
2
_
. (6.278)
15. The matrix representation of the Hamiltonian in the basis of [; S
z
states is given by
H =
2
( x ) , (6.279)
where is the Pauli matrix vector
x
=
_
0 1
1 0
_
,
y
=
_
0 i
i 0
_
,
z
=
_
1 0
0 1
_
. (6.280)
a) The time evolution operator is given by
u(t) = exp
_
iHt
_
=exp
_
it
2
( x )
_
. (6.281)
Using the identity
exp(iu ) = 1cos +i u sin , (6.282)
where u = u is a three-dimensional real vector and u is a three-
dimensional real unit vector, one nds
u(t) = 1cos
t
2
i
1
sin
t
2
=
_
cos
t
2
i sin
t
2
i sin
t
2
cos
t
2
_
.
(6.283)
b) Let P
++
(t) be the probability to measure S
z
= +/2 at time t > 0
given that at time t = 0 the spin was found to have S
z
= +/2.
Similarly, P
(t) is the probability to measure S

z
= /2 at time
t > 0 given that at time t = 0 the spin was found to have S
z
= /2.
These probabilities are given by
P
++
(t) =
_
1 0
_
u(t)
_
1
0
_
2
= cos
2
t
2
, (6.284)
P
(t) =
_
0 1
_
u(t)
_
0
1
_
2
= cos
2
t
2
. (6.285)
Thus, assuming that the rst measurement has yielded S
z
= +/2
one nds p
same
=
_
P
++
_
T
N
_
N
, whereas assuming that the rst mea-
surement has yielded S
z
= /2 one nds p
same
=
_
P
_
T
N
_
N
.
Thus in general independently on the result of the rst measurement
one has
p
same
= cos
2N
T
2N
. (6.286)
6.7. Solutions
c) Using
p
same
= exp
_
2N log
_
cos
T
2N
__
= exp
_
2N log
_
1
1
2
_
T
2N
_
2
+O
_
1
N
_
4
__
= exp
_
(T)
2
4N
+O
_
1
N
_
3
_
,
(6.287)
one nds
lim
N
p
same
= 1 . (6.288)
This somewhat surprising result is called the quantum Zeno eect or
the watched pot never boils eect.
16. The eigenvectors of S u with eigenvalues /2 are given by
[+; u = cos

2
e
i
2
[+ + sin

2
e
i
2
[ , (6.289a)
[; u = sin

2
e
i
2
[+ + cos

2
e
i
2
[ , (6.289b)
where the states [ are eigenvectors of S z. Let P (
3
,
2
[
1
) be the
probability to measure S u =
2
(/2) in the second measurement and to
measure Sz =
3
(/2) in the third measurement given that the result of
the rst measurement was S z =
1
(/2), and where
n
+, . The
following holds
P (+, +[+) = [+[+; u[
2
[+[+; u[
2
= cos
4

2
, (6.290a)
P (+, [+) = [+[; u[
2
[+[; u[
2
= sin
4
2
, (6.290b)
P (, [) = [[; u[
2
[[; u[
2
= cos
4

2
, (6.290c)
P (, +[) = [[+; u[
2
[[+; u[
2
= sin
4
2
, (6.290d)
thus independently on what was the result of the rst measurement one
has
p
same
= cos
4

2
+ sin
4
2
= 1
1
2
sin
2
. (6.291)
H = B . (6.292)
Using Eq. (4.38) for
z
one has
d
z
dt
=
1
i
[
z
, H]
=
2
i
B
x
[S
z
, S
x
] +B
y
[S
z
, S
y
]
=
2
B
y
S
x
B
x
S
y
= ( B) z .
(6.293)
Similar expressions are obtained for
x
and
y
that together can be
written in a vector form as
d
dt
(t) = (t) B(t) . (6.294)
18. Using Eq. (6.137), which is given by
( a) ( b) = a b +i (a b) , (6.295)
one has
_

_
p
q
c
A
__
2
=
_
p
q
c
A
_
2
+i ((p qA) (p qA))
=
_
p
q
c
A
_
2
i
q
c
(Ap +p A) .
(6.296)
The z component of the term (Ap +p A) can be expressed as
(Ap +p A) z = A
x
p
y
A
y
p
x
+p
x
A
y
p
y
A
x
= [A
x
, p
y
] [A
y
, p
x
] ,
(6.297)
thus, with the help of Eq. (3.76) one nds that
(Ap +p A) z = i
_
dA
x
dy

dA
y
dx
_
= i (A) z . (6.298)
Similar results can be obtained for the x and y components, thus
_

_
p
q
c
A
__
2
=
_
p
q
c
A
_
2
q
c
B . (6.299)
19. Since
j, m[ J
x
[j, m = j, m[ J
y
[j, m = 0 , (6.300)
and
J
2
x
+J
2
y
= J
2
J
2
z
, (6.301)
6.7. Solutions
one nds that
j, m[
_
(J
x
)
2
+ (J
y
)
2
_
[j, m = j, m[ J
2
[j, m j, m[ J
2
z
[j, m
=
2
_
j
2
+j m
2
_
.
(6.302)
20. The condition is
p A = A p , (6.303)
or
[p
x
, A
x
] + [p
y
, A
y
] + [p
z
, A
z
] = 0 , (6.304)
or using Eq. (3.76)
A
x
x
+
A
y
y
+
A
z
z
= 0 , (6.305)
or
A = 0 . (6.306)
21. The Hamiltonian is given by Eq. (1.62)
H =
_
p
q
c
A
_
2
2m
+q , (6.307)
thus, the coordinates representation of H[ = E[ is given by
r
[ H[ = Er
[ . (6.308)
Using the notation
r
[ = (r
) (6.309)
for the wavefunction together with Eqs. (3.23) and (3.29) one has
_
1
2m
_
i
q
c
A
_
2
+q
_
(r
) = E (r
) . (6.310)
H =
_
p
e
c
A
_
2
2m
=
p
2
x
+p
2
z
2m
+
_
p
y
eBx
c
_
2
2m
=
p
2
x
2m
+
1
2
m
2
c
_
x
cp
y
eB
_
2
+
p
2
z
2m
,
(6.311)
where
c
=
eB
mc
. (6.312)
Using the clue
y
y) exp(ik
z
z) (6.313)
one nds that the time independent Schrdinger equation for the wave
function (x) is thus given by
_
p
2
x
2m
+
1
2
m
2
c
_
x
ck
y
eB
_
2
_
(x) =
_
E

2
k
2
z
2m
_
(x) , (6.314)
where p
x
= i/x, thus the eigen energies are given by
E
n,k
=
c
_
n +
1
2
_
+

2
k
2
z
2m
, (6.315)
where n is integer and k is real, and the ground state energy is
E
n=0,k=0
=

c
2
. (6.316)
23. Using the gauge A = Bx y the Hamiltonian is given by [see Eq. (1.62)]
H =
_
p
q
c
A
_
2
2m
qEx
=
p
2
x
+p
2
z
2m
+
_
p
y
qBx
c
_
2
2m
qEx .
(6.317)
The last two terms can be written as
_
p
y
qBx
c
_
2
2m
qEx =
p
2
y
2m
+
1
2
m
2
c
_
(x x
0
)
2
x
2
0
_
, (6.318)
where
c
=
qB
mc
, (6.319)
and
x
0
=
mc
2
q
2
B
2
_
qE +
qp
y
mc
B
_
. (6.320)
Substituting the trial wavefunction
6.7. Solutions
y
y) exp(ik
z
z) , (6.321)
into the three dimensional Schrdinger equation yields a one dimensional
Schrdinger equation
_
p
2
x
2m
+
1
2
m
2
c
(x x
0
)
2
1
2
m
2
c
x
2
0
+

2
k
2
y
+
2
k
2
z
2m
_
(x) = E(x) ,
(6.322)
where p
x
= i/x and where
x
0
=
mc
2
q
2
B
2
_
qE +
qk
y
mc
B
_
. (6.323)
This equation describes a harmonic oscillator with a minimum potential
at x = x
0
, with added constant terms that give rise to a shift in the
energy level, which are thus given by
E
n,k
y
,k
z
=
c
_
n +
1
2
_
1
2
m
2
c
x
2
0
+

2
k
2
y
+
2
k
2
z
2m
=
c
_
n +
1
2
_
mc
2
E
2
2B
2

ck
y
E
B
+

2
k
2
z
2m
,
(6.324)
where n = 0, 1, 2, and where the momentum variables k
y
and k
z
can
take any real value.
24. The Schrdinger equation reads
_
_
p
e
c
A
_
2
2m
+U (y)
_
(x, y) = E (x, y) , (6.325)
where
p = i.
Employing the gauge A = By x one has
_
_
p
x
+
e
c
By
_
2
2m
+
p
2
y
2m
+U (y)
_
(x, y) = E (x, y) , (6.326)
where p
x
= i/x and p
y
= i/y. By substituting the trial wave-
function
(x, y) = exp(ikx) (y) , (6.327)
one obtains a one dimensional Schrdinger equation for (y)
_
p
2
y
2m
+
_
e
c
By +k
_
2
2m
+
1
2
m
2
0
y
2
_
(y) = E(y) , (6.328)
or
_
p
2
y
2m
+

2
k
2
2m
+
1
2
m
2
c0
y
2
eBk
mc
y
_
(y) = E(y) , (6.329)
where
2
c0

2
c
+
2
0
and
c
= [e[ B/mc. This can also be written as
_
p
2
y
2m
+
1
2
m
2
c0
_
y
eBk
m
2
c
2
c0
_
2
+

2
k
2
2m

2
0
2
c0
_
(y) = E(y) . (6.330)
This is basically a one-dimensional Schrdinger equation with a parabolic
potential of a harmonic oscillator and the eigenenergies are thus given
by:
E(n, k) =
c0
_
n +
1
2
_
+

2
k
2
2m

2
0
2
c0
,
where n = 0, 1, 2, and k is real.
25. It is convenient to choose a gauge having cylindrical symmetry, namely
A =
1
2
r B . (6.331)
For this gauge A = 0, thus according to Eq. (6.158) the Hamiltonian
is given by
H =
1
2
p
2
q
c
p A+
q
2
2c
2
A
2
. (6.332)
The Schrdinger equation in cylindrical coordinates (, z, ) is given by
(note that A = (B/2)
2
2
_
1
_
+
1
2
+

2
z
2
_
+
iqB
2c

+
q
2
2c
2
_
B
2
_
2
= E .
(6.333)
The particle is constrained to move along the ring, which is located at
z = 0 and = a, thus the eective one dimensional Schrdinger equation
of the system is given by

2
2a
2
2
+
iqB
2c

+
q
2
a
2
B
2
8c
2
= E . (6.334)
6.7. Solutions
a) Consider a solution of the form
() =
1
2a
exp(im) , (6.335)
where the pre factor (2a)
1/2
ensures normalization. The continu-
ity requirement that (2) = (0) implies that m must be an inte-
ger. Substituting this solution into the Schrdinger equation (6.334)
yields
E
m
=

2
m
2
2a
2

qBm
2c
+
q
2
a
2
B
2
8c
2
=

2
2a
2
_
m
2
qBa
2
c
m+
1
4
_
qBa
2
c
_
2
_
=

2
2a
2
_
m
qBa
2
2c
_
2
=

2
2a
2
_
m

0
_
2
,
(6.336)
where
= Ba
2
, (6.337)
is the magnetic ux threading the ring and
0
=
ch
q
. (6.338)
b) In general the current density is given by Eq. (4.181). For a wave-
function having the form
(r) = (r) e
i(r)
, (6.339)
where both and are real, one has
J =

Im[(() +(i))]
q
c
(A)
=

2
()
q
c
2
A
=
[[
2
_
()
q
c
A
_
.
(6.340)
In the present case one has
A =
B
2
, (6.341)
=
m
a

, (6.342)
and the normalized wavefunctions are
m
() =
1
2a
exp(im) , (6.343)
thus
J
m
=
1
2a
_
m
a
q
c
aB
2
_
=

2a
2
_
m

0
_
. (6.344)
Note that the following holds
[J
m
[ =
c
q
E
m
. (6.345)
26. The Hamiltonian can be written as
H =
L
2
L
2
z
2I
xy
+
L
2
z
2I
z
=
L
2
2I
xy
+
_
1
2I
z

1
2I
xy
_
L
2
z
,
(6.346)
Thus the states [l, m (the standard eigenstates of L
2
and L
z
) are eigen-
states of H and the following holds
H[l, m = E
l,m
[l, m , (6.347)
where
E
l,m
=
2
_
l (l + 1)
2I
xy
+
_
1
2I
z

1
2I
xy
_
m
2
_
. (6.348)
Using the expression
Y
1
1
(, ) =
_
3
8
sine
i
, (6.349)
one nds that
sin cos =
_
2
3
_
Y
1
1
Y
1
1
_
, (6.350)
thus the normalized state at t = 0 can be written as
[(0) =
1
2
([1, 1 [1, 1) . (6.351)
Since E
1,1
= E
1,1
the state [ (0) is stationary. Moreover
6.7. Solutions
(t)[ L
z
[ (t) = (0)[ L
z
[ (0)
=
1
2
((1, 1[ 1, 1[)) L
z
(([1, 1 [1, 1))
=
1
2
((1, 1[ 1, 1[)) (([1, 1 [1, 1))
= 0 .
(6.352)
27. With the help of the relations
L
x
=
L
+
+L
2
, (6.353)
L
+
[l, m =
_
l (l + 1) m(m+ 1) [l, m+ 1 , (6.354)
L
[l, m =
_
l (l + 1) m(m1) [l, m1 . (6.355)
one nds
a)
L
x
=

2
_
_
0 1 0
1 0 1
0 1 0
_
_
. (6.356)
b)
L
x
=

2
_
1
2
1
2
1
2
_
_
_
0 1 0
1 0 1
0 1 0
_
_
_
_
1
2
1
2
1
2
_
_
= . (6.357)
c)
L
x
=

2
_
2
0
1
2
_
_
_
0 1 0
1 0 1
0 1 0
_
_
_
_
2
0
1
2
_
_
= 0 . (6.358)
d)
D
z
() = exp
_
iL
z
_
=
_
_
exp(i) 0 0
0 1 0
0 0 exp(i)
_
_
. (6.359)
e) In general
D
n
(d) = exp
_
i (d) L n
_
= 1
i (d) L n
+O
_
(d)
2
_
,
(6.360)
thus
D
x
(d) =
_
_
_
_
1
i(d)
2
0
i(d)
2
1
i(d)
2
0
i(d)
2
1
_
_
_
_
+O
_
(d)
2
_
. (6.361)
28. Using
L
z
= xp
y
yp
x
, (6.362)
x =
_

2m
_
a
x
+a
x
_
, (6.363)
y =
_

2m
_
a
y
+a
y
_
, (6.364)
p
x
= i
_
m
2
_
a
x
+a
x
_
, (6.365)
p
y
= i
_
m
2
_
a
y
+a
y
_
, (6.366)
one nds
L
z
=
i
2
__
a
x
+a
x
_ _
a
y
+a
y
_
_
a
y
+a
y
_ _
a
x
+a
x
_
= i
_
a
x
a
y
a
x
a
y
_
.
(6.367)
a) Thus
L
z
= i
_
y
_
. (6.368)
b) Using the commutation relations
_
a
x
, a
= 1 , (6.369)
_
a
y
, a
= 1 , (6.370)
one nds
L
2
z
_
=
2
x
,
y
,
z
[
_
a
x
x
a
y
_ _
x
a
y
a
y
_
[
x
,
y
,
z
=
2
_
[
x
[
2
_
1 +[
y
[
2
_
+[
y
[
2
_
1 +[
x
[
2
_
y
_
2
(
x
a
y
)
2
_
,
(6.371)
thus
(L
z
)
2
=
2
_
[
x
[
2
_
1 +[
y
[
2
_
+[
y
[
2
_
1 +[
x
[
2
_
+
_
y
_
2
y
_
2
(
x
a
y
)
2
_
=
2
_
[
x
[
2
_
1 +[
y
[
2
_
+[
y
[
2
_
1 +[
x
[
2
_
2 [
x
[
2
[
y
[
2
_
=
2
_
[
x
[
2
+[
y
[
2
_
,
(6.372)
and
L
z
=
_
[
x
[
2
+[
y
[
2
. (6.373)
6.7. Solutions
29. The normalization constant can be chosen to be A = 1/
2. In general:
L
x
=
L
+
+L
2
, (6.374)
L
+
[l, m =
_
l (l + 1) m(m+ 1) [l, m+ 1 , (6.375)
L
[l, m =
_
l (l + 1) m(m1) [l, m1 . (6.376)
L
x
[ =
(L
[1, 1 L
+
[1, 1)
2
2
=
([1, 0 [1, 0)
2
= 0 ,
(6.377)
thus
L
x
= 0 . (6.378)
b) Using L
x
[ = 0 one nds
_
(L
x
)
2
_
=

L
2
x
_
L
x
2
= 0 0 = 0 . (6.379)
30. The Hamiltonian can be expressed as
H =
L
2
2I
1
+
L
2
z
2I
2

L
2
z
2I
1
=
L
2
2I
1
+
L
2
z
2I
e
, (6.380)
where
I
e
=
I
1
I
2
I
1
I
2
. (6.381)
Thus, the angular momentum states [l, m, which satisfy
L
2
[l, m = l (l + 1)
2
[l, m , (6.382)
L
z
[l, m = m [l, m , (6.383)
are eigenvector of H
H[l, m = E
l,m
[l, m , (6.384)
where
E
l,m
=
l (l + 1)
2
2I
1
+
m
2
2
2I
e
=

2
2I
1
_
l (l + 1) m
2
+m
2
I
1
I
2
_
. (6.385)
a) Since [H, L
z
] = 0 one has
exp
_
iL
z
_
Hexp
_
iL
z
_
= H , (6.386)
thus for the ground state l = m = 0
A
z
() =
0
[ H[
0
= E
0,0
= 0 . (6.387)
b) The operator L
x
can be expressed as
L
x
=
L
+
+L
2
. (6.388)
In general
L
+
[l, m =
_
l (l + 1) m(m+ 1) [l, m+ 1 , (6.389)
L
[l, m =
_
l (l + 1) m(m1) [l, m1 , (6.390)
thus
L
+
[0, 0 = L
[0, 0 = 0 , (6.391)
and consequently
exp
_
iL
x
_
[
0
= [
0
, (6.392)
thus
A
x
() =
0
[ H[
0
= E
0,0
= 0 . (6.393)
31. The wavefunction of a point particle is given by
(r) = (x +y + 2z) f (r) , (6.394)
where f (r) is a function of the radial coordinate r =
_
x
2
+y
2
+z
2
. As
can be see from Eqs. (6.131) and (6.132), which are given by
Y
1
1
(, ) =
_
3
8
sine
i
, (6.395)
Y
0
1
(, ) =
_
3
4
cos . (6.396)
the following holds
x = r
_
2
3
_
Y
1
1
+Y
1
1
_
(6.397)
y = ir
_
2
3
_
Y
1
1
+Y
1
1
_
(6.398)
z = r
_
4
3
Y
0
1
. (6.399)
and thus
(r) = 2
_
3
_
1 +i
2
Y
1
1
+
1 +i
2
Y
1
1
+ 2Y
0
1
_
rf (r) . (6.400)
a) In a measurement of L
2
the only possible outcome is 2
2
.
6.7. Solutions
b) In a measurement of L
z
the outcome and have both probability
1/6, whereas the outcome 0 has probability 2/3.
32. The notation [
1
,
2
is used to label the common eigenvectors of the
operator S
1z
, S
2z
, S
2
1
and S
2
2
, where
1
+, and
2
+, . The
following holds [see Eqs. (6.69) and (6.76)]
S
1z
[
1
,
2
=
1
2
[
1
,
2
, (6.401)
S
2z
[
1
,
2
=
2
2
[
1
,
2
, (6.402)
and
S
2
1
[
1
,
2
= S
2
2
[
1
,
2
=
3
2
4
[
1
,
2
. (6.403)
S
2
= S
2
1
+S
2
2
+S
1
S
2
+S
2
S
1
. (6.404)
Any operator of the rst particle commutes with any operator of the
second one thus
S
2
= S
2
1
+S
2
2
+ 2S
1
S
2
= S
2
1
+S
2
2
+ 2 (S
1x
S
2x
+S
1y
S
2y
+S
1z
S
2z
) .
(6.405)
In terms of the operators S
1
and S
2
, which are related to S
1x
, S
2x
,
S
1y
and S
2y
by
S
1x
=
S
1+
+S
1
2
, S
1y
=
S
1+
S
1
2i
, (6.406)
S
2x
=
S
2+
+S
2
2
, S
2y
=
S
2+
S
2
2i
, (6.407)
S
2
is given by
S
2
= S
2
1
+S
2
2
+S
1+
S
2
+S
1
S
2+
+ 2S
1z
S
2z
. (6.408)
With the help Eqs. (6.24) and (6.41) one nds that
_
S
2
, S
z
=
_
S
2
1
+S
2
2
+S
1+
S
2
+S
1
S
2+
+ 2S
1z
S
2z
, S
1z
+S
2z
= [S
1+
S
2
+S
1
S
2+
, S
1z
+S
2z
]
= [S
1+
, S
1z
] S
2
+ [S
1
, S
1z
] S
2+
+S
1+
[S
2
, S
2z
] +S
1
[S
2+
, S
2z
]
= (S
1+
S
2
+S
1
S
2+
+S
1+
S
2
S
1
S
2+
) ,
(6.409)
thus
_
S
2
, S
z
= 0 . (6.410)
b) The following holds [see Eqs. (6.71) and (6.72)]
S
2
_
_
_
_
[+, +
[+,
[, +
[,
_
_
_
_
=
2
_
_
_
_
2 0 0 0
0 1 1 0
0 1 1 0
0 0 0 2
_
_
_
_
_
_
_
_
[+, +
[+,
[, +
[,
_
_
_
_
, (6.411)
and
S
z
_
_
_
_
[+, +
[+,
[, +
[,
_
_
_
_
=
_
_
_
_
1 0 0 0
0 0 0 0
0 0 0 0
0 0 0 1
_
_
_
_
_
_
_
_
[+, +
[+,
[, +
[,
_
_
_
_
. (6.412)
It is thus easy to show that the following set of 4 ket vectors
[S = 0, M = 0 =
[+, [, +
2
, (6.413)
[S = 1, M = 1 = [+, + , (6.414)
[S = 1, M = 0 =
[+, +[, +
2
, (6.415)
[S = 1, M = 1 = [, , (6.416)
forms the desired complete and orthonormal basis of common eigen-
vectors of S
2
and S
z
, and the following holds
S
2
[S, M = S (S + 1)
2
[S, M , (6.417)
S
z
[S, M = M [S, M . (6.418)
7. Central Potential
Consider a particle having mass m in a central potential, namely a potential
V (r) that depends only on the distance
r =
_
x
2
+y
2
+z
2
(7.1)
from the origin. The Hamiltonian is given by
H =
p
2
2m
+V (r) . (7.2)
[H, L
z
] = 0 , (7.3)
_
H, L
2
= 0 . (7.4)
Solution 7.0.1. Using
[x
i
, p
j
] = i
ij
, (7.5)
L
z
= xp
y
yp
x
, (7.6)
one has
_
p
2
, L
z
=
_
p
2
x
, L
z
+
_
p
2
y
, L
z
+
_
p
2
z
, L
z
=
_
p
2
x
, xp
y
_
p
2
y
, yp
x
= i (2p
x
p
y
+ 2p
y
p
x
)
= 0 ,
(7.7)
and
_
r
2
, L
z
=
_
x
2
, L
z
+
_
y
2
, L
z
+
_
z
2
, L
z
= y
_
x
2
, p
x
+
_
y
2
, p
y
x
= 0 .
(7.8)
Thus L
z
commutes with any smooth function of r
2
, and consequently
[H, L
z
] = 0. In a similar way one can show that [H, L
x
] = [H, L
y
] = 0,
and therefore
_
H, L
2
= 0.
Chapter 7. Central Potential
In classical physics the corresponding Poissons brackets relations hold
H, L
x
= H, L
y
= H, L
z
= 0 , (7.9)
and
_
H, L
2
_
= 0 . (7.10)
These relations imply that classically the angular momentum is a constant of
the motion [see Eq. (1.40)]. On the other in quantum mechanics, as we have
seen in section 2.12 of chapter 2, the commutation relations
[H, L
z
] = 0 , (7.11)
_
H, L
2
= 0 , (7.12)
imply that it is possible to nd a basis for the vector space made of common
eigenvectors of the operators H, L
2
and L
z
.
7.1 Simultaneous Diagonalization of the Operators H,
L
2
and L
z
We start by proving some useful relations:
L
2
= r
2
p
2
(r p)
2
+ir p . (7.13)
Solution 7.1.1. The following holds
L
2
z
= (xp
y
yp
x
)
2
= x
2
p
2
y
+y
2
p
2
x
xp
y
yp
x
yp
x
xp
y
= x
2
p
2
y
+y
2
p
2
x
xp
x
([p
y
, y] +yp
y
) yp
y
([p
x
, x] +xp
x
)
= x
2
p
2
y
+y
2
p
2
x
xp
x
yp
y
yp
y
xp
x
+i (xp
x
+yp
y
) .
(7.14)
Using the relation
xp
x
xp
x
= x([p
x
, x] +xp
x
) p
x
= ixp
x
+x
2
p
2
x
, (7.15)
or
ixp
x
= x
2
p
2
x
xp
x
xp
x
, (7.16)
one has
7.1. Simultaneous Diagonalization of the Operators H, L
2
and L
z
L
2
z
= x
2
p
2
y
+y
2
p
2
x
xp
x
yp
y
yp
y
xp
x
+
i
2
(xp
x
+yp
y
)
+
1
2
_
x
2
p
2
x
xp
x
xp
x
+y
2
p
2
y
yp
y
yp
y
_
.
(7.17)
By cyclic permutation one obtains similar expression for L
2
x
and for L
2
y
. Com-
bining these expressions lead to
L
2
= L
2
x
+L
2
y
+L
2
z
= y
2
p
2
z
+z
2
p
2
y
yp
y
zp
z
zp
z
yp
y
+
i
2
(yp
y
+zp
z
) +
1
2
_
y
2
p
2
y
yp
y
yp
y
+z
2
p
2
z
zp
z
zp
z
_
+z
2
p
2
x
+x
2
p
2
z
zp
z
xp
x
xp
x
zp
z
+
i
2
(zp
z
+xp
x
) +
1
2
_
z
2
p
2
z
zp
z
zp
z
+x
2
p
2
x
xp
x
xp
x
_
+x
2
p
2
y
+y
2
p
2
x
xp
x
yp
y
yp
y
xp
x
+
i
2
(xp
x
+yp
y
) +
1
2
_
x
2
p
2
x
xp
x
xp
x
+y
2
p
2
y
yp
y
yp
y
_
=
_
x
2
+y
2
+z
2
_ _
p
2
x
+p
2
y
+p
2
z
_
(xp
x
+yp
y
+zp
z
)
2
+i(xp
x
+yp
y
+zp
z
)
= r
2
p
2
(r p)
2
+ir p .
(7.18)
r
[ p
2
[ =
2
_
1
r
2
r
2
r
[
1
2
r
2
r
[ L
2
[
_
. (7.19)
Solution 7.1.2. Using the identities
L
2
= r
2
p
2
(r p)
2
+ir p , (7.20)
r
[ r [ = r
[ , (7.21)
and
r
[ p[ =

i
r
[ , (7.22)
one nds that
r
[ L
2
[ = r
[ r
2
p
2
[ r
[ (r p)
2
[ +i r
[ r p[ . (7.23)
The following hold
r
[ r p[ = ir
[ = ir
[ , (7.24)
r
[ (r p)
2
[ =
2
_
r
_
2
r
[
=
2
_
r
2

2
r
2
+r
_
r
[ ,
(7.25)
r
[ r
2
p
2
[ = r
2
r
[ p
2
[ , (7.26)
thus
r
[ p
2
[ =
2
__

2
r
2
+
2
r
_
r
[
1
2
r
2
r
[ L
2
[
_
, (7.27)
or
r
[ p
2
[ =
2
_
1
r
2
r
2
r
[
1
2
r
2
r
[ L
2
[
_
. (7.28)
The time-independent Schrdinger equation in the coordinates represen-
tation
r
[ H[ = Er
[ , (7.29)
where the Hamiltonian H is given by Eq. (7.2), can thus be written using the
above results as
r
[ H[ =

2
2m
_
1
r
2
r
2
r
[
1
2
r
2
r
[ L
2
[
_
+V (r
) r
[ . (7.30)
7.2 The Radial Equation
Consider a solution having the form
r
[ = (r
) = R(r
) Y
m
l
_
_
. (7.31)
r
[ L
2
[ =
2
l (l + 1) (r
) . (7.32)
Substituting into Eq. (7.30) yields an equation for R(r)
2
2m
_
1
r
d
2
dr
2
rR(r)
1
r
2
l (l + 1) R(r)
_
+V (r) R(r) = ER(r) . (7.33)
The above equation, which is called the radial equation, depends on the quan-
tum number l, however, it is independent on the quantum number m. The
dierent solutions for a given l are labeled using the index k
2
2m
_
1
r
d
2
dr
2
rR
kl
1
r
2
l (l + 1) R
kl
_
+V R
kl
= ER
kl
. (7.34)
It is convenient to introduce the function u
kl
(r), which is related to R
kl
(r)
by the following relation
R
kl
(r) =
1
r
u
kl
(r) . (7.35)
7.2. The Radial Equation
Substituting into Eq. (7.34) yields an equation for u
kl
(r)
_
2
2m
d
2
dr
2
+V
e
(r)
_
u
kl
(r) = E
kl
u
kl
(r) , (7.36)
where the eective potential V
e
(r) is given by
V
e
(r) =
l (l + 1)
2
2mr
2
+V (r) . (7.37)
The total wave function is thus given by
klm
(r) =
1
r
u
kl
(r) Y
m
l
(, ) . (7.38)
Since the spherical harmonic Y
m
l
(, ) is assumed to be normalized [see Eq.
(6.114)], to ensure that
klm
(r) is normalized we require that
1 =
_
0
drr
2
[R
kl
(r)[
2
=
_
0
dr [u
kl
(r)[
2
. (7.39)
In addition solutions with dierent k are expected to be orthogonal, thus
_
0
dru
l
(r) u
kl
(r) =
kk
. (7.40)
The wave functions
klm
(r) represent common eigenstates of the operators
H, L
z
and L
2
, which are denoted as [klm and which satisfy the following
relations
klm
(r
) = r
[klm , (7.41)
and
H[klm = E
kl
[klm , (7.42)
L
2
[klm = l (l + 1)
2
[klm , (7.43)
L
z
[klm = m [klm . (7.44)
The following claim reveals an important property of the radial wavefunc-
tion near the origin (r = 0):
Claim. If the potential energy V (r) does not diverge more rapidly than 1/r
near the origin then
lim
r0
u(r) = 0 . (7.45)
Proof. Consider the case where near the origin u(r) has a dominant power
term having the form r
s
(namely, all other terms are of order higher than s,
and thus become negligibly small for suciently small r). Substituting into
Eq. (7.36) and keeping only the dominant terms (of lowest order in r) lead
to
2
2m
s (s 1) r
s2
+
l (l + 1)
2
2m
r
s2
= 0 , (7.46)
thus s = l or s = l + 1. However, the solution s = l for l 1 must
be rejected since for this case the normalization condition (7.39) cannot be
satised as the integral diverges near r = 0. Moreover, also for l = 0 the
solution s = l must be rejected. For this case (r) 1/r near the origin,
however, such a solution contradicts Eq. (7.30), which can be written as
2
2m
2
(r) +V (r) (r) = E(r) . (7.47)
since
2
1
r
= 4 (r) . (7.48)
We thus conclude that only the solution s = l + 1 is acceptable, and conse-
quently lim
r0
u(r) = 0.
7.3 Hydrogen Atom
The hydrogen atom is made of two particles, an electron and a proton. It
is convenient to employ the center of mass coordinates system. As is shown
below, in this reference frame the two body problem is reduced into a central
potential problem of eectively a single particle.
Exercise 7.3.1. Consider two point particles having mass m
1
and m
2
re-
spectively. The potential energy V (r) depends only on the relative coordi-
nate r = r
1
r
2
. Show that the Hamiltonian of the system in the center of
mass frame is given by
H =
p
2
2
+V (r) , (7.49)
where the reduced mass is given by
=
m
1
m
2
m
1
+m
2
. (7.50)
7.3. Hydrogen Atom
Solution 7.3.1. The Lagrangian is given by
/ =
m
1
r
2
1
2
+
m
2
r
2
2
2
V (r
1
r
2
) . (7.51)
In terms of center of mass r
0
and relative r coordinates, which are given by
r
0
=
m
1
r
1
+m
2
r
2
m
1
+m
2
, (7.52)
r = r
1
r
2
, (7.53)
the Lagrangian is given by
/ =
m
1
_
r
0
+
m2
m1+m2
r
_
2
2
+
m
2
_
r
0
m1
m1+m2
r
_
2
2
V (r)
=
M r
2
0
2
+
r
2
2
V (r) ,
(7.54)
where the total mass M is given by
M = m
1
+m
2
, (7.55)
and the reduced mass by
=
m
1
m
2
2
+m
2
m
2
1
(m
1
+m
2
)
2
=
m
1
m
2
m
1
+m
2
. (7.56)
Note that the Euler Lagrange equation for the coordinate r
0
yields that r
0
= 0
(since the potential is independent on r
0
). In the center of mass frame r
0
= 0.
The momentum canonically conjugate to r is given by
p =
/
r
. (7.57)
Thus the Hamiltonian is given by
H = p r/ =
p
2
2
+V (r) . (7.58)
For the case of hydrogen atom the potential between the electron having
charge e and the proton having charge e is given by
V (r) =
e
2
r
. (7.59)
Since the protons mass m
p
is signicantly larger than the electrons mass
m
e
(m
p
1800m
e
) the reduced mass is very close to m
e
=
m
e
m
p
m
e
+m
p
m
e
. (7.60)
The radial equation (7.36) for the present case is given by
_
2
2
d
2
dr
2
+V
e
(r)
_
u
kl
(r) = E
kl
u
kl
(r) , (7.61)
where
V
e
(r) =
e
2
r
+
l (l + 1)
2
2r
2
. (7.62)
In terms of the dimensionless radial coordinate
=
r
a
0
, (7.63)
where
a
0
=

2
e
2
= 0.53 10
10
m , (7.64)
is the Bohrs radius, and in terms of the dimensionless parameter
kl
=
_
E
kl
E
I
, (7.65)
where
E
I
=
e
4
2
2
= 13.6 eV , (7.66)
is the ionization energy, the radial equation becomes
_
d
2
d
2
+V
l
() +
2
kl
_
u
kl
= 0 (7.67)
where
V
l
() =
2
+
l (l + 1)
2
. (7.68)
-10
-8
-6
-4
-2
0
2
4
6
8
10
1 2 3 4 5
The function V
l
() for l = 0 (solid line) and l = 1 (dashed line).
7.3. Hydrogen Atom
We seek solutions of Eq. (7.67) that represent bound states, for which E
kl
is negative, and thus
kl
is a nonvanishing real positive. In the limit
the potential V
l
() 0, and thus it becomes negligibly small in comparison
with
kl
[see Eq. (7.67)]. Therefore, in this limit the solution is expected to
be asymptotically proportional to e
kl
. To ensure that the solution is nor-
malizable the exponentially diverging solution e
+
kl
is excluded. Moreover,
as we have seen above, for small the solution is expected to be proportional
to
l+1
. Due to these considerations we express u
kl
(r) as
u
kl
(r) = y ()
l+1
e
kl
. (7.69)
Substituting into Eq. (7.67) yields an equation for the function y ()
_
d
2
d
2
+ 2
_
(l + 1)

kl
_
d
d
+
2 (1
kl
(l + 1))
_
y = 0 . (7.70)
Consider a power series expansion of the function y ()
y () =

q=0
c
q
q
. (7.71)
Substituting into Eq. (7.70) yields
q=0
q (q 1) c
q
q2
+ 2 (l + 1)

q=0
qc
q
q2
2
kl
q=0
qc
q
q1
+ 2 (1
kl
(l + 1))

q=0
c
q
q1
= 0 ,
(7.72)
thus
c
q
c
q1
=
2 [
kl
(q +l) 1]
q (q + 2l + 1)
. (7.73)
We argue below that for physically acceptable solutions y () must be a poly-
nomial function [i.e. the series (7.71) needs to be nite]. To see this note that
for large q Eq. (7.73) implies that
lim
q
c
q
c
q1
=
2
kl
q
. (7.74)
Similar recursion relation holds for the coecients of the power series expan-
sion of the function e
2
kl
e
2
kl
q=0
c
q
q
, (7.75)
where
c
q
=
(2
kl
)
q
q!
, (7.76)
thus
c
q
c
q1
=
2
kl
q
. (7.77)
This observation suggests that for large the function u
kl
asymptotically
becomes proportional to e
kl
. However, such an exponentially diverging so-

lution must be excluded since it cannot be normalized. Therefore, to avoid
such a discrepancy, we require that y () must be a polynomial function.
As can be see from Eq. (7.73), this requirement is satised provided that
kl
(q +l) 1 = 0 for some q. A polynomial function of order k 1 is ob-
tained when
kl
is taken to be given by
kl
=
1
k +l
, (7.78)
where k = 1, 2, 3, . With the help of Eq. (7.73) the polynomial function
can be evaluated. Some examples are given below
y
k=1,l=0
() = c
0
, (7.79)
y
k=1,l=1
() = c
0
, (7.80)
y
k=2,l=0
() = c
0
_
1

2
_
, (7.81)
y
k=2,l=1
() = c
0
_
1

6
_
. (7.82)
The coecient c
0
can be determined from the normalization condition.
As can be seen from Eqs. (7.65) and (7.78), all states having the same
sum k +l, which is denoted as
n = k +l , (7.83)
have the same energy. The index n is called the principle quantum number.
Due to this degeneracy, which is sometimes called accidental degeneracy, it
is common to label the states with the indices n, l and m, instead of k, l and
m. In such labeling the eigenenergies are given by
E
n
=
E
I
n
2
, (7.84)
where
n = 1, 2, . (7.85)
For a given n the quantum number l can take any of the possible values
l = 0, 1, 2, , n 1 , (7.86)
7.4. Problems
and the quantum number m can take any of the possible values
m = l, l + 1, , l 1, l . (7.87)
The level of degeneracy of the level E
n
is given by
g
n
= 2
n1
l=0
(2l + 1) = 2
_
2 (n 1) n
2
+n
_
= 2n
2
. (7.88)
Note that the factor of 2 is due to spin. The normalized radial wave functions
of the states with n = 1 and n = 2 are found to be given by
R
10
(r) = 2
_
1
a
0
_
3/2
e
r/a0
, (7.89)
R
20
(r) =
_
2
r
a
0
__
1
2a
0
_
3/2
e
r
2a
0
, (7.90)
R
21
(r) =
_
1
2a
0
_
3/2
r
3a
0
e
r
2a
0
. (7.91)
The wavefunction
n,l,m
(r) of an eigenstate with quantum numbers n, l and
m is given by
nlm
(r, , ) = R
nl
(r) Y
m
l
(, ) . (7.92)
While the index n labels the shell number, the index l labels the sub-shell.
In spectroscopy it is common to label dierent sub-shells with letters:
l = 0 s
l = 1 p
l = 2 d
l = 3 f
l = 4 g
7.4 Problems
1. Consider the wave function with quantum numbers n, l, and m of a
hydrogen atom
n,l,m
(r).
a) Show that the probability current in spherical coordinates r, , is
given by
J
n,l,m
(r) =

n,l,m
(r)
2
r sin

, (7.93)
where is the reduced mass and

is a unit vector orthogonal to z
and r.
b) Use the result of the previous section to show that the total angular
momentum expectation value is given by L = mz.
2. Show that the average electrostatic potential in the neighborhood of an
hydrogen atom in its ground state is given by
= e
_
1
a
0
+
1
r
_
exp
_
2r
a
0
_
, (7.94)
where a
0
is the Bohr radius.
3. An hydrogen atom is in its ground state. The distance r between the
electron and the proton is measured. Calculate the expectation value r
and the most probable value r
0
(at which the probability density obtains
a maximum).
4. Tritium, which is labeled as
3
H, is a radioactive isotope of hydrogen.
The nucleus of tritium contains 1 proton and two neutrons. An atom
of tritium is in its ground state, when the nucleus suddenly decays into
a Helium nucleus, with the emission of a fast electron, which leaves the
atom without perturbing the extra-nuclear electron. Find the probability
that the resulting He
+
ion will be left in:
a) 1s state.
b) 2s state.
c) a state with l ,= 0.
5. At time t = 0 an hydrogen atom is in the state
[(t = 0) = A([2, 1, 1 +[2, 1, 1) ,
where A is a normalization constant and where [n, l, m denotes the eigen-
state with quantum numbers n, l and m. Calculate the expectation value
x at time t.
6. Find the ground state energy E
0
of a particle having mass m in a central
potential V (r) given by
V (r) =
_
0 a r b
else
, (7.95)
where r =
_
x
2
+y
2
+z
2
.
7. Consider a particle having mass m in a 3D potential given by
V (r) = A (r a) , (7.96)
where r =
_
x
2
+y
2
+z
2
is the radial coordinate, the length a is a
constant and () is the delta function. For what range of values of the
constant A the particle has a bound state.
8. Consider a particle having mass m in a 3D central potential given by
V (r) =
_
U
0
r r
0
0 r > r
0
. (7.97)
7.5. Solutions
where r =
_
x
2
+y
2
+z
2
is the radial coordinate, U
0
is real and r
0
is
positive. For what range of values of the potential depth U
0
the particle
has a bound state.
9. A spinless point particle is in state [. The state vector [ is an eigen-
vector of the operators L
x
, L
y
and L
z
(the x, y and z components of
the angular momentum vector operator). What can be said about the
wavefunction (r
) of the state [?
10. Consider two (non-identical) particles having the same mass m moving
under the inuence of a potential U (r), which is given by
U (r) =
1
2
m
2
r
2
.
In addition, the particles interact with each other via a potential given
by
V (r
1
, r
2
) =
1
2
m
2
(r
1
r
2
)
2
,
where r
1
and r
2
are the (three dimensional) coordinate vectors of the rst
and second particle respectively. Find the eigenenergies of the system.
7.5 Solutions
1. In general the current density is given by Eq. (4.181). For a wavefunction
having the form
(r) = (r) e
i(r)
, (7.98)
where both and are real, one has
J =

Im[( +i)]
=

2

=
[[
2
.
(7.99)
a) The wavefunction
n,l,m
(r) is given by
nlm
(r, , ) = R
nl
(r) Y
m
l
(, ) = R
nl
(r) F
m
l
() e
im
, (7.100)
where both R
nl
and F
m
l
are real, thus
J
n,l,m
(r) =

n,l,m
(r)
(m) . (7.101)
In spherical coordinates one has
= x

x
+ y

y
+z

z
= r

r
+
1
r

1
r sin

,
(7.102)
thus
J
n,l,m
(r) =

n,l,m
(r)
2
r sin

. (7.103)
b) The contribution of the volume element d
3
r to the angular momen-
tum with respect to the origin is given by r J
n,l,m
(r) d
3
r. In
spherical coordinates the total angular momentum is given by
L =
_
rJ
n,l,m
(r) d
3
r = m
_

n,l,m
(r)
2
r sin
r d
3
r . (7.104)
By symmetry, only the component along z of r contributes, thus
L = mz . (7.105)
2. The charge density of the electron in the ground state is given by
= e
1,0,0
(r)
2
=
e
a
3
0
exp
_
2r
a
0
_
. (7.106)
The Poissons equation is given by
2
= 4 . (7.107)
To verify that the electrostatic potential given by Eq. (7.94) solves this
equation we calculate
2
=
1
r
d
2
dr
2
(r)
=
e
r
d
2
dr
2
__
r
a
0
+ 1
_
exp
_
2r
a
0
__
=
4e exp
_
2r
a0
_
a
3
0
= 4 .
(7.108)
Note also that
lim
r
(r) = 0 , (7.109)
as is required for a neutral atom.
7.5. Solutions
3. The radial wave function of the ground state is given by
R
10
(r) = 2
_
1
a
0
_
3/2
exp
_
r
a
0
_
(7.110)
thus the probability density of the variable r is given by
f (r) = [rR
10
(r)[
2
=
4
r
_
r
a
0
_
3
exp
_
2r
a
0
_
. (7.111)
Thus
r =
_

0
rf (r) dr = 4a
0
_

0
x
3
exp(2x) dx =
3
2
a
0
. (7.112)
The most probable value r
0
is found from the condition
0 =
df
dr
=
8r
0
a
4
0
exp
_
2r
0
a
0
_
(a
0
r
0
) , (7.113)
thus
r
0
= a
0
. (7.114)
4. The radial wave function of a hydrogen-like atom with a nucleus having
charge Ze is found by substituting e
2
by Ze
2
in Eqs. (7.89), (7.90) and
(7.91), namely
R
(Z)
10
(r) = 2
_
Z
a
0
_
3/2
e
Zr/a0
,
R
(Z)
20
(r) = (2 Zr/a
0
)
_
Z
2a
0
_
3/2
e
Zr
2a
0
,
R
(Z)
21
(r) =
_
Z
2a
0
_
3/2
Zr
3a
0
e
Zr
2a
0
.
The change in reduced mass is neglected. Therefore
a) For the 1s state
Pr (1s) =
_
_
_
0
drr
2
R
(Z=1)
10
R
(Z=2)
10
_
_
2
=
2
7
a
3
0
_
2a
3
0
_
2
3
3
= 0.702 .
b) For the 2s state
Pr (2s) =
_
_
_
0
drr
2
R
(Z=1)
10
R
(Z=2)
20
_
_
2
=
16
a
6
0
_
a
3
0
8
(2 3)
_
2
= 0.25 .
c) For this case the probability vanishes due to the orthogonality be-
tween spherical harmonics with dierent l.
5. The normalization constant is chosen to be A = 1/
2. Since both states

[2, 1, 1 and [2, 1, 1 have the same energy the state [ is stationary.
The following holds
nlm
(r, , ) = R
nl
(r) Y
m
l
(, ) , (7.115a)
R
21
(r) =
_
1
2a
0
_
3/2
r
3a
0
e
r
2a
0
, (7.115b)
Y
1
1
(, ) =
1
2
_
3
2
sine
i
, (7.115c)
Y
1
1
(, ) =
1
2
_
3
2
sine
i
, (7.115d)
x = r sin cos . (7.115e)
In general
n
[ x[nlm =
_
0
dr r
3
R
n
l
R
nl
1
_
1
d(cos )
2
_
0
d sin cos
_
Y
m
Y
m
l
.
(7.116)
thus
2, 1, 1[x[2, 1, 1
2
_
0
d cos = 0 , (7.117)
2, 1, 1[x[2, 1, 1
2
_
0
d cos = 0 , (7.118)
2, 1, 1[x[2, 1, 1
2
_
0
d cos e
2i
= 0 , (7.119)
2, 1, 1[x[2, 1, 1
2
_
0
d cos e
2i
= 0 , (7.120)
and therefore
x (t) = 0 . (7.121)
6. The radial equation is given by [see Eq. (7.36)]
_
2
2m
d
2
dr
2
+
l (l + 1)
2
2mr
2
+V (r)
_
u
k,l
(r) = E
k,l
u
k,l
(r) . (7.122)
7.5. Solutions
Since the centrifugal term l (l + 1)
2
/2mr
2
is non-negative the ground
state is obtained with l = 0. Thus the ground state energy is [see Eq.
(4.162)]
E
0
=

2
2
2m(b a)
2
. (7.123)
7. The radial equation is given by
_
2
2m
d
2
dr
2
+
l (l + 1)
2
2mr
2
+V (r)
_
u
k,l
(r) = E
k,l
u
k,l
(r) . (7.124)
The boundary conditions imposed upon u(r) by the potential are
u(0) = 0 , (7.125)
u
_
a
+
_
= u
_
a
_
(7.126)
du(a
+
)
dr

du(a
)
dr
=
2
a
0
u(a) . (7.127)
where
a
0
=

2
mA
. (7.128)
Since the centrifugal term l (l + 1)
2
/2mr
2
state is obtained with l = 0. We seek a solution for that case having the
form
u(r) =
_
sinh(r) r < a
sinh(a) exp((r a)) r > a
, (7.129)
where
=
2mE
. (7.130)
The condition (7.127) yields
sinh(a) cosh(a) =
2
a
0
sinh(a) , (7.131)
or
a
0
2
=
1
1 + coth(a)
.
A real solution exists only if
a
0
2
< a , (7.132)
or
A >

2
2ma
. (7.133)
8. The radial equation is given by
_
2
2m
d
2
dr
2
+
l (l + 1)
2
2mr
2
+V (r)
_
u
k,l
(r) = E
k,l
u
k,l
(r) . (7.134)
The boundary condition that is imposed upon u(r) at the origin is u(0) =
0. Since the centrifugal term l (l + 1)
2
/2mr
2
state is obtained with l = 0. For that case the solution in the range r r
0
has the form u(r) = sinkr, where k is related to the energy E by
h
2
k
2
2m
= E +U
0
. (7.135)
In the range r > r
0
the general solution has the form u(r) = Ae
r
+
Be
r
, where
h
2
2
2m
= E . (7.136)
A bound state can be obtained provided that E < 0 (to ensure that
is real) and B = 0 (to ensure that lim
r
u(r) = 0; it is assumed that
is non-negative). The requirements that both u(r) and du/dr [see Eq.
(4.122)] are continuous at r = r
0
yield (for the case B = 0)
sinkr
0
= Ae
r
0
, (7.137)
k
cos kr
0
= Ae
r
0
, (7.138)
thus the following must hold
tankr
0
=
k
. (7.139)
Since both k and are required to be nonnegative, the above condition
can be satised only if tankr
0
0, which implies that
kr
0
=
_
2m(E +U
0
)
h
2
r
0

2
. (7.140)
This together with the requirement that E < 0 yield
_
2mU
0
h
2
r
0

2
, (7.141)
or
U
0

2
h
2
8mr
2
0
. (7.142)
7.5. Solutions
9. The state vector [ is an eigenvector of the operators L
x
, L
y
, therefore it
is easy to see that it consequently must be an eigenvector of the operator
[L
x
, L
y
] with a zero eigenvalue. Thus, since [L
x
, L
y
] = iL
z
, one has
L
z
[ = 0. Similarly, one nds that L
x
[ = L
y
[ = 0. Therefore, [
is also an eigenvector of the operator L
2
= L
2
x
+ L
2
y
+ L
2
z
with a zero
eigenvalue. Therefore the wavefunction has the form
(r
) = R(r
) Y
m=0
l=0
_
_
=
R(r
4
, (7.143)
where the radial function R(r
) is an arbitrary normalized function.

10. The Lagrangian is given by
/ =
m
_
r
2
1
+ r
2
2
_
2

1
2
m
2
_
r
2
1
+r
2
2
_
1
2
m
2
(r
1
r
2
)
2
. (7.144)
In terms of center of mass r
0
and relative r coordinates, which are given
by
r
0
=
r
1
+r
2
2
, (7.145)
r = r
1
r
2
, (7.146)
the Lagrangian is given by
/ =
m
_
_
r
0
+
1
2
r
_
2
+
_
r
0
1
2
r
_
2
_
2
1
2
m
2
_
_
r
0
+
1
2
r
_
2
+
_
r
0
1
2
r
_
2
_
1
2
m
2
r
2
=
m
_
2 r
2
0
+
1
2
r
2
_
2

1
2
m
2
_
2r
2
0
+
1
2
r
2
_
1
2
m
2
r
2
=
M r
2
0
2

1
2
M
2
r
2
0
+
r
2
2

1
2
2
+ 2
2
_
r
2
,
(7.147)
where the total mass M is given by
M = 2m , (7.148)
and the reduced mass [see also Eq. (7.50)] by
=
m
2
. (7.149)
The Lagrangian / describes two decoupled three dimensional harmonic
oscillators. The rst, which is associated with the center of mass motion,
has mass M = 2m and angular resonance frequency , whereas the
second one, which is associated with the relative coordinate r, has mass
= m/2 and angular resonance frequency

2
+ 2
2
. The quantum
energy eigenvectors are denoted by [n
0x
, n
0y
, n
0z
, n
x
, n
y
, n
z
, where all
six quantum numbers n
0x
, n
0y
, n
0z
, n
x
, n
y
and n
z
are integers, and the
corresponding eigenenergies are given by
E
n0x,n0y,n0z,nx,ny,nz
=
_
3
2
+n
0x
+n
0y
+n
0z
_
+
_
2
+ 2
2
_
3
2
+n
x
+n
y
+n
z
_
.
(7.150)
8. Density Operator
Consider an ensemble of N identical copies of a quantum system. The en-
semble can be divided into subsets, where all systems belonging to the same
subset have the same state vector. Let Nw
i
be the number of systems having
state vector

(i)
_
, where
0 w
i
1 , (8.1)
and where
i
w
i
= 1 . (8.2)
The state vectors are all assumed to be normalized
_
(i)

(i)
_
= 1 . (8.3)
Consider a measurement of an observable A, having a set of eigenvalues
a
n
and corresponding set of eigenvectors [a
n
A[a
n
= a
n
[a
n
. (8.4)
The set of eigenvectors [a
n
is assumed to be orthonormal and complete
a
m
[a
n
=
nm
, (8.5)
n
[a
n
a
n
[ = 1 . (8.6)
Consider a measurement of the observable A done on a system that is ran-
domly chosen from the ensemble. The probability to choose a system having
state vector

(i)
_
is w
i
. Given that the state vector is

(i)
_
, the expectation
value of A is

(i)
(i)
_
[see Eq. (2.84)]. Thus, the expectation (average)
value of such a measurement done on a system that is randomly chosen from
the ensemble is given by
A =
i
w
i
_
(i)
(i)
_
=
n
w
i
a
n
(i)
_
2
a
n
. (8.7)
Chapter 8. Density Operator
Claim. The expectation value can be expressed as
A = Tr (A) , (8.8)
where
=

i
w
i
(i)
__
(i)
(8.9)
is the density operator.
Proof. Let [b
m
be an orthonormal and complete basis for the vector space
m
[b
m
b
m
[ = 1 . (8.10)
The following holds
A =

i
w
i
_
(i)
(i)
_
=

i
m
w
i
_
(i)
A[b
m
b
m
(i)
_
=

m
b
m
[
i
w
i
(i)
__
(i)
A[b
m
= Tr (A) ,
(8.11)
where
=

i
w
i
(i)
__
(i)
.
Below we discuss some basic properties of the density operator:
= .
Solution 8.0.1. Trivial by the denition (8.9).
Exercise 8.0.2. Show that Tr () = 1.
Solution 8.0.2. Using a complete orthonormal basis

m
[b
m
b
m
[ = 1 one
has
Tr () =

m
b
m
[
_
i
w
i
(i)
__
(i)
_
[b
m
=

i
w
i
_
(i)
m
[b
m
b
m
[
_
(i)
_
=

i
w
i
_
(i)

(i)
_
=

i
w
i
= 1 .
(8.12)
Exercise 8.0.3. Show that for any normalized state [ the following holds
0 [ [ 1 . (8.13)
Solution 8.0.3. Clearly, 0 [ [ since
[ [ =

i
w
i
(i)
__
(i)
[ =

i
w
i
(i)
[
2
0 . (8.14)
On the other hand, according to the Schwartz inequality [see Eq. (2.166)],
which is given by
[u [v[
_
u [u
_
v [v , (8.15)
one has
(i)
[

_
[
_
(i)

(i)
_
= 1 . (8.16)
Moreover,

i
w
i
= 1, thus
[ [ =

i
w
i
(i)
[
2
1 . (8.17)
Exercise 8.0.4. Show that Tr
_
2
_
1.
Solution 8.0.4. The fact that is Hermitian (i.e.,
= ) guaranties the
existence of a complete orthonormal basis [q
m
of eigenvectors of , which
satisfy
q
m
[q
m
=
mm
, (8.18)
m
[q
m
q
m
[ = 1 , (8.19)
and
[q
m
= q
m
[q
m
, (8.20)
where the eigenvalues q
m
are real. Using this basis one has
Tr
_
2
_
=
m
q
m
[
2
[q
m
=

m
q
2
m
. (8.21)
According to inequality (8.13)
0 q
m
= q
m
[ [q
m
1 , (8.22)
thus
Tr
_
2
_
=
m
q
2
m

_
m
q
m
_
2
= (Tr ())
2
= 1 . (8.23)
Denition 8.0.1. An ensemble is said to be pure if its density operator can
be expressed as
= [ [ .
Exercise 8.0.5. Show that Tr
_
2
_
= 1 i represents a pure ensemble.
Solution 8.0.5. (i) Assuming that represents a pure ensemble, one has
2
= , thus Tr
_
2
_
= Tr () = 1. (ii) Assume that Tr
_
2
_
= 1. Since
is Hermitian (i.e.,
= ), there is a complete orthonormal basis [q

m
of
eigenvectors of , such that
q
m
[q
m
=
mm
, (8.24)
m
[q
m
q
m
[ = 1 , (8.25)
and
[q
m
= q
m
[q
m
, (8.26)
where the eigenvalues q
m
are real. Moreover, according to inequality (8.22)
0 q
m
1 . (8.27)
For this basis the assumption Tr
_
2
_
= 1 yields
1 = Tr
_
2
_
=

m
q
2
m
. (8.28)
Moreover, also Tr () = 1, thus
1 =
m
q
m
. (8.29)
Both equalities can be simultaneously satised only if
q
m
=
_
1 m = m
0
0 m ,= m
0
. (8.30)
For this case = [q
m0
q
m0
[, thus represents a pure ensemble.
8.2. Quantum Statistical Mechanics
8.1 Time Evolution
Consider a density operator
(t) =

i
w
i
(i)
(t)
__
(i)
(t)
, (8.31)
where the state vectors
_
(i)
(t)
__
evolve in time according to
i
d
(i)
_
dt
= H
(i)
_
, (8.32)
i
d
(i)
dt
=
_
(i)
H , (8.33)
where H is the Hamiltonian. Taking the time derivative yields
d
dt
=
1
i
_
i
w
i
H
(i)
__
(i)
(t)
i
w
i
(i)
(t)
__
(i)
H
_
, (8.34)
thus
d
dt
=
1
i
[, H] . (8.35)
This result resembles the equation of motion (4.37) of an observable in the
Heisenberg representation, however, instead of a minus sign on the right hand
side, Eq. (4.37) has a plus sign.
Alternatively, the time evolution of the operator can be expressed in
terms of the time evolution operator u(t, t
0
), which relates the state vector
at time

(i)
(t
0
)
_
with its value

(i)
(t)
_
at time t [see Eq. (4.4)]
(i)
(t)
_
= u(t, t
0
)
(i)
(t
0
)
_
. (8.36)
With the help of this relation Eq. (8.31) becomes
(t) = u(t, t
0
) (t
0
) u
(t, t
0
) . (8.37)
8.2 Quantum Statistical Mechanics
Consider an ensemble of identical copies of a quantum system. Let H be the
Hamiltonian having a set of eigenenergies E
i
and a corresponding set of
eigenstates [i, which forms an orthonormal and complete basis
H[i = E
i
[i , (8.38)
i
[i i[ = 1 . (8.39)
Consider the case where the ensemble is assumed to be a canonical ensemble
in thermal equilibrium at temperature T. According to the laws of statistical
mechanics the probability w
i
to nd an arbitrary system in the ensemble in
a state vector [i having energy E
i
is given by
w
i
=
1
Z
e
Ei
, (8.40)
where = 1/k
B
T, k
B
is Boltzmanns constant, and where
Z =
i
e
E
i
(8.41)
is the partition function.
Exercise 8.2.1. Show that the density operator can be written as
=
e
H
Tr (e
H
)
. (8.42)
Solution 8.2.1. According to the denition (8.9) one has
=

i
w
i
[i i[ =
1
Z
i
e
E
i
[i i[ . (8.43)
Moreover, the following hold
Z =
i
e
Ei
=

i
i[ e
H
[i = Tr
_
e
H
_
, (8.44)
and
i
e
Ei
[i i[ =
i
e
H
[i i[ = e
H
i
[i i[ = e
H
, (8.45)
thus
=
e
H
Tr (e
H
)
. (8.46)
As will be demonstrated below [see Eq. (8.330)], the last result for can
also be obtained from the principle of maximum entropy.
8.3 Problems
1. Consider a spin 1/2 in a magnetic eld B = Bz and in thermal equilib-
rium at temperature T. Calculate S u , where S is the vector operator
of the angular momentum and where u is a unit vector, which can be
described using the angles and
u =(sin cos , sin sin, cos ) . (8.47)
8.3. Problems
2. A spin 1/2 particle is an eigenstate of the operator S
y
with eigenvalue
+/2.
a) Write the density operator in the basis of eigenvectors of the operator
S
z
.
b) Calculate
n
, where n is integer.
c) Calculate the density operator (in the same basis) of an ensemble of
particles, half of them in an eigenstate of S
y
with eigenvalue +/2,
and half of them in an eigenstate of S
y
with eigenvalue /2.
d) Calculate
n
for this case.
3. A spin 1/2 is at time t = 0 in an eigenstate of the operator S
= S
x
sin+
S
z
cos with an eigenvalue +/2, where is real and S
x
and S
z
are
the x and z components, respectively, of the angular momentum vector
operator. A magnetic eld B is applied in the x direction between time
t = 0 and time t = T.
a) The z component of the angular momentum is measured at time
t > T. Calculate the probability P
+
to measure the value /2.
b) Calculate the density operator of the spin at times t = T.
4. A spin 1/2 electron is put in a constant magnetic eld given by B =
Bz, where B is a constant. The system is in thermal equilibrium at
temperature T.
a) Calculate the correlation function
C
z
(t) = S
z
(t) S
z
(0) . (8.48)
b) Calculate the correlation function
C
x
(t) = S
x
(t) S
x
(0) . (8.49)
5. Consider a harmonic oscillator with frequency . Show that the variance
of the number operator N =
_
N
2
N
2
(where N = a
a ) is given
by
a) N = 0 for energy eigenstates.
b) N =
_
N for coherent states.
c) N =
_
N (N + 1) for thermal states.
6. Consider a harmonic oscillator having angular resonance frequency .
The oscillator is in thermal equilibrium at temperature T. Calculate the
expectation value

x
2
_
.
7. Consider a harmonic oscillator in thermal equilibrium at temperature T,
whose Hamiltonian is given by
H =
p
2
2m
+
m
2
x
2
2
. (8.50)
Show that the density operator is given by
=
_ _
d
2
[ [ P () , (8.51)
where [ is a coherent state, d
2
denotes innitesimal area in the
complex plane,
P () =
1
N
exp
_
[[
2
N
_
, (8.52)
and where N is the expectation value of the number operator N.
H =
p
2
2m
+
m
2
x
2
2
. (8.53)
Calculate the probability density f (x) of the random variable x.
9. An LC oscillator (see gure) made of a capacitor C in parallel with an
inductor L, is in thermal equilibrium at temperature T. The charge in
the capacitor q is being measured.
L C L C
a) Calculate the expectation value q of q.
b) Calculate the variance
_
(q)
2
_
.
10. Consider an observable A having a set of eigenvalues a
n
. Let P
n
be
a projector operator onto the eigensubspace corresponding to the eigen-
value a
n
. A given physical system is initially described by the density op-
erator
0
. A measurement of the observable A is then performed. What
is the density operator
1
of the system immediately after the measure-
ment?
11. A model that was proposed by von Neumann describes an indirect mea-
surement process of a given observable A. The observable A is assumed
to be a function of the degrees of freedom of a subsystem, which we refer
to as the measured system (MS). The indirect measurement is performed
by rst letting the MS to interact with a measuring device (MD), hav-
ing its own degrees of freedom, and then in the nal step, performing a
quantum measurement on the MD. The MS is assumed to initially be in
a pure state [ (i.e. its density operator is assumed to initially be given
by
0
= [ [). Let A be an observable operating on the Hilbert space
8.3. Problems
of the MS. The initial state of the MS can be expanded in the basis of
eigenvectors [a
n
of the observable A
[ =

n
c
n
[a
n
, (8.54)
where c
n
= a
n
[ and where
A[a
n
= a
n
[a
n
. (8.55)
For simplicity, the Hamiltonian of the MS is taken to be zero. The MD is
assumed to be a one dimensional free particle, whose Hamiltonian van-
ishes, and whose initial state is labeled by [
i
. The position wavefunction
(x
) = x
[
i
of this state is taken to be Gaussian having width x
0
(x
) =
1
1/4
x
1/2
0
exp
_
1
2
_
x
x
0
_
2
_
. (8.56)
The interaction between the MS and the MD is taken to be given by
V (t) = f (t) xA , (8.57)
where f (t) is assumed to have compact support with a peak near the
time of the measurement.
a) Express the vector state of the entire system [ (t) at time t in the
basis of states [p
[a
n
. This basis spans the Hilbert space of
the entire system (MS and MD). The state [p
[a
n
is both, an
eigenvector of A (with eigenvalue a
n
) and of the momentum p of the
MD (with eigenvalue p
).
b) In what follows the nal state of the system after the measurement
will be evaluated by taking the limit t . The outcome of the
measurement of the observable A, which is labeled by /, is deter-
mined by performing a measurement of the momentum variable p of
the MD. The outcome, which is labeled by T, is related to / by
/ =
T
p
i
, (8.58)
where
p
i
=
_

dt
f (t
) . (8.59)
Calculate the probability distribution g (/) of the random variable
/.
c) Consider another measurement that is performed after the entan-
glement between the MS and the MD has been fully created. The
additional measurement is associated with the observable B, which
is assumed to be a function of the degrees of freedom of the MS only.
Show that the expectation value

B of the observable B is given by
B =

n
a
n
[ B
R
[a
n
, (8.60)
where the operator
R
, which is called the reduced density operator,
is given by
R
=

n
,n
c
n
c
n
e
2
_
a
n
a
n
2
_
2
[a
n
a
n
[ . (8.61)
12. A particle having mass m moves in the xy plane under the inuence of
a two dimensional potential V (x, y), which is given by
V (x, y) =
m
2
2
_
x
2
+y
2
_
+m
2
xy , (8.62)
where both and are real constants. Calculate in thermal equilibrium
at temperature T the expectation values x,

x
2
_
.
13. Consider a harmonic oscillator having angular resonance frequency and
mass m. Calculate the correlation function G(t) =

x
(H)
(t) x
(H)
(0)
_
,
where x
(H)
(t) is the Heisenberg representation of the position operator,
for the cases where
a) the oscillator is in its ground state.
b) the oscillator is in thermal equilibrium at temperature T.
14. In general, the Wigner function of a point particle moving in one dimen-
sion is given by
W (x
, p
) =
1
2
_

exp
_
i
p
__
x
+
x
2
_
dx
, (8.63)
where is the density operator of the system, and where [x
represents
an eigenvector of the position operator x having eigenvalue x
, i.e. x[x
=
x
[x
. As can be seen from Eq. (4.196), the Wigner function is the inverse
Weyl transformation of the density operator divided by the factor of 2.
Consider the case of a point particle having mass m in a potential of a
harmonic oscillator having angular frequency . Calculate the Wigner
function W (x
, p
) for the case where the system is in a coherent sate

[.
15. Consider a point particle having mass m in a potential of a harmonic
oscillator having angular frequency . Calculate the Wigner function
W (x
, p
) for the case where the system is in thermal equilibrium at

temperature T.
8.3. Problems
16. Consider a point particle having mass m in a potential of a harmonic
oscillator having angular frequency . Calculate the Wigner function
W (x
, p
) for the case where the system is in the number state [n = 1.

17. The Wigner function of a point particle moving in one dimension is
given by Eq. (8.63). Show that the marginal distributions x
[ [x
and
p
[ [p
of the position x and momentum p observables, respectively, are

given by
x
[ [x
=
1
_

dp
W (x
, p
) , (8.64)
p
[ [p
=
1
_

dx
W (x
, p
) . (8.65)
18. Homodyne Tomography - Consider a point particle having mass m
in a potential of a harmonic oscillator having angular frequency . The
normalized homodyne observable X
with a real phase is dened by

X
=
a
e
i
+ae
i
2
, (8.66)
where a and a
are annihilation and creation operators [see Eqs. (5.9)

and (5.10)]. Let w
_
X
_
be the normalized probability density of the
observable X
. The technique of homodyne detection can be used to

measure w
_
X
_
for any given value of the phase . Show that the Wigner
function (8.63) can be extracted from the measured distributions w
_
X
_
for all values of .
H =
p
2
2m
+
m
2
x
2
2
. (8.67)
Calculate the matrix elements x
[ [x
of the density operator in the

basis of eigenvectors of the position operator x.
20. A point particle having mass m is conned by the three dimensional
potential
V (r) =
1
2
m
2
r
2
, (8.68)
where r =
_
x
2
+y
2
+z
2
and where is a real constant. Calculate x
and

x
2
_
in thermal equilibrium at temperature T.
21. The entropy is dened by
= Tr (log ) . (8.69)
Show that is time independent.
22. Consider the 2 2 matrix , which is given by
=
1
2
(1 +k ) , (8.70)
where k = (k
x
, k
y
, k
z
) is a three dimensional vector of complex numbers
and where = (
x
,
y
,
z
) is the Pauli matrix vector.
a) Under what conditions on k the matrix can represent a valid density
operator of a spin 1/2 particle?
b) Under what conditions on k the matrix can represent a valid density
operator of a spin 1/2 particle in a pure state?
c) Calculate the term Tr ( u ), where u is a unit vector, i.e. u u = 1.
23. The matrix representation in the basis of eigenvectors of S
z
of the density
operator of a spin 1/2 particle is given by
=
1
2
(1 +k ) , (8.71)
where k = (k
x
, k
y
, k
z
) is a three dimensional vector of real numbers, and
= (
x
,
y
,
z
) is the Pauli matrix vector. The entropy is dened by
= Tr (log ) . (8.72)
a) Calculate .
b) A measurement of S
z
is performed. Calculate the entropy after the
measurement.
24. The maximum entropy principle - The entropy is dened by
() = Tr ( log ) . (8.73)
Consider the case where the density matrix is assumed to satisfy a set of
contrarians, which are expressed as
g
l
() = 0 , (8.74)
where l = 0, 1, L. The functionals g
l
() maps the density operator
to a complex number, i,e. g
l
() (.
a) Find an expression for a density matrix that satisfy all these con-
strains, for which the entropy obtains a stationary point (maxi-
mum, minimum or a saddle point). Assume that the constrain l = 0
is the requirement that Tr () = 1, i.e. g
0
() can be taken to be given
by
g
0
() = Tr () 1 = 0 . (8.75)
Moreover, assume that the other constrains l = 1, L are the re-
quirements that the expectation values of the Hermitian operators
X
1
, X
2
, X
L
are the following real numbers A
1
, A
2
, ,A
L
respec-
tively, i.e. g
l
() for l 1 can be taken to be given by
g
l
() = Tr (X
l
) A
l
= 0 . (8.76)
8.3. Problems
b) Express for the case of a microcanonical ensemble, for which the
only required constrain is (8.75).
c) Express for the case of a canonical ensemble, for which in addition
to the constrain is (8.75) the expectation value of the Hamiltonian
H is required to have a given value, which is labeled by H.
d) Express for the case of a grandcanonical ensemble, for which in ad-
dition to the constrain is (8.75) the expectation values of the Hamil-
tonian H and of the operator N are required to have given values,
which are labeled by H and N respectively. The operator N,
which will be dened in chapter 16, is called the number of particles
operator.
25. Consider a point particle having mass m moving in one dimension under
the inuence of the potential V (x). Calculate the canonical partition
function Z
c
[see Eq. (8.331)] in the classical limit, i.e. in the limit of high
temperature.
26. Let be the density operator of a given system. The system is composed
of two subsystems, each having its own degrees of freedom, which are
labeled as 1 and 2 (e.g. a system of two particles). Let [n
1
1
([n
2
2
)
be an orthonormal basis spanning the Hilbert space of subsystem 1
(2). The set of vectors [n
1
, n
2
, where [n
1
, n
2
= [n
1
1
[n
2
2
, forms an
orthonormal basis spanning the Hilbert space of the combined system.
For a general operator O the partial trace over subsystem 1 is dened
by the following relation
Tr
1
(O)
n
1
1
n
1
[ O[n
1
1
. (8.77)
Similarly, the partial trace over subsystem 2 is dened by
Tr
2
(O)
n2
2
n
2
[ O[n
2
2
. (8.78)
The observable A
1
is a given Hermitian operator on the Hilbert space of
subsystem 1. Show that the expectation value of a measurement of A
1
that is performed on subsystem 1 is given by
A
1
= Tr
1
(
1
A
1
) . (8.79)
where the operator
1
, which is given by
1
= Tr
2
, (8.80)
is called the reduced density operator of subsystem 1.
27. Let be the density operator of a given system. The total entropy of the
system is given by
= Tr (log ) . (8.81)
As in the previous exercise, the system is composed of two subsystems,
which are labeled as 1 and 2. Let [n
1
1
([n
2
2
) be an orthonormal
basis spanning the Hilbert space of subsystem 1 (2). The set of vec-
tors [n
1
, n
2
, where [n
1
, n
2
= [n
1
1
[n
2
2
, forms an orthonormal basis
spanning the Hilbert space of the combined system. The reduced density
operators
1
and
2
of subsystems 1 and 2 respectively are giving by
1
=

n
2
2
n
2
[ [n
2
2
= Tr
2
, (8.82)
2
=

n1
1
n
1
[ [n
1
1
= Tr
1
, (8.83)
and the subsystems entropies
1
and
2
are given by
1
= Tr
1
(
1
log
1
) , (8.84)
2
= Tr
2
(
2
log
2
) . (8.85)
Show that
1
+
2
. (8.86)
28. Consider a spin 1/2 in a magnetic eld B = Bz and in thermal equilib-
rium at temperature T. Calculate the entropy , which is dened by
= Tr (log ) , (8.87)
where is the density operator of the system.
8.4 Solutions
H = S
z
, (8.88)
where
=
[e[ B
m
e
c
(8.89)
is the Larmor frequency. In the basis of the eigenvectors of S
z
S
z
[ =
2
[ , (8.90)
one has
H[ =
2
[ , (8.91)
8.4. Solutions
thus
=
e
H
Tr (e
H
)
=
e

2
[+ +[ +e

2
[ [
e

2
+e

2
,
(8.92)
where = 1/k
B
T, and therefore with the help of Eqs. (2.102) and (2.103),
which are given by
S
x
=

2
([+ [ +[ +[) , (8.93)
S
y
=

2
(i [+ [ +i [ +[) , (8.94)
one has
S
x
= Tr (S
x
) = 0 , (8.95)
S
y
= Tr (S
y
) = 0 , (8.96)
and with the help of Eq. (2.99), which is given by
S
z
=

2
([+ +[ [ [) , (8.97)
one has
S
z
= Tr (S
z
)
= Tr
_
e

2
[+ +[ +e

2
[ [
e

2
+e

2
2
([+ +[ [ [)
_
=

2
e

2
e

2
e

2
+e

2
=
2
tanh
_
2
_
,
(8.98)
thus
S u =
cos
2
tanh
_
2
_
. (8.99)
2. Recall that
[; y =
1
2
([+ i [) , (8.100)
a) thus
=
1
2
_
1
i
_
_
1 i
_
=
1
2
_
1 i
i 1
_
. (8.101)
b) For a pure state
n
= .
c) For this case
=
1
2
_
_
[+; y +; y[ +[; y ; y[
. .
=1
_
_
=
1
2
_
1 0
0 1
_
, (8.102)
d) and
n
=
1
2
n
_
1 0
0 1
_
. (8.103)
3. The state at time t = 0 is given by
[(t = 0) =
_
cos

2
sin

2
_
, (8.104)
and the one at time t = T is
[(t = T) = exp
_
iT
x
2
_
[ (t = 0) , (8.105)
where
x
is a Pauli matrix, and
=
eB
m
e
c
. (8.106)
Using the identity
exp
_
i n
2
_
= cos

2
i nsin

2
, (8.107)
one nds
exp
_
i n
2
_
= cos
T
2
+i
x
sin
T
2
=
_
cos
T
2
i sin
T
2
i sin
T
2
cos
T
2
_
, (8.108)
thus
[(t = T) =
_
cos
T
2
i sin
T
2
i sin
T
2
cos
T
2
__
cos

2
sin

2
_
=
_
cos
T
2
cos

2
+i sin
T
2
sin

2
i sin
T
2
cos

2
+ cos
T
2
sin

2
_
.
(8.109)
a) The probabilities to measured /2 are thus given by
P
+
= cos
2
T
2
cos
2

2
+ sin
2
T
2
sin
2

2
=
1 + cos (T) cos
2
, (8.110)
8.4. Solutions
and
P
= cos
2
T
2
sin
2
2
+ sin
2
T
2
cos
2
2
=
1 cos (T) cos
2
. (8.111)
b) The density operator is given by
11
= P
+
,
22
= P
21
=
_
cos
T
2
cos

2
+i sin
T
2
sin

2
__
i sin
T
2
cos

2
+ cos
T
2
sin

2
_
=
sin
2

i
2
sinT cos ,
12
=
21
.
H = S
z
, (8.112)
where
=
eB
m
e
c
, (8.113)
thus, the density operator is given by
=
1
Z
_
_
exp
_

2k
B
T
_
0
0 exp
_

2kBT
_
_
_
, (8.114)
where
Z = exp
_

2k
B
T
_
+ exp
_

2k
B
T
_
. (8.115)
a) Using
S
z
(t) = exp
_
iHt
_
S
z
(0) exp
_
iHt
_
= S
z
(0) , (8.116)
one nds
C
z
(t) =

S
2
z
(0)
_
= Tr
_
S
2
z
(0)
_
=

2
4
. (8.117)
b) The following holds
S
x
(t) = exp
_
iS
z
t
_
S
x
(0) exp
_
iS
z
t
_
= S
x
cos (t) +S
y
sin(t) ,
(8.118)
thus
C
x
(t) = cos (t)
S
2
x
(0)
_
+ sin(t) S
y
(0) S
x
(0) (8.119)
=
cos (t)
2
4
+ sin(t) S
y
(0) S
x
(0) .
In terms of Pauli matrices
S
y
(0) S
x
(0) =

2
4Z
Tr
_
_
_
_
exp
_

2k
B
T
_
0
0 exp
_

2kBT
_
_
_
_
0 i
i 0
__
0 1
1 0
_
_
_
=

2
4Z
Tr
_
_
_
_
i exp
_

2kBT
_
0
0 i exp
_

2kBT
_
_
_
_
_
=
i
2
4
tanh
_

2k
B
T
_
,
(8.120)
thus
C
x
(t) =

2
4
_
cos (t) i sin(t) tanh
_

2k
B
T
__
. (8.121)
5. The variance N is given by
a) For an energy eigenstate [n one has
N [n = n[n , (8.122)
thus
N = n[ N [n = n , (8.123)
and
N
2
_
= n[ N
2
[n = n
2
, (8.124)
therefore
N = 0 . (8.125)
b) For a coherent state [ one has
a [ = [ , (8.126)
thus
N = [ N [ = [ a
a [ = [[
2
, (8.127)
and
8.4. Solutions
N
2
_
= [ a
aa
a[ = [ a
_
_
_
_
a, a
. .
=1
+a
a
_
_
_a [ = [[
2
+[[
4
,
(8.128)
therefore
N =
_
N . (8.129)
c) In general, for a thermal state one has
O = Tr (O) , (8.130)
where O is an operator,
=
1
Z
e
H
, (8.131)
Z = Tr
_
e
H
_
, (8.132)
and = 1/k
B
T and H is the Hamiltonian. For the present case
H =
_
N +
1
2
_
, (8.133)
thus
N = Tr (N)
=
n=0
n[ e
H
N [n
n=0
n[ e
H
[n
=
n=0
ne
n
n=0
e
n
=
1
log
_

n=0
e
n
_
=
e
1 e
,
(8.134)
and
N
2
_
= Tr
_
N
2
_
=
n=0
n[ e
H
N
2
[n
n=0
n[ e
H
[n
=
n=0
n
2
e
n
n=0
e
n
=
_
1
_
2

2
n=0
e
n
n=0
e
n
=
_
e
+ 1
_
e
(1 e
)
2
,
(8.135)
and therefore
(N)
2
=

N
2
_
N
2
=
e
(1 e
)
2
= N (N + 1) . (8.136)
6. The density operator is given by
=
1
Z
e
H
. (8.137)
where
Z = tr
_
e
H
_
=

n=0
e
(n+
1
2
)
=
e

2
1 e
=
1
2 sinh
_
2
_ , (8.138)
and = 1/k
B
T. Thus using
x
2
=

2m
_
a
2
+
_
a
_
2
+ 2a
a + 1
_
, (8.139)
one nds
8.4. Solutions
x
2
_
= Tr
_
x
2
_
=
1
Z
n=0
n[ x
2
e
H
[n
=
1
Z
n=0
e
(n+
1
2
)
n[ x
2
[n
=

m
1
Z
n=0
_
n +
1
2
_
e
(n+
1
2
)
=

m
1
Z
_
_
d
d
n=0
e
(n+
1
2
)
.
(8.140)
However
n=0
e
(n+
1
2
)
= Z , (8.141)
thus
x
2
_
=
1
m
2
d
d
log Z
1
=
1
m
2
d
d
sinh
_
2
_
sinh
_
2
_
=
1
m
2
2
coth
_
2
_
.
(8.142)
Note that at high temperatures 1
x
2
_
k
B
T
m
2
, (8.143)
as is required by the equipartition theorem of classical statistical mechan-
ics.
7. In the basis of number states the density operator is given by
=
e
H
Tr (e
H
)
=
n=0
e
H
[n n[
n=0
n[ e
H
[n
=
n=0
e
(N+
1
2
)
[n n[
n=0
n[ e
(N+
1
2
)
[n
=
n=0
e
n
[n n[
n=0
e
n
=
_
1 e
n=0
e
n
[n n[ ,
(8.144)
where = 1/k
B
T. Thus, N is given by
N = Tr (N)
=
_
1 e
n=0
ne
n
=
_
1 e
n=0
e
n
=
_
1 e
1
1 e
=
e
1 e
.
(8.145)
Moreover, the following holds
N + 1 =
1
1 e
, (8.146)
N
N + 1
= e
, (8.147)
thus, can be rewritten as
=
_
1 e
n=0
e
n
[n n[
=
1
N + 1
n=0
_
N
N + 1
_
n
[n n[ .
(8.148)
8.4. Solutions
To verify the validity of Eq. (8.51) we calculate
n[ [m =
_ _
d
2
P () n [ [m . (8.149)
[ = e
||
2
2
n=0
n!
[n , (8.150)
one nds that
n[ [m =
1
N
_ _
d
2
exp
_
[[
2
N
_
e
||
2
n
n!
m!
. (8.151)
Employing polar coordinates in the complex plane = re
i
, where r is
non-negative real and is real,
n[ [m =
1
N
n!m!
_
0
dre
(1+
1
N
)r
2
r
n+m+1
2
_
0
de
i(nm)
. .
2nm
=
2
nm
N n!
_
0
dre
(1+
1
N
)r
2
r
2n+1
,
(8.152)
and the transformation of the integration variable
t =
_
1 +
1
N
_
r
2
, (8.153)
dt =
_
1 +
1
N
_
2rdr , (8.154)
lead to
n[ [m =

nm
N
_
1 +
1
N
_
n+1
n!
_
0
dte
t
t
n
. .
(n+1)=n!
=

nm
N
_
1 +
1
N
_
n+1
=
N
n
nm
(1 +N)
n+1
,
(8.155)
in agreement with Eq. (8.148).
8. The density operator [see Eq. (8.51)] is given by
=
_ _
d
2
[ [ P () , (8.156)
2
complex plane,
P () =
1
N
exp
_
[[
2
N
_
, (8.157)
and where
N =
e
1 e
(8.158)
is the expectation value of the number operator N. Thus,
f (x
) = x
[ [x
=
_ _
d
2
P () x
[ [x
.
By employing the expression for the wave function
(x
) = x
[ of a
coherent state which is given by [see Eq. (5.51)]
(x
) = x
[
= exp
_
2
4
_
_
m
_
1/4
exp
_
_
x
2x
_
2
+i p
_
,
(8.159)
where
x
= [ x[ =
_
2
m
Re () , (8.160)
x
=
_
[ (x)
2
[ =
_

2m
, (8.161)
one nds that
8.4. Solutions
f (x
) = x
[ [x
=
1
N
_
m
_
1/2
_ _
d
2
exp
_
[[
2
N
_
exp
_
2
_
x
2x
_
2
_
=
1
N
_
m
_
1/2
_ _
d
2
exp
_
[[
2
N
_
exp
_
_
_2
_
_
x
_
2
m
Re ()
_
_
2
_
_
_
=
_
m
_
1/2
_
1 + 2 N
e
2
_
_
_
x
2
m
_
_
_
2
1+2N
=
1
_
1
m
(1 + 2 N)
e
2
_
_
_
x
2
m
_
_
_
2
1+2N
=
1
_
x
_
2
,
where
=
_

m
(1 + 2 N) ,
and where
1 + 2 N = 1 + 2
e
1 e
= coth
_
2
_
. (8.162)
9. Recall that the LC circuit is a harmonic oscillator.
a) In terms of the annihilation and creation operators
a =
_
L
2
_
q +
ip
L
_
, (8.163)
a
=
_
L
2
_
q
ip
L
_
, (8.164)
one has
q =
_

2L
_
a +a
_
, (8.165)
H =
_
a
a +
1
2
_
. (8.166)
The density operator is given by
=
1
Z
e
H
, (8.167)
where
=
1
k
B
T
, (8.168)
and
Z = Tr e
H
=

n=0
e
(n+
1
2
)
=
e

2
1 e
=
1
2 sinh

2
, (8.169)
thus
q = Tr (q) =
1
Z
_

2L
n=0
n[
_
a +a
_
e
H
[n = 0 . (8.170)
b) Similarly
q
2
_
= Tr
_
q
2
_
=

2L
1
Z
n=0
n[
_
a +a
_
2
e
H
[n
=
1
L
2
1
Z
n=0
_
n +
1
2
_
e
(n+
1
2
)
=
1
L
2
1
Z
dZ
d
=
C
2
coth

2k
B
T
.
(8.171)
10. In general,
0
can be expressed as
0
=
i
w
i
(i)
__
(i)
, (8.172)
where 0 w
i
1,

i
w
i
= 1, and where

(i)

(i)
_
= 1. Assume rst
that the system is initially in the state

(i)
_
. The probability for this to
be the case is w
i
. In general, the possible results of a measurement of the
observable A are the eigenvalues a
n
. The probability p
n
to measure the
eigenvalue a
n
given that the system is initially in state

(i)
_
is given by
p
n
=
_
(i)
P
n
(i)
_
. (8.173)
After a measurement of A with an outcome a
n
the state vector collapses
onto the corresponding eigensubspace and becomes
8.4. Solutions
(i)
_
P
n
(i)
_
_
(i)
P
n
(i)
_
. (8.174)
Thus, given that the system is initially in state

(i)
_
the nal density
operator is given by
(i)
1
=

n
_
(i)
P
n
(i)
_
P
n
(i)
_
_
(i)
P
n
(i)
_
(i)
P
n
_
(i)
P
n
(i)
_
=

n
P
n
(i)
__
(i)
P
n
.
(8.175)
Averaging over all possible initial states thus yields
1
=

i
w
i
(i)
1
=

n
P
n
i
w
i
(i)
__
(i)
P
n
=

n
P
n
0
P
n
. (8.176)
11. Since [V (t) , V (t
)] = 0 the time evolution operator from initial time t

0
to time t is given by
u(t, t
0
) = exp
_
_
t
t0
dt
V (t
)
_
= exp
_
ip
i
A
x
_
,
(8.177)
where
p
i
=
_
t
t0
dt
f (t
) . (8.178)
While the initial state of the entire system at time t
0
is given by [ (t
0
) =
[
i
[, the nal state at time t is given by
[ (t) = u(t, t
0
) [ (t
0
)
=

n
c
n
J
n
[
i
[a
n
,
(8.179)
where the operator J
n
is given by
J
n
= exp
_
ip
i
a
n
x
_
. (8.180)
a) By introducing the identity operator
_
dp
[p
[ = 1
MD
on the
Hilbert space of the MD, where [p
are eigenvectors of the momen-

tum operator p, which is canonically conjugate to x, the state [ (t)
can be expressed as
[ (t) =

n
c
n
_
dp
[ J
n
[
i
[p
[a
n
. (8.181)
With the help of the general identity (3.76), which is given by
[p, A(x)] = i
dA
dx
, (8.182)
where A(x) is a function of the operator x, one nds that
pJ
n
[p
= ([p, J
n
] +J
n
p) [p
= (p
i
a
n
+p
) J
n
[p
,
(8.183)
thus the vector J
n
[p
is an eigenvector of p with eigenvalue p

i
a
n
+p
.
Moreover, note that this vector, which is labeled as [p
+p
i
a
n

J
n
[p
, is normalized, provided that [p
is normalized, since J
n
is
unitary. The momentum wavefunction (p
) = p
[
i
of the state
[
i
is related to the position wavefunction x
[
i
by a Fourier trans-
form [see Eq. (3.60)]
(p
) =
dx
ip
[
i
2
=
1
1/4
p
1/2
0
exp
_
1
2
_
p
p
0
_
2
_
,
(8.184)
where
p
0
=

x
0
. (8.185)
In terms of (p
) the state [ (t) thus can be expressed as

[ (t) =

n
c
n
_
dp
p
i
a
n
[
i
[p
[a
n
=

n
c
n
_
dp
(p
p
i
a
n
) [p
[a
n
.
(8.186)
8.4. Solutions
b) The probability distribution g (/) of the random variable / can be
calculated using Eq. (8.186)
g (/) = p
i
[(a
n
[ p
i
/[) [ (t)[
2
= p
i
[c
n
[
2
[(p
i
(/a
n
))[
2
=

1/2
[c
n
[
2
e
2
(Aa
n
)
2
,
(8.187)
where
=
p
i
p
0
=
x
0
_

t
0
dt
f (t
) . (8.188)
The expectation value of / is given by
/ =
d/
g (/
) /
=

n
[c
n
[
2

1/2
d/
e
(A
)
2
(/
+a
n
)
=

n
[c
n
[
2
a
n
.
(8.189)
c) The density operator of the entire system is taken to be given by
f
= [ () ()[ for this case. The additional measurement is
associated with the observable B, which is assumed to be a function
of the degrees of freedom of the MS only. This assumption allows
expressing the expectation value

B of the observable B as
B = Tr (B
f
)
=

n
_
dp
a
n
[ p
[ B
f
[p
[a
n

=

n
a
n
[ B
R
[a
n
,
(8.190)
where
R
, which is given by
R
=
_
dp
[
f
[p
, (8.191)
is called the reduced density operator. Note that
R
is an operator
on the Hilbert space of the MS. With the help of the expressions
[ () =

n
c
n
_
dp
(p
p
i
a
n
) [p
[a
n
, (8.192)
()[ =

n
_
dp
(p
p
i
a
n
) a
n
[ p
[ , (8.193)
one nds that
R
=

n
,n
c
n
c
_
dp
(p
p
i
a
n
)
(p
p
i
a
n
) [a
n
a
n
[ .
(8.194)
By employing the transformation of integration variable
x =
2p
p
i
(a
n
+a
n
)
2p
0
, (8.195)
and its inverse
p
= p
0
_
x +
p
i
p
0
a
n
+a
n
2
_
, (8.196)
one nds that
_
dp
(p
p
i
a
n
)
(p
p
i
a
n
) = e
2
_
a
n
a
n
2
_
2
, (8.197)
thus
R
=

n
,n
c
n
c
n
e
2
_
a
n
a
n
2
_
2
[a
n
a
n
[ . (8.198)
12. It is convenient to employ the coordinate transformation
x
=
x +y
2
, (8.199)
y
=
x y
2
. (8.200)
The Lagrangian of the system can be written using these coordinates [see
Eq. (9.173)] as
/ = /
+
+/
, (8.201)
where
8.4. Solutions
/
+
=
m x
2
2

m
2
2
(1 +) x
2
, (8.202)
and
/
=
m y
2
2

m
2
2
(1 ) y
2
. (8.203)
Thus, the system is composed of two decoupled harmonic oscillators with
angular resonance frequencies
1 + (for x
) and
1 (for y
). In
thermal equilibrium according to Eq. (8.142) one has
x
2
_
=

2m
1 +
coth
_
1 +
2
_
, (8.204)
y
2
_
=

2m
1
coth
_
1
2
_
, (8.205)
where = 1/k
B
T. Moreover, due to symmetry, the following holds
x
= y
= 0 , (8.206)
x
= 0 . (8.207)
With the help of the inverse transformation, which is given by
x =
x
+y
2
, (8.208)
y =
x
2
, (8.209)
one thus nds
x = 0 , (8.210)
and
x
2
_
=
1
2
x
2
+y
2
_
=

4m
_
_
coth
_
1+
2
_
1 +
+
coth
_
1
2
_
1
_
_
.
(8.211)
13. Using Eq. (5.138), which is given by
x
(H)
(t) = x
(H)
(0) cos (t) +
p
(H)
(0)
m
sin(t) , (8.212)
one nds that
G(t) = cos (t)
_
x
2
(H)
(0)
_
+
sin(t)
m
p
(H)
(0) x
(H)
(0)
_
. (8.213)
Using the relations
x =
_

2m
_
a +a
_
, (8.214)
p = i
_
m
2
_
a +a
_
, (8.215)
_
a, a
= 1 , (8.216)
one nds that
x
2
=

2m
_
a
2
+
_
a
_
2
+ 2a
a + 1
_
, (8.217)
px
m
= i

2m
_
a
2
+
_
a
_
2
1
_
. (8.218)
a) Thus, for the ground state [see Eqs. (5.28) and (5.29)]
G(t) =

2m
[cos (t) i sin(t)] =

2m
exp(it) . (8.219)
b) The density operator is given by Eq. (8.148)
=
1
N + 1
n=0
_
N
N + 1
_
n
[n n[ , (8.220)
where
N = Tr (N) =
e
1 e
, (8.221)
N = a
a, and where = 1/k

B
T. Using the fact that is diagonal
in the basis of number states one nds that

a
2
_
=
_
_
a
_
2
_
= 0.
Combining all these results leads to
G(t) =

2m
[(2 N + 1) cos (t) i sin(t)]
=

2m
_
coth

2
cos (t) i sin(t)
_
.
(8.222)
14. The wave function of the coherent state [ is given by Eq. (5.51)
(x
) = x
[
= exp
_
2
4
_
_
m
_
1/4
exp
_
_
x
2x
_
2
+i p
_
.
(8.223)
where
8.4. Solutions
x
= [ x[ =
_
2
m
, (8.224)
p
= [ p [ =
2m
, (8.225)
= Re () , (8.226)
= Im() , (8.227)
x
=
_
[ (x)
2
[ =
_

2m
, (8.228)
Using the denition (8.63) and the identity
exp
_
ax
2
+bx +c
_
dx =
_
a
e
1
4
4ca+b
2
a
, (8.229)
one has
W (x
, p
) =
1
2
_

exp
_
i
p
__
x
2
[
+
x
2
_
dx
=
_
m
_
1/2
2
_

dx
_
x
2
x
2x
_
2
_
x
+
x
2
x
2x
_
2
+i
_
p
_
x
=
_
m
_
1/2
2
_

dx
(x
)
2
+
(
x
2
)
2
2(x)
2
+i
_
p
_
x
,
(8.230)
thus
W (x
, p
) =
1
1
2
_
x
x
_
2
1
2
_
p
p
_
2
, (8.231)
where [see Eq. (5.49)]
p
=
_
[ (p)
2
[ =
_
m
2
=

2x
. (8.232)
=
_ _
d
2
[ [ P () , (8.233)
2
complex plane,
P () =
1
N
exp
_
[[
2
N
_
, (8.234)
and where
N =
e
1 e
(8.235)
is the expectation value of the number operator N. Thus
W (x
, p
) =
_ _
d
2
P () W
(x
, p
) . (8.236)
where
W
(x
, p
) =
1
2
_

exp
_
i
p
__
x
2
[
+
x
2
_
dx
,
(8.237)
which is the Wigner function of a coherent state [, was found to be
given by [see Eq. (8.231)]
W (x
, p
) =
1
1
2
_
x
x
_
2
1
2
_
p
p
_
2
, (8.238)
where
x
= [ x[ =
_
2
m
= 2x
, (8.239)
p
= [ p [ =
2m
= 2p
, (8.240)
= Re () , (8.241)
= Im() , (8.242)
x
=
_
[ (x)
2
[ =
_

2m
, (8.243)
p
=
_
[ (p)
2
[ =
_
m
2
=

2x
. (8.244)
Thus W (x
, p
) is given by
W (x
, p
) =
1
2
N
_ _
d
2
e
||
2
N
e
1
2
_
x
x
_
2
1
2
_
p
p
_
2
=
1
2
N
_
d
2
N
1
2
_
x
2x
x
_
2
_
d
2
N
1
2
_
p
2x
p
_
2
.
(8.245)
With the help of the identity (8.229) one thus nds that
8.4. Solutions
W (x
, p
) =
1
1
2 N + 1
e
1
2
1
2N+1
_
_
x
x
_
2
+
_
p
p
_
2
_
, (8.246)
where
2 N + 1 = 1 + 2
e
1 e
= coth
_
2
_
, (8.247)
and where = 1/k
B
T.
16. With the help of Eqs. (5.113) one nds that the wavefunction of the
number state [n = 1 is given by
n=1
(x
) = x
[n = 1 =
2
x
x
0
exp
_
x
2
2x
2
0
_
1/4
x
1/2
0
, (8.248)
where
x
0
=
_

m
. (8.249)
thus
W (x
, p
) =
1
e
i
p
2
x0
e
(
x
2
)
2
2x
2
0
x
+
x
2
x0
e
(
x
+
x
2
)
2
2x
2
0
1/2
x
0
dx
,
(8.250)
or
W (x
, p
) =
e
_
x
x
0
_
2
1/2
_

_
_
x
x
0
_
2
X
2
4
_
e
ip
p
0
X
X
2
4
dX . (8.251)
where X = x
/x
0
and where p
0
= /x
0
. The integration, which is per-
formed with the help of Eq. (8.229), yields
W (x
, p
) =
2
_
x
x
0
_
2
_
p
p
0
_
2
_
_
x
x
0
_
2
+
_
p
p
0
_
2
1
2
_
. (8.252)
Note that near the origin the Wigner function W (x
, p
) becomes nega-
tive.
17. The relation (8.64) is proven by
1
_

dp
W (x
, p
)
=
_

_
x
+
x
2
_
dx
1
2
_

dp
e
i
p
. .
(x
)
= x
[ [x
.
(8.253)
With the help of the identities (3.45) and (3.52) W (x
, p
) can be ex-
pressed as
W (x
, p
)
=
1
(2)
2
dx
dp
dp
e
i
p
+p
(
x
2
)
p
(
x
+
x
2
)
[ [p
=
1
2
_

dp
dp
e
i
x
(p
[ [p
1
2
_

dx
e
i
x
_
p
2

p
2
_
. .
_
p
2

p
2
_
.
(8.254)
The above result easily leads to (8.65)
1
_

dx
W (x
, p
)
=
_

dp
dp
[ [p

_
p
2

p
2
_
1
2
_

dx
e
i
x
(p
. .
(p
)
= p
[ [p
.
(8.255)
18. The normalized homodyne observable X
can be expressed in terms of

the dimensionless position and momentum operators X and P, which are
given by
X =
a +a
2
, P =
a a
2
, (8.256)
and which satisfy [X, P] = i [see Eq. (5.13)], as
X
=
X iP
2
e
i
+
X +iP
2
e
i
= X cos P sin .
(8.257)
The associated dimensionless momentum operator P
is dened as P
=
X sin +P cos , i.e.
_
X
_
=
_
cos sin
sin cos
__
X
P
_
, (8.258)
and the inverse transformation is given by
_
X
P
_
=
_
cos sin
sin cos
__
X
_
. (8.259)
8.4. Solutions
The marginal distributions Pr (X
) for the dimensionless position X and

Pr (P
) for the dimensionless momentum P are related to the Wigner

function (8.63) W (X
, P
) by [see Eqs. (8.64) and (8.65)]

Pr (X
) =
_

dP
W (X
, P
) , (8.260)
Pr (P
) =
_

dX
W (X
, P
) , (8.261)
where Pr (A
) denotes the probability density of the random variable

A
. Thus, the probability density Pr

_
X
_
of X
, which is denoted as
w
_
X
_
, is given by
w
_
X
_
=
_

dP
W
_
X
cos +P
sin, X
sin +P
cos
_
.
(8.262)
The last result (18) can be inverted using the so-called inverse Radon
transform. The Fourier transformed Wigner function

W (, ) is ex-
pressed as
W (, ) =
1
2
_

dX
dP
W (X
, P
) e
iX
iP
. (8.263)
The Fourier transform w() of the probability distribution w
_
X
_
is
given by [see Eq. (18)]
w() =
1
2
_

dX
w
_
X
_
e
iX
=
1
2
_

dX
dP
W
_
X
cos +P
sin, X
sin +P
cos
_
e
iX
.
(8.264)
The variable transformation (8.259) leads to
w() =
1
2
_

dX
dP
W (X
, P
) e
i(X
cos P
sin)
,
(8.265)
thus according to Eq. (8.263) the following holds
w() =
2

W ( cos , sin) . (8.266)
The last result (8.265) allows evaluating the Wigner function by inverse
Fourier transform
W (X
, P
) =
1
2
_

dd

W (, ) e
iX
+iP
. (8.267)
In polar coordinates this becomes
W (X
, P
) =
1
2
_

d
_

0
d [[

W ( cos , sin) e
i(X
cos +P
sin)
,
(8.268)
W (X
, P
) =
1
(2)
3/2
_

d
_

0
d [[ w() e
i(X
cos +P
sin)
,
(8.269)
thus [see Eq. (8.266)]
W (X
, P
) =
1
(2)
2
_

d
_

0
d
_

dX
[[ w
_
X
_
e
i(X
cos +P
sinx
)
.
(8.270)
=
_ _
d
2
[ [ P () , (8.271)
2
complex plane,
P () =
1
N
exp
_
[[
2
N
_
, (8.272)
and where
N =
e
1 e
(8.273)
is the expectation value of the number operator N. By employing the
expression for the wave function
(x
) = x
[ of a coherent state
which is given by [see Eq. (5.51)]
8.4. Solutions
(x
) = x
[
= exp
_
2
4
_
_
m
_
1/4
exp
_
_
x
2x
_
2
+i p
_
,
(8.274)
where
x
= [ x[ =
_
2
m
, (8.275)
p
= [ p [ =
2m
, (8.276)
= Re () , (8.277)
= Im() , (8.278)
x
=
_
[ (x)
2
[ =
_

2m
, (8.279)
one nds that
x
[ [x
=
_ _
d
2
P () x
[ [x
=
_
m
_
1/2
N
_ _
d
2
exp
_
[[
2
N
_
exp
_
_
x
2x
_
2
_
x
2x
_
2
+i p
(x
_
=
_
m
_
1/2
N
_ _
d
2
exp
_
2
+
2
N
_
exp
_
_
X
2
_
2
_
X
2
_
2
+i
(X
)
_
=
_
m
_
1/2
N
_
d
exp
_
2 N + 1
N

2
+ (X
+X
X
2
+X
2
4
_
_
d
exp
_
2
N
+i
(X
)
_
.
(8.280)
where
X
=
_
2m
, (8.281)
X
=
_
2m
. (8.282)
With the help of the identity
exp
_
ax
2
+bx +c
_
dx =
_
a
e
1
4
4ca+b
2
a
, (8.283)
one nds that
x
[ [x
=
_
m
_
1/2
_
1
2 N + 1
e
(
X
2
+X
2
)
N(X
)
2
+
N(X
+X
)
2
2N+1
4
,
(8.284)
or using the identity
X
2
+X
2
=
(X
+X
)
2
+ (X
)
2
2
, (8.285)
one has
x
[ [x
=
_
m
_
1/2
_
1
2 N + 1
e
_
X
+X
2
_
2
2(2N+1)

_
X
2
_
2
2N+1
2
.
(8.286)
In terms of x
and x
this result can be written as

x
[ [x
=
1
_
x
+x
2
_
2
(2N+1)
2
_
x
2
_
2
, (8.287)
where
=
_

m
(2 N + 1) , (8.288)
2 N + 1 = 1 + 2
e
1 e
= coth
_
2
_
, (8.289)
and where = 1/k
B
T. Alternatively, the result can be expressed as
x
[ [x
=
e
tanh(

2
)
_
x
+x
2x
0
_
2
coth(

2
)
_
x
2x
0
_
2
x
0
_
coth
_
2
_
, (8.290)
where
x
0
=
_

m
. (8.291)
8.4. Solutions
20. The dynamics along the x direction is governed by the Hamiltonian H
x
of a harmonic oscillator
H
x
=
p
2
x
2m
+
1
2
m
2
x
2
. (8.292)
By symmetry x = 0. The expectation value

x
2
_
was calculated in Eq.
(8.142) and found to be given by
x
2
_
=
1
m
2
2
coth
_
2
_
. (8.293)
21. In general, for any smooth function f () of the following holds
f ( (t)) = u(t, t
0
) f ( (t
0
)) u
(t, t
0
) . (8.294)
This can be shown by Taylor expanding the function f () as a power
series
f ( (t)) =

n=0
a
n
((t))
n
, (8.295)
using Eq. (8.37) and the fact that u
(t, t
0
) u(t, t
0
) = 1, i.e. the unitarity
of the time evolution operator. By using this result for the function log
together with the general identity Tr (XY ) = Tr (Y X) [see Eq. (2.129)]
one easily nds that is time independent. This somewhat surprising
result can be attributed to the fact that the unitary time evolution that is
governed by the Schrdinger equation is symmetric under time reversal.
In the language of statistical mechanics it corresponds to a reversible
process, for which entropy is preserved.
22. Using the denition of the Pauli matrices (6.136) one nds that
=
1
2
_
1 +k
z
k
x
ik
y
k
x
+ik
y
1 k
z
_
, (8.296)
and
2
=
1
4
_
1 + 2k
z
+k
2
2 (k
x
ik
y
)
2 (k
x
+ik
y
) 1 2k
z
+k
2
_
,
where k
2
= k
2
x
+k
2
y
+k
2
z
.
a) Note that for any k the following holds Tr () = 1. The requirement
that is Hermitian, i.e. the requirement that
= , implies that
k
z
= k
z
and k
x
k
x
+i
_
k
y
k
y
_
= 0, thus k
x
, k
y
and k
z
are all real.
Moreover, the requirement that Tr
_
2
_
=
_
1 +k
2
_
/2 1 implies
that k
2
1.
b) For this case Tr
_
2
_
= 1, thus k
2
= 1.
c) With the help of Eq. (6.137), which is given by
( a) ( b) = a b +i (a b) , (8.297)
and the fact that all three Pauli matrices have a vanishing trace, one
nds that
Tr ( u ) =
1
2
Tr ( u ) +
1
2
Tr (( u ) (k ))
=
1
2
Tr (( u ) (k ))
=
1
2
Tr ( u k) +
i
2
Tr ( ( u k))
=
1
2
Tr ( u k)
= u k .
(8.298)
23. Using the denition of the Pauli matrices (6.136) one nds that
=
1
2
_
1 +k
z
k
x
ik
y
k
x
+ik
y
1 k
z
_
. (8.299)
a) Let
be the two eigenvalues of . The following holds

Tr () =
+
+
= 1 , (8.300)
and
Det () =
+
=
_
1 k
2
_
/4 , (8.301)
where k
2
= k
2
x
+k
2
y
+k
2
z
. Thus
=
1 [k[
2
, (8.302)
and therefore
= f ([k[) , (8.303)
where
f (x) =
1 x
2
log
1 x
2

1 +x
2
log
1 +x
2
. (8.304)
8.4. Solutions
0.1
0.2
0.3
0.4
0.5
0.6
0.7
y
0.2 0.4 0.6 0.8 1
x
The function f (x) =
1x
2
log
1x
2

1+x
2
log
1+x
2
.
b) As can be seen from Eq. (8.176), after the measurement becomes
diagonal in the basis of eigenvectors of the measured observable,
namely, after the measurement the density matrix is given by
c
=
1
2
_
1 +k
z
0
0 1 k
z
_
, (8.305)
and thus the entropy after the measurement is
c
= f (k
z
) =
1 k
z
2
log
1 k
z
2

1 +k
z
2
log
1 +k
z
2
. (8.306)
24. First, consider a general functional g () of the density operator having
the form
g () = Tr (f ()) , (8.307)
where the function f () can be Taylor expanded as a power series
f () =

k=0
a
k
k
, (8.308)
and where a
k
are complex constants. Consider an innitesimal change in
the density operator + d. To rst order in d the corresponding
change dg in the functional g () can be expressed as
dg = g ( + d) g ()
= Tr
_

k=0
a
k
_
( + d)
k
k
_
_
= Tr
_
_

k=0
a
k
_
_
k1
d +
k2
(d) +
k3
(d)
2
+
. .
k terms
_
_
_
_
+O
_
(d)
2
_
.
(8.309)
By exploiting the general identity Tr (XY ) = Tr (Y X) the above result
can be simplied (note that generally needs not to commute with d)
dg = Tr
__

k=0
a
k
k
k1
_
d
_
+O
_
(d)
2
_
, (8.310)
thus to rst order in d the following holds
dg = Tr
_
df
d
d
_
. (8.311)
In the above expression the term df/d is calculated by simply taking the
derivative of the function f (x) (where x is considered to be a number)
and substituting x = . Alternatively, the change dg can be expressed in
terms of the innitesimal change d
nm
in the matrix elements
nm
of .
To rst order in the innitesimal variables d
nm
one has
dg =

n,m
g
nm
d
nm
. (8.312)
It is convenient to rewrite the above expression as
dg =

g d , (8.313)
where the vector elements of the nabla operator

and of d are given
by
_
_
n,m
=

nm
, (8.314)
and
_
d
_
n,m
= d
nm
. (8.315)
Thus, to rst order one has
d =

d , (8.316)
and
dg
l
=

g
l
d , (8.317)
where l = 0, 1, 2, ...L.
a) In general, the technique of Lagrange multipliers is very useful for
nding stationary points of a function, when constrains are applied.
A stationary point of occurs i for every small change d, which is
orthogonal to all vectors

g
0
,

g
1
,

g
2
, ...,

g
L
(i.e. a change which
does not violate the constrains) one has
8.4. Solutions
0 = d =

d . (8.318)
This condition is fullled only when the vector

belongs to the
subspace spanned by the vectors
_
g
0
,

g
1
,

g
2
, ...,

g
L
_
. In other
words, only when
=
0

g
0
+
1

g
1
+
2

g
2
+... +
L

g
L
, (8.319)
where the numbers
0
,
1
, ...,
L
, which are called Lagrange multipli-
ers, are constants. By multiplying by d the last result becomes
d =
0
dg
0
+
1
dg
1
+
2
dg
2
+... +
L
dg
L
. (8.320)
Using Eqs. (8.311), (8.73, (8.75) and (8.76) one nds that
d = Tr ((1 + log ) d) , (8.321)
dg
0
= Tr (d) , (8.322)
dg
l
= Tr (X
l
d) , (8.323)
thus
0 = Tr
__
1 + log +
0
+
L
l=1
l
X
l
_
d
_
. (8.324)
The requirement that the last identity holds for any d implies that
1 + log +
0
+
L
l=1
l
X
l
= 0 , (8.325)
thus
= e
1
0
exp
_
l=1
l
X
l
_
. (8.326)
The Lagrange multipliers
0
,
1
, ...,
L
can be determined from Eqs.
(8.75) and (8.76). The rst constrain (8.75) is satisfy by replacing
the factor e
1
0
by the inverse of the partition function Z
=
1
Z
exp
_
l=1
l
X
l
_
. (8.327)
where
Z = Tr
_
l=1
l
X
l
_
. (8.328)
As can be seen from the above expression for Z, the following holds
X
l
=
1
Z
Tr
_
X
l
e
H
_
=
log Z
l
. (8.329)
b) For the case of a microcanonical ensemble Eq. (8.327) yields = 1/Z,
i.e. is proportional to the identity operator.
c) For the case of a canonical ensemble Eq. (8.327) yields
c
=
1
Z
c
e
H
, (8.330)
where the canonical partition function Z
c
is given by
Z
c
= Tr
_
e
H
_
, (8.331)
and where labels the Lagrange multiplier associated with the given
expectation value H. By solving Eq. (8.76), which for this case is
given by [see also Eq. (8.329)]
H =
1
Z
c
Tr
_
He
H
_
=
log Z
c
. (8.332)
the Lagrange multiplier can be determined. Note that the tem-
perature T is dened by the relation = 1/k
B
T, where k
B
is the
Boltzmanns constant.
d) For the case of a grandcanonical ensemble Eq. (8.327) yields
gc
=
1
Z
gc
e
H+N
, (8.333)
where the grandcanonical partition function Z
gc
is given by
Z
gc
= Tr
_
e
H+N
_
. (8.334)
Here the Lagrange multiplier associated with the given expectation
value N is given by , where is known as the chemical poten-
tial. The average energy H is given by
H = Tr
_
H
gc
_
=
Tr
_
He
(HN)
_
Tr
_
e
(HN)
_
=
Tr
_
(HN) e
(HN)
_
Tr
_
e
(HN)
_ +

Tr
_
Ne
H+N
_
Tr (e
H+N
)
,
thus
H =
_
log Z
gc
_
log Z
gc
. (8.335)
Similarly, the average number of particles N is given by
N = Tr
_
N
gc
_
=
Tr
_
Ne
H+N
_
Tr (e
H+N
)
. (8.336)
8.4. Solutions
In terms of the fugacity , which is dened by
= e
, (8.337)
N can be expressed as
N =
log Z
gc
. (8.338)
25. The Hamiltonian can be expressed as a function of the operators p and
x as
H(p, x) =
p
2
2m
+V (x) . (8.339)
Evaluating Z
c
according to Eq. (8.331) by tracing over momentum states
yields
Z
c
= Tr
_
e
H
_
=
_
dp
[ e
H
[p
=
_
dx
_
dp
[x
[ e
H
[p
.
(8.340)
In the classical limit the parameter , which is inversely proportional to
the temperature, is considered as small. Using Eq. (12.120) from chapter
12, which states that for general operators A and B the following holds
e
(A+B)
= e
A
e
B
+O
_
2
_
, (8.341)
one nds that
e
H
= e
V (x)
e
p
2
2m
+O
_
2
_
. (8.342)
This result together with Eq. (3.52), which is given by
x
[p
=
1
2
exp
_
ip
_
, (8.343)
yield in the classical limit
Z
c
=
_
dx
_
dp
[x
[ e
V (x)
e
p
2
2m
[p
=
_
dx
_
dp
e
V (x
)
e
p
2
2m
p
[x
[p
=
1
2
_
dx
_
dp
e
H(p
,x
)
.
(8.344)
Note that the this result can be also obtained by taking the limit 0,
for which the operator x and p can be considered as commuting operators
(recall that [x, p] = i), and consequently in this limit e
H
can be
factored in the same way [see Eq. (8.342)].
26. The measurement of the observable A
1
is describe by the its extension,
which is given by A
1
1
2
, where 1
2
is the identity operator on subsystem
2. Thus with the help of Eq. (8.8) one nds that
A
1
= Tr (A
1
1
2
)
=

n1,n2
n
1
, n
2
[ A
1
1
2
[n
1
, n
2
=

n1
1
n
1
[
_
n2
2
n
2
[ [n
2
2
_
A
1
[n
1
1
= Tr
1
(
1
A
1
) .
(8.345)
27. In terms of the matrix elements
n1,n2,m1,m2
of the operator , which are
given by
(n
1
,n
2
),(m
1
,m
2
)
= n
1
, n
2
[ [m
1
, m
2
, (8.346)
the matrix elements of
1
and
2
are given by
(
1
)
n1,m1
=

n
2
(n
1
,n
2
),(m
1
,n
2
)
, (8.347)
and
(
2
)
n2,m2
=

n1
(n
1
,n
2
),(n
1
,m
2
)
. (8.348)
In general is Hermitian, i.e.
_
(n
1
,n
2
),(m
1
,n
2
)
_
=
(m
1
,n
2
),(n
1
,n
2
)
, (8.349)
and therefore
_
(
1
)
n1,m1
_
=

n2
_
(n1,n2),(m1,n2)
_
=

n2
(m
1
,n
2
),(n
1
,n
2
)
= (
1
)
m1,n1
,
(8.350)
i.e.
1
is also Hermitian, and similarly
2
is also Hermitian. Thus the
eigenvalues of
1
and
2
are all real. Moreover, these eigenvalues represent
8.4. Solutions
probabilities, and therefore they are expected to be all nonnegative and
smaller than unity. In what follows it is assumed that the set of vectors
[n
1
1
([n
2
2
) are chosen to be eigenvectors of the operator
1
(
2
).
Thus
1
and
2
can be expressed as
1
=
n
1
w
(1)
n
1
[n
1
1
1
n
1
[ , (8.351)
and
2
=
n2
w
(2)
n1
[n
2
2
2
n
2
[ , (8.352)
where the eigenvalues satisfy 0 w
(1)
n
1
1 and 0 w
(2)
n
1
1. Similarly,
can be diagonalized as
=
k
w
k
[k k[ , (8.353)
where 0 w
k
1. In terms of these eigenvalues the entropies are given
by
1
= Tr
1
(
1
log
1
) =
n1
w
(1)
n1
log w
(1)
n1
, (8.354)
2
= Tr
2
(
2
log
2
) =
n2
w
(2)
n2
log w
(2)
n2
, (8.355)
and
= Tr (log ) =
k
w
k
log w
k
. (8.356)
As can be seen from Eqs. (8.346), (8.347) and (8.351), the following holds
w
(1)
n1
= (
1
)
n
1
,n
1
=

n
2
(n
1
,n
2
),(n
1
,n
2
)
=

n2
n
1
, n
2
[ [n
1
, n
2
=

n
2
k
n
1
, n
2
[k w
k
k [n
1
, n
2
,
(8.357)
thus
w
(1)
n1
=

n2
w
n1,n2
, (8.358)
and similarly
w
(2)
n
2
=

n
1
w
n
1
,n
2
, (8.359)
where
w
n1,n2
=
k
n
1
, n
2
[k w
k
k [n
1
, n
2
. (8.360)
Note that
n1,n2
w
n1,n2
=

k
w
k
k[
_

n1,n2
[n
1
, n
2
n
1
, n
2
[
_
[k
=

k
w
k
k [k ,
(8.361)
thus the normalization condition k [k = 1 together with the require-
ment that
Tr =

k
w
k
= 1 , (8.362)
imply that
n1,n2
w
n1,n2
= 1 , (8.363)
i.e.
Tr
1
1
=

n
1
w
(1)
n
1
= 1 , (8.364)
and
Tr
2
2
=

n2
w
(2)
n2
= 1 . (8.365)
Consider the quantity y
_
w
n1,n2
/w
(1)
n1
w
(2)
n2
_
, where the function y (x) is
given by
y (x) = xlog x x + 1 . (8.366)
The following holds
dy
dx
= log x , (8.367)
8.4. Solutions
and
d
2
y
dx
2
=
1
x
, (8.368)
thus the function y (x) has a single stationary point at x = 1, which is a
minima point. Moreover y (1) = 0, thus one concludes that
y (x) 0 (8.369)
for x 0. For x = w
n1,n2
/w
(1)
n1
w
(2)
n2
the inequality (8.369) implies that
w
n
1
,n
2
w
(1)
n
1
w
(2)
n
2
log
w
n
1
,n
2
w
(1)
n
1
w
(2)
n
2
w
n
1
,n
2
w
(1)
n
1
w
(2)
n
2
+ 1 0 . (8.370)
Multiplying by w
(1)
n1
w
(2)
n2
and summing over n
1
and n
2
yields
n
1
,n
2
w
n
1
,n
2
log
w
n
1
,n
2
w
(1)
n
1
w
(2)
n
2

n
1
,n
2
w
n
1
,n
2
+
n
1
w
(1)
n1
n
2
w
(2)
n2
0 , (8.371)
thus with the help of Eqs. (8.363), (8.364) and (8.365) one nds that
n
1
,n
2
w
n
1
,n
2
log
w
n1,n2
w
(1)
n1
w
(2)
n2
0 , (8.372)
and with the help of Eqs. (8.354) and (8.355) that
1
+
2

n
1
,n
2
w
n
1
,n
2
log w
n
1
,n
2
. (8.373)
Using Eq. (8.360) one obtains

n
1
,n
2
w
n
1
,n
2
log w
n
1
,n
2
=

n
1
,n
2
k
[n
1
, n
2
[k[
2
w
k
log
1
w
n
1
,n
2
=

n1,n2
_
k
[n
1
, n
2
[k[
2
w
k
log
w
k
w
n1,n2
_

n
1
,n
2
k
[n
1
, n
2
[k[
2
w
k
log w
k
=

n1,n2
_
k
[n
1
, n
2
[k[
2
w
k
log
w
k
w
n1,n2
_
k
w
k
log w
k
n
1
,n
2
[n
1
, n
2
[k[
2
. .
=1
,
(8.374)
thus

n1,n2
w
n1,n2
log w
n1,n2
=

n1,n2
_
k
[n
1
, n
2
[k[
2
w
k
log
w
k
w
n1,n2
_
+ .
(8.375)
Furthermore, according to inequality (8.369) the following holds
k
[n
1
, n
2
[k[
2
w
k
log
w
k
w
n1,n2
=

k
[n
1
, n
2
[k[
2
w
n
1
,n
2
w
k
w
n
1
,n
2
log
w
k
w
n
1
,n
2
k
[n
1
, n
2
[k[
2
w
n
1
,n
2
_
w
k
w
n1,n2
1
_
,
k
[n
1
, n
2
[k[
2
w
k
w
n1,n2
k
[n
1
, n
2
[k[
2
= 0 .
(8.376)
These results together with inequality (8.373) yield
1
+
2
. (8.377)
28. With the help of Eq. (8.92), which is given by
=
e
H
Tr (e
H
)
= p
+
[+ +[ +p
[ [ , (8.378)
where the probabilities p
+
and p
are given by
p
=
e

2
e

2
+e

2
, (8.379)
= [e[ B/m
e
c is the Larmor frequency [see Eq. (4.22)] and where =
1/k
B
T, one nds that
= p
+
log p
+
p
log p
=
1 tanh

2
2
log
1 tanh

2
2

1 + tanh

2
2
log
1 + tanh

2
2
.
(8.380)
9. Time Independent Perturbation Theory
Consider a Hamiltonian H
0
having a set of eigenenergies E
k
. Let g
k
be
the degree of degeneracy of eigenenergy E
k
, namely g
k
is the dimension of
the corresponding eigensubspace , which is denoted by T
k
. The set [k, i
of eigenvectors of H
0
is assumed to form an orthonormal basis for the vector
space, namely
H
0
[k, i = E
k
[k, i , (9.1)
and
k
, i
[k, i =
kk

ii
. (9.2)
For a given k the degeneracy index i can take the values 1, 2, , g
k
. The
set of vectors [k, 1 , [k, 2 , , [k, g
k
forms an orthonormal basis for the
eigensubspace T
k
. The closure relation can be written as
1 =
k
g
k
i=1
[k, i k, i[ =

k
P
k
, (9.3)
where
P
k
=
g
k
i=1
[k, i k, i[ (9.4)
is a projector onto eigen subspace T
k
. The orthogonality condition (9.2)
implies that
P
k
P
k
= P
k
kk
. (9.5)
A perturbation V =
V is being added to the Hamiltonian

H = H
0
+
V , (9.6)
where . We wish to nd the eigenvalues and the eigenvectors of the
Hamiltonian H
H[ = E[ . (9.7)
Chapter 9. Time Independent Perturbation Theory
In many cases nding an analytical solution to the above equation is either
very hard or impossible. In such cases one possibility is to employ numerical
methods. However, another possibility arrises provided that the eigenvalues
and eigenvectors of H
0
are known and provided that the perturbation
V can
be considered as small, namely, provided the eigenvalues and eigenvectors of
H do not signicantly dier from those of H
0
. In such a case an approximate
solution can be obtained by the time independent perturbation theory.
9.1 The Level E
n
Consider the level E
n
of the unperturbed Hamiltonian H
0
. Let P
n
be the
projector onto the eigensubspace T
n
, and let
Q
n
= 1 P
n
=

k=n
P
k
. (9.8)
Equation (9.7) reads
V [ = (E H
0
) [ . (9.9)
It is useful to introduce the operator R, which is dened as
R =

k=n
P
k
E E
k
. (9.10)
Claim. The eigenvector [ of the Hamiltonian H is given by
[ =
_
1 R
V
_
1
P
n
[ . (9.11)
Proof. Using Eq. (9.5) it is easy to show that
P
n
R = RP
n
= 0 . (9.12)
Moreover, the following holds
Q
n
R =

k=n
=n
P
k
P
k
E E
k
=

k=n
P
k
E E
k
= R , (9.13)
and similarly
RQ
n
= R . (9.14)
Furthermore, by expressing H
0
as
H
0
=

k
g
k
i=1
E
k
[k, i k, i[ = E
n
P
n
+
k=n
E
k
P
k
, (9.15)
9.1. The Level E
n
one nds that
R(E H
0
) =

k=n
P
k
_
E E
n
P
n
=n
E
k
P
k
_
E E
k
=

k=n
P
k
(E E
k
)
E E
k
= Q
n
,
(9.16)
and similarly
(E H
0
) R = Q
n
. (9.17)
The last two results suggest that the operator R can be considered as the
inverse of E H
0
in the subspace of eigenvalue zero of the projector P
n
(which is the subspace of eigenvalue unity of the projector Q
n
). Multiplying
Eq. (9.9) from the left by R yields
R
V [ = R(E H
0
) [ . (9.18)
R
V [ = Q
n
[ . (9.19)
Since P
n
= 1 Q
n
[see Eq. (9.8)] the last result implies that
P
n
[ = [ R
V [ =
_
1 R
V
_
[ , (9.20)
which leads to Eq. (9.11)
[ =
_
1 R
V
_
1
P
n
[ . (9.21)
Note that Eq. (9.11) can be expanded as power series in
[ =
_
1 +R
V +
2
R
V R
V +
_
P
n
[ . (9.22)
9.1.1 Nondegenerate Case
In this case g
n
= 1 and
P
n
= [n n[ . (9.23)
In general the eigenvector [ is determined up to multiplication by a con-
stant. For simplicity we choose that constant to be such that
P
n
[ = [n , (9.24)
namely
n [ = 1 . (9.25)
Multiplying Eq. (9.9), which is given by
V [ = (E H
0
) [ , (9.26)
from the left by n[ yields
n[
V [ = n[ (E H
0
) [ , (9.27)
or
n[ E[ = n[ H
0
[ +n[
V [ , (9.28)
thus
E = E
n
+n[
V [ . (9.29)
Equation (9.22) together with Eq. (9.24) yield
[ =
_
1 +R
V +
2
R
V R
V +
_
[n
= [n +
k=n
i
[k, i k, i[
V [n
E E
k
+
2

k=n
i
=n
i
[k, i k, i[
V [k
, i k
, i[

V [n
(E E
k
) (E E
k
)
+ .
(9.30)
Substituting Eq. (9.30) into Eq. (9.29) yields
E = E
n
+n[
V [n
+
2

k=n
i
n[

V [k, i k, i[
V [n
E E
k
+
3

k=n
i
=n
i
n[

V [k, i k, i[
V [k
, i k
, i[

V [n
(E E
k
) (E E
k
)
+ .
(9.31)
9.2. Example
Note that the right hand side of Eq. (9.31) contains terms that depend on E.
To second order in one nds
E = E
n
+n[ V [n +
k=n
i
[k, i[ V [n[
2
E
n
E
k
+O
_
3
_
. (9.32)
Furthermore, to rst order in Eq. (9.30) yields
[ = [n +
k=n
i
[k, i k, i[ V [n
E
n
E
k
+O
_
2
_
. (9.33)
9.1.2 Degenerate Case
The set of vectors [n, 1 , [n, 2 , , [n, g
n
forms an orthonormal basis for
the eigensubspace T
n
. Multiplying Eq. (9.9) from the left by P
n
yields
P
n
V [ = P
n
(E H
0
) [ , (9.34)
thus with the help of Eq. (9.15) one has
P
n
V [ = (E E
n
) P
n
[ . (9.35)
Substituting Eq. (9.22), which is given by
[ =
_
P
n
+R
V P
n
+
2
R
V R
V P
n
+
_
[ , (9.36)
into this and noting that P
2
n
= P
n
yield
P
n
V P
n
[ +
2
P
n
V R
V P
n
[ + = (E E
n
) P
n
[ . (9.37)
Thus, to rst order in the energy correction E E
n
is found by solving
P
n
V P
n
[ = (E E
n
) P
n
[ . (9.38)
The solutions are the eigenvalues of the g
n
g
n
matrix representation of the
operator V in the subspace T
n
.
9.2 Example
Consider a point particle having mass m whose Hamiltonian is given by
H = H
0
+V , (9.39)
where
H
0
=
p
2
2m
+
m
2
x
2
2
. (9.40)
and where
V =
_
m
x . (9.41)
The eigenvectors and eigenvalues of the Hamiltonian H
0
, which describes a
one dimensional harmonic oscillator, are given by
H
0
[n = E
n
[n , (9.42)
where n = 0, 1, , 2 , and where
E
n
( = 0) =
_
n +
1
2
_
. (9.43)
Note that, as was shown in chapter 5 [see Eq. (5.147)], the eigenvectors
and eigenvalues of H can be found analytically for this particular case. For
the sake of comparison we rst derive this exact solution. Writing H as
H =
p
2
2m
+
m
2
x
2
2
+
_
m
x
=
p
2
2m
+
m
2
2
_
x +
_

m
_
2
1
2
2
,
(9.44)
one sees that H describes a one dimensional harmonic oscillator (as H
0
also
does). The exact eigenenergies are given by
E
n
() = E
n
( = 0)
1
2
2
, (9.45)
and the corresponding exact wavefunctions are
x
[n() =
_
x
+
_

m
[n . (9.46)
Using identity (3.19), which is given by
J (x) [x
= [x
+x , (9.47)
where J (x) is the translation operator, the exact solution (9.46) can be
rewritten as
x
[n() = x
[ J
_
_

m
_
[n , (9.48)
9.2. Example
or simply as
[n() = J
_
_

m
_
[n . (9.49)
Next we calculate an approximate eigenvalues and eigenvectors using per-
turbation theory. Using the identity
x =
_

2m
_
a +a
_
, (9.50)
one has
V =

2
_
a +a
_
. (9.51)
Furthermore, using the identities
a[n =
n[n 1 , (9.52)
a
[n =
n + 1 [n + 1 , (9.53)
one has
m[ V [n =

2
_
m[ a [n +m[ a
[n
_
=

2
_
n
m,n1
+
n + 1
m,n+1
_
.
(9.54)
Thus E
n
() can be expanded using Eq. (9.32) as
E
n
() = E
n
+n[ V [n
. .
=0
+
k=n
i
[k, i[ V [n[
2
E
n
E
k
+O
_
3
_
= E
n
+
[n 1[ V [n[
2
E
n
E
n1
+
[n + 1[ V [n[
2
E
n
E
n+1
+O
_
3
_
=
_
n +
1
2
_
+
n
2
2

(n + 1)
2
2
+O
_
3
_
=
_
n +
1
2
_
2
2
+O
_
3
_
,
(9.55)
in agreement (to second order) with the exact result (9.45), and [n() can
be expanded using Eq. (9.30) as
[n() = [n +
k=n
i
[k, i k, i[ V [n
E
n
E
k
+O
_
2
_
= [n +
[n 1 n 1[ V [n
E
n
E
n1
+
[n + 1 n + 1[ V [n
E
n
E
n+1
+O
_
2
_
= [n +
[n 1

[n + 1
n + 1
+O
_
2
_
= [n +

2
a[n

2
a
[n +O
_
2
_
.
(9.56)
Note that with the help of the following identify
p = i
_
m
2
_
a +a
_
, (9.57)
the last result can be written as
[n() =
_
1 +
_

m
ip
_
[n +O
_
2
_
. (9.58)
Alternatively, in terms of the translation operator J (x), which is given by
J (x) = exp
_
ipx
_
, (9.59)
one has
[n() = J
_
_

m
_
[n +O
_
2
_
, (9.60)
in agreement (to second order) with the exact result (9.49).
9.3 Problems
1. The volume eect: The energy spectrum of the hydrogen atom was
calculated in chapter 8 by considering the proton to be a point particle.
Consider a model in which the proton is instead assumed to be a sphere
of radius
0
where
0
<< a
0
(a
0
is Bohrs radius), and the charge of
the proton +e is assumed to be uniformly distributed in that sphere.
Show that the energy shift due to such perturbation to lowest order in
perturbation theory is given by
E
n,l
=
e
2
10
2
0
[R
n,l
(0)[
2
, (9.61)
where R
n,l
(r) is the radial wave function.
9.3. Problems
2. Consider an hydrogen atom. A perturbation given by
V = Ar , (9.62)
where r =
_
x
2
+y
2
+z
2
is the radial coordinate and A is a constant is
added.
a) Calculate to rst order in A the energy of the ground state.
b) Calculate to rst order in A the energy of the rst excited state.
3. A weak uniform electric eld E = Ez, where E is a constant, is applied
to a hydrogen atom. Calculate to 1st order in perturbation theory the
correction to the energy of the
a) level n = 1 (n is the principle quantum number).
b) level n = 2.
4. Consider two particles, both having the same mass m, moving in a one-
dimensional potential with coordinates x
1
and x
2
respectively. The po-
tential energy is given by
V (x
1
, x
2
) =
1
2
m
2
(x
1
a)
2
+
1
2
m
2
(x
2
+a)
2
+m
2
(x
1
x
2
)
2
,
(9.63)
where is real. Find the energy of the ground state to lowest non-
vanishing order in .
5. A particle having mass m is conned in a potential well of width l, which
is given by
V (x) =
_
0 for 0 x l
+ elsewhere
. (9.64)
Find to lowest order in perturbation theory the correction to the ground
state energy due to a perturbation given by
W (x) = w
0
_
x
l
2
_
, (9.65)
where w
0
is a real constant.
6. Consider a particle having mass m in a two dimensional potential well of
width a that is given by
V (x, y) =
_
0 if 0 x a and 0 y a
+ elsewhere
. (9.66)
A perturbation given by
W (x, y) =
_
w
0
if 0 x
a
2
and 0 y
a
2
0 elsewhere
, (9.67)
is added.
a) Calculate to lowest non-vanishing order in w
0
the energy of the
ground state.
b) The same for the rst excite state.
7. Consider a particle having mass m moving in a potential energy given by
V (x, y) =
m
2
2
_
x
2
+y
2
_
+m
2
xy , (9.68)
where the angular frequency is a constant and where the dimensionless
real constant is assumed to be small.
a) Calculate to rst order in the energy of the ground state.
b) Calculate to rst order in the energy of the rst excited state.
8. Consider a harmonic oscillator having angular resonance frequency . A
perturbation given by
V =
_
a
+aa
_
(9.69)
is added, where a is the annihilation operator and is a real constant.
Calculate the energies of the system to second order in .
9. The Hamiltonian of a spin S = 1 is given by
H = S
2
z
+
_
S
2
x
S
2
y
_
, (9.70)
where and are both constants.
a) Write the matrix representation of Hin the basis [S = 1, m = 1 , [S = 1, m = 0 [S = 1, m = 1.
b) Calculate (exactly) the eigenenergies and the corresponding eigen-
vectors.
c) For the case << use perturbation theory to calculate to lowest
order in and the eigen energies of the system.
10. Consider a system composed of a harmonic oscillator having angular
resonance frequency
r
> 0 and a two-level system. The Hamiltonian of
the system is given by
H = H
r
+H
a
+V . (9.71)
The term H
r
is the Hamiltonian for the harmonic oscillator
H
r
=
r
_
a
a +
1
2
_
, (9.72)
where a and a
are annihilation and creation operators respectively. The

term H
a
is the Hamiltonian for the two-level system
H
a
=

a
2
([+ +[ [ [) , (9.73)
where the ket vectors [ represent the two levels and where
a
> 0. The
coupling term between the oscillator and the two-level system is given by
9.3. Problems
V = g
_
a
+
[ +[ +a [+ [
_
. (9.74)
Assume the case where [g[ <<
r
and where [g[ <<
a
. Calculate to
lowest non-vanishing order in g the eigen energies of the system for the
following cases: (a)
r
,=
a
; (b)
r
=
a
.
11. Consider a particle having mass m in a two-dimensional potential given
by
V
0
=
1
2
m
2
_
x
2
+y
2
_
. (9.75)
The following perturbation is added
V
1
=

L
2
z
, (9.76)
where L
z
is the z component of the angular momentum operator.
a) Find to second orders in the energy of the ground state.
b) Find to rst order in the energy of the rst excited level.
12. A particle having mass m moves in a one dimensional potential
V (x) =
_
V
0
sin
2x
l
0 x l
else
. (9.77)
Consider the constant V
0
to be small. Calculate the systems eigenenergies
E
n
to rst order in V
0
.
13. Consider a particle having mass m conned by the one-dimensional po-
tential well, which is given by
V (x) =
_
_
_
x < 0
x
L
0 x L
x > L
.
Find to rst order in the energy of the ground state.
14. A particle of mass m is trapped in an innite 2 dimensional well of width
l
V (x, y) =
_
0 0 x l and 0 y l
else
. (9.78)
A perturbation given by
W (x, y) =
2
m
(x l
x
) (y l
y
) . (9.79)
is added, where
0 l
x
l , (9.80)
and
0 l
y
l . (9.81)
Calculate to 1st order in perturbation theory the correction to the energy
of the:
a) ground state.
b) rst excited state.
15. Consider a rigid rotator whose Hamiltonian is given by
H =
L
2
x
+L
2
y
2I
xy
+
L
2
z
2I
z
+
L
2
x
L
2
y
2I
xy
, (9.82)
where L is the angular momentum vector operator. Use perturbation
theory to calculate the energy of the ground state to second order in .
16. Consider two particles having the same mass m moving along the x axis.
The Hamiltonian of the system is given by
H =
p
2
1
2m
+
p
2
2
2m
(x
1
) (x
2
) + (x
1
x
2
) , (9.83)
where x
1
and x
2
are the coordinates of the rst and second particle
respectively, p
1
and p
2
are the corresponding canonically conjugate mo-
mentums, and are both real positive constants and () denotes the
delta function. Calculate to rst order in the energy of the ground state
of the system.
17. In this problem the main results of time independent perturbation theory
are derived using an alternative approach. Consider a general square
matrix
W = D +V , (9.84)
where , D is diagonal
D[n
0
=
n0
[n
0
, (9.85a)
n
0
[ D =
n0
n
0
[ , (9.85b)
and we assume that none of the eigenvalues of D is degenerate. The set
of eigenvectors of D is assumed to be orthonormal
n
0
[m
0
=
nm
, (9.86)
and complete (the dimensionality is assumed to be nite)
1 =

n
[n
0
n
0
[ . (9.87)
Calculate the eigenvalues of W
W [n = [n (9.88)
to second order in .
9.4. Solutions
18. Consider the Hamiltonian H, which is given by
H = H
0
+
V , (9.89)
where . The set [k of eigenvectors of H
0
with corresponding
eigenvalues E
k
, which satisfy
H
0
[k = E
k
[k , (9.90)
is assumed to form an orthonormal basis for the vector space, i.e.
k
[k =
kk
. (9.91)
Consider the transformation
H
R
= e
L
He
L
, (9.92)
where the operator L is assumed to be anti Hermitian, i.e. L
= L, in
order to ensure that e
L
is unitary. Show that to second order in the
matrix elements k[ H
R
[k
are given by
k[ H
R
[k
= E
k
k,k
+

2
2
k[

V [k
V [k
_
1
E
k
E
k

1
E
k
E
k
_
,
(9.93)
provided that the following condition is satised
V + [L, H
0
] = 0 . (9.94)
Note that to rst order in the following holds k[ H
R
[k
= k[ H
0
[k
,
thus, in spite of the fact that the perturbation
V is rst order in , the

transformed Hamiltonian H
R
depends on only to second order.
9.4 Solutions
1. The radial force acting on the electron is found using Gauss theorem
F
r
(r) =
_
_
_
e
2
r
2
r >
0
e
2
r
2
_
r
0
_
3
r
0
. (9.95)
The potential energy V (r) is found by integrating F
r
(r) and by requiring
that V (r) is continuous at r =
0
V (r) =
_
_
_
e
2
r
r >
0
e
2
2
0
_
_
r
0
_
2
3
_
r
0
. (9.96)
Thus, the perturbation term in the Hamiltonian is given by
V
p
(r) = V (r)
_
e
2
r
_
=
_
_
_
0 r >
0
e
2
2
0
_
_
r
0
_
2
+
2
0
r
3
_
r
0
. (9.97)
To rst order one has
E
n,l
= nlm[ V
p
[nlm . (9.98)
The wavefunctions for the unperturbed case are given by
nlm
(r, , ) = R
nl
(r) Y
m
l
(, ) , (9.99)
Since V
p
depends on r only, one nds that
E
n,l
=
_
0
drr
2
[R
nl
(r)[
2
V
p
(r)
=
0
_
0
drr
2
[R
nl
(r)[
2
e
2
2
0
_
_
r
0
_
2
+
2
0
r
3
_
.
(9.100)
In the limit where
0
<< a
0
the term [R
nl
(r)[
2
can approximately be
replaced by [R
nl
(0)[
2
, thus
E
n,l
= [R
nl
(0)[
2
0
_
0
drr
2
e
2
2
0
_
_
r
0
_
2
+
2
0
r
3
_
=
e
2
2
0
10
[R
nl
(0)[
2
.
(9.101)
2. The wavefunctions for the unperturbed case are given by
nlm
(r, , ) = R
nl
(r) Y
m
l
(, ) , (9.102)
where for the states relevant to this problem
9.4. Solutions
R
10
(r) = 2
_
1
a
0
_
3/2
e
r/a
0
, (9.103a)
R
20
(r) = (2 r/a
0
)
_
1
2a
0
_
3/2
e
r
2a
0
, (9.103b)
R
21
(r) =
_
1
2a
0
_
3/2
r
3a
0
e
r
2a
0
, (9.103c)
Y
0
0
(, ) =
_
1
4
, (9.103d)
Y
1
1
(, ) =
1
2
_
3
2
sine
i
, (9.103e)
Y
0
1
(, ) =
1
2
_
3
cos , (9.103f)
Y
1
1
(, ) =
1
2
_
3
2
sine
i
, (9.103g)
and the corresponding eigenenergies are given by
E
(0)
n
=
E
I
n
2
, (9.104)
where
E
I
=
m
e
e
4
2
2
. (9.105)
The perturbation term V in the Hamiltonian is given by V = Ar. The
matrix elements of V are expressed as
n
[ V [nlm = A
_
0
dr r
3
R
n
l
R
nl
1
_
1
d(cos )
2
_
0
d
_
Y
m
Y
m
l
= A
l,l

m,m
_
0
dr r
3
R
n
l
R
nl
.
(9.106)
a) Thus, to rst order
E
1
= E
(0)
1
+100[ V [100 +O(A
2
) , (9.107)
where
100[ V [100 = A
_
0
dr r
3
R
2
10
(r) =
3Aa
0
2
. (9.108)
b) The rst excited state is degenerate, however, as can be seen from
Eq. (9.106) all o-diagonal elements are zero. The diagonal elements
are given by
200[ V [200 = A
_
0
dr r
3
R
2
20
= 6Aa
0
, (9.109a)
21m[ V [21m = A
_
0
dr r
3
R
21
= 5Aa
0
. (9.109b)
Thus, the degeneracy is lifted
E
2,l=0
= E
(0)
2
+ 6Aa
0
+O(A
2
) , (9.110)
E
2,l=1
= E
(0)
2
+ 5Aa
0
+O(A
2
) . (9.111)
3. The wavefunctions for the unperturbed case are given by
nlm
(r, , ) = R
nl
(r) Y
m
l
(, ) , (9.112)
where for the states relevant to this problem
R
10
(r) = 2
_
1
a
0
_
3/2
e
r/a
0
, (9.113)
R
20
(r) = (2 r/a
0
)
_
1
2a
0
_
3/2
e
r
2a
0
, (9.114)
R
21
(r) =
_
1
2a
0
_
3/2
r
3a
0
e
r
2a
0
, (9.115)
Y
0
0
(, ) =
_
1
4
, (9.116)
Y
1
1
(, ) =
1
2
_
3
2
sine
i
, (9.117)
Y
0
1
(, ) =
1
2
_
3
cos , (9.118)
Y
1
1
(, ) =
1
2
_
3
2
sine
i
, (9.119)
and the corresponding eigenenergies are given by
E
(0)
n
=
E
I
n
2
, (9.120)
where
E
I
=
m
e
e
4
2
2
. (9.121)
9.4. Solutions
The perturbation term V in the Hamiltonian is given by
V = eEz = eEr cos . (9.122)
The matrix elements of V are expressed as
n
[ V [nlm = eE
_
0
dr r
3
R
n
l
R
nl
1
_
1
d(cos )
2
_
0
d cos
_
Y
m
Y
m
l
.
(9.123)
a) Disregarding spin this level is non degenerate. To 1st order
E
1
= E
(0)
1
+1, 0, 0[ V [1, 0, 0 = E
(0)
1
,
since
1
_
1
d(cos ) cos = 0 ,
thus the energy of the ground state is unchanged to 1st order.
b) The level n = 2 has degeneracy 4. The matrix of the perturbationV
in the degenerate subspace is given by
M =
_
_
_
_
2, 0, 0[ V [2, 0, 0 2, 0, 0[ V [2, 1, 1 2, 0, 0[ V [2, 1, 0 2, 0, 0[ V [2, 1, 1
2, 1, 1[ V [2, 0, 0 2, 1, 1[ V [2, 1, 1 2, 1, 1[ V [2, 1, 0 2, 1, 1[ V [2, 1, 1
2, 1, 0[ V [2, 0, 0 2, 1, 0[ V [2, 1, 1 2, 1, 0[ V [2, 1, 0 2, 1, 0[ V [2, 1, 1
2, 1, 1[ V [2, 0, 0 2, 1, 1[ V [2, 1, 1 2, 1, 1[ V [2, 1, 0 2, 1, 1[ V [2, 1, 1
_
_
_
_
.
(9.124)
Using
1
_
1
d(cos ) cos = 0 , (9.125)
1
_
1
d(cos ) cos sin = 0 , (9.126)
1
_
1
d(cos ) cos sin
2
= 0 , (9.127)
1
_
1
d(cos ) cos
3
= 0 , (9.128)
2
_
0
d e
i
= 0 , (9.129)
one nds
M =
_
_
_
_
0 0 0
0 0 0 0
0 0 0
0 0 0 0
_
_
_
_
, (9.130)
where
= 2, 0, 0[ V [2, 1, 0
= eE
_
0
dr r
3
R
2,0
R
2,1
1
_
1
d(cos )
2
_
0
d cos
_
Y
0
0
_
Y
0
1
=
eE
8
_
0
dr
_
2
r
a
0
__
r
a
0
_
4
e
r
a
0
1
4
1
_
1
d(cos ) cos
2
2
_
0
d .
(9.131)
Using
1
_
1
d(cos ) cos
2
=
2
3
, (9.132)
and
_

0
x
4
e
x
dx = 24 (9.133)
_

0
x
5
e
x
dx = 120 (9.134)
one nds
= 2, 0, 0[ V [2, 1, 0
=
eE
24
_
0
dr
_
2
r
a
0
__
r
a
0
_
4
e
r
a
0
=
a
0
eE
24
_
0
dx (2 x) x
4
e
x
= 3a
0
eE .
(9.135)
The eigenvalues of the matrix M are 0, 0, 3a
0
eE and 3a
0
eE. Thus
to 1st order the degeneracy is partially lifted with subspace of dimen-
sion 2 having energy E
(0)
2
, and another 2 nondegenerate subspaces
having energies E
(0)
2
3a
0
eE.
9.4. Solutions
4. To lowest order in perturbation theory the ground state energy is given
by
E
gs
= +m
2
dx
1
dx
2
2
0
(x
1
a)
2
0
(x
2
+a) (x
1
x
2
)
2
+O
_
2
_
,
(9.136)
where
0
(x) is the ground state wavefunction of a particle having mass
m conned by a potential (1/2) m
2
x
2
centered at x = 0. Employing the
transformation
x
1
= x
1
a , (9.137)
x
2
= x
2
+a , (9.138)
E
gs
=
+m
2
dx
2
0
(x
1
) (x
1
+a)
2
+m
2
dx
2
0
(x
2
) (x
2
+a)
2
2m
2
dx
2
0
(x
1
) (x
1
+a)
dx
2
0
(x
2
) (x
2
a)
+O
_
2
_
= + 2m
2
_

2m
+a
2
_
+ 2m
2
a
2
+O
_
2
_
= +
_
+ 4m
2
a
2
_
+O
_
2
_
.
(9.139)
Note that this problem can be also solved exactly by employing the co-
ordinate transformation
x
+
=
x
1
+x
2
2
, (9.140)
x
=
x
1
x
2
2
. (9.141)
x
1
=
x
+
+x
2
, (9.142)
x
2
=
x
+
x
2
. (9.143)
The following holds
x
2
1
+x
2
2
= x
2
+
+x
2
, (9.144)
and
x
2
1
+ x
2
2
= x
2
+
+ x
2
. (9.145)
Thus, the Lagrangian of the system can be written as
/ =
m
_
x
2
1
+ x
2
2
_
2
V (x
1
, x
2
)
=
m
_
x
2
+
+ x
2
_
2

1
2
m
2
_
x
2
+
+x
2
2a
2x
+ 2a
2
+ 4x
2
_
= /
+
+/
,
(9.146)
where
/
+
=
m x
2
+
2

1
2
m
2
x
2
+
, (9.147)
and
/
=
m x
2
2

1
2
m
2
_
_
(1 + 4)
_
x
2
1 + 4
_
2
+
8a
2
1 + 4
_
_
. (9.148)
Thus, the system is composed of two decoupled harmonic oscillators, and
therefore, the exact eigenenergies are given by
E
n+,n
=
_
n
+
+
1
2
_
+
1 + 4
_
n
+
1
2
_
+
4m
2
a
2
1 + 4
, (9.149)
where n
+
, n
= 0, 1, 2, . To rst order in one thus has

E
n+,n
=
_
n
+
+
1
2
_
+
_
n
+
1
2
_
+
_
(2n
+ 1) + 4m
2
a
2
+O
_
2
_
.
(9.150)
5. For w
0
= 0 the normalized wavefunctions
(0)
n
(x) are given by
(0)
n
(x) = x
[n =
_
2
l
sin
nx
l
, (9.151)
E
(0)
n
=

2
2
n
2
2ml
2
. (9.152)
9.4. Solutions
The matrix elements of the perturbation are given by
n[W[m =
2w
0
l
l
_
0
sin
nx
l
sin
mx
l

_
x
l
2
_
dx
=
2w
0
l
sin
n
2
sin
m
2
.
(9.153)
For the ground state
1[V [1 =
2w
0
l
, (9.154)
thus
E
1
=

2
2
2ml
2
+
2w
0
l
+O
_
w
2
0
_
. (9.155)
6. For w
0
= 0 the normalized wavefunctions
(0)
n
x
,n
y
(x, y) are given by
(0)
nx,ny
(x
, y
) = x
, y
[n
x
, n
y
=
2
a
sin
n
x
x
a
sin
n
y
y
a
, (9.156)
E
(0)
nx,ny
=

2
2
_
n
2
x
+n
2
y
_
2ma
2
, (9.157)
where n
x
= 1, 2, and n
y
= 1, 2, .
a) The ground state (n
x
, n
y
) = (1, 1) is nondegenerate, thus to rst
order in w
0
E
0
=

2
2
ma
2
+1, 1[W[1, 1
=

2
2
ma
2
+
4w
0
a
2
a/2
_
0
sin
2
x
a
dx
a/2
_
0
sin
2
y
a
dy
2
ma
2
+
w
0
4
,
(9.158)
b) The rst excite state is doubly degenerate. The matrix of the per-
turbation in the corresponding subspace is given by
_
1, 2[W[1, 2 1, 2[W[2, 1
2, 1[W[1, 2 2, 1[W[2, 1
_
=
4w
0
a
2
_
_
_
_
_
a/2
_
0
sin
2 x
a
dx
a/2
_
0
sin
2 2y
a
dy
a/2
_
0
sin
x
a
sin
2x
a
dx
a/2
_
0
sin
2y
a
sin
y
a
dy
a/2
_
0
sin
2x
a
sin
1x
a
dx
a/2
_
0
sin
y
a
sin
2y
a
dy
a/2
_
0
sin
2 2x
a
dx
a/2
_
0
sin
2 y
a
dy
_
_
_
_
_
= w
0
_
1
4
16
9
2
16
9
2
1
4
_
,
(9.159)
To rst order in perturbation theory the eigenenergies are found
by adding the eigenvalues of the above matrix to the unperturbed
eigenenergy E
(0)
1,2
= E
(0)
2,1
. Thus, to rst order in w
0
E
1,
=
5
2
2
2ma
2
+
w
0
4

16w
0
9
2
+O
_
w
2
0
_
. (9.160)
7. For the unperturbed case = 0 one has
H
0
[n
x
, n
y
= (n
x
+n
y
+ 1) [n
x
, n
y
, (9.161)
where n
x
, n
y
= 0, 1, 2, . Using the identities
x =
_

2m
_
a
x
+a
x
_
, (9.162)
y =
_

2m
_
a
y
+a
y
_
, (9.163)
the perturbation term V
1
= m
2
xy can be expressed as
V
1
=
2
_
a
x
+a
x
_ _
a
y
+a
y
_
.
a) For the ground state [0, 0, which is nondegenerate, one has
E
0,0
() = +0, 0[ V
1
[0, 0
. .
=0
+

nx,ny=0,0
[n
x
, n
y
[ V
1
[0, 0[
2
E
0,0
(0) E
nx,ny
= +
[1, 1[ V
1
[0, 0[
2
2
=
_
2
_
2
2
=
_
1

2
8
_
.
(9.164)
9.4. Solutions
b) The rst excited state is doubly degenerate, thus the eigenenergies
are found by diagonalizing the matrix of V
1
in the corresponding
subspace
_
1, 0[ V
1
[1, 0 1, 0[ V
1
[0, 1
0, 1[ V
1
[1, 0 0, 1[ V
1
[0, 1
_
=

2
_
0 1
1 0
_
. (9.165)
Thus the degeneracy is lifted and the energies are given by 2 (1 /4).
Note that this problem can be also solved exactly by employing the
coordinate transformation
x
=
x +y
2
, (9.166)
y
=
x y
2
. (9.167)
x =
x
+y
2
, (9.168)
y =
x
2
. (9.169)
The following hold
x
2
+y
2
= x
2
+y
2
, (9.170)
x
2
+ y
2
= x
2
+ y
2
, (9.171)
xy =
1
2
_
x
2
y
2
_
. (9.172)
Thus, the Lagrangian of the system can be written as
/ =
m
_
x
2
+ y
2
_
2
V (x
1
, x
2
)
=
m
_
x
2
+ y
2
_
2

m
2
2
_
x
2
+y
2
_
m
2
2
_
x
2
y
2
_
= /
+
+/
,
(9.173)
where
/
+
=
m x
2
2

m
2
2
(1 +) x
2
, (9.174)
and
/
=
m y
2
2

m
2
2
(1 ) y
2
. (9.175)
Thus, the system is composed of two decoupled harmonic oscillators,
and therefore, the exact eigenenergies are given by
E
n+,n
=
_
_
1 +
_
n
x
+
1
2
_
+
_
1
_
n
y
+
1
2
__
, (9.176)
where n
x
, n
y
= 0, 1, 2, . To second order in one thus has
E
n+,n
=
_
n
x
+n
y
+ 1 +
n
x
n
y
2

n
x
+n
y
+ 1
8

2
_
+O
_
3
_
.
(9.177)
8. Using Eqs. (5.28) and (5.29) one nds that
m[ V [n =
_
n(n 1)
m,n2
+
_
(n + 1) (n + 2)
m,n+2
, (9.178)
thus
E
n
() =
_
n +
1
2
_
+n[ V [n
. .
=0
+

m=n
[m[ V [n[
2
E
n
(0) E
m
(0)
+O
_
3
_
=
_
n +
1
2
_
+

2
2
[n(n 1) (n + 1) (n + 2)] +O
_
3
_
=
_
n +
1
2
_
_
1 2
_

_
2
_
+O
_
3
_
.
(9.179)
9. In general the subspace of angular momentum states with J = 1 is
spanned by the basis
[j = 1, m = 1 , [j = 1, m = 0 , [j = 1, m = 1 , (9.180)
j
, m
[ J
z
[j, m = m
j
,j
,m
, (9.181)
j
, m
[ J
2
[j, m = j (j + 1)
2
,j
,m
, (9.182)
j
, m
[ J
[j, m =
_
(j m) (j m+ 1)
j
,j
,m1
, (9.183)
J
= J
x
iJ
y
. (9.184)
In matrix form
J
z
=
_
_
1 0 0
0 0 0
0 0 1
_
_
, (9.185)
J
2
= 2
2
_
_
1 0 0
0 1 0
0 0 1
_
_
, (9.186)
J
+
=
2
_
_
0 1 0
0 0 1
0 0 0
_
_
, (9.187)
J
2
_
_
0 0 0
1 0 0
0 1 0
_
_
. (9.188)
9.4. Solutions
a) The Hamiltonian is given by
H = S
2
z
+
_
S
2
x
S
2
y
_
= S
2
z
+

4
_
(S
+
+S
)
2
+ (S
+
S
)
2
_
= S
2
z
+

2
_
S
2
+
+S
2
_
.
(9.189)
Thus, in matrix form
H =
2
_
_
1 0 0
0 0 0
0 0 1
_
_
+
2
_
_
_
_
0 0 1
0 0 0
0 0 0
_
_
+
_
_
0 0 0
0 0 0
1 0 0
_
_
_
_
=
2
_
_
0
0 0 0
0
_
_
.
(9.190)
b) The eigenvalues and eigenvectors are given by
2
_
_
0
0 0 0
0
_
_
_
_
1
0
1
_
_
=
2
( +)
_
_
1
0
1
_
_
, (9.191)
2
_
_
0
0 0 0
0
_
_
_
_
1
0
1
_
_
=
2
( )
_
_
1
0
1
_
_
, (9.192)
2
_
_
0
0 0 0
0
_
_
_
_
0
1
0
_
_
=
2
0
_
_
0
1
0
_
_
. (9.193)
c) The Hamiltonian is written as H = H
0
+V where in matrix form
H
0
=
2
_
_
1 0 0
0 0 0
0 0 1
_
_
, (9.194)
V =
2
_
_
0 0 1
0 0 0
1 0 0
_
_
. (9.195)
For the nondegenerate eigenenergy E
0
m=0
= 0 on has to second order
in perturbation expansion
E
m=0
= E
0
m=0
+1, 0[ V [1, 0+

m
=1
[1, m
[ V [1, 0[
2
E
0
m=0
E
0
m
= 0 . (9.196)
For the degenerate eigenenergy E
0
m=1
=
2
the perturbation in
the subspace spanned by [1, 1 , [1, 1 is given in matrix form by
V
m=1
=
2
_
0 1
1 0
_
, (9.197)
thus to rst order in perturbation expansion
E
m=1
=
2
( ) . (9.198)
10. For the unperturbed case V = 0, the eigenvectors and eigenenergies are
related by
(H
r
+H
a
) [n, = E
0
n,
[n, , (9.199)
where n = 0, 1, 2, is the quantum number of the harmonic oscillator,
and 1, +1 is the quantum number associated with the two-level
particle, and
E
0
n,
=
r
_
n +
1
2
_
+
a
2
. (9.200)
Consider rst the nondegenerate case where
r
,=
a
. To second order
in perturbation theorem
E
n,
= E
0
n,
+n, [ V [n, +

n
=n,
[n
[ V [n, [
2
E
0
n,
E
0
n
. (9.201)
Using
V [n, + = ga
[n, = g
n + 1 [n + 1, , (9.202)
V [n, = ga[n, + = g
n[n 1, + , (9.203)
one nds for = +1
E
n,+1
=
r
_
n +
1
2
_
+

a
2
+
g
2
(n + 1)
r
, (9.204)
and for = 1
E
n,1
=
r
_
n +
1
2
_

a
2

g
2
n
r
. (9.205)
For the general case this can be written as
E
n,
=
_
r
+
g
2
r
__
n +
1
2
_
+

a
2
+
1
2
g
2
r
. (9.206)
In the degenerate case
r
=
a
the eigenenergies for the case V = 0
are given by
E
0
n,
=
_
n +
1
2
+

2
_
, (9.207)
thus the pairs of states [n, + and [n + 1, are degenerate. In the subset
of such a pair the perturbation is given by
9.4. Solutions
_
n, +[ V [n, + n, +[ V [n + 1,
n + 1, [ V [n, + n + 1, [ V [n + 1,
_
=
_
0 g
n + 1
g
n + 1 0
_
,
(9.208)
thus to rst order in g the eigenenergies are given by
E =
_
(n + 1) g
n + 1
. (9.209)
11. Using creation and annihilation operators one has
H
0
=
p
2
x
+p
2
y
2m
+
1
2
m
2
_
x
2
+y
2
_
= (N
x
+N
y
+ 1) , (9.210)
where N
x
= a
x
a
x
, N
y
= a
y
a
y
, and
V =

L
2
z
=

(xp
y
yp
x
)
2
=

_
i
_
a
x
a
y
a
x
a
y
_
2
=
__
a
2
x
_
a
y
_
2
+
_
a
x
_
2
a
2
y
a
x
a
x
a
y
a
y
a
x
a
x
a
y
a
y
__
=
_
a
2
x
_
a
y
_
2
+
_
a
x
_
2
a
2
y
(1 +N
x
) N
y
N
x
(1 +N
y
)
_
.
(9.211)
a) For the case = 0 the ground state [0, 0 is nondegenerate and has
energy E
0,0
= . Since V [0, 0 = 0 one nds to second order in
E
0,0
= +0, 0[ V [0, 0
1
nx,ny=0,0
[n
x
, n
y
[ V [0, 0[
2
n
x
+n
y
= +O
_
3
_
.
(9.212)
b) For the case = 0 the rst excited state is doubly degenerate
H
0
[1, 0 = 2 [1, 0 , (9.213)
H
0
[0, 1 = 2 [0, 1 . (9.214)
The matrix of V in the basis [1, 0 , [0, 1 is given by
_
1, 0[ V [1, 0 1, 0[ V [0, 1
0, 1[ V [1, 0 0, 1[ V [0, 1
_
=
_
1, 0[ [(1 +N
x
) N
y
+N
x
(1 +N
y
)] [1, 0 0
0 0, 1[ [(1 +N
x
) N
y
+N
x
(1 +N
y
)] [0, 1
_
=
_
1 0
0 1
_
.
(9.215)
Thus to rst order in the rst excited state remains doubly de-
generate with energy 2 (1 +). Note - The exact solution can be
found using the transformation
a
d
=
1
2
(a
x
ia
y
) , (9.216)
a
g
=
1
2
(a
x
+ia
y
) . (9.217)
The following holds
_
a
d
, a
d
_
=
_
a
g
, a
= 1 ,
a
d
a
d
+a
g
a
g
=
1
2
_
a
x
+ia
y
_
(a
x
ia
y
) +
1
2
_
a
x
ia
y
_
(a
x
+ia
y
)
= a
x
a
x
+a
y
a
y
,
(9.218)
and
a
d
a
d
a
g
a
g
=
1
2
_
a
x
+ia
y
_
(a
x
ia
y
)
1
2
_
a
x
ia
y
_
(a
x
+ia
y
)
= i
_
a
x
a
y
a
x
a
y
_
,
(9.219)
thus
H
0
= (N
d
+N
g
+ 1) , (9.220)
V = (N
d
N
g
)
2
, (9.221)
and the exact eigen vectors and eigen energies are given by
(H
0
+V ) [n
d
, n
g
=
_
n
d
+n
g
+ 1 + (n
d
n
g
)
2
_
[n
d
, n
g
.
(9.222)
12. For V
0
= 0 the wavefunctions
(0)
n
(x) are given by
(0)
n
(x) = x
[n =
_
2
l
sin
nx
l
, (9.223)
E
(0)
n
=

2
2
n
2
2ml
2
. (9.224)
The matrix elements of the perturbation are given by
n[V [m =
2V
0
l
l
_
0
sin
nx
l
sin
mx
l
sin
2x
l
dx . (9.225)
For the diagonal case n = m
9.4. Solutions
n[V [n =
2V
0
l
l
_
0
sin
2
nx
l
sin
2x
l
dx (9.226)
=
2V
0
l
l/2
_
l/2
sin
2
_
ny
l
+
n
2
_
sin
_
2y
l
+
_
dy (9.227)
=
2V
0
l
l/2
_
l/2
1 cos
_
2ny
l
+n
_
2
sin
2y
l
dy (9.228)
= 0 , (9.229)
(9.230)
since the integrand is clearly an odd function of y. Thus to rst order in
V
0
the energies are unchanged
E
n
=

2
2
n
2
2ml
2
+O
_
V
2
0
_
. (9.231)
13. For the case = 0 the exact wave functions are given by
(0)
n
(x) =
_
2
L
sin
_
nx
L
_
, (9.232)
and the corresponding eigen energies are
E
(0)
n
=

2
2
n
2
2mL
2
, (9.233)
where n is integer. To rst order in the energy of the ground state n = 1
is given by
E
1
= E
(0)
1
+

L
_
L
0
dx
_
(0)
1
(x)
_
2
x +O
_
2
_
= E
(0)
1
+
2
L
2
_
L
0
dx sin
2
_
x
L
_
x +O
_
2
_
= E
(0)
1
+

2
+O
_
2
_
(9.234)
14. For the case = 0 the exact wave functions of the eigenstates are given
by
(0)
nx,ny
(x, y) =
2
l
sin
n
x
x
l
sin
n
y
y
l
, (9.235)
and the corresponding eigen energies are
E
(0)
nx,ny
=

2
2
_
n
2
x
+n
2
y
_
2ml
2
, (9.236)
where n
x
and n
y
are non-zero integers.
a) The ground state is non degenerate thus to 1st order the energy is
given by
E
0
= E
(0)
1,1
+
_
l
0
_
l
0
_
(0)
1,1
_
2
W dxdy
=

2
2
ml
2
+
2
ml
2
4
_
l
0
_
l
0
sin
2
x
l
sin
2
y
l
(x l
x
) (y l
y
) dxdy
=

2
2
ml
2
_
1 + 4sin
2
l
x
l
sin
2
l
y
l
_
.
(9.237)
b) The rst excited state is doubly degenerate. The matrix of the per-
turbation W in the eigen subspace is given by
W =
_
2, 1[ W [2, 1 2, 1[ W [1, 2
1, 2[ W [2, 1 1, 2[ W [1, 2
_
= 4
2
ml
2
_
sin
2 2l
x
l
sin
2 l
y
l
sin
2l
x
l
sin
l
x
l
sin
l
y
l
sin
2l
y
l
sin
lx
l
sin
2lx
l
sin
2ly
l
sin
ly
l
sin
2 lx
l
sin
2 2ly
l
_
= 4
2
ml
2
_
4 sin
2 lx
l
cos
2 lx
l
sin
2 ly
l
4 cos
lx
l
sin
2 lx
l
cos
ly
l
sin
2 ly
l
4 cos
lx
l
sin
2 lx
l
cos
ly
l
sin
2 ly
l
4 sin
2 lx
l
sin
2 ly
l
cos
2
ly
l
_
=
16
2
2
sin
2 lx
l
sin
2 l
y
l
ml
2
_
cos
2 lx
l
cos
lx
l
cos
ly
l
cos
l
x
l
cos
ly
l
cos
2
ly
l
_
.
(9.238)
The eigenvalues of W are
w
1
= 0 , (9.239)
and
w
2
=
16
2
2
sin
2 lx
l
sin
2 ly
l
_
cos
2 lx
l
+ cos
2
ly
l
_
ml
2
. (9.240)
15. The unperturbed Hamiltonian ( = 0) can be written as
H =
L
2
L
2
z
2I
xy
+
L
2
z
2I
z
=
L
2
2I
xy
+
_
1
2I
z

1
2I
xy
_
L
2
z
,
(9.241)
9.4. Solutions
thus the states [l, m (the standard eigenstates of L
2
and L
z
) are eigen-
states of H and the following holds
H[l, m = E
l,m
[l, m , (9.242)
where
E
l,m
=
2
_
l (l + 1)
2I
xy
+
_
1
2I
z

1
2I
xy
_
m
2
_
. (9.243)
Since the unperturbed Hamiltonian is positive-denite, it is clear that
the state [l = 0, m = 0 is the (nondegenerate) ground state of the system
since its energy vanishes E
0,0
= 0. Using
L
x
=
L
+
+L
2
, (9.244)
L
y
=
L
+
L
2i
, (9.245)
one nds that the perturbation term V can be written as
V =
L
2
+
+L
2
4I
xy
. (9.246)
To second order in the energy of the ground state is found using Eq.
(9.32)
E
0
= E
0,0
+0, 0[ V [0, 0+

l
,m
=0,0
[l
, m
[ V [0, 0[
2
E
0,0
E
l
,m
+O
_
3
_
. (9.247)
Using the relations
L
+
[l, m =
_
l (l + 1) m(m+ 1)[l, m+ 1 , (9.248)
L
[l, m =
_
l (l + 1) m(m1)[l, m1 , (9.249)
it is easy to see that all terms to second order in vanish, thus
E
0
= 0 +O
_
3
_
. (9.250)
16. The Hamiltonian can be written as
H = H
1
+H
2
+V , (9.251)
where
H
1
=
p
2
1
2m
(x
1
) , (9.252)
H
2
=
p
2
2
2m
(x
2
) , (9.253)
and
V = (x
1
x
2
) . (9.254)
First consider H
1
only. A wavefunction
(1)
(x
1
) of an eigenstate of H
1
must satisfy the following Schrdinger equation
_
d
2
dx
2
1
+
2m
2
(E + (x
1
))
_
(1)
(x
1
) = 0 . (9.255)
Integrating around x
1
= 0 yields the condition
d
(1)
(0
+
)
dx
1

d
(1)
(0
)
dx
1
+
2m
2

(1)
(0) = 0 . (9.256)
Requiring also that the wavefunction is normalizable leads to
(1)
(x
1
) =
_
m
2
exp
_
2
[x
1
[
_
.
The corresponding eigenenergy is
E
(1)
0
=
m
2
2
2
.
The ground state of H
2
can be found in a similar way. Thus, the normal-
ized wavefunction of the only bound state of H
1
+H
2
, which is obviously
the ground state, is given by
0
(x
1
, x
2
) =
m
2
exp
_
2
[x
1
[
_
exp
_
2
[x
2
[
_
, (9.257)
and the corresponding energy is given by
E
0
=
m
2
2
. (9.258)
Therefore, to rst order in the energy of the ground state of H is given
by Eq. (9.32)
E
gs
=
m
2
2
+
dx
1
dx
2

0
(x
1
, x
2
) (x
1
x
2
)
0
(x
1
, x
2
) +O
_
2
_
=
m
2
2
+
_
m
2
_
2

_
dx
1
exp
_
4m
2
[x
1
[
_
+O
_
2
_
=
m
2
2
+
m
2
2
+O
_
2
_
.
(9.259)
9.4. Solutions
17. Substituting the expansions
[n = [n
0
+[n
1
+
2
[n
2
+O
_
3
_
, (9.260)
and
=
n0
+
n1
+
2
n2
+O
_
3
_
, (9.261)
into Eq. (9.88) and collecting terms having the same order in (up to
second order) yield
(D
n0
) [n
0
= 0 , (9.262)
(D
n0
) [n
1
+ (V
n1
) [n
0
= 0 , (9.263)
(D
n0
) [n
2
+ (V
n1
) [n
1

n2
[n
0
= 0 . (9.264)
We further require normalization
n[n = 1 , (9.265)
and choose the phase of n
0
[n such that
n
0
[n . (9.266)
Expressing the normalization condition using Eq. (9.260) and collecting
terms having the same order in yield
n
0
[n
0
= 1 , (9.267)
n
0
[n
1
+n
1
[n
0
= 0 , (9.268)
n
0
[n
2
+n
2
[n
0
+n
1
[n
1
= 0 . (9.269)
These results together with Eq. (9.266) yield
n
0
[n
1
= n
1
[n
0
= 0 , (9.270)
n
0
[n
2
= n
2
[n
0
=
1
2
n
1
[n
1
. (9.271)
Multiplying Eq. (9.263) by m
0
[ yields
n1
m
0
[n
0
= (
m0
n0
) m
0
[n
1
+m
0
[ V [n
0
, (9.272)
thus for m = n
n1
= n
0
[ V [n
0
. (9.273)
Using this result for m ,= n yields
m
0
[n
1
=
m
0
[ V [n
0
n0
m0
, (9.274)
thus with the help of Eq. (9.87) one has
[n
1
=

m
m
0
[ V [n
0
n0
m0
[m
0
. (9.275)
Multiplying Eq. (9.264) by n
0
[ yields
n2
= n
0
[ V [n
1

n1
n
0
[n
1
, (9.276)
or using Eq. (9.275)
n2
=

m
n
0
[ V [m
0
m
0
[ V [n
0
n0
m0
. (9.277)
Thus, using this result together with Eq. (9.273) one nds
=
n0
+n
0
[ V [n
0
+
2
m
n
0
[ V [m
0
m
0
[ V [n
0
n0
m0
+O
_
3
_
.
(9.278)
18. The condition [9.94] together with Eq. (9.90) can be used to evaluate the
matrix elements of L
k[ L[k
=
k[
V [k
E
k
E
k
. (9.279)
H
R
= H
0
+
V +[L, H
0
]+
_
L,
V
_
+
1
2!
_
L,
_
L, H
0
+
V
__
+ . (9.280)
Thus, for the case where condition (9.94) is satised the following holds
[note that according to Eq. (9.279) L = O()]
H
R
= H
0
+
_
L,
V
_
+
1
2!
[L, [L, H
0
]] +O
_
3
_
= H
0
+
1
2
_
L,
V
_
+O
_
3
_
.
(9.281)
where Eq. (9.94) has been employed in the last step. The last result
together with Eq. (9.279) and the closure relation 1 =
k
[k
[ [see
Eq. (2.23)] lead to
k[ H
R
[k
= k[ H
0
[k
+
1
2
k[
_
L,
V
_
[k
+O
_
3
_
= E
k
k,k
+

2
k
k[

V [k
[

V [k
_
1
E
k
E
k

1
E
k
E
k
_
2
+O
_
3
_
.
(9.282)
10. Time-Dependent Perturbation Theory
Recall that the time evolution of a state vector [ is governed by the
Schrdinger equation (4.1)
i
d[
dt
= H[ , (10.1)
where the Hermitian operator H = H
is the Hamiltonian of the system. The

time evolution operator u(t, t
0
) [see Eq. (4.4)] relates the state vector [(t
0
)
at time t
0
with its value [(t) at time t
[(t) = u(t, t
0
) [(t
0
) . (10.2)
As we have seen in chapter 4, when the Hamiltonian is time independent
u(t, t
0
) is given by
u(t, t
0
) = exp
_
i (t t
0
)
H
_
. (10.3)
In this chapter we consider the more general case where H is allowed to vary
in time. We rst derive a formal expression for the time evolution operator
u(t, t
0
) applicable for general H. Then we present the perturbation theory
expansion of the time evolution operator, and discuss approximation schemes
to evaluated u(t, t
0
).
10.1 Time Evolution
Dividing the time interval (t
0
, t) into N sections of equal duration allows
expressing the time evolution operator as
u(t, t
0
) =
N
n=1
u(t
n
, t
n1
) , (10.4)
where
t
n
= t
0
+n , (10.5)
Chapter 10. Time-Dependent Perturbation Theory
and where
=
t t
0
N
. (10.6)
Furthermore, according to the Schrdinger equation (4.7), the following holds
u(t
n1
+, t
n1
) = 1
i
H(t
n
) +O
_
2
_
. (10.7)
In the limit where N higher than rst order terms in , i.e. O
_
2
_
terms, are not expected to contribute, thus the time evolution operator can
be expressed as
u(t, t
0
) = lim
N
N
n=1
_
1
i
H(t
n
)
_
. (10.8)
10.2 Perturbation Expansion
Consider the case where
H = H
0
+H
1
, (10.9)
where is real. The perturbation expansion expresses the time evolution
operator u(t, t
0
) of the full Hamiltonian H as
u(t, t
0
) = u
0
(t, t
0
) +u
1
(t, t
0
) +
2
u
2
(t, t
0
) +O
_
3
_
, (10.10)
where u
0
(t, t
0
) is the time evolution of the Hamiltonian H
0
. Such an expan-
sion can be very useful for cases where u
0
(t, t
0
) can be exactly calculated and
where the parameter is small, i.e. [[ 1. For such cases only low order
terms in this expansion are needed for approximately evaluating u(t, t
0
).
By employing Eq. (10.8)
u(t, t
0
) = lim
N
N
n=1
_
1
i
(H
0
(t
n
) +H
1
(t
n
))
_
, (10.11)
one easily obtain the terms u
0
, u
1
and u
2
u
0
(t, t
0
) = lim
N
N
n=1
_
1
i
H
0
(t
n
)
_
, (10.12)
u
1
(t, t
0
) = lim
N
N
n=1
i
u
0
(t, t
n
) H
1
(t
n
) u
0
(t
n
, t
0
)
=
i
t
_
t
0
dt
u
0
(t, t
) H
1
(t
) u
0
(t
, t
0
) ,
(10.13)
10.3. The Operator O(t) = u
0
(t, t
0
) u(t, t
0
)
and
u
2
(t, t
0
) = lim
N
N1
n=1
N
m=n+1
_
_
2
u
0
(t, t
n
) H
1
(t
n
) u
0
(t
n
, t
m
) H
1
(t
m
) u
0
(t
m
, t
0
)
=
1
2
t
_
t
0
dt
_
t
0
dt
u
0
(t, t
) H
1
(t
) u
0
(t
, t
) H
1
(t
) u
0
(t
, t
0
) .
(10.14)
10.3 The Operator O(t) = u
0
(t, t
0
) u(t, t
0
)
As can be seen from its denition, the operator O(t) = u
0
(t, t
0
) u(t, t
0
)
combines forward time evolution governed by the full Hamiltonian H from
time t
0
to time t, and backward time evolution governed by the unperturbed
Hamiltonian H
0
from time t back to time t
0
. Using the above expansion one
can calculate the operator O(t) to second order in
O(t) = 1
i
t
_
t0
dt
u
0
(t
0
, t
) H
1
(t
) u
0
(t
, t
0
)
2
t
_
t0
dt
_
t0
dt
u
0
(t
0
, t
) H
1
(t
) u
0
(t
, t
) H
1
(t
) u
0
(t
, t
0
)
+O
_
3
_
,
(10.15)
or
O(t) = 1
i
t
_
t0
dt
H
1I
(t
t
_
t0
dt
_
t0
dt
H
1I
(t
) H
1I
(t
)
+O
_
3
_
,
(10.16)
where H
1I
(t), which is dened by
H
1I
(t) u
0
(t, t
0
) H
1
(t
0
) u
0
(t, t
0
) , (10.17)
is the so called interaction representation of H
1
with respect to u
0
.
Exercise 10.3.1. Calculate the expectation value squared [O(t)[
2
to low-
est nonvanishing order in .
Solution 10.3.1. Since H
1
(t) is Hermitian one nds that
[O(t)[
2
=
_
1
i
t
_
t0
dt
H
1I
(t
)

2
2
t
_
t0
dt
_
t0
dt
H
1I
(t
) H
1I
(t
)
_
_
1 +
i
t
_
t0
dt
H
1I
(t
)

2
2
t
_
t0
dt
_
t0
dt
H
1I
(t
) H
1I
(t
)
_
= 1 +

2
2
_
t
_
t0
dt
H
1I
(t
)
_
2
2
t
_
t0
dt
_
t0
dt
(H
1I
(t
) H
1I
(t
) +H
1I
(t
) H
1I
(t
))
+O
_
3
_
,
(10.18)
or
[O(t)[
2
= 1 +

2
2
_
t
_
t0
dt
H
1I
(t
)
_
2
2
t
_
t0
dt
t
_
t0
dt
H
1I
(t
) H
1I
(t
) ,
(10.19)
thus
[O(t)[
2
= 1

2
2
t
_
t0
dt
t
_
t0
dt
[H
1I
(t
) H
1I
(t
) H
1I
(t
) H
1I
(t
)] ,
(10.20)
or
[O(t)[
2
= 1

2
2
t
_
t
0
dt
t
_
t
0
dt
H
1I
(t
) H
1I
(t
) , (10.21)
where
H
1I
(t) = H
1I
(t) H
1I
(t) . (10.22)
10.4 Transition Probability
Consider the case where the unperturbed Hamiltonian H
0
is time indepen-
dent. The eigenvectors of H
0
are denoted as [a
n
, and the corresponding
eigenenergies are denoted as E
n
10.4. Transition Probability
H
0
[a
n
= E
n
[a
n
, (10.23)
where
a
n
[a
n
=
nn
. (10.24)
In this basis u
0
(t, t
0
) is given by
u
0
(t, t
0
) = exp
_
i (t t
0
)
H
0
_
=

n
exp
_
iE
n
(t t
0
)
_
[a
n
a
n
[ .
(10.25)
Assuming that initially at time t
0
the system is in state [a
n
, what is the
probability to nd it later at time t > t
0
in the state [a
m
? The answer to
this question is the transition probability p
nm
, which is given by
p
nm
= [a
m
[ u(t, t
0
) [a
n
[
2
. (10.26)
e
iEm(tt
0
)
a
m
[ u(t, t
0
) [a
n
= a
m
[ O(t) [a
n
=
nm
t
_
t0
dt
a
m
[ H
1I
(t
) [a
n
t
_
t0
dt
_
t0
dt
a
m
[ H
1I
(t
) H
1I
(t
) [a
n
+O
_
3
_
,
(10.27)
thus
p
nm
=
nm
t
_
t0
dt
a
m
[ H
1I
(t
) [a
n
t
_
t0
dt
_
t0
dt
a
m
[ H
1I
(t
) H
1I
(t
) [a
n
+O
_
3
_
2
.
(10.28)
In what follows, we calculate the transition probability p
nm
to lowest non-
vanishing order in for the case where n ,= m, for which the dominant
contribution comes from the term of order in Eq. (10.28). For simplicity
the initial time t
0
, at which the perturbation is turned on, is taken to be zero,
i.e. t
0
= 0. We consider below the following cases:
10.4.1 The Stationary Case
In this case H
1
is assumed to be time independent (after being turned on at
t
0
= 0). To lowest nonvanishing order in Eq. (10.28) yields
p
nm
=

2
t
_
0
dt
e
imnt
2
[a
m
[ H
1
(t
) [a
n
[
2
, (10.29)
where
mn
=
E
m
E
n
. (10.30)
Using the identity
t
_
0
dt
e
it
= 2e
i
t
2
sin
_
t
2
_
, (10.31)
one nds that
p
nm
=
4
2
sin
2 mnt
2
2
mn
[a
m
[ H
1
(t
) [a
n
[
2
. (10.32)
Note that in the limit t one nds with the help of Eq. (10.31) that
lim
t
4 sin
2
_
t
2
_
2
= lim
t
t
_
0
e
it
dt
2
= lim
t
t
_
0
dt
t
_
0
dt
e
i(t
)
= 2 ()
t
_
0
dt
= 2t () .
(10.33)
In this limit p
nm
is proportional to the time t, i.e. p
nm
can be expressed as
p
nm
= w
nm
t, where w
nm
is the transition rate, which is given by
w
nm
=
2
2
(
mn
) [a
m
[ H
1
(t
) [a
n
[
2
. (10.34)
The delta function (
mn
) ensures that energy is conserved in the limit
of long time, and transitions between states having dierent energies are
excluded. However, such transitions have nite probability to occur for any
nite time interval t. On the other hand, as can be see from Eq. (10.32)
(see also the gure below, which plots the function f (x) = sin
2
x/x
2
), the
probability is signicant only when
mn
t 1, or alternatively when
Et , (10.35)
where E =
mn
.
0
0.2
0.4
0.6
0.8
-10 -8 -6 -4 -2 2 4 6 8 10
x
The function f (x) = sin
2
x/x
2
.
10.4.2 The Near-Resonance Case
In this case H
1
is assumed to be given by
H
1
(t
) = /e
it
+/
e
it
, (10.36)
where / is an operator that is assumed to be time independent (after being
turned on at t
0
= 0), and where the angular frequency is a positive constant.
The transition probability is given by [see Eq. (10.28)]
p
nm
=
4
e
i
(mn)t
2
sin
_
(mn)t
2
_
a
m
[ /[a
n
mn
+
e
i
(mn+)t
2
sin
_
(mn+)t
2
_
a
m
[ /
[a
n
mn
+
2
,
(10.37)
We refer to the case where =
mn
as absorption resonance, and to the
case where =
mn
as stimulated emission resonance. Near any of these
resonances
mn
the dominant contribution to p
nm
comes from only
one out of the two terms in Eq. (10.37), thus
p
nm

_
_
_
4
2
sin
2 (mn)t
2
(mn)
2
[a
m
[ /[a
n
[
2
mn

4
2
sin
2 (
mn
+)t
2
(
mn
+)
2
a
m
[ /
[a
n
mn

. (10.38)
In the long time limit, i.e. in the limit t , the probability p
nm
is
found using Eq. (10.33) to be proportional to the time t, i.e. p
nm
= w
nm
t,
where the transition rate w
nm
is given by
w
nm

_
2
2
(
mn
) [a
m
[ /[a
n
[
2
mn

2
2
(
mn
+)
a
m
[ /
[a
n
mn

. (10.39)
In many cases of interest the nal state [a
m
lie in a band of dense states. Let
w
n
be the total transition rate from the initial state [a
n
. Assume that the
matrix element a
m
[ /[a
n
does not vary signicantly as a function of the
energy E
m
. For this case the total rate w
n
can be expressed in terms of the
density of states g (E
m
) (i.e. number of states per unit energy) in the vicinity
of the nal state [a
m
[see Eq. (10.39)] as
w
n
=
2
g (E
m
) [a
m
[ /[a
n
[
2
, (10.40)
where E
m
= E
n
+. This result is known as the Fermis golden rule.
10.4.3 H
1
is Separable
For this case it is assumed that H
1
can be expressed as
H
1
(t
) = f (t
)

H
1
, (10.41)
where f (t
) is a real function of time and where

H
1
is time independent
Hermitian operator. To lowest nonvanishing order in Eq. (10.28) yields
p
nm
=
1
t
_
0
dt
e
imnt
f (t
a
m
[

H
1
[a
n
2
. (10.42)
10.5 Problems
1. Find the exact time evolution operator u(t, 0) of the Hamiltonian H,
which is given by
H = H
0
+H
p
, (10.43)
where
H
0
= a
a , (10.44)
H
p
= i (t)
_
e
2i(t)
a
2
e
2i(t)
a
2
_
, (10.45)
a and a
are the annihilation and creation operators (as dened in chapter

5), is positive, is real and (t) is an arbitrary real function of time t.
10.5. Problems
2. Consider a particle having mass m moving under the inuence of a one
dimensional potential given by
V (x) =
m
2
0
x
2
2
, (10.46)
where the angular resonance frequency
0
is a constant. A perturbation
given by
H
1
(t
) = 2xcos (t
) , (10.47)
where the real constant is assumed to be small, is turned on at time
t = 0. Given that the system was initially at time t = 0 in the ground
state [0 of the unperturbed Hamiltonian, calculate the transition proba-
bility p
n0
(t) to the number state [n to lowest nonvanishing order in the
perturbation expansion.
3. Repeat the previous exercise with the perturbation
H
1
(t
) = xf (t
) , (10.48)
where the force f (t
) is given by
f (t
) =
exp
_
t
2
2
_
, (10.49)
and where both and are real. Given that the system was initially at
time t in the ground state [0 of the unperturbed Hamiltonian,
nd the transition probability p
n0
to the number state [n in the limit
t . Compare your approximated result with the exact result given
by Eq. (5.325).
4. Consider a spin 1/2 particle. The Hamiltonian is given by
H = S
x
, (10.50)
where is a Larmor frequency and where S
x
is the x component of the
angular momentum operator. Given that the spin is initially at time t = 0
in the eigenstate [+; z of the operator S
z
(having eigenvalue +/2), what
is the probability p
++
(t) to nd the spin at the same state [+; z at a
later time t. Compare the exact result with the approximated value that
is obtained from Eq. (10.21).
5. Consider a particle having mass m conned in a potential well given by
V (x) =
_
0 if 0 x a
if x < 0 or x > a
. (10.51)
The particle is initially at time t in the ground state of the well.
A small perturbation
H
1
(t) =

a
xe
(
t
)
2
, (10.52)
where 1 and where is a positive constant having the dimensions
of time, is applied. Calculate the probability to nd the particle in the
rst excited state in the limit t .
10.6 Solutions
1. Expressing the ket vector state as
[ = e
iH0t/
[
I
, (10.53)
and substituting into the Schrdinger equation, which is given by
i
d[
dt
= (H
0
+H
p
) [ , (10.54)
yield
i
d[
I
dt
= H
I
[
I
. (10.55)
where H
I
, which is given by
H
I
= e
iH0t/
H
p
e
iH0t/
, (10.56)
is the interaction picture representation of H
p
. With the help of the vector
identity (2.173), which is given by
e
L
Ae
L
= A+[L, A] +
1
2!
[L, [L, A]] +
1
3!
[L, [L, [L, A]]] + , (10.57)
and the relations
it
_
H
0
, a
2
= 2ita
2
, (10.58)
and
it
_
H
0
, a
2
= 2ita
2
, (10.59)
one nds that
e
iH0t/
a
2
e
iH0t/
= a
2
e
2it
, (10.60)
e
iH0t/
a
2
e
iH0t/
= a
2
e
2it
, (10.61)
thus
10.6. Solutions
H
I
= i (t)
_
e
2i
a
2
e
2i
a
2
_
. (10.62)
Since [H
I
(t) , H
I
(t
)] = 0 the solution of Eq. (10.55) is given by

[
I
(t) = exp
_
_
t
0
dt
H
I
(t
)
_
[
I
(0)
= S (, ) [
I
(0) ,
(10.63)
where
S (, ) = exp
_
_
e
2i
a
2
e
2i
a
2
_
, (10.64)
and where
=
_
t
0
dt
(t
) , (10.65)
and thus the time evolution operator is thus given by
u(t, 0) = e
iH0t/
S (, ) . (10.66)
2. To lowest nonvanishing order in the perturbation expansion one nds
using Eq. (10.38) together with Eqs. (5.11), (5.28) and (5.29) that
p
n0
(t) =
2
2
m
sin
2 (0)t
2
(
0
)
2

n,1
. (10.67)
3. To lowest nonvanishing order in perturbation expansion Eq. (10.42) yields
p
n0
=
n,1
, (10.68)
where
=
1
2m
0
dt
e
i0t
f (t
2
=

2
2m
0
e
1
2
2
0
.
(10.69)
The exact result is found from Eq. (5.325)
p
n
=
e
n
n!
. (10.70)
To rst order in both results agree.
4. While the exact result is [see Eq. (6.284)]
p
++
(t) = cos
2
t
2
, (10.71)
Eq. (10.21) yields
p
++
(t) = 1
1
2
t
_
0
dt
t
_
0
dt
+; z[ (S
x
+; z[ S
x
[+; z)
2
[+; z
= 1

2
+; z[ S
2
x
[+; z
2
t
_
0
dt
t
_
0
dt
= 1
_
t
2
_
2
.
(10.72)
5. The normalized wavefunctions
n
(x
) of the wells energy eigenstates are

given by
n
(x
) =
_
2
a
sin
nx
a
, (10.73)
E
n
=

2
2
n
2
2ma
2
, (10.74)
where n = 1, 2, . The transition probability is calculated to lowest
nonvanishing order in

with the help of Eq. (10.42)
p
2,1
=
_

a
_
2
t
_
0
dt
e
i(E
2
E
1
)t
/
e
_
t
_
2
2
a
a
_
0
dx
sin
2x
a
sin
x
2
,
(10.75)
thus [see Eq. (8.229)
p
2,1
=
256
2
81
3
exp
_
9
4
2
8m
2
a
4
_
. (10.76)
11. WKB Approximation
The theory of geometrical optics provides an approximated solution to
Maxwells equation that is valid for systems whose typical size scales are
much larger than the wavelength of electromagnetic waves. In 1926 using a
similar approach the physicists Wentzel, Kramers and Brillouin (WKB) in-
dependently found an approximated solution to the Schrdinger equation in
the coordinate representation for the case where the wavelength associated
with the wavefunction (to be dened below) can be considered as short. Be-
low the WKB approximation is discussed for the time independent and one
dimensional case. This chapter is mainly based on Ref. [3].
11.1 WKB Wavefunction
Consider a point particle having mass m moving under the inuence of a one-
dimensional potential V (x). The time independent Schrdinger equation for
the wavefunction (x) is given by [see Eq. (4.50)]
d
2
(x)
dx
2
+
2m
2
(E V (x)) (x) = 0 , (11.1)
where E is the energy. In terms of the local momentum p (x), which is dened
by
p (x) =
_
2m(E V (x)) , (11.2)
the Schrdinger equation becomes
d
2
(x)
dx
2
+
_
p
_
2
(x) = 0 . (11.3)
Using the notations
(x) = e
iW(x)/
, (11.4)
and the relation
d
2
(x)
dx
2
=
_
i
d
2
W
dx
2

_
1
dW
dx
_
2
_
(x) , (11.5)
Chapter 11. WKB Approximation
one nds that the Schrdinger equation can be written as
i
d
2
W
dx
2

_
dW
dx
_
2
+p
2
= 0 . (11.6)
In the WKB approach the Planks constant is treated as a small para-
meter. Expanding W as a power series in
W = W
0
+W
1
+
2
W
2
+ (11.7)
one nds that
_
dW
0
dx
_
2
+i
d
2
W
0
dx
2
2
dW
0
dx
dW
1
dx
+p
2
+O
_
2
_
= 0 . (11.8)
The terms of order zero in yield
_
dW
0
dx
_
2
+p
2
= 0 . (11.9)
thus
W
0
(x) =
x
_
x
0
dx
p (x
) , (11.10)
where x
0
is a constant.
What is the range of validity of the zero order approximation? As can be
seen by comparing Eq. (11.6) with Eq. (11.9), the approximation W W
0
is valid when the rst term in Eq. (11.6) is negligibly small in absolute value
in comparison with the second one, namely when
d
2
W
dx
2
dW
dx
2
. (11.11)
It is useful to express this condition in terms of the local wavelength (x),
which is given by
(x) =
2
p (x)
. (11.12)
By employing the lowest order approximation dW/dx = p the condition
(11.11) becomes
d
dx
2 . (11.13)
This means that the approximation is valid provided that the change in
wavelength over a distance of one wavelength is small.
11.1. WKB Wavefunction
The terms of 1st order in of Eq. (11.8) yield an equation for W
1
dW
1
dx
=
i
2
d
2
W0
dx
2
dW0
dx
=
i
2
d
dx
log
_
dW
0
dx
_
. (11.14)
Using Eq. (11.9) one thus has
d
dx
_
iW
1
log
1
p
_
= 0 . (11.15)
Therefor, to 1st order in the wave function is given by
(x) = C
+
+
(x) +C
(x) , (11.16)
where
(x) =
1
p
exp
_
_
x
_
x0
dx
p (x
)
_
_
, (11.17)
and where both C
+
and C
are constants.
In general, the continuity equation (4.70), which is given by
d
dt
+
dJ
dx
= 0 , (11.18)
relates the probability density = [[
2
and the current density J =
(/m) Im(
d/dx) associated with a given one dimensional wavefunction

(x). For a stationary (x) the probability density is time independent,
and thus J is a constant. Consider a region where E > V (x). In such a region,
which is classically accessible, the momentum p (x) is real and positive, and
thus the probability density (x) of the WKB wavefunctions
(x) is pro-
portional to 1/p. This is exactly what is expected from a classical analysis
of the dynamics, where the time spent near a point x is inversely propor-
tional to the local classical velocity at that point v (x) = p (x) /m. With the
help of Eq. (4.181) one nds that the current density J associated with the
wavefunction (11.16) is given by
J =

m
Im
_
_
C
+
+C
_
_
C
+
d
+
dx
+C
dx
__
=

m
_
[C
+
[
2
Im
_
+
d
+
dx
_
+[C
[
2
Im
_
dx
_
+Im
_
C
+
C
+
d
dx
+C
+
C
d
+
dx
__
.
(11.19)
As can be seen from Eq. (11.17), the last term vanishes since
(x) =
+
(x).
Therefor, with the help of Eq. (6.340) one nds that
J =
1
m
_
[C
+
[
2
[C
[
2
_
. (11.20)
Thus, the current density J associated with the state
+
(x) is positive,
whereas J < 0 for
(x). Namely,
+
(x) describes a state propagating from
left to right, whereas
(x) describes a state propagating in the opposite

direction.
11.2 Turning Point
Consider a point x = a for which E = V (a), namely p (a) = 0 [see Fig.
11.1 (a)]. Such a point is called a turning point since a classical particle that
reaches the point x = a momentarily stops and changes its direction. Near
a turning point the local wavelength diverges, and consequently, as can be
seen from Eq. (11.13), the WKB approximation breaks down. Consider the
case where E > V (x) for x > a and where E < V (x) for x < a. In the region
x > a the WKB wave function is expressed using Eq. (11.16), where, for
convenient, the constant x
0
is chosen to be a. However, on the other side of
the turning point, namely for x < a, the momentum p (x) becomes imaginary
since E < V (x). Thus, in this region, which is classically forbidden, the wave
function given by Eq. (11.16) contains one exponentially decaying term in
the limit x and another exponentially diverging term in the same
limit. To ensure that the wavefunction remains normalizable, the coecient
of the exponentially diverging term is required to vanish, and thus we seek a
solution having the form
(x) =
_
_
C
|p|
exp
_
1
x
_
a
dx
[p[
_
x < a
C+
p
exp
_
i
x
_
a
dx
p
_
+
C
p
exp
_
x
_
a
dx
p
_
x > a
. (11.21)
Note that the pre-factor 1/
p in the classically forbidden region x < a is

substituted in Eq. (11.21) by 1/
_
[p[. The ratio between these two factors
in the region x < a is a constant, which is assumed to be absorbed by the
constant C. For given value of C, what are the values of C
+
and C
? It
should be kept in mine that Eq. (11.21) becomes invalid close to the turning
point x = a where the WKB approximation breaks down. Thus, this question
cannot be simply answered by requiring that (x
) and its rst derivative

are continuous at x = a [e.g., see Eq. (4.122)].
As we have seen above, the WKB approximation breaks down near the
turning point x = a. However the two regions x < a and x > a can be tailored
together by the technique of analytical continuation. In the vicinity of the
turning point, namely for x a, the potential V (x) can be approximated by
V (x) V (a) (x a) , (11.22)
11.2. Turning Point
x
a
( ) E x V
( ) x Re
a
( ) x Im
( ) x Re
a
( ) x Im
( ) x Re
a
( ) x Im
(a) (b)
(c) (d)
x
a
( ) E x V
( ) x Re
a
( ) x Im
( ) x Re
a
( ) x Im
( ) x Re
a
( ) x Im
(a) (b)
(c) (d)
Fig. 11.1. (a) The turning point at x = a. (b) The integration trajectory
. The
singly connected region
+
(c) and
(d).
where
=
dV
dx
x=a
, (11.23)
and thus for x a
p (x)
_
2m(x a) . (11.24)
Formally, the coordinate x can be considered as complex. Consider a circle
in the complex plane centered at x = a having a radius . The radius > 0
is assumed to be suciently large to ensure the validity of the WKB approx-
imation outside it. On the other hand, it is also assumed to be suciently
small to allow the employment of the approximation (11.24), namely, for any
point on that circle
x = a +e
i
, (11.25)
where is real, it is assumed that
p (x)
_
2me
i/2
. (11.26)
We consider below analytical continuation of the wavefunction given by Eq.
(11.21) for the case x > a into a region in the complex plane. Such a region
must excludes the vicinity of the turning point x = a where the WKB approx-
imation breaks down and in addition it is required to be singly connected to
allow analytical continuation. Two such regions are considered below, the rst
one, which is labeled as
+
(see Fig. 11.1 (c)), excludes the circle [x a[
and also excludes the negative imaginary line x = a ib, whereas the sec-
ond one, which is labeled as
(see Fig. 11.1 (d)), also excludes the circle

[x a[ and in addition excludes the positive imaginary line x = a + ib,
where in both cases the parameter b is assumed to be real and positive.
To perform the tailoring it is convenient to dene the term
I
() =
i
dx
p ,
where the integration trajectory
[see Fig. 11.1 (b)] contains two sections,

the rst along the real axis from x = a to x = a+ and the second along the
arc x = a +e
i
from
= 0 to
= . With the help of the approximation

(11.26) one nds that
I
() =
i
2m
_
_

_
0
d
+i
3/2
_
0
d
e
i3
/2
_
_
=
i
2m
_
_
2
3
3/2
i
3/2
2i
_
e
3
2
i
1
_
3
_
_
=
2i
2m
3/2
3
e
3
2
i
=
2
2m
3/2
3
e
i((1
1
2
)+
3
2
)
,
(11.27)
thus
I
() =
2
2m
3/2
3
, (11.28)
I
() =
2
2m
3/2
3
. (11.29)
The last result allows expressing the analytical continuation of the wavefunc-
tion given by Eq. (11.21) for the case x > a and evaluate its value at the
point x = a . For the case where the singly connected region
+
(
) is
employed, this is done using integration along the trajectory
), and
the result is labeled as
+
(a ) [
(a )]
11.2. Turning Point
+
(a ) =
C
+
exp
_
2
2m
3/2
3
_
+C
exp
_
2m
3/2
3
_
(2m)
1/4
e
i/4
, (11.30)
(a ) =
C
+
exp
_
2m
3/2
3
_
+C
exp
_
2
2m
3/2
3
_
(2m)
1/4
e
i/4
. (11.31)
Note that the denominators of Eqs. (11.30) and (11.31) are evaluated by
analytical continuation of the factor
p [see Eq. (11.26)] along the trajectories
and
respectively. On the other hand, according to Eq. (11.21) in the

region x < a one nds by integration along the real axis that
(a ) =
C
(2m)
1/4
exp
_
2m
3
2
3
_
. (11.32)
Comparing Eqs. (11.30) and (11.31) with Eq. (11.32) shows that for each of
the two choices
+
and
the analytical continuation yields one exponential

term having the same form as the one in Eq. (11.32), and another one, which
diverges in the limit x . Excluding the diverging terms one nds that
continuity of the non diverging term requires that
C =
C
+
e
i/4
=
C
e
i/4
, (11.33)
and thus the tailored wavefunction is given by
(x) =
_
_
C
|p|
exp
_
1
x
_
a
dx
[p[
_
x < a
2C
p
cos
_
1
x
_
a
dx
p

4
_
x > a
. (11.34)
The fact that analytical continuation of the wavefunction in the region x < a
along the trajectory
) yields only the right to left (left to right) prop-

agating term in the region x > a, and the other term is getting lost along the
way, can be attributed to the limited accuracy of the WKB approximation.
As can be seen from Eq. (11.27), along the integration trajectory
near the
point = /3 one term becomes exponentially larger than the other, and
consequently, within the accuracy of this approximation the small term gets
lost.
It is important to keep in mind that the above result (11.34) is obtained
by assuming a particular form for the solution in the region x < a, namely
by assuming that in the classically forbidden region the coecient of the
exponentially diverging term vanishes. This tailoring role will be employed in
the next section that deals with bound states in a classically accessible region
between two turning points [see Fig. 11.2(a)]. On the other hand, a modied
tailoring role will be needed when dealing with quantum tunneling. For this
case, which will be discussed below, we seek a wave function having the form
x
a
( ) E x V
(a)
b
x
( ) E x V
(b)
a
b
x
a
( ) E x V
(a)
b
x
( ) E x V
(b)
a
b
Fig. 11.2. The region a x b bounded by the two turning points at x = a
and x = b is classically accessible in panel (a), whereas it is classically forbidden in
panel (b) .
(x) =
_
_
C+
|p|
exp
_
1
x
_
a
dx
[p[
_
+
C
|p|
exp
_
x
_
a
dx
[p[
_
x < a
C
p
exp
_
i
x
_
a
dx
p +
i
4
_
x > a
. (11.35)
Thus, in this case only the term describing propagation from left to right
is kept in the region x > a, and the coecient of the other term in that
region that describes propagation in the opposite direction is assumed to
vanish. Using the same tailoring technique as in the previous case one nd
that C
+
= 0 and C
= C, and thus
(x) =
_
_
C
|p|
exp
_
x
_
a
dx
[p[
_
x < a
C
p
exp
_
i
x
_
a
dx
p +
i
4
_
x > a
. (11.36)
11.3 Bohr-Sommerfeld Quantization Rule
Consider a classical accessible region a x b bounded by two turning
points at x = a and x = b, namely, consider the case where E > V (x) for
a x b and where E < V (x) for x < a and for x > b [see Fig. 11.2(a)].
We seek a normalizable solution, thus the wave function in the classically
forbidden regions x < a and for x > b is assumed to vanish in the limit
x . Employing the tailoring role (11.34) with respect to the turning
point at x = a yields the following wave function for the region a x b
a
(x) =
2C
a
p
cos
_
_
1
x
_
a
dx
p

4
_
_
, (11.37)
11.3. Bohr-Sommerfeld Quantization Rule
where C
a
is a constant. Similarly, employing the tailoring role (11.34) with
respect to the turning point at x = b yields
b
(x) =
2C
b
p
cos
_
_
1
b
_
x
dx
p

4
_
_
. (11.38)
The requirement
a
(x) =
b
(x) can be satised for any x in the region
a x b only if
1
b
_
a
dx
p =

2
+n . (11.39)
where n is integer. Alternatively, this result, which is known as Bohr-
Sommerfeld quantization rule, can be expressed as
1
2
_
dx
p = n +
1
2
, (11.40)
where
_
dx
p = 2
b
_
a
dx
p . (11.41)
To normalize the wavefunction
a
(x) =
b
(x) we assume that (a) only
the accessible region a x b contributes, since outside this region the
wavefunction exponentially decays; and (b) in the limit of large n the co-
sine term rapidly oscillates and therefore the average of its squared value is
approximately 1/2. Applying these assumptions to
a
(x), which is given by
Eq. (11.37), implies that
1
b
_
a
dx
[
a
(x)[
2
2 [C
a
[
2
b
_
a
dx
p
. (11.42)
Note that the time period T of classical oscillations between the turning
points x = a and x = b is given by
T = 2
b
_
a
dx
v
. (11.43)
where v (x) = p (x) /m is the local classical velocity. Thus, by choosing the
pre-factor to be real, one nds that the normalized wavefunction is given by
(x) = 2
_
m
pT
cos
_
_
1
x
_
a
dx
p

4
_
_
. (11.44)
The Bohr-Sommerfeld quantization rule (11.40) can be used to relate the
classical time period T with the energy spacing E = E
n+1
E
n
between
consecutive quantum eigen energies. As can be seen from the validity condi-
tion of the WKB approximation (11.13), the integer n is required to be large
to ensure the validity of Eq. (11.40). In this limit E E, and thus by
taking the derivative of Eq. (11.40) with respect to energy one nds that
E
_
dx
_
E
p
_
1
= 2 . (11.45)
In classical mechanics E/p is the velocity of the particle v, therefor
_
dx
_
E
p
_
1
= T , (11.46)
thus
E =
2
T
. (11.47)
11.4 Tunneling
In this case we consider a classical forbidden region a x b bounded by
two turning points at x = a and x = b, namely, it is assumed that E < V (x)
for a x b and E > V (x) for x < a and for x > b [see Fig. 11.2(b)].
In classical mechanics a particle cannot penetrate into the potential barrier
in the region a x b, however such a process is possible in quantum
mechanics. Consider a solution having the form
(x) =
_
_
1
p
exp
_
i
x
_
a
dx
p
_
+
r
p
exp
_
x
_
a
dx
p
_
x < a
C+
|p|
exp
_
1
x
_
b
dx
[p[
_
+
C
|p|
exp
_
x
_
b
dx
[p[
_
a x b
t
p
exp
_
i
x
_
b
dx
p +
i
4
_
x > b
,
where we have introduced the transmission and reection coecients t and
r respectively. Such a solution describes an incident wave of unit amplitude
propagating in the region x < a from left to right, a reected wave having
amplitude r in the same region, and a transmitted wave having amplitude t
in the opposite side of the barrier x > b.
Employing the tailoring role (11.36) yields C
+
= 0 and C
= t. Moreover,
employing the tailoring role (11.34) and using the identity
exp
_
_
x
_
b
dx
[p[
_
_
=
1/2
exp
_
_
x
_
a
dx
[p[
_
_
, (11.48)
11.6. Solutions
where
= exp
_
_
b
_
a
dx
[p[
_
_
, (11.49)
yield [t[
1/2
= 1, thus the transmission probability is given by
[t[
2
= = exp
_
_
b
_
a
dx
[p[
_
_
. (11.50)
It is important to keep in mind that this approximation is valid only when
1. One way of seeing this is by noticing that the second tailoring step, as
can be seen from Eq. (11.34), also leads to the conclusion that [r[ = 1. This
apparently contradicts Eq. (11.50), which predicts a nonvanishing value for
[t[, whereas current conservations, on the other hand, requires that [t[
2
+[r[
2
=
1. This apparent contradiction can be attributed to limited accuracy of the
WKB approximation, however, Eq. (11.50) can be considered to be a good
approximation only provided that 1.
11.5 Problems
1. Calculate the transmission probability of a particle having mass m
and energy E through the potential barrier V (x), which vanishes in the
region x < 0 and which is given by V (x) = U ax in the region x 0,
where a > 0 and where U > E.
2. Calculate the transmission probability of a particle having mass m and
energy E through the potential barrier V (x) = m
2
x
2
/2, where > 0.
Consider the general case without assuming 1.
3. Consider a particle having mass m moving in a one dimensional double
well potential (see Fig. 11.3), which is assumed to be symmetric, i.e.
V (x) = V (x). In the limit where the barrier separating the two wells
can be considered as impenetrable, each well is characterized by a set of
eigenstates having eigen energies E
n
. To lowest nonvanishing order in
the penetrability of the barrier calculate the eigenenergies of the system.
4. Employ the WKB approximation to derive the eigenenergies of the hy-
drogen atom.
11.6 Solutions
1. The classical turning points are x = 0 and x = (U E) /. Thus with
the help of Eq. (11.50) one nds that
x
( ) x V
n
E
a a
x
( ) x V
n
E
a a
Fig. 11.3. Double well potential.
= exp
_
_
_
2
2m
(UE)/
_
0
dx
_
U E
x
_
_
_
= exp
_
2m
3
(U E)
3
2
_
.
(11.51)
2. The factor p/ can be expressed as
p (x)
=
_
2m
_
E +
m
2
x
2
2
_
=
=
x
x
2
0
_
1 +
2Ex
2
0
E
0
x
2
,
(11.52)
where x
0
=
_
/m and where E
0
= . For suciently large [x[, namely
for x
2
Ex
2
0
/E
0
, one has
p (x)

x
x
2
0
+
E
E
0
x
, (11.53)
where x is assumed to be positive. The corresponding WKB wavefunc-
tions (11.17) in the same limit of large large [x[ are given (up to multi-
plication by a constant) by
11.6. Solutions
(x) =
1
_
x
0
p/
exp
_
_
x
_
dx
p (x
)
_
_
exp
_
i
x
2
0
x
_
0
dx
_
exp
_
i
E
E0
x
_
x
0
dx
_
_
x
x
2
0
_
1/2
_
1 +
2Ex
2
0
E
0
x
2
_
1/4
_
x
x
0
_
i
E
E
0
1
2
exp
_
ix
2
2x
2
0
_
.
(11.54)
Consider a solution having the asymptotic form
(x
) =
_
_
_
x
x0
_
i
E
E
0
1
2
exp
_
ix
2
2x
2
0
_
+r
_
x
x0
_
i
E
E
0
1
2
exp
_
ix
2
2x
2
0
_
x
t
_
x
x0
_
i
E
E
0
1
2
exp
_
ix
2
2x
2
0
_
x
,
(11.55)
where t and r are transmission and reection coecients respectively,
which can be related one to another by the technique of analytical con-
tinuation. Consider x as a complex variable
x
x
0
= e
i
, (11.56)
where > 0 and is real. The transmitted term in the limit x
along the upper semicircle x/x
0
= e
i
, where 0 is given by
t
_
e
i
_
i
E
E
0
1
2
exp
_
i
2
e
2i
2
_
, (11.57)
thus for = this term becomes identical to the reected term at x/x
0
=
, which is given by
r ()
i
E
E
0
1
2
exp
_
i
2
2
_
, (11.58)
provided that
t
_
e
i
_
i
E
E
0
1
2
exp
_
i
2
e
2i
2
_
= r ()
i
E
E
0
1
2
exp
_
i
2
2
_
, (11.59)
or
ite
E
E
0
= r . (11.60)
Moreover, current conservation requires that [t[
2
+[r[
2
= 1, thus
[t[
2
+
ite
E
E
0
2
= 1 , (11.61)
and therefor the transmission probability = [t[
2
is given by
=
1
1 +e
2E
E
0
. (11.62)
As we have seen above, the analytical continuation of the transmitted
term in the region x leads to the reected term in the region
x . What about the incident term in the region x (the rst
term)? Note that this term (the incident one) becomes exponentially
small compared with the reected term in a section near = 3/4 along
the upper semicircle [due to the exponential factors exp
_
ix
2
/2x
2
0
_
].
Consequently, within the accuracy of the WKB approximation it does
not contribute to the analytically continued value.
3. Consider one of the single-well eigenenergies E
n
. The associated eigen-
state corresponding to the left well is labeled as [n, L and the one corre-
sponding to the right well as [n, R. The eect of nite penetrability of
the barrier can be evaluated using time independent perturbation theory
for the degenerate case [see Eq. (9.38]. For the unperturbed case, where
the barrier separating the two wells can be considered as impenetrable,
the level E
n
is doubly degenerate, and the corresponding eigen space
is spanned by the vectors [n, L , [n, R. The projection of the Hamil-
tonian of the system H = p
2
/2m+V on this eigen space is represented
by the 2 2 matrix H
n
, which is given by
H
n
=
_
n, L[ H[n, L n, L[ H[n, R
n, R[ H[n, L n, R[ H[n, R
_
. (11.63)
Employing the approximations
H[n, L E
n
[n, L , (11.64)
H[n, R E
n
[n, R , (11.65)
one nds that
H
n
= E
n
_
1
1
_
, (11.66)
where
= n, L [n, R , (11.67)
or in the coordinate representation
11.6. Solutions
=
dx
n,L
(x)
n,R
(x) , (11.68)
where
n,L
(x) and
n,R
(x) are the wavefunctions of the states [n, L
and [n, R respectively, i.e.
n,L
(x) = x [n, L , (11.69)
n,R
(x) = x [n, R . (11.70)
The main contribution to the overlap integral (11.68) comes from the
classically forbidden region [x[ a, where x = a are turning points
(i.e., E
n
= V (a) = V (a)). With the help of Eq. (11.36) one nds that

a
_
a
dx
[C[
2
exp
_
x
_
a
dx
[p[
_
exp
_
a
_
x
dx
[p[
_
[p[
(11.71)
= [C[
2
exp
_
_
a
_
a
dx
[p[
_
_
a
_
a
dx
[p[
, (11.72)
(11.73)
where C is the normalization factor of the WKB wavefunction, which is
approximately given by C = 2
_
m/T (T is the time period of classical
oscillations of a particle having energy E
n
in a well) in the limit of large
n [see Eq. (11.44)], thus

4
a
_
a
dx
|p/m|
T
exp
_
_
a
_
a
dx
[p[
_
_
. (11.74)
Finally, By diagonalizing the matrix H
n
one nds that the two eigenen-
ergies are E
n
(1 ).
4. The radial equation for the case of hydrogen is given by [see Eq. (7.61)]
_
2
2
d
2
dr
2
+V
e
(r)
_
u
kl
(r) = E
kl
u
kl
(r) , (11.75)
where m
e
is the reduced mass (m
e
is the electrons mass), and where
V
e
(r) =
e
2
r
+
l (l + 1)
2
2r
2
. (11.76)
The eigenenergies E
kl
are calculated using the Bohr-Sommerfeld quanti-
zation rule (11.40)
1
r2
_
r1
dr
_
2(E
kl
V
e
(r)) = k +
1
2
. (11.77)
where k is required to be an integer. The points r
1,2
are classical turning
points that satisfy
E
kl
= V
e
(r
1,2
) . (11.78)
Using the notation
1,2
=
r
1,2
a
0
, (11.79)
=
E
kl
E
I
, (11.80)
where
a
0
=

2
e
2
(11.81)
is the Bohrs radius and where
E
I
=
e
4
2
2
(11.82)
is the ionization energy, Eq. (11.78) becomes
=
2

l (l + 1)
2
, (11.83)
thus
1,2
=
1
_
1
_
1 l (l + 1)
_
. (11.84)
Similarly Eq. (11.77) becomes
2
_
1
d
_
2

l (l + 1)
2
=
_
k +
1
2
_
, (11.85)
or
I =
_
k +
1
2
_
, (11.86)
where the integral I, which is given by
I =
2
_
1
d
_
(
1
) (
2
)
, (11.87)
11.6. Solutions
can be calculated using the residue theorem
I =
1
+
2
2
_
1
_
4
1
2
(
1
+
2
)
2
_
. (11.88)
Thus the quantization condition (11.86) becomes
=
E
kl
E
I
=
1
_
_
l (l + 1) +k +
1
2
_
2
. (11.89)
Comparing with the exact result (7.84) shows that the WKB result is a
good approximation provided that the quantum numbers are large.
12. Path Integration
In this chapter, which is mainly based on Ref. [4], the technique of Feynmans
path integration is briey reviewed.
12.1 Charged Particle in Electromagnetic Field
Consider a point particle having mass m and charge q moving under the
inuence of electric eld E and magnetic eld B, which are related to the
scalar potential and to the vector potential A by
E =
1
c
A
t
, (12.1)
and
B = A . (12.2)
The classical Lagrangian of the system is given by Eq. (1.43)
/ =
1
2
m r
2
q +
q
c
A r , (12.3)
and the classical Hamiltonian is given by Eq. (1.62)
H =
_
p
q
c
A
_
2
2m
+q . (12.4)
The solution of the Euler Lagrange equations (1.8) yields the classical equa-
tion of motion of the system, which is given by Eq. (1.60)
mr = q
_
E+
1
c
r B
_
. (12.5)
In what follows, we consider for simplicity the case where both and A are
time independent. For this case H becomes time independent, and thus the
quantum dynamics is governed by the time evolution operator, which is given
by Eq. (4.9)
Chapter 12. Path Integration
u(t) = exp
_
iHt
_
. (12.6)
The propagator K(r
b
, t; r
a
) is dened by
K(r
b
, t; r
a
) = r
b
[ u(t) [r
a
, (12.7)
where [r
denotes a common eigenvector of the position operators x, y, and z

with vector of eigenvalues r
= (x
, y
, z
). As can be seen from the denition,

the absolute value squared of the propagator K(r
b
, t; r
a
) is the probability
density to nd the particle at point r
b
at time t given that it was initially
localized at point r
a
at time t = 0.
Dividing the time interval (0, t) into N sections of equal duration allows
expressing the time evolution operator as
u(t) =
_
u
_
t
N
__
N
. (12.8)
The identity operator in the position representation [see Eq. (3.65)] is given
by
1
r
=
_
d
3
r
[r
[ . (12.9)
Inserting 1
r
between any two factors in Eq. (12.8) and using the notation
r
a
= r
0
, (12.10)
r
b
= r
N
, (12.11)
=
t
N
, (12.12)
one nds that
K(r
b
, t; r
a
) = r
N
[ u() u() u() u() [r
=
_
d
3
r
1
_
d
3
r
2

_
d
3
r
N1
r
N
[ u()
N1
_
r
N1
u()
N2
_
r
N2
u() [r
1
r
1
[ u() [r
0
,
(12.13)
thus
K(r
b
, t; r
a
) =
N1
n=1
_
d
3
r
n
N1
m=0
K
_
r
m+1
, ; r
m
_
. (12.14)
In what follows the limit N will be taken, and therefor it is sucient
to calculate the innitesimal propagator K
_
r
m+1
, ; r
m
_
to rst order only
in .
12.1. Charged Particle in Electromagnetic Field
With the help of the relation (12.120), which is given by
e
(A+B)
= e
A
e
B
+O
_
2
_
, (12.15)
one has
u() = exp
_
iH
_
= exp
_
i
_
p
q
c
A
_
2
2m
_
exp
_
iq
_
+O
_
2
_
.
(12.16)
Equation (12.122), which is given by
exp
_
iV
2
2m
_
=
1
(2i)
3/2
_
d
3
r
exp
_
ir
2
2
i
_

m
V r
_
, (12.17)
allows expressing the rst term in Eq. (12.16) as
exp
_
i
_
p
q
c
A
_
2
2m
_
=
1
(2i)
3/2
_
d
3
r
exp
_
ir
2
2
i
_

m
_
p
q
c
A
_
r
_
.
(12.18)
Moreover, with the help of Eq. (12.121), which is given by
e
(A+B)
= e
B/2
e
A
e
B/2
+O
_
3
_
, (12.19)
one nds that
exp
_
i
_

m
_
p
q
c
A
_
r
_
= exp
_
i
_

m
q
c
A r
2
_
exp
_
i
_

m
p r
_
exp
_
i
_

m
q
c
A r
2
_
+O
_
3/2
_
.
(12.20)
Combining these results yields
K
_
r
m+1
, ; r
m
_
=

r
m+1
u() [r
=
1
(2i)
3/2
_
d
3
r
exp
_
ir
2
2
_
exp
_
i
_

m
q
c
_
A(r
m
) +A
_
r
m+1
_
r
2
_
exp
_
iq(r
m
)
m+1
exp
_
i
_

m
p r
_
[r
+O
_
3/2
_
.
(12.21)
In the next step the identity operator in the momentum representation [see
Eq. (3.71)], which is given by
1
p
=
_
d
3
p
[p
[ , (12.22)
is inserted to the left of the ket vector [r
m
. With the help of Eq. (3.75),
which is given by
r
[p
=
1
(2)
3/2
exp
_
ip
_
, (12.23)
one nds that
m+1
exp
_
i
_

m
p r
_
[r
=
1
(2)
3
_
d
3
p
exp
_
ip
_
r
m+1
r
m
_
_
exp
_
i
_

m
p
_
.
(12.24)
Thus, by using Eq. (3.84), which is given by
1
(2)
3
_
d
3
p
exp
_
ip
_
= (r
) , (12.25)
one nds that
m+1
exp
_
i
_

m
p r
_
[r
m
=
_
r
m+1
r
m
r
_
, (12.26)
and thus
K
_
r
m+1
, ; r
m
_
=
_
m
2i
_
3/2
exp
_
i
L
m
_
+O
_
3/2
_
, (12.27)
where
L
m
=
m
_
r
m+1
r
_
2
2
q(r
m
)+
q
c
A(r
m
) +A
_
r
m+1
_
2

r
m+1
r
. (12.28)
Comparing Eq. (12.28) with the classical Lagrangian of the system, which
is given by Eq. (1.43)
/ =
1
2
m r
2
q +
q
c
A r , (12.29)
shows that L
m
is nothing but the Lagrangian at point r
m
12.2. Classical Limit
L
m
= /(r
m
) . (12.30)
As we have discussed above, the terms of order
3/2
in Eq. (12.27) are not
expected to contribute to K (r
b
, t; r
a
) in the limit of N . By ignoring
these terms Eq. (12.14) becomes
K(r
b
, t; r
a
) = lim
N
_
Nm
2it
_
N/2 N1
n=1
_
d
3
r
n
exp
_
i
t
N
N1
m=0
/(r
m
)
_
.
(12.31)
Recall that the action in classical physics [see Eq. (1.4)] associated with a
given path is given by
S =
_
dt / . (12.32)
Thus, by dening the integral operator
_
r
b
r
a
T[r
(t)] = lim
N
_
Nm
2it
_
N/2 N1
n=1
_
d
3
r
n
, (12.33)
the propagator K(r
b
, t; r
a
) can be written as
K(r
b
, t; r
a
) =
_
r
b
r
a
T[r
(t
)] exp
_
i
S
r
(t
)
_
, (12.34)
where
S
r
(t)
=
t
_
0
dt /[r
(t)] . (12.35)
Equation (12.34), which is known as Feynmans path integral, expresses the
propagator K(r
b
, t; r
a
) in terms of all possible paths r
(t
) satisfying r
(0) =
r
a
and r
(t) = r
b
, where each path r
(t
) contributes a phase factor given by

exp
_
iS
r
(t
)
/
_
, where S
r
(t
)
is the classical action of the path r
(t
).
A note regarding notation: In the above derivation of Eq. (12.34) eigenval-
ues and eigenvectors were denoted with prime (e.g., r
, [r
, r
[, p
) to make
them distinguishable from the corresponding operators (e.g., r and p). This
distinction is no longer needed for the rest of this chapter, since no quan-
tum operators are used to evaluate path integrals, and therefore, to make the
notation less cumbersome, we omit the prime notation.
12.2 Classical Limit
Recall that the Hamiltons principle of least action states that the path taken
by a classical system is the one for which the action S obtains a local mini-
mum. This implies that for any innitesimal change in the path the resultant
change in the action S vanishes (i.e., S = 0). As we have seen in chapter 1,
this principle leads to Lagranges equations of motion (1.8), which are given
by
d
dt
/
q
n
=
/
q
n
. (12.36)
While in classical mechanics a denite path is associated with the systems
dynamics, in quantum mechanics all possible paths are included in Feynmans
path integral. However, as we show below, in the classical limit the dominant
contribution to the path integral comes only from paths near the classical
one. The classical limit is dened to be the limit where the Planks constant
approaches zero 0. In this limit the exponent exp(iS/) in the path
integral rapidly oscillates, and consequently contributions from neighboring
paths tend to cancel each other. However, near the classical path, such av-
eraging out does not occur since according to the principle of least action
S = 0 for the classical path. Consequently, constructive interference between
neighboring paths is possible near the classical path, and as a result the main
contribution to the path integral in the classical limit comes from the paths
near the classical path.
-1
-0.8
-0.6
-0.4
-0.2
0
0.2
0.4
0.6
0.8
1
-4 -2 2 4
x
Graphical demonstration of the stationary phase approximation The plot
shows the function cos
_
x
2
_
for the case = 1. According to the
stationary phase approximation, in the limit , the main contribution
to the integral
_
dx cos
_
x
2
_
comes from the region near the point
x = 0, where d
_
x
2
_
/dx = 0.
12.3 Aharonov-Bohm Eect
Using Eq. (1.43) for the classical Lagrangian of a charged particle in station-
ary electromagnetic eld one nds that the classical action (12.35) associated
with a path r (t) in the time interval (0, t) is given by
12.3. Aharonov-Bohm Eect
S =
t
_
0
dt
_
1
2
m r
2
q +
q
c
A r
_
. (12.37)
Consider rst the case where the vector potential vanishes, i.e. A = 0. For
this case, the system is said to be conservative, and therefor, as we have seen
in chapter 1 [see Eq. (1.29)], the energy of the system
E =
1
2
m r
2
+q (12.38)
is a constant of the motion (see exercise 5 below). In terms of E the action
S (12.37), which is labeled as S
0
for this case where A = 0, can be expressed
as
S
0
=
t
_
0
dt
_
1
2
m r
2
q
_
=
t
_
0
dt
_
E +m r
2
_
= Et +m
rb
_
ra
dr r .
(12.39)
where r
a
= r (0) and r
b
= r (t). Employing Eq. (12.38) again allows rewriting
S
0
as
S
0
= Et +
r
b
_
ra
dr p (r) , (12.40)
where p (r) is the local classical momentum
p (r) =
_
2m(E q(r)) . (12.41)
The phase factor in the path integral corresponding to S
0
is given by
exp
_
iS
0
_
= exp
_
iEt
_
exp
_
_
i
rb
_
ra
dr p (r)
_
_
. (12.42)
Note the similarity between the second factor in the above equation and
between the WKB wavefunction [see Eq. (11.17)]. In the general case, where
A can be nonzero, the phase factor in the path integral becomes [see Eq.
(12.37)]
exp
_
iS
_
= exp
_
iS
0
_
exp
_
_
iq
c
r
b
_
ra
dr A
_
_
. (12.43)
source
collector
0 B
impenetrable
long coil
0 ,
a
r
t ,
b
r
source
collector
0 B 0 B
impenetrable
long coil
0 ,
a
r
t ,
b
r
Fig. 12.1. Two-slit interference experiment with a very long impenetrable cylinder
placed near the gap between the slits.
12.3.1 Two-slit Interference
Consider a two-slit interference experiment where electrons having charge
q = e are injected from a point source at r
a
(see Fig. 12.1). A collector at
point r
b
measures the probability density to detect an electron at that point.
A very long impenetrable cylinder is placed near the gap between the slits in
order to produce a magnetic eld inside the cylinder in the direction normal
to the plane of the gure. The eld outside the cylinder, however, can be
made arbitrarily small, and in what follows we assume that it vanishes.
The probability density P
b
to detect the electron at time t by the collector
located at point r
b
is given by
P
b
= [K(r
b
, t; r
a
)[
2
, (12.44)
where the propagator (12.34) is given for this case by
K(r
b
, t; r
a
) =
_
r
b
ra
T[r (t)] exp
_
iS
0,r(t)
_
exp
_
_
_
ie
c
_
r(t)
dr A
_
_
_ , (12.45)
where the trajectories r (t) satisfy r (0) = r
a
and r (t) = r
b
.
How P
b
is modied when the magnetic eld is turned on, and consequently
the last factor in Eq. (12.45) starts to play a role? To answer this question
it is convenient to divide the sum over all paths into two groups, one for all
12.3. Aharonov-Bohm Eect
paths going through the left slit, and another for all paths going through the
right one. Here we disregard paths crossing a slit more than one time, as their
contribution is expected to be small. In general, the dierence
12
between
the vector potential phase factor in Eq. (12.45) associated with two dierent
paths r
1
(t) and r
2
(t) is given by
12
=
e
c
_
_
_
_
r1(t)
dr A
_
r2(t)
dr A
_
_
_
=
e
c
_
dr A ,
(12.46)
where the closed path integral is evaluated along the path r
1
(t) in the forward
direction from r
a
to r
b
, and then along the path r
2
(t) in the backward
direction from r
b
back to r
a
. This integral can be calculated using Stokes
theorem [see Eq. (12.2)]
12
=
e
c
_
dr A =
e
c
_
ds B = 2

0
, (12.47)
where is the magnetic ux threaded through the area enclosed by the closed
path, and where
0
=
hc
e
(12.48)
is the so called ux quantum. While
12
vanishes for pairs of paths going
through the same slit, it has the same value
12
= 2/
0
(
12
= 2/
0
)
for all the pairs where r
1
(t) goes through the left (right) path and where r
2
(t)
goes through the right (left) one. Thus, we come to the somewhat surprising
conclusion that the probability density P
b
is expected to be dependent on
the magnetic eld. The expected dependence is periodic in the magnetic
ux with ux quantum
0
period. Such dependence cannot be classically
understood, since in this example the electrons can never enter the region in
which the magnetic eld B is nite, and thus the Lorentz force vanishes in
the entire region accessible for the electrons outside the impenetrable coil.
12.3.2 Gauge Invariance
Consider the following gauge transformation
A

A = A+ , (12.49)
= , (12.50)
where = (r) is an arbitrary smooth and continuous function of r, which
is assumed to be time independent. As can be seen from Eqs. (12.1) and
(12.2), this transformation leaves E and B unchanged, since () = 0.
In chapter 1 we have seen that such a gauge transformation [see Eqs. (1.44)
and 1.45)] modies the Lagrangian [see Eq. (1.43)]
/

/ = / +
q
c
r , (12.51)
and also the action [see Eq. (12.37)]
S

S = S +
t
_
0
dt
q
c
r
= S +
q
c
rb
_
ra
dr
= S +
q
c
[(r
b
) (r
a
)] ,
(12.52)
however, the classical motion is unaected.
In quantum mechanics, the propagator is expressed as a path integral
[see Eq. (12.34)], where each path r (t) contributes a phase factor given by
exp
_
iS
r(t)
/
_
. As can be seen from Eq. (12.52), this phase factor is generally
not singly determined, since it depends on the chosen gauge. This result,
however, should not be considered as paradoxical, since only phase dierence
between dierent paths has any physical meaning. Indeed, as we have seen
above [see Eq. (12.47)], phase dierence
12
, which determines the relative
phase between two dierent paths, is evaluated along a closed path, which is
singly determined, and therefore gauge invariant.
Exercise 12.3.1. Given that the wavefunction (r
, t
) solves the Schrdinger

equation with vector A and scalar potentials, show that the wavefunction
(r
, t
), which is given by
(r
, t
) = exp
_
iq(r
)
c
_
(r
, t
) , (12.53)
solves the Schrdinger equation with vector

A = A+ and scalar =
potentials.
Solution 12.3.1. Using Eq. (3.76) one nds that
exp
_
iq
c
_
pexp
_
iq
c
_
= exp
_
iq
c
__
p, exp
_
iq
c
__
+p
= p+
q
c
.
(12.54)
12.4. One Dimensional Path Integrals
This result implies that
exp
_
iq
c
_
_
p
q
c
A
q
c
_
exp
_
iq
c
_
=
_
p
q
c
A
_
, (12.55)
and therefore the following holds
exp
_
iq
c
_

Hexp
_
iq
c
_
= H , (12.56)
H =
_
p
q
c
A
_
2
2m
+q (12.57)
is the Hamiltonian corresponding to the vector potential A, whereas
H =
_
p
q
c
A
q
c
_
2
2m
+q , (12.58)
is the Hamiltonian corresponding to the vector potential

A. Thus, one nds
that the state vector
_
= exp
_
iq
c
_
[ (12.59)
solves the Schrdinger equation with

H, provided that the state vector [
solves the Schrdinger equation with H, and therefore
(r
, t
) = exp
_
iq(r
)
c
_
(r
, t
) . (12.60)
12.4 One Dimensional Path Integrals
Consider a point particle having mass m moving in one dimension along the
x axis under the inuence of the potential V (x). The path integral (12.34)
for this case becomes
K (x
b
, t; x
a
) = lim
N
_
Nm
2it
_
N/2 N1
n=1
_
dx
n
exp
_
i
t
N
N1
m=0
/
_
x
m
,
x
m+1
x
m
t
N
_
_
.
(12.61)
where the Lagrangian is
/(x, x) =
1
2
m x
2
V (x) . (12.62)
The solution of the Euler Lagrange equation, which is given by Eq. (1.8)
d
dt
/
x
=
/
x
, (12.63)
yields the classical equation of motion of the system
m x =
dV
dx
. (12.64)
12.4.1 One Dimensional Free Particle
For this case V (x) = 0.
K(x
b
, t; x
a
) =
_
m
2it
exp
_
im
2t
(x
b
x
a
)
2
_
. (12.65)
Solution 12.4.1. The path integral (12.61) for this case becomes
K (x
b
, t; x
a
) = lim
N
_
imN
2t
_
N/2 N1
n=1
_
dx
n
exp
_
imN
2t
N1
m=0
(x
m+1
x
m
)
2
_
,
(12.66)
or
K(x
b
, t; x
a
) = lim
N
_
_
N/2
N1
n=1
_
dx
n
exp
_
N1
m=0
(x
m+1
x
m
)
2
_
,
(12.67)
where
=
imN
2t
. (12.68)
The rst integral
_
dx
1
can be calculated using the identity
dx
1
exp
_
(x
2
x
1
)
2
(x
1
x
0
)
2
_
=
_

2
exp
_
2
(x
2
x
0
)
2
_
,
(12.69)
The second integral
_
dx
2
can be calculated using the identity
dx
2
exp
_
(x
3
x
2
)
2

2
(x
2
x
0
)
2
_
=
_
2
3
exp
_
3
(x
3
x
0
)
2
_
.
(12.70)
Similarly, the nth integral
_
dx
n
yields
_
n
(n + 1)
exp
_

n + 1
(x
n+1
x
0
)
2
_
. (12.71)
Therefor, the propagator is given by
K (x
b
, t; x
a
) = lim
N
_
_
N/2
_

2
2
3

(N 1)
N
exp
_
N
(x
b
x
a
)
2
_
,
(12.72)
or
K(x
b
, t; x
a
) =
_
m
2it
exp
_
im
2t
(x
b
x
a
)
2
_
. (12.73)
As can be seen from the classical equation of motion (12.64), a free particle
moves at a constant velocity. Thus, the classical path satisfying x(0) = x
a
and x(t) = x
b
is given by
x
c
(t
) = x
a
+
(x
b
x
a
) t
t
. (12.74)
The corresponding classical action S
c
is
S
c
=
_
x
c
(t
)
dt
/(x, x) =
m(x
b
x
a
)
2
2t
. (12.75)
Note that the following holds
d
2
S
c
dx
a
dx
b
=
m
t
. (12.76)
Thus the propagator can be expressed in terms of the classical action S
c
as
K(x
b
, t; x
a
) =
_
i
2
d
2
S
c
dx
a
dx
b
exp
_
i
S
c
_
. (12.77)
As we will see below, a similar expression for the propagator is obtained also
for other cases.
12.4.2 Expansion Around the Classical Path
Motivated by the previous example of a free particle, we attempt below to
relate the propagator for the more general case, where V (x) is allowed to be
x dependent, with the classical path x
c
(t
) and the corresponding classical

action S
c
. Consider a general path x(t
) satisfying the boundary conditions

x(0) = x
a
and x(t) = x
b
. It is convenient to express the path as
x(t
) = x
c
(t
) + (t
) , (12.78)
where the deviation (t
) from the classical path x

c
(t
) vanishes at the end

points (0) = (t) = 0. The action associated with the path x(t
) can be
expressed as
S =
_
x(t
)
dt
/(x, x) , (12.79)
where the Lagrangian is given by Eq. (12.62).
Expanding S is orders of yields
S = S
c
+S
1
+S
2
+ , (12.80)
where
S
c
=
_
dt
/(x, x) , (12.81)
S
1
=
_
dt
_
/
x
x=x
c
+
/
x
x=x
c
_
, (12.82)
S
2
=
_
dt
2
/
x
2
x=xc
2
+ 2

2
/
x x
x=xc
+

2
/
x
2
x=xc
2
_
. (12.83)
In the general case, higher orders in such an expansion may play an important
role, however, as will be discussed below, in the classical limit the dominant
contribution to the path integral comes from the lowest order terms.
Claim. S
1
= 0.
Proof. Integrating by parts the term proportional to

in the expression for
S
1
yields
S
1
=
_
/
x
x=xc
t
0
+
_
dt
_
/
x
x=xc
d
dt
/
x
x=xc
_
. (12.84)
The rst term in Eq. (12.84) vanishes due to the boundary conditions (0) =
(t) = 0, whereas the second one vanishes because x
c
(t
) satises the Euler

Lagrange equation (12.63), thus S
1
= 0. The fact that S
1
vanishes is a direct
consequence of the principle of least action of classical mechanics that was
discussed in chapter 1.
Employing the coordinate transformation (12.78) and the expansion of S
around the classical path allows rewriting the path integral (12.61) as
K(x
b
, t; x
a
) = P
c
(x
b
, t; x
a
) /(t) , (12.85)
where
P
c
(x
b
, t; x
a
) = exp
_
iS
c
_
, (12.86)
/(t) =
_
T[ (t
)] exp
_
i
_
S
2
+O
_
3
__
_
, (12.87)
and where
_
T[ (t
)] = lim
N
_
Nm
2it
_
N/2 N1
n=1
_
d
n
. (12.88)
The term /(t) is evaluated by integrating over all paths (t
) satisfying the
boundary conditions (0) = (t) = 0.
_
dx
P
c
(x
b
, t
2
; x
) P
c
(x
, t
1
; x
a
)
P
c
(x
b
, t
1
+t
2
; x
a
)
=
/(t
1
+t
2
)
/(t
1
) /(t
2
)
. (12.89)
Solution 12.4.2. As can be seen from the denition of the propagator
(12.7), the following holds
_
dx
K(x
b
, t
2
; x
) K (x
, t
1
; x
a
) =
_
dx
x
b
[ u(t
2
) [x
[ u(t
1
) [x
a
= x
b
[ u(t
1
+t
2
) [x
a
= K(x
b
, t
1
+t
2
; x
a
) .
(12.90)
Requiring that this property is satised by the propagator K (x
b
, t; x
a
) that
is given by Eq. (12.85) leads to
_
dx
P
c
(x
b
, t
2
; x
) P
c
(x
, t
1
; x
a
)
P
c
(x
b
, t
1
+t
2
; x
a
)
=
/(t
1
+t
2
)
/(t
1
) /(t
2
)
. (12.91)
12.4.3 One Dimensional Harmonic Oscillator
For this case the Lagrangian is taken to be given by
/(x, x) =
m x
2
2
+
m
1
x x
2

m
2
x
2
2
, (12.92)
where m, and
1
are assumed to be real constants. As we will see below,
the term (m
1
/2) x x doesnt aect the dynamics, however, it is taken into
account in order to allows studying the more general case where the La-
grangian contains all possible types of quadratic (in x and x) terms (though,
for simplicity, all coecients in the Lagrangian are assumed to be time inde-
pendent). Consider a general path x(t
) satisfying the boundary conditions

x(0) = x
a
and x(t) = x
b
. Using the notation
x(t
) = x
c
(t
) + (t
) ,
the Lagrangian becomes
/(x, x) =
m
2
_
_
x
c
+

_
2
+
1
(x
c
+)
_
x
c
+

2
(x
c
+)
2
_
, (12.93)
thus the action associated with the path x(t
) can be expressed as
S =
_
x(t
)
dt
/(x, x) = S
c
+S
1
+S
2
, (12.94)
where
S
c
=
m
2
t
_
0
dt
_
x
2
c
+
1
x
c
x
c
2
x
2
c
_
, (12.95)
S
1
= m
t
_
0
dt
_
x
c
+

1
2
_
x
c
+ x
c
_
2
x
c
_
, (12.96)
S
2
=
m
2
t
_
0
dt
2
+
1
2
_
. (12.97)
As we have seen above, the principle of least action implies that S
1
= 0. Note
that in this case the expansion to second order in is exact and all higher
order terms vanish. Thus, the exact solution of this problem will also provide
an approximate solution for systems whose Lagrangian can be approximated
by a quadratic one.
Exercise 12.4.3. Find the classical action S
c
of a classical path satisfying
x(0) = x
a
and x(t) = x
b
.
Solution 12.4.3. The Euler Lagrange equation (12.63)
d
dt
/
x
=
/
x
, (12.98)
for this case yields
x =
2
x , (12.99)
thus, indeed the term (m
1
/2) x x doesnt aect the dynamics. Requiring also
boundary conditions x(0) = x
a
and x(t) = x
b
leads to
x
c
(t
) =
x
b
sin(t
) x
a
sin( (t
t))
sin(t)
. (12.100)
To evaluate the corresponding action we calculate the following integrals
t
_
0
dt
_
x
2
c
2
x
2
c
_
=
2
sin
2
(t)
t
_
0
dt
_
(x
b
cos (t
) x
a
cos ( (t
t)))
2
(x
b
sin(t
) x
a
sin (t
t))
2
_
=
_
_
x
2
a
+x
2
b
_
cot (t)
2x
a
x
b
sin(t)
_
=
_
(x
a
x
b
)
2
cot (t) 2x
a
x
b
tan
_
t
2
__
,
(12.101)
and
t
_
0
dt
x
c
x
c
=

t
_
0
dt
(x
b
sin(t
) x
a
sin( (t
t))) (x
b
cos (t
) x
a
cos ( (t
t)))
sin
2
(t)
=
x
2
b
x
2
a
2
,
(12.102)
thus, the action is given by
S
c
=
_
xc(t
)
dt
/(x, x)
=
m
2
_
(x
a
x
b
)
2
cot (t) 2x
a
x
b
tan
_
t
2
__
+
m
1
_
x
2
b
x
2
a
_
4
.
(12.103)
To evaluate the propagator according to Eq. (12.85) the factor /(t) has
to be determined. This can be done by employing relation (12.89) for the
case where x
a
= x
b
= 0
_
dx
P
c
(0, t
2
; x
) P
c
(x
, t
1
; 0)
P
c
(0, t
1
+t
2
; 0)
=
/(t
1
+t
2
)
/(t
1
) /(t
2
)
. (12.104)
/(t) =
_
m
2i sin(t)
. (12.105)
Solution 12.4.4. By using Eqs. (12.103) and (12.104) one nds that
_
dx
exp
_
i
m
2
(cot (t
2
) + cot (t
1
)) x
2
_
=
/(t
1
+t
2
)
/(t
1
) /(t
2
)
, (12.106)
thus, using the general integral identity
_

dx
exp
_
ix
2
_
=
_
i
, (12.107)
where is real, one nds that
_
2i
m (cot (t
2
) + cot (t
1
))
=
/(t
1
+t
2
)
/(t
1
) /(t
2
)
. (12.108)
Alternatively, using the identity
1
cot (t
2
) + cot (t
1
)
=
sin(t
1
) sin(t
2
)
sin( (t
1
+t
2
))
, (12.109)
this can be rewritten as
_
2i
m
sin(t
1
) sin(t
2
)
sin( (t
1
+t
2
))
=
/(t
1
+t
2
)
/(t
1
) /(t
2
)
. (12.110)
Consider a solution having the form
/(t) = e
f(t)
_
m
2isin(t)
, (12.111)
where f (t) is an arbitrary function of time. Substituting this into Eq. (12.110)
yields
f (t
1
) +f (t
2
) = f (t
1
+t
2
) , (12.112)
thus f (t) = At, where A is a constant. Combining all these results the prop-
agator (12.85) for the present case becomes
12.5. Semiclassical Limit
K(x
b
, t; x
a
) = e
At
_
m
2i sin(t)
exp
_
i
m
2
_
(x
a
x
b
)
2
cot (t) 2x
a
x
b
tan
_
t
2
__
+
m
1
_
x
2
b
x
2
a
_
4
_
.
(12.113)
In addition we require that in the limit ,
1
0 the above result will
approach the result given by Eq. (12.65) for the propagator of a free particle.
This requirement yields A = 0. Note that
(x
a
x
b
)
2
cot (t) 2x
a
x
b
tan
_
t
2
_
=
_
x
2
a
+x
2
b
_
cot (t)
2x
a
x
b
sin(t)
.
(12.114)
For the case
1
= 0 the propagator K (x
b
, t; x
a
) becomes
K(x
b
, t; x
a
) =
_
m
2i sin(t)
exp
_
im
2sin(t)
__
x
2
a
+x
2
b
_
cos (t) 2x
a
x
b
_
.
(12.115)
As can be seen from Eq. (12.103), the following holds
d
2
S
c
dx
a
dx
b
=
m
sin(t)
, (12.116)
thus, similar to the case of a free particle [see Eq. (12.77)], also for the present
case of a harmonic oscillator, the propagator can be expressed in terms of
the classical action S
c
as
K(x
b
, t; x
a
) =
_
i
2
d
2
S
c
dx
a
dx
b
exp
_
i
S
c
_
. (12.117)
12.5 Semiclassical Limit
In the semiclassical limit the Planks constant is considered to be small.
In this limit the dominant contribution to the path integral comes only from
paths near the classical one, which has the least action. This implies that in
the expansion of S around the classical path (12.80) terms of order O
_
3
_
can be approximately neglected. Thus, as can be seen from Eq. (12.87), in
this limit [see also Eq. (12.85)] the propagator K (x
b
, t; x
a
) is evaluated by
path integration over the quadratic terms S
2
only of the action [see Eq.
(12.83)]. In the previous section we have exactly calculated the propagator
associated with the quadratic Lagrangian of a harmonic oscillator. The result
was expressed in Eq. (12.117) in terms of the classical action S
c
. As can be
seen from Eq. (12.77), the same expression is applicable also for the case
of a free particle. It can be shown that the same form is also applicable
for expressing the propagator K (x
b
, t; x
a
) in the semiclassical limit for the
general case
K(x
b
, t; x
a
) =
_
i
2
d
2
S
c
dx
a
dx
b
exp
_
i
S
c
_
. (12.118)
The proof of the above result, which requires generalization of the derivation
that led to Eq. (12.117) for the case of a general quadratic Lagrangian, will
not be given here. Another important result, which also is given here without
a proof, generalizes Eq. (12.118) for the case of motion in n spacial dimensions
K(r
b
, t; r
a
) =
_
det
_
i
2
d
2
S
c
dr
a
dr
b
_
exp
_
i
S
c
_
. (12.119)
12.6 Problems
1. Show that
e
(A+B)
= e
A
e
B
+O
_
2
_
, (12.120)
where A and B are operators.
2. Show that
e
(A+B)
= e
B/2
e
A
e
B/2
+O
_
3
_
, (12.121)
where A and B are operators.
3. Show that
exp
_
iV
2
2m
_
=
1
(2i)
3/2
_
d
3
r
exp
_
ir
2
2
i
_

m
V r
_
, (12.122)
where V is a vector operator.
4. Show that the energy (12.38) is indeed a constant of the motion.
5. Consider a quantum system having time independent Hamiltonian Hand
a discrete energy spectrum. Express its partition function Z in terms of
the systemss propagator K (x
b
, t; x
a
).
6. Consider a one dimensional harmonic oscillator having mass m and res-
onance angular frequency in thermal equilibrium at temperature T.
Calculate the matrix elements x
[ [x
of the density operator in the

basis of eigenvectors [x
of the position operator x.

12.7. Solutions
7. Consider a free particle in one dimension having mass m. Calculate the
position wavefunction (x
, t) at time t given that the position wave-

function (x
, 0) at time t = 0 is given by
(x
, 0) =
1
1/4
x
1/2
0
exp
_
1
2
_
x
x
0
_
2
_
. (12.123)
where x
0
is a constant.
8. A particle having mass m is in the ground state of the potential well
V
0
(x) = (1/2) m
2
x
2
for times t < 0 . At time t = 0 the potential
suddenly changes and becomes V
1
(x) = mgx.
a) Calculate the propagator K (x
b
, t; x
a
) from point x
a
to point x
b
in
the semiclassical limit for the case where the potential is V
1
(x) (i.e.
for the Hamiltonian after the change at t = 0).
b) Use the result of the previous section to calculate the variance
_
(x)
2
(t)
_
=

x
2
(t)
_
x(t)
2
of the position operator x at time t.
12.7 Solutions
1. Consider the operator
C () = e
A
e
(A+B)
e
B
. (12.124)
Clearly, C (0) = 1. Moreover, with the help of Eq. (2.170) one nds that
dC
d
= e
A
Ae
(A+B)
e
B
+e
A
e
(A+B)
(A+B) e
B
e
A
e
(A+B)
e
B
B ,
(12.125)
thus
dC
d
=0
= A+ (A+B) B = 0 , (12.126)
namely
C () = 1 +O
_
2
_
,
and therefor
e
(A+B)
= e
A
e
B
+O
_
2
_
. (12.127)
2. Consider the operator
C () = e
B/2
e
(A+B)
e
B/2
e
A
. (12.128)
As in the previous exercise, it is straightforward (though, somewhat te-
dious) to show that
C (0) = 1 , (12.129)
dC
d
=0
= 0 , (12.130)
d
2
C
d
2
=0
= 0 , (12.131)
thus
C () = 1 +O
_
3
_
, (12.132)
and therefor
e
(A+B)
= e
B/2
e
A
e
B/2
+O
_
3
_
. (12.133)
3. The proof is trivial using the identity
_

e
x
2
+x
dx
=
_
2
/4
. (12.134)
4. By taking the time derivative of E one has
dE
dt
= m r r +q r = r (mr qE) . (12.135)
However, according to the equation of motion (1.60) the term in the
brackets vanishes, and therefor dE/dt = 0.
5. Assume that the energy eigenstates of the Hamiltonian H are labeled by
[a
n
and the corresponding eigenenergies by E
n
, i.e.
H[a
n
= E
n
[a
n
, (12.136)
where
a
n
[a
n
=
nn
. (12.137)
With the help of the closure relation
1 =

n
[a
n
a
n
[ , (12.138)
one nds that the propagator K (x
b
, t; x
a
) can be expressed as
K(x
b
, t; x
a
) = x
b
[ exp
_
iHt
_
[x
a
=

n
x
b
[a
n
exp
_
iE
n
t
_
a
n
[x
a
.
(12.139)
12.7. Solutions
Taking x
b
= x
a
and integrating over x
a
yields
_

dx
a
K(x
a
, t; x
a
) =
n
exp
_
iE
n
t
_
. (12.140)
Thus, the partition function Z, which is given by Eq. (8.41)
Z =

n
e
En
, (12.141)
where = 1/k
B
T, can be expressed as
Z =
_

dx
K(x
, i; x
) . (12.142)
6. Using Eq. (8.42) one nds that
x
[ [x
=
x
[ e
H
[x
Z
, (12.143)
where the partition function Z = Tr
_
e
H
_
can be expressed in terms
of the propagator K(x
, t; x
) [see Eq. (12.142)]

Z =
_

dx
K(x
, i; x
) . (12.144)
Furthermore, as can be seen from the denition of the propagator [see
Eq. (12.7)]
K(x
, t; x
) = x
[ e
iHt
[x
, (12.145)
the following holds
x
[ e
H
[x
= K (x
, i; x
) . (12.146)
Thus, with the help of Eq. (12.115) one nds for the case of a harmonic
oscillator that (recall that sin(ix) = i sinhx and cos (ix) = coshx)
Z =
_

dx
K(x
, i; x
)
=
_
m
2 sinh()
_

dx
exp
_
m [cosh() 1] x
2
sinh()
_
=
_
1
2 [cosh() 1]
=
1
2 sinh

2
,
(12.147)
and therefore one nds, in agreement with Eq. (8.290), that
x
[ [x
=
K(x
, i; x
)
Z
= sinh

2
_
2m
sinh()
exp
_
m
2 sinh()
__
x
2
+x
2
_
cosh() 2x
_
=
e
tanh(

2
)
_
x
+x
2x
0
_
2
coth(

2
)
_
x
2x
0
_
2
x
0
_
coth
_
2
_
,
(12.148)
where
x
0
=
_

m
. (12.149)
7. Denoting the state ket vector of the system by [ (t) and the time evo-
lution operator by u(t) one has
(x
, t) = x
[ (t)
= x
[ u(t) [ (0)
=
_

dx
[ u(t) [x
[ (0)
=
_

dx
K(x
, t; x
) (x
, 0) ,
(12.150)
where the propagator K(x
, t; x
) is given by Eq. (12.73)

K(x
, t; x
) =
_
1
2itx
2
0
exp
_
i
2t
(x
)
2
x
2
0
_
, (12.151)
and where
=

mx
2
0
, (12.152)
thus
(x
, t) =
1
1/4
x
1/2
0
_
1
2itx
2
0
_

dx
exp
_
1
2
_
1
i
t
__
x
x
0
_
2
i
t
x
x
2
0
+
i
2t
_
x
x
0
_
2
_
.
(12.153)
12.7. Solutions
1
exp
_
ax
2
+bx +c
_
dx =
_
1
a
e
1
4
4ca+b
2
a
, (12.154)
one nds that
(x
, t) =
1
1/4
x
1/2
0
_
1
1 +it
exp
_
1
2 (1 +it)
_
x
x
0
_
2
_
. (12.155)
8. The Lagrangian for times t > 0 is given by
/(x, x) =
1
2
m x
2
mgx . (12.156)
The Euler Lagrange equation yields the classical equation of motion of
the system
x = g . (12.157)
The general solution reads
x = x
0
+v
0
t
gt
2
2
, (12.158)
where the constants x
0
and v
0
are the initial values of the position and
velocity at time t = 0. Given that x = x
a
at time t = 0 and x = x
b
at
time t one nds that x
0
= x
a
and
v
0
=
x
b
x
a
t
+
gt
2
, (12.159)
thus the classical trajectory x
c
(t
) is given by
x
c
(t
) = x
a
+
_
x
b
x
a
t
+
g
2
t
_
t
g
2
t
2
. (12.160)
Using the notation
x
t
=
gt
2
2
, (12.161)
the trajectory x
c
(t
) is expressed as
x
c
(t
) = x
a
+ (x
b
x
a
x
t
)
t
t
+x
t
t
2
t
2
, (12.162)
and the corresponding velocity x
c
(t
) is expressed as
x
c
(t
) =
x
b
x
a
x
t
t
+
2x
t
t
t
2
. (12.163)
The Lagrangian along the classical trajectory is given by
/(x
c
, x
c
) =
1
2
m x
2
c
mgx
c
=
m
_
x
b
x
a
x
t
t
+
2xtt
t
2
_
2
2
mg
_
x
a
+ (x
b
x
a
x
t
)
t
t
+x
t
t
2
t
2
_
,
(12.164)
and the corresponding action S
c
is given by
S
c
=
_
xc(t
)
dt
/(x, x)
= m
t
_
0
dt
_
_
x
b
xaxt
t
+
2x
t
t
t
2
_
2
2
g
_
x
a
+ (x
b
x
a
x
t
)
t
t
+x
t
t
2
t
2
_
_
_
= m
(x
b
x
a
)
2
+ 2x
t
(x
b
+x
a
)
x
2
t
3
2t
(12.165)
a) In general, the propagator in the semiclassical limit is given by Eq.
(12.118)
K(x
b
, t; x
a
) =
_
i
2
d
2
S
c
dx
a
dx
b
exp
_
i
S
c
_
, (12.166)
where for the present case S
c
is given by Eq. (12.165) and
d
2
S
c
dx
a
dx
b
=
m
t
, (12.167)
thus
K(x
b
, t; x
a
) =
_
m
2it
exp
_
im
(x
b
x
a
)
2
+ 2x
t
(x
b
+x
a
)
x
2
t
3
2t
_
=
1
x
0
_
1
2it
exp
_
i
t
(x
b
x
a
)
2
+ 2x
t
(x
b
+x
a
)
x
2
t
3
2x
2
0
_
=
1
x
0
_
1
2it
exp
_
i
t
8
3
x
2
t
+x
2
a
+ 2 (2x
t
x
a
) (x
b
x
t
) + (x
b
x
t
)
2
2x
2
0
_
,
(12.168)
where
12.7. Solutions
x
0
=
_

m
. (12.169)
b) Initially at time t = 0 the wavefunction (x
) is given by [see Eq.

(5.110)]
(x
, t = 0) =
1
1/4
x
1/2
0
exp
_
1
2
_
x
x
0
_
2
_
, (12.170)
where
x
0
=
_

m
. (12.171)
The wave function at time t is evaluated using the propagator
(x
, t) =
_

dx
K (x
, t; x
) (x
, 0)
=
1
x
0
_
1
2it
1
1/4
x
1/2
0
dx
e
i
t
8
3
x
2
t
+(x
)
2
+2(2x
t
x
)(x
x
t)+(x
x
t)
2
2x
2
0

1
2
_
x
x
0
_
2
=
1
x
0
_
1
2it
e
i
t
8
3
x
2
t
+4x
t(x
x
t)+(x
x
t)
2
2x
2
0
1/4
x
1/2
0
dx
(
1
i
t
)(
x
x
0
)
2
2
+
ix
(x
x
t)
tx
2
0
.
(12.172)
1
exp
_
ax
2
+bx
_
dx =
_
1
a
e
b
2
4a
, (12.173)
one nds that
(x
, t) =
e
1
2
1
1+it
_
x
x
t
x
0
_
2
+
8
3
ix
2
t
+4ix
t(x
x
t)
2tx
2
0
1/4
x
1/2
0
1 +it
. (12.174)
The probability density f (x
) = [ (x
, t)[
2
to nd the particle near
point x
at time t is thus given by

f (x
) =
e
1
1+(t)
2
_
x
x
t
x
0
_
2
x
0
_
1 + (t)
2
, (12.175)
therefore f (x
) has a Gaussian distribution with a mean value x

t
and
variance given by
_
(x)
2
(t)
_
=
x
2
0
2
_
1 + (t)
2
_
. (12.176)
Note that this result is in agreement with Eq. (5.256).
13. Adiabatic Approximation
The adiabatic approximation can be employed for treating systems having
slowly varying Hamiltonian. This chapter is mainly based on Ref. [5].
13.1 Momentary Diagonalization
The Schrdinger equation (4.1) is given by
i
d[
dt
= H[ . (13.1)
For any given value of the time t the Hamiltonian H(t) is assumed to have
a discrete spectrum
H(t) [n(t) = E
n
(t) [n(t) , (13.2)
where n = 1, 2, , the momentary eigenenergies E
n
(t) are real, and the set
of momentary eigenvectors is assumed to be orthonormal
n(t) [m(t) =
nm
. (13.3)
The general solution can be expanded using the momentary eigenvectors as
a momentary basis
[(t) =

n
a
n
(t) e
i
n
(t)
[n(t) . (13.4)
The phase factors
n
(t) in the expansion (13.4) are chosen to be given by
n
(t) =
n
(t) +
n
(t) , (13.5)
where the phase factors
n
(t) =
1
_
t
dt
E
n
(t
) (13.6)
are the so-called dynamical phases, and the other phase factors
Chapter 13. Adiabatic Approximation
n
(t) = i
_
t
dt
n(t
) [ n(t
) (13.7)
are the so-called geometrical phases. As we will see below, choosing the phase
factor
n
(t) to be given by Eq. (13.5) ensures that the coecients a
n
(t)
become constants in the adiabatic limit.
Exercise 13.1.1. Show that the term n(t
) [ n(t
) is pure imaginary.
Solution 13.1.1. Note that by taking the derivative with respect to t (de-
noted by upper-dot) of the normalization condition (13.3) one nds that
n [m +n [ m = 0 , (13.8)
thus
n [ m = m [ n
. (13.9)
The last result for the case n = m implies that n(t
) [ n(t
) is pure imagi-
nary, and consequently
n
(t) are pure real.
Substituting Eq. (13.4) into Eq. (13.1) leads to
i
n
a
n
(t) e
i
n
(t)
[n(t)
n
a
n
(t)

n
(t) e
i
n
(t)
[n(t)
+i
n
a
n
(t) e
i
n
(t)
[ n(t)
=

n
a
n
(t) e
i
n
(t)
E
n
(t) [n(t) .
(13.10)
Taking the inner product with m(t)[ e
i
m
(t)
yields
a
m
(t)+i
m
(t) a
m
(t)+
n
a
n
(t) e
i
n
(t)
e
i
m
(t)
m(t) [ n(t) =
E
m
(t)
i
a
m
(t) .
(13.11)
Since, by denition, the following holds
i
m
(t) =
E
m
(t)
i
m(t) [ m(t) , (13.12)
Eq. (13.11) can be rewritten as
a
m
=

n=m
e
i(
n
(t)
m
(t))
m(t) [ n(t) a
n
. (13.13)
13.3. Adiabatic Limit
Exercise 13.1.2. Show that for n ,= m the following holds
m(t) [ n(t) =
m(t)[

H[n(t)
E
n
(t) E
m
(t)
. (13.14)
Solution 13.1.2. Taking the time derivative of Eq. (13.2)
H[n +H[ n =

E
n
[n +E
n
[ n , (13.15)
and the inner product with m(t)[, where m ,= n, yields the desired identity.
13.2 Gauge Transformation
The momentary orthonormal basis [n(t
)
n
, which is made of eigenvec-
tors of H(t), is clearly not singly determined. Consider the following gauge
transformation [see for comparison Eq. (12.49)]
[n(t
) [ n(t
) = e
i(t
)
[n(t
) , (13.16)
where (t
) is arbitrary real function of time. The geometrical phase

n
(t),
which is given by Eq. (13.7)
n
(t) = i
_
t
t0
dt
n(t
) [ n(t
) , (13.17)
is transferred to
n
(t)
n
(t) =
n
(t) +(t) (t
0
) . (13.18)
Thus, in general the geometrical phase is not singly determined. However, it
becomes singly determined, and thus gauge invariant, if the path is closed,
namely if H(t) = H(t
0
), since for such a case (t) = (t
0
).
13.3 Adiabatic Limit
In the adiabatic limit the terms m(t) [ n(t) are considered to be negligibly
small. As can be seen from Eq. (13.14), this limit corresponds to the case
where the rate of change in time of the Hamiltonian approaches zero. In
this limit the coecients a
n
(t) become constants [see Eq. (13.13)], and the
solution (13.4) thus becomes
[(t) =

n
a
n
e
i
n
(t)
[n(t) . (13.19)
13.4 The Case of Two Dimensional Hilbert Space
In this case the Hilbert space is two dimensional and the Hamiltonian can be
represented by a 2 2 matrix, which is conveniently expressed as a combina-
tion of Pauli matrices
H =h
0
I +h , (13.20)
where I is the 2 2 identity matrix, h
0
is a real scalar, h =(h
1
, h
2
, h
3
) is a
three-dimensional real vector, and the components of the Pauli matrix vector
are given by
1
=
_
0 1
1 0
_
,
2
=
_
0 i
i 0
_
,
3
=
_
1 0
0 1
_
. (13.21)
Using the notation h = H
h, where H =
h h, and where

h is a unit vector,
given in spherical coordinates by
h = (cos sin, sinsin, cos ) , (13.22)

one nds that
H =h
0
I +H
_
cos sin exp(i)
sin exp(i) cos
_
. (13.23)
The orthonormal eigenvectors are chosen to be given by [see Eqs. (6.199) and
(6.200)]
[+ =
_
cos

2
exp
_
i
2
_
sin

2
exp
_
i
2
_
_
, (13.24)
[ =
_
sin

2
exp
_
i
2
_
cos

2
exp
_
i
2
_
_
, (13.25)
and the following holds +[+ = [ = 1, +[ = 0, and
H[ = (h
0
H) [ . (13.26)
Note that the eigenstates [ are independent of both h
0
and H.
The geometrical phase (13.7) can be evaluated by integration along the
path h(t)
n
(t) = i
_
t
0
dt
n(t
) [ n(t
) = i
_
h(t)
h(0)
dh n(h)[
h
[n(h) . (13.27)
[
h
[ =
i
2H
ctg . (13.28)
13.4. The Case of Two Dimensional Hilbert Space
Solution 13.4.1. Using the expression for a gradient in spherical coordi-
nates (the radial coordinate r in the present case is H)
f =
f
r
r +
1
r
f
+
1
r sin
f
, (13.29)
one nds that
h
[+ =

2H
_
sin

2
exp
_
i
2
_
cos

2
exp
_
i
2
_
_
+
i
2Hsin
_
cos

2
exp
_
i
2
_
sin

2
exp
_
i
2
_
_
,
(13.30)
thus
+[
h
[+ =
i
2H sin
_
cos

2
exp
_
i
2
_
sin

2
exp
_
i
2
_ _
_
cos

2
exp
_
i
2
_
sin

2
exp
_
i
2
_
_
=
i
2H
ctg .
(13.31)
The term [
h
[ is calculated in a similar way.
For the case of a close path, Stocks theorem can be used to express the
integral in terms of a surface integral over the surface bounded by the close
curve h(t)
n
= i
_
dh n[
h
[n = i
_
S
da (n[
h
[n) . (13.32)
[
h
[ =
ih
2 [h[
3
. (13.33)
Solution 13.4.2. Using the general expression for the curl operator in spher-
ical coordinates (again, note that the radial coordinate r in the present case
is H)
A =
1
r sin
_
(sinA
_
r
+
1
r
_
1
sin
A
r

r
(rA
)
_
+
1
r
_
(rA
)
r

A
r
_
,
(13.34)
one nds that
[
h
[ =
i
h
2H
2
sin
cos
=
ih
2 [h[
3
. (13.35)
With the help of the last result one thus nds that
=
1
2
_
S
da
h
[h[
3
=
1
2
, (13.36)
where is the solid angle subtended by the close path h(t) as seen from the
origin. Due to the geometrical nature of the last result, these phase factors
were given the name geometrical phases.
13.5 Transition Probability
The set of equations of motion (13.13) can be rewritten in a matrix form as
i
d
dt
[a) = H[a) , (13.37)
where
[a) =
_
_
_
a
1
a
2
.
.
.
_
_
_ (13.38)
is a column vector of the coecients a
n
(, and where the matrix elements
of H are given by
H
mn
= H
nm
= ie
i(
n
(t)
m
(t))
m(t) [ n(t) (13.39)
for the case n ,= m and H
nn
= 0 otherwise.
The inner product between the vectors
[a) =
_
_
_
a
1
a
2
.
.
.
_
_
_ , [b) =
_
_
_
b
1
b
2
.
.
.
_
_
_ , (13.40)
is dened by
(a [b) = (b [a)
m
a
m
b
m
. (13.41)
The set of vectors [n) (n = 1, 2, ), having coecients a
m
=
nm
, forms
an orthonormal basis for the vector space
(n
1
[n
2
) =
n
1
n
2
. (13.42)
Consider the case where the system is initially at time t
0
in the state [n).
What is the probability p
nn
(t) to nd it later at time t > t
0
at the same state
[n)? The adiabatic approximation is valid only when p
nn
1. Considering
the matrix H as a perturbation, the probability p
nn
can be approximated
using time dependent perturbation theory.
Exercise 13.5.1. Show that to lowest nonvanishing order the following
holds
p
nn
(t) = 1
t
_
t0
dt
H
nm
(t
2
. (13.43)
Solution 13.5.1. By employing Eqs. (10.21) and (10.27) one nds that (re-
call that H
nn
= 0)
p
nn
(t) = 1
t
_
t0
dt
t
_
t0
dt
(n[ H (t
) H(t
) [n) . (13.44)
Inserting the identity operator 1 =

m
[m) (m[ between H (t
) and H(t
)
and recalling that H
mn
= H
nm
lead to
p
nn
(t) = 1
m
p
mn
(t) , (13.45)
where
p
mn
(t) =
t
_
t0
dt
H
nm
(t
2
. (13.46)
As can be seen from Eq. (13.39), the matrix elements H
nm
(t
) are pro-
portional to the oscillatory dynamical phase factors
H
mn
exp(i (
n
(t)
m
(t))) = exp
_
_
t
dt
(E
n
(t
) E
m
(t
))
_
.
(13.47)
In the adiabatic limit these terms rapidly oscillate and consequently the prob-
abilities p
mn
(t) are exponentially small. From the same reason, the dominant
contribution to the integral is expected to come from regions where the en-
ergy gap E
n
(t
)E
m
(t
) is relatively small. Moreover, it is also expected that

the main contribution to the total survival probability p
nn
will come from
those states whose energy E
m
(t
) is close to E
n
(t
). Having this is mind, we

study below the transition probability for the case of a two level system. As
we will see below, the main contribution indeed comes from the region near
the point where the energy gap obtains a minimum.
13.5.1 The Case of Two Dimensional Hilbert Space
We calculate below p
+
for the case H =h , where h(t) is the straight line
h(t) = (0, 1, t) , (13.48)
where is a positive constant, is a real constant, and where the initial time
is taken to be t
0
= and the nal one is taken to be t = . In spherical
coordinates h(t) is given by
h(t) = H (0, sin, cos ) , (13.49)
where
H =
_
1 + (t)
2
, (13.50)
cot = t , (13.51)
and where = /2. Thus, the energy gap 2H obtains a minimum at time
t = 0. As can be seen from Eqs. (13.24) and (13.25), for any curve lying on
a plane with a constant azimuthal angle , the following holds

+
_
=

2
[ , (13.52)
and therefor

+
_
=

2
, (13.53)
and
+

+
_
=
_
= 0 . (13.54)
For the present case one nds using Eq. (13.51) that

+
_
=
1
2

1 + (t)
2
. (13.55)
This together with Eqs. (13.39) and (13.46) leads to
p
+
=
dt
e
i(
+
(t
(t
))
(t
)

+(t
)
_
2
=
dt
exp
_
i
_
2
_
t
0
dt
_
1 + (t
)
2
__
1 + (t
)
2
2
=
dt
exp
_
2i
_
t
0
d

1 +
2
_
1 + (t
)
2
2
=
1
2
d
exp
_
1 +
2
ln
_
+
1 +
2
_
_
1 +
2
2
.
(13.56)
Exercise 13.5.2. Show that if / 1 then
p
+
exp
_
_
.
Solution 13.5.2. The variable transformation
= sinhz , (13.57)
and the identities
_
1 +
2
= coshz , (13.58)
_
1 +
2
=
1
2
sinh(2z) , (13.59)
ln
_
+
_
1 +
2
_
= z , (13.60)
d = coshz dz , (13.61)
yield
p
+
=
1
2
dz
exp
_
_
1
2
sinh(2z) +z
_
_
coshz
2
. (13.62)
In the limit / the phase oscillates rapidly and consequently p
+
0.
The stationary phase points z
n
in the complex plane are found from the
condition
0 =
d
dz
_
1
2
sinh2z +z
_
= cosh2z + 1 , (13.63)
thus
z
n
= i
_
n +
1
2
_
, (13.64)
where n is integer. Note, however that the term 1/ coshz has poles at the same
points. Using the Cauchys theorem the path of integration can be deformed
to pass close to the point z
1
= i/2. Since the pole at z
1
is a simple one,
the principle value of the integral exists. To avoid passing through the pole at
z
1
a trajectory forming a half circle "above" the pole with radius is chosen
were 0. This section gives the dominant contribution which is iR, where
R is the residue at the pole. Thus the probability p
+
is approximately given
by
p
+
exp
_
_
. (13.65)
The last result can be used to obtain a validity condition for the adia-
batic approximation. In the adiabatic limit p
+
1, and thus the condition
/ 1 is required to ensure the validity of the approximation.
13.6 Slow and Fast Coordinates
Consider a system whose Hamiltonian is given by
H = H
0
+H
1
. (13.66)
The Hamiltonian H
0
is assume to depend on a set of coordinates x =
(x
1
, x
2,...
) and on their canonically conjugate variables p = (p
1
, p
2,...
), i.e.
H
0
= H
0
( x, p). In what follows the coordinates x = (x
1
, x
2,...
) will be con-
sidered as slow, and thus H
0
will be considered as the Hamiltonian of the
slow subsystem. The other part of the system is a fast subsystem, which is
assumed to have a much faster dynamics and its energy spectrum is assumed
to be discrete. The Hamiltonian of the fast subsystem H
1
is assumed to para-
metrically depend on the slow degrees of freedom x, i.e. H
1
= H
1
( x). This
dependence gives rise to the coupling between the slow and fast subsystems.
An adiabatic approximation can be employed in order to simplify the
equations of motion of the combined system. In what follows, for simplicity,
this method will be demonstrated for the case where the slow subsystem is
assumed to be composed of a set of decoupled harmonic oscillators. For this
case the Hamiltonian H
0
H
0
=

l
_
p
2
l
2m
l
+
m
l
2
l
x
2
l
2
_
, (13.67)
where m
l
and
l
are the mass and angular frequency of mode l, respectively.
The Hamiltonian of the fast subsystem H
1
( x), which depends paramet-
rically on x, has a set of eigenvectors and corresponding eigenvalues for any
given value of x
H
1
[n( x) =
n
( x) [n( x) , (13.68)
where n = 1, 2, . The set of local eigenvectors [n( x) is assumed to
form an orthonormal basis of the Hilbert space of the fast subsystem, and
thus the following is assumed to hold
m( x) [n( x) =
mn
, (13.69)
and [see Eq. (2.23)]
k
[k ( x) k ( x)[ = 1
F
, (13.70)
where 1
F
is the identity operator on the Hilbert space of the fast subsystem.
The state of the entire system (t) at time t is expanded at any point x
using the local basis [n( x)
(t) =

n
n
( x, t) [n( x) . (13.71)
13.6. Slow and Fast Coordinates
In the above expression a mixed notation is being employed. On one hand,
the ket notation is used to denote the state of the fast subsystem (the terms
[n( x)). On the other hand, a wavefunction in the position representation
(the terms
n
( x, t)) is employed to denote the state of the slow subsystem.
Substituting the expansion (13.71) into the Schrdinger equation (4.1)
i
d
dt
= H , (13.72)
leads to
n
[H
0
+
n
( x)]
n
( x, t) [n( x) = i
n
( x, t) [n( x) , (13.73)
where overdot represents time derivative. Projecting m( x)[ leads to
n
m( x)[ H
0
n
( x, t) [n( x) +
m
( x)
m
( x, t) = i
m
( x, t) . (13.74)
By calculating the term [see Eq. (13.69)]
m( x)[ p
2
l
n
( x, t) [n( x) =
n
( x, t) m( x)[ p
2
l
[n( x)
+ 2 (p
l
n
( x, t)) m( x)[ p
l
[n( x) +
mn
p
2
l
n
( x, t) ,
(13.75)
introducing the notation
A
m,n;l
m( x)[ p
l
[n( x) , (13.76)
and using [see Eq. (13.70)]
m( x)[ p
2
l
[n( x) =
k
m( x)[ p
l
[k ( x) k ( x)[ p
l
[n( x)
= p
l
A
m,n;l
+
k
A
m,k;l
A
k,n;l
, (13.77)
one obtains
m( x)[ p
2
l
n
( x, t) [n( x)
=
n
( x, t)
_
p
l
A
m,n;l
+
k
A
m,k;l
A
k,n;l
_
2A
m,n;l
p
l
n
( x, t) +
mn
p
2
l
n
( x, t) .
(13.78)
With the help of Eqs. (13.67) and (13.74) one nds that
l
1
2m
l
n
( x, t)
_
p
l
A
m,n;l
+
k
A
m,k;l
A
k,n;l
_
l
1
2m
l
n
_
2A
m,n;l
p
l
n
( x, t) +
mn
p
2
l
n
( x, t)
l
m
l
2
l
x
2
l
2

m
( x, t) +
m
( x)
m
( x, t) = i
m
( x, t) .
(13.79)
Dening the matrices
_
A
l
_
m,n
= A
m,n;l
, ()
m,n
=
m
mn
, and the vector
_
_
n
=
n
, the above can be written in a matrix form as
_
l
_
1
2m
l
_
p
l

A
l
_
2
+
m
l
2
l
x
2
l
2
_
+
_
= i
. (13.80)
To calculate the o-diagonal matrix elements of

A
l
we apply p
l
to Eq.
(13.68) and project m( x)[, where m ,= n
m( x)[ p
l
H
1
[n( x) = m( x)[ p
l
n
( x) [n( x) . (13.81)
Using Eq. (13.69), the denition (13.76) and p
l
= i

x
l
[see Eq. (3.29)] one
nds that [compare with Eq. (13.14)]
A
m,n;l
= i
m( x)[
H1
x
l
[n( x)
m
. (13.82)
In the adiabatic approximation the o diagonal elements of

A
l
[see Eq.
(13.82)] are considered as negligible small. In this case the set of equations
of motion (13.80) becomes decoupled
i
m
=
_
l
_
(p
l
A
m,m;l
( x))
2
2m
l
+
m
l
2
l
x
2
l
2
_
+
m
( x)
_
m
. (13.83)
As can be seen from the above result (13.83), the adiabatic approximation
greatly simplies the systems equations of motion. The eect of the fast
subsystem on the dynamics of the slow one is taken into account by adding a
vector potential A
m,m;l
( x) and a scalar potential
m
( x) to the Schrdinger
equation of the slow subsystem [compare with Eq. (4.175)]. However, both
potential terms depend on the state m that is being occupied by the fast
subsystem.
Exercise 13.6.1. Show that if m( x)[ /x
l
[m( x) is pure real then
A
mm;l
( x) = 0 . (13.84)
13.8. Solutions
Solution 13.6.1. Note that in general the diagonal elements A
m,m;l
are real
since p
l
is Hermitian [see Eq. (13.76)]. On the other hand, if m( x)[ /x
l
[m( x)
is pure real then A
mm;l
( x) is pure imaginary, thus for this case A
mm;l
( x) = 0.
13.7 Problems
1. Consider the following gauge transformation
[+ =
_
cos

2
exp
_
i
2
_
sin

2
exp
_
i
2
_
_

+
_
=
_
cos

2
sin

2
exp(i)
_
, (13.85)
[ =
_
sin

2
exp
_
i
2
_
cos

2
exp
_
i
2
_
_
_
=
_
sin

2
cos

2
exp(i)
_
. (13.86)
Find an expression for the transformed geometrical phase
(t).
2. Consider a particle having mass mconned by a time dependent potential
well given by
V (x) =
_
0 if 0 x a
if x < 0 or x > a
. (13.87)
where the width of the well a oscillates in time according to
a(t) = a
0
_
1 sin
2
(
p
t)
_
, (13.88)
where a
0
, and
p
are positive, and where < 1.
a) Under what condition the adiabatic approximation is expected to be
valid.
b) Calculate the geometrical phases
n
[see Eq. (13.7)] for a cyclic evo-
lution from time t = 0 to time t = /
p
.
13.8 Solutions

+
_
= exp
_
i
2
_
[+ , (13.89)
_
= exp
_
i
2
_
[ , (13.90)
thus the transformed geometrical phase
(t) [see Eq. (13.18)] becomes
n
(t)
n
(t) =
n
(t)
(t)
2
+
(t
0
)
2
. (13.91)
2. The momentary eigenenergies of the system are given by
E
n
(t) =

2
2
n
2
2ma
2
0
_
1 sin
2
(
p
t)
_
2
, (13.92)
where n = 1, 2, . The corresponding eigenvectors are denoted by
[n(t).
a) In general, the main contribution to interstate transitions come from
time periods when the energy gap between neighboring eigenener-
gies is relatively small. At any given time the smallest energy gap
between momentary eigenenergies is the one between the two lowest
states H
21
(t) = E
2
(t) E
1
(t). Furthermore, the main contribution
to the transition probability is expected to come fromthe regions near
minima points of the energy gap H
21
(t). Near the minima point at
time t = 0 the energy gap H
21
(t) is given by
H
21
(t) =
3
2
2
2ma
2
0
_
1 sin
2
(
p
t)
_
2
=
3
2
2
2ma
2
0
_
1 + 2(
p
t)
2
_
+O
_
t
3
_
=
3
2
2
2ma
2
0
_
1 + 4(
p
t)
2
+O
_
t
3
_
.
(13.93)
The estimated transition probability for the two dimensional case
is given by Eq. (13.65). In view of the fact that all other energy
gaps between momentary eigenenergies is at least 5/3 larger than
H
21
, it is expected that this estimate is roughly valid for the present
case. The requirement that the transition probability given by Eq.
(13.65) is small is taken to be the validity condition for the adiabatic
approximation. Comparing the above expression for H
21
(t) near t =
0 with Eq. (13.50) yields the following validity condition
2
p
1/2
3
3
2ma
2
0
. (13.94)
b) In general, the term n(t
) [ n(t
) is pure imaginary [see Eq. (13.9)].

On the other hand, the fact that the wavefunctions of one dimensional
bound states can be chosen to be real (see exercise 7 of chapter 4),
implies that n(t
) [ n(t
) is pure real. Thus n(t
) [ n(t
) = 0 and
therefore all geometrical phases vanish.
14. The Quantized Electromagnetic Field
This chapter discusses the quantization of electromagnetic (EM) eld for the
relatively simple case of a free space cavity.
14.1 Classical Electromagnetic Cavity
Consider an empty volume surrounded by conductive walls having innite
conductivity. The Maxwells equations (in Gaussian units) are given by
B =
1
c
E
t
, (14.1)
E =
1
c
B
t
, (14.2)
E = 0 , (14.3)
B = 0 , (14.4)
where c = 2.99 10
8
ms
1
is the speed of light in vacuum. In the Coulomb
gauge the vector potential A is chosen such that
A = 0 , (14.5)
and the scalar potential vanishes in the absence of sources (charge and
current). In this gauge both electric and magnetic elds E and B can be
expressed in terms of A only as [see Eqs. (1.41) and (1.42)]
E =
1
c
A
t
, (14.6)
and
B = A . (14.7)
2
A =
1
c
2
2
A
t
2
. (14.8)
Chapter 14. The Quantized Electromagnetic Field
Solution 14.1.1. The gauge condition (14.5) and Eqs. (14.6) and (14.7)
guarantee that Maxwells equations (14.2), (14.3), and (14.4) are satised
E =
1
c
(A)
t
=
1
c
B
t
, (14.9)
E =
1
c
( A)
t
= 0 , (14.10)
B = (A) = 0 , (14.11)
where in the last equation the general vector identity (A) = 0 has
been employed. Substituting Eqs. (14.6) and (14.7) into the only remaining
nontrivial equation, namely into Eq. (14.1), leads to
(A) =
1
c
2
2
A
t
2
. (14.12)
Using the vector identity
(A) = ( A)
2
A , (14.13)
and the gauge condition (14.5) one nds that
2
A =
1
c
2
2
A
t
2
. (14.14)
Exercise 14.1.2. Consider a solution having the form
A = q (t) u(r) , (14.15)
where q (t) is independent on position r and u(r) is independent on time t.
Show that q (t) and u(r) must satisfy
2
u+
2
u = 0 , (14.16)
and
d
2
q
dt
2
+
2
q = 0 , (14.17)
where is a constant and where
= c . (14.18)
Solution 14.1.2. The gauge condition (14.5) leads to
u = 0 . (14.19)
From Eq. (14.8) one nds that
14.1. Classical Electromagnetic Cavity
q
2
u =
1
c
2
u
d
2
q
dt
2
. (14.20)
Multiplying by an arbitrary unit vector n leads to
_
2
u
_
n
u n
=
1
c
2
q
d
2
q
dt
2
. (14.21)
The left hand side of Eq. (14.21) is a function of r only while the right hand
side is a function of t only. Therefore, both should equal a constant, which is
denoted as
2
, thus
2
u+
2
u = 0 , (14.22)
and
d
2
q
dt
2
+
2
q = 0 , (14.23)
where
= c . (14.24)
Exercise 14.1.3. Show that the general solution can be expanded as
A =

n
q
n
(t) u
n
(r) . (14.25)
where the set u
n
forms a complete orthonormal basis spanning the vector
space of all solutions of Eq. (14.16) satisfying the proper boundary conditions
on the conductive walls having innite conductivity.
Solution 14.1.3. Equation (14.16) should be solved with the boundary con-
ditions of a perfectly conductive surface. Namely, on the surface S enclosing
the cavity we have B s = 0 and Es = 0, where s is a unit vector normal
to the surface. To satisfy the boundary condition for E we require that u be
normal to the surface, namely, u =s (u s) on S. This condition guarantees
also that the boundary condition for B is satised. To see this we calculate
the integral of the normal component of B over some arbitrary portion S
of
S. Using Eq. (14.7) and Stokes theorem one nds that
_
S
(B s) dS = q
_
S
[(u) s] dS
= q
_
C
u dl ,
(14.26)
where the close curve C encloses the surface S
. Thus, since u is normal to

the surface, one nds that the integral along the close curve C vanishes, and
therefore
_
S
(B s) dS = 0 . (14.27)
Since S
is arbitrary we conclude that B s =0 on S. Each solution of Eq.

(14.16) that satises the boundary conditions is called an eigen mode. As can
be seen from Eq. (14.23), the dynamics of a mode amplitude q is the same
as the dynamics of a harmonic oscillator having angular frequency
=
c. Inner product between dierent solutions of Eq. (14.16) that satisfy the
boundary conditions can be dened as
u
1
, u
2

_
V
(u
1
u
2
) dV , (14.28)
where the integral is taken over the volume of the cavity. Using Eq. (14.16)
one nds that
_
2
2
2
1
_
_
V
(u
1
u
2
) dV =
_
V
_
u
1

2
u
2
u
2

2
u
1
_
dV . (14.29)
Using Greens theorem one nds that
_
2
2
2
1
_
_
V
(u
1
u
2
) dV =
_
S
(u
1
[(s ) u
2
] u
2
[(s ) u
1
]) dS .
(14.30)
Using Eq. (14.19), the boundary condition u =s (u s) on S, and writing =
s (s ) s(s ), we nd that the right hand side of (14.30) vanishes.
Thus, solutions with dierent
2
are orthogonal to each other. Let u
n
be a
complete orthonormal basis spanning the vector space of all solutions of Eq.
(14.16) satisfying the boundary conditions. For any two vectors in this basis
the orthonormality condition is
u
n
, u
m
=
_
V
(u
n
u
m
) dV =
n,m
. (14.31)
Using such a basis we can expand the general solution as
A =

n
q
n
(t) u
n
(r) . (14.32)
Exercise 14.1.4. Show that the total electric energy in the cavity is given
by
U
E
=
1
8c
2
n
q
2
n
, (14.33)
and the total magnetic energy is given by
U
B
=
1
8
2
n
q
2
n
. (14.34)
14.1. Classical Electromagnetic Cavity
Solution 14.1.4. Using Eqs. (14.6),(14.7) and (14.25) one nds that the
elds are given by
E =
1
c
n
q
n
u
n
, (14.35)
and
B =

n
q
n
u
n
. (14.36)
The total energy of the eld is given by U
E
+ U
B
, where U
E
(U
B
) is the
energy associated with the electric (magnetic) eld, namely,
U
E
=
1
8
_
V
E
2
dV , (14.37)
and
U
B
=
1
8
_
V
B
2
dV . (14.38)
U
E
=
1
8c
2
n
q
2
n
, (14.39)
and using Eq. (14.36) one nds that
U
B
=
1
8
n,m
q
n
q
m
_
V
(u
n
) (u
m
) dV . (14.40)
The last integral can be calculated by using the vector identity
(F
1
F
2
) = (F
1
) F
2
F
1
(F
2
) , (14.41)
applied to u
n
(u
m
), thus
(u
n
)(u
m
) = (u
n
(u
m
))+u
n
[(u
m
)] . (14.42)
Using the divergence theorem and the fact that u
n
and (u
m
) are or-
thogonal to each other on S one nds that the volume integral of the rst
term vanishes. To calculate the integral of the second term it is convenient
to use the identity
(u
m
) = ( u
m
)
2
u
m
. (14.43)
This together with Eqs. (14.19), (14.16), and (14.31) lead to
U
B
=
1
8
2
n
q
2
n
. (14.44)
The Lagrangian of the system is given by [see Eq. (1.16)]
/ = U
E
U
B
=
1
4c
2
n
_
q
2
n
2

2
n
q
2
n
2
_
, (14.45)
where the symbol overdot is used for derivative with respect to time, and
where
n
= c
n
. The Euler-Lagrange equations, given by
d
dt
_
/
q
n
_
/
q
n
= 0 , (14.46)
lead to Eq. (14.23).
The variable canonically conjugate to q
n
is [see Eq. (1.20)]
p
n
=
/
q
n
=
1
4c
2
q
n
. (14.47)
The classical Hamiltonian H
F
of the eld is given by [see Eq. (1.22)]
H
F
=

n
p
n
q
n
/ =
n
_
4c
2
p
2
n
2
+
1
4c
2
2
n
q
2
n
2
_
. (14.48)
The Hamilton-Jacobi equations of motion, which are given by
q
n
=
H
F
p
n
= 4c
2
p
n
, (14.49)
p
n
=
H
F
q
n
=

2
n
4c
2
q
n
, (14.50)
lead also to Eq. (14.23). Note that, as expected, the following holds
H
F
= U
E
+U
B
, (14.51)
namely the Hamiltonian expresses the total energy of the system.
14.2 Quantum Electromagnetic Cavity
The coordinates q
n
and their canonically conjugate variables p
n
are regarded
as Hermitian operators satisfying the following commutation relations [see
Eqs. (3.5) and (4.41)]
[q
n
, p
m
] = i
n,m
, (14.52)
and
[q
n
, q
m
] = [p
n
, p
m
] = 0 . (14.53)
14.2. Quantum Electromagnetic Cavity
In general, the Heisenberg equation of motion (4.37) of an operator A
(H)
is
given by
dA
(H)
dt
=
1
i
_
A
(H)
, H
(H)
F
_
+
A
(H)
t
. (14.54)
Thus, with the help of Eq. (14.48) one nds that
q
n
= 4c
2
p
n
, (14.55)
and
p
n
=

2
n
4c
2
q
n
. (14.56)
It is useful to introduce the annihilation and creation operators
a
n
= e
int
_

n
8c
2
_
q
n
+
4ic
2
p
n
n
_
, (14.57)
a
n
= e
int
_

n
8c
2
_
q
n
4ic
2
p
n
n
_
. (14.58)
The phase factor e
i
n
t
in the denition of a
n
is added in order to make it
time independent. The inverse transformation is given by
q
n
=
_
2c
2
n
_
e
i
n
t
a
n
+e
i
n
t
a
n
_
, (14.59)
p
n
= i
_
n
8c
2
_
e
i
n
t
a
n
+e
i
n
t
a
n
_
. (14.60)
The commutation relations for the these operators are derived directly from
Eqs. (14.52) and (14.53)
_
a
n
, a
=
n,m
, (14.61)
[a
n
, a
m
] =
_
a
n
, a
= 0 . (14.62)
The Hamiltonian (14.48) can be expressed using Eqs. (14.59) and (14.60) as
H
F
=

n
n
_
a
n
a
n
+
1
2
_
. (14.63)
The eigenstates are the photon-number states [s
1,
s
2
, ..., s
n
, ..., which satisfy
[see chapter 5]
H
F
[s
1,
s
2
, , s
n
, =

n
n
_
s
n
+
1
2
_
[s
1,
s
2
, , s
n
, . (14.64)
The following holds [see Eqs. (5.28) and (5.29)]
a
n
[s
1,
s
2
, , s
n
, =
s
n
[s
1,
s
2
, , s
n
1, , (14.65)
a
n
[s
1,
s
2
, , s
n
, =
s
n
+ 1 [s
1,
s
2
, , s
n
+ 1, . (14.66)
The non-negative integer s
n
is the number of photons occupying mode n.
The vector potential A (14.25) becomes
A(r, t) =
n
_
2c
2
n
_
e
int
a
n
+e
int
a
n
_
u
n
(r) . (14.67)
14.3 Periodic Boundary Conditions
Consider the case where the EM eld is conned to a nite volume V , which
for simplicity is taken to have a cube shape with edge L = V
1/3
. The
eigen modes and eigen frequencies of the EM eld are found in exercise 1
of this chapter for the case where the walls of the cavity are assumed to
have innite conductance [see Eq. (14.84)]. It is however more convenient to
assume instead periodic boundary conditions, since the spatial dependence
of the resulting eigen modes [denoted by u
n
(r)], can be expressed in terms of
exponential functions, rather than trigonometric functions [see Eqs. (14.77),
(14.78) and (14.79)]. For this case Eq. (14.63) becomes
H
F
=

k,
k
_
a
k,
a
k,
+
1
2
_
, (14.68)
and Eq. (14.67) becomes
A(r, t) =

k,
_
2c
2
k
V
_
k,
e
i(kr
k
t)
a
k,
+
k,
e
i(kr
k
t)
a
k,
_
, (14.69)
where the eigen frequencies are given by
k
= c [k[. In the limit of large
volume the discrete sum over wave vectors k can be replaced by an integral
V
(2)
3
_

dk
x
_

dk
y
_

dk
z
, (14.70)
and the commutation relations between eld operators become
[a
k,
, a
k
,
] =
_
a
k,
, a
,
_
= 0 , (14.71)
_
a
k,
, a
_
=
,

_
k k
_
. (14.72)
The sum over contains two terms corresponding to two polarization
vectors
k,
, which are normalized to unity and mutually orthogonal, i,e.
k,

k,
=
,
. Furthermore, the Coulomb gauge condition requires that
14.5. Solutions
k,
k =
k,
k = 0, i.e. the polarization vectors are required to be orthogonal
to the wave vector k. Linear polarization can be represented by two mutually
orthogonal real vectors
k,1
and
k,2
, which satisfy
k,1
k,2
= k/ [k[. For
the case of circular polarization the polarization vectors can be chosen to be
given by
k,+
=
1
2
(
k,1
+i
k,2
) , (14.73)
k,
=
1
2
(
k,1
i
k,2
) . (14.74)
For this case of circular polarization the following holds
k,

k,
=
,
, (14.75)
k,
k,
= i
k
[k[
,
. (14.76)
where +, .
14.4 Problems
1. Find the eigen modes and eigen frequencies of a cavity having a pizza
box shape with volume V = L
2
d.
2. Consider two perfectly conducting metallic plates placed in parallel to
each other. The gap between the plates is d and the temperature is as-
sumed to be zero. Calculate the force per unit area acting between the
plates (the Casimir force).
3. Find the average energy per unit volume of the electromagnetic eld in
thermal equilibrium at temperature T.
14.5 Solutions
1. We seek solutions of Eq. (14.16) satisfying the boundary condition that
the tangential component of u vanishes on the walls. Consider a solution
having the form
u
x
(r) =
_
8
V
a
x
cos (k
x
x) sin(k
y
y) sin(k
z
z) , (14.77)
u
y
(r) =
_
8
V
a
y
sin(k
x
x) cos (k
y
y) sin(k
z
z) , (14.78)
u
z
(r) =
_
8
V
a
z
sin(k
x
x) sin(k
y
y) cos (k
z
z) . (14.79)
While the boundary condition on the walls x = 0, y = 0, and z = 0 is
guaranteed to be satised, the boundary condition on the walls x = L,
y = L, and z = d yields
k
x
=
n
x
L
, (14.80)
k
y
=
n
y
L
, (14.81)
k
z
=
n
z
d
, (14.82)
where n
x
, n
y
and n
z
are integers. This solution clearly satises Eq.
(14.16), where the eigen value is given by
=
_
k
2
x
+k
2
y
+k
2
z
. (14.83)
Alternatively, using the notation n = (n
x
, n
y
, n
z
) one has = (/L) n,
where n =
_
n
2
x
+n
2
y
+n
2
z
. Using Eq. (14.24) one nds that the angular
frequency of a mode characterized by the vector of integers n is given by
n
= c
_
_
n
x
L
_
2
+
_
n
y
L
_
2
+
_
n
z
d
_
2
. (14.84)
In addition to Eq. (14.16) and the boundary condition, each solution has
to satisfy also the transversality condition u = 0 (14.19), which in
the present case reads
k a = 0 , (14.85)
where k = (k
x
, k
y
, k
z
) and a = (a
x
, a
y
, a
z
). Thus, for each set of integers
n
x
, n
y
, n
z
there are two orthogonal modes (polarizations), unless n
x
=
0 or n
y
= 0 or n
z
= 0. In the latter case, only a single solution exists.
The inner product between two solutions u
1
and u
1
having vectors of
integers n
1
= (n
x1
, n
y1
, n
z1
) and n
2
= (n
x2
, n
y2
, n
z2
), and vectors of
amplitudes a
1
= (a
x1
, a
y1
, a
z1
) and a
2
= (a
x2
, a
y2
, a
z2
), respectively, can
be calculated using Eq. (14.28)
u
1
, u
2
=
_
V
(u
1
u
2
) dV
=
_
V
(u
x1
u
x2
+u
y1
u
y2
+u
z1
u
z2
) dV .
(14.86)
The following holds
14.5. Solutions
_
V
u
x1
u
x2
dV
=
8
V
a
x1
a
x2
_
L
0
cos
_
n
x1
L
x
_
cos
_
n
x2
L
x
_
dx
_
L
0
sin
_
n
y1
L
y
_
sin
_
n
y2
L
y
_
dy
_
d
0
sin
_
n
z1
d
z
_
sin
_
n
z2
d
z
_
dz ,
=
8
V
a
x1
a
x2
L
2
d
8

n
x1
,n
x2
n
y1
,n
y2
n
z1
,n
z2
.
(14.87)
Similar results are obtained for the contribution of the y and z compo-
nents. Thus
u
1
, u
2
= (a
1
a
2
)
nx1,nx2
ny1,ny2
nz1,nz2
, (14.88)
and therefore the vectors of amplitudes a are required to be normalized,
i.e. to satisfy a a = 1, in order to ensure that the solutions u are
normalized.
2. Employing the results of the previous exercise, the eigen frequencies
n
are taken to be given by Eq. (14.84), where L is assumed to be much
larger than d. As can be seen from Eq. (14.64), each mode contributes
energy of
n
/2 to the total energy of the ground state of the system,
which is denoted by E(d). Let E() be the ground state energy in the
limit where d and let U (d) = E(d)E() be the potential energy
of the system. Formally, both E(d) and E() are innite, however, as
we will show below, the divergence can be regulated when evaluating the
dierence U (d). The assumption that L is large allows substituting the
discrete sums over n
x
and n
y
by integrals when evaluating E(d) and
E(). Moreover the discrete sum over n
z
is substituted by an integral
in the expression for E(). The prime on the summation symbol over
n
z
in the expression for E(d) below implies that a factor of 1/2 should
be inserted if n
z
= 0, when only one polarization exists (see previous
exercise). Using these approximations and notation one has
U (d) = E(d) E()
= c
_
L
_
2
nz
_

0
dk
x
_

0
dk
y
_
k
2
x
+k
2
y
+
_
n
z
d
_
2
c
_
L
_
2
d
_

0
dk
x
_

0
dk
y
_

0
dk
z
_
k
2
x
+k
2
y
+k
2
z
.
(14.89)
In polar coordinates u =
_
k
2
x
+k
2
y
and = tan
1
(k
y
/k
x
) one has
dk
x
dk
y
= udud, thus
U (d) = c
_
L
_
2
n
z
_

0
du u
_
u
2
+
_
n
z
d
_
2
c
_
L
_
2
d

2
_

0
du u
_

0
dk
z
_
u
2
+k
2
z
.
(14.90)
Changing the integration variables
x =
_
ud
_
2
, (14.91)
N
z
=
k
z
d
, (14.92)
leads to
U (d) =

2
cL
2
4d
3
_

nz
F (n
z
)
_

0
dN
z
F (N
z
)
_
=

2
cL
2
4d
3
_
1
2
F (0) +

nz=1
F (n
z
)
_

0
dN
z
F (N
z
)
_
,
(14.93)
where the function F () is given by
F () =
_

0
dx
_
x +
2
=
_

2
dy
y . (14.94)
Formally, the function F () diverges. However, the following physical
argument can be employed in order to regulate this divergency. The as-
sumption that the walls of the cavity perfectly conduct is applicable at
low frequencies. However, any metal becomes eectively transparent in
the limit of high frequencies. Thus, the contribution to the ground state
energy of high frequency modes is expected to be eectively d indepen-
dent, and consequently U (d) is expected to become nite. Based on this
argument the divergency in F () is removed by introducing a cuto func-
tion f (y) into the integrand in Eq. (14.94)
F () =
_

2
dy
yf (y) . (14.95)
While near y = 0 (low frequencies) the cuto function is assumed to be
given by f (y) = 1, in the limit of large y (high frequencies) the function
f (y) is assumed to approach zero suciency fast to ensure that F () is
14.5. Solutions
nite for any . Moreover, it is assumed that F() 0. In this case the
Euler-Maclaurin summation formula, which is given by
1
2
F (0) +

n=1
F (n)
_

0
dN F (N) =
1
12
F
(0) +
1
720
F
(0) + ,
(14.96)
can be employed to evaluate U (d). The following holds
F
() = 2
2
f
_
2
_
, (14.97)
thus for small [where the cuto function is assumed to be given by
f (y) = 1] F
() = 4 and F
() = 4, and therefore
U (d) =
2
cL
2
720d
3
. (14.98)
The force per unit area (pressure) P (d) is found by taking the derivative
with respect to d and by dividing by the area L
2
P (d) =

2
c
240d
4
. (14.99)
The minus sign indicates that the force is attractive.
3. The average energy U in thermal equilibrium is given by Eq. (8.332),
which is given by
U =
log Z
c
, (14.100)
where Z
c
= Tr
_
e
H
_
is the canonical partition function, H is the
Hamiltonian [see Eq. (14.64)], = 1/k
B
T and k
B
is the Boltzmanns
constant. The partition function is found by summing over all photon-
number states [s
1,
s
2
, ...
Z
c
=

s1,s2,...=0
s
1,
s
2
, ...[ e
H
[s
1,
s
2
, ...
=

s1,s2,...=0
e
n
n(sn+
1
2
)
=

n
_

sn=0
e
n(sn+
1
2
)
_
=

n
_
1
2 sinh

n
2
_
,
(14.101)
where n labels the cavity modes. Using the last result one nds that
U =
log Z
c
n
log
_
1
2 sinh
n
2
_
=

n
n
2
coth

n
2
.
(14.102)
It is easy to see that the above sum diverges since the number of modes
in the cavity is innite. To obtain a nite result we evaluate below the
dierence U
d
= U (T) U (T = 0) between the energy at temperature
T and the energy at zero temperature, which is given by (recall that
coth(x) 1 in the limit x )
U
d
=

n
n
2
_
coth

n
2
1
_
=

n
n
e
n
1
.
(14.103)
The angular frequencies
n
of the modes are given by Eq. (14.84). For
simplicity a cubical cavity having volume V = L
3
is considered. For this
case U
d
is given by (the factor of 2 is due to polarization degeneracy)
U
d
= 2k
B
T

nx=0
ny=0
nz=0
n
e
n
1
. (14.104)
where n =
_
n
2
x
+n
2
y
+n
2
z
, and where the dimensionless parameter is
given by
=
c
L
. (14.105)
In the limit where 1 (macroscopic limit) the sum can be approxi-
mated by the integral
14.5. Solutions
U
d
= 2k
B
T
4
8
_
0
dn n
2
n
e
n
1
=
k
B
T
_
0
x
3
dx
e
x
1
. .
4
15
,
(14.106)
thus the energy per unit volume is given by
U
d
V
=

2
(k
B
T)
4
15
3
c
3
. (14.107)
15. Light Matter Interaction
In this chapter the transitions between atomic states that result from inter-
action with an electromagnetic (EM) eld are discussed.
15.1 Hamiltonian
Consider an atom in an EM eld. The classical Hamiltonian H
F
of the EM
eld is given by Eq. (14.48). For the case of hydrogen, and in the absence
of EM eld, the Hamiltonian of the atom is given by Eq. (7.2). In general,
the classical Hamiltonian of a point particle having charge e and mass m
e
in an EM eld having scalar potential and vector potential A is given by
Eq. (1.62). In the Coulomb gauge the vector potential A is chosen such that
A = 0, and the scalar potential vanishes provided that no sources
(charge and current) are present. The EM eld is assumed to be suciently
small to allows employing the following approximation
_
p
e
c
A
_
2
p
2
2
e
c
A p , (15.1)
where p is the momentum vector. Recall that in the Coulomb gauge the vector
operators p and A satisfy the relation p A = A p, as can be seen from
Eqs. (6.158) and (6.303). These results and approximation allow expressing
the Hamiltonian of the system as
H = H
0
+H
F
+H
p
, (15.2)
where H
0
is the Hamiltonian of the atom in the absence of EM eld, and
where H
p
, which is given by
H
p
=
e
m
e
c
A p , (15.3)
is the coupling Hamiltonian between the atom and the EM eld.
In the quantum case The Hamiltonian H
F
of the EM eld is given by Eq.
(14.68)
H
F
=

k,
k
_
a
k,
a
k,
+
1
2
_
, (15.4)
Chapter 15. Light Matter Interaction
and the vector potential A is given by Eq. (14.69)
A(r, t) =
k,
_
2c
2
k
V
_
k,
e
i(kr
k
t)
a
k,
+
k,
e
i(kr
k
t)
a
k,
_
. (15.5)
15.2 Transition Rates
While the Hamiltonian H
p
is considered as a perturbation, the unperturbed
Hamiltonian is taken to be H
0
+ H
F
. The eigenvectors of H
0
+ H
F
are la-
beled as [s
k,
, . While the integers s
k,
represent the number of photons
occupying each of the modes of the EM eld, the index labels the atomic
energy eigenstate. The following holds
H
0
[s
k,
, = E
[s
k,
, ,
where E
is the energy of the atomic state, and

H
F
[s
k,
, =

k,
k
_
s
k,
+
1
2
_
[s
k,
, . (15.6)
15.2.1 Spontaneous Emission
Consider the case where the system is initially in a state [i = [s
k,
= 0 ,
i
,
for which all photon occupation numbers are zero, and the atomic state is la-
beled by the index
i
. The nal state is taken to be [f = a
k,
[s
k,
= 0 ,
f
,
for which one photon is created in mode k, , and the atomic state is labeled
by the index
f
. To lowest nonvanishing order in perturbation theory the
transition rate w
i,f
is given by Eq. (10.34)
w
i,f
=
2
2
(
k
i,f
) [f[ H
p
[i[
2
, (15.7)
where
i,f
=
_
E
i
E
f
_
/. With the help of Eqs. (14.69) and (15.3) w
i,f
can be rewritten as
w
i,f
=
_
e
m
e
c
_
2
4
2
c
2
k
V
(
k
i,f
)
f[
k,
pe
ikr
a
k,
[i
2
. (15.8)
As can be seen from Eq. (7.2), the following holds
[H
0
, r] =
1
m
e
(i) p , (15.9)
thus
15.2. Transition Rates
w
i,f
=
4
2
e
2
k
V
(
k
i,f
)
f[
k,
re
ikr
a
k,
[i
2
=
4
2
e
2
k
V
(
k
i,f
) [M
i,f
[
2
,
(15.10)
where the atomic matrix element M
i,f
is given by
M
i,f
=
f
[
k,
re
ikr
[
i
. (15.11)
15.2.2 Stimulated Emission and Absorption
The process of spontaneous emission of a photon in mode k, can be labeled
as (i,s
k,
) (f,s
k,
+ 1), where s
k,
= 0. In the case of stimulated emission,
on the other hand, the initial photon occupation is assumed to be nonzero,
i.e. s
k,
1. Let w
(e)
(i,s
k,
)(f,s
k,
+1),
be the rate of emission of photons in
mode k, , given that the initial photon occupation number is s
k,
. With
the help of Eq. (14.66) the expression for the case of spontaneous emission
(15.10) can be easily generalized for arbitrary initial photon occupation s
k,
w
(e)
(i,s
k,
)(f,s
k,
+1),
=
4
2
e
2
k
(s
k,
+ 1)
V
(
k
i,f
) [M
i,f
[
2
. (15.12)
Note that for the case of emission it is assumed that the energy of the
atomic state i is larger than the energy of the atomic state f, i.e.
i,f
=
_
E
i
E
f
_
/ > 0.
Absorption is the reverse process. Let w
(a)
(i,s
k,
)(f,s
k,
1),
be the rate of
absorption of photons in mode k, , given that the initial photon occupation
number is s
k,
. With the help of Eq. (14.65) one nds using a derivation
similar to the one that was used above to obtain Eq. (15.12) that
w
(a)
(i,sk,)(f,sk,1),
=
4
2
e
2
k
s
k,
V
(
k
+
i,f
) [M
i,f
[
2
. (15.13)
Note that in this case it is assumed that
i,f
< 0.
The emission (15.12) and absorption (15.13) rates provide the contribu-
tion of a single mode of the EMeld. Let d
(e)
(i,s)(f,s+1),
/d (d
(a)
(i,s)(f,s1),
/d)
be the total emitted (absorbed) rate in the innitesimal solid angle d having
polarization . For both cases s denotes the photon occupation number of the
initial state. To calculate these rates the contributions from all modes in the
EM eld should be added. In the limit of large volume the discrete sum over
wave vectors k can be replaced by an integral according to Eq. (14.70). Note
that both the emitted and absorbed radiations are expected to be isotropic.
Thus, by using the relation
k
= ck, where k = [k[, one nds that
d
(e)
(i,s)(f,s+1),
d
=
V
(2)
3
_

0
dk k
2
w
(e)
(i,s
k,
)(f,s
k,
+1),
=
e
2
(s + 1)
2c
3
[M
i,f
[
2
_

0
dx x
3
(x
i,f
)
=

fs
(s + 1)
3
i,f
2c
2
[M
i,f
[
2
.
(15.14)
where
fs
= e
2
/c 1/137 is the ne-structure constant. In a similar way,
one nds for the case of absorption that
d
(a)
(i,s)(f,s+1),
d
=

fs
s
3
i,f
2c
2
[M
i,f
[
2
. (15.15)
15.2.3 Selection Rules
While the size of an atom a
atom
is on the order of the Bohrs radius a
0
=
0.5310
10
m (7.64), the energy dierence E
i
E
f
is expected to be on the
order of the ionization energy of hydrogen atom E
I
= 13.6 eV (7.66). Using
the relation
k
=
_
E
i
E
f
_
/ = ck one nds that a
atom
k 10
3
. Thus,
to a good approximation the term e
ikr
in the expression for the matrix
element M
i,f
can be replaced by unity
M
i,f

f
[
k,
r [
i
. (15.16)
This approximation is called the dipole approximation.
The atomic energy eigenstates [ can be chosen to be also eigenvectors
of the angular momentum operators L
z
and L
2
. It is convenient to employ
the notation [k, l, m, to label these states, where k, l and m are orbital
quantum numbers and where labels the spin state. As can be seen from
Eqs. (7.42), (7.43) and (7.44) the following holds
H
0
[k, l, m, = E
kl
[k, l, m, , (15.17)
L
2
[k, l, m, = l (l + 1)
2
[k, l, m, , (15.18)
L
z
[k, l, m, = m [k, l, m, . (15.19)
Since it is assumed that no magnetic eld is externally applied, the eigenen-
ergies E
kl
are taken to be independent on the quantum numbers m and .
Radiation transitions between a pair of states [k
i
, l
i
, m
i
,
i
and [k
f
, l
f
, m
f
,
f
can occur only when the corresponding matrix element (15.16) is nonzero.
This requirement yields some conditions known as selection rules. The rst
one refers to the spin quantum number . Note that M
i,f
is a matrix element
of an orbital operator (15.16), and consequently it vanishes unless
f
=
i
, or
alternatively, unless =
f

i
= 0. It is important to keep in mind that
this selection rule is valid only when spin-orbit interaction can be neglected.
15.2. Transition Rates
Exercise 15.2.1. Show that the selection rule for the magnetic quantum
number m is given by
m = m
f
m
i
1, 0, 1 . (15.20)
Solution 15.2.1. Using the relations L
z
= xp
y
yp
x
and [x
i
, p
j
] = i
ij
it is easy to show that [L
z
, z] = 0 and [L
z
, x iy] = (x iy). The rst
relation together with Eq. (15.19) imply that
0 = k
f
, l
f
, m
f
,
f
[ [L
z
, z] [k
i
, l
i
, m
i
,
i
= (m
f
m
i
) k
f
, l
f
, m
f
,
f
[ z [k
i
, l
i
, m
i
,
i
,
(15.21)
whereas the second relation together with Eq. (15.19) imply that
k
f
, l
f
, m
f
,
f
[ [L
z
, x iy] [k
i
, l
i
, m
i
,
i
= (m
f
m
i
) k
f
, l
f
, m
f
,
f
[ x iy [k
i
, l
i
, m
i
,
i
= k
f
, l
f
, m
f
,
f
[ (x iy) [k
i
, l
i
, m
i
,
i
,
(15.22)
thus
(m
f
m
i
1) k
f
, l
f
, m
f
,
f
[ x iy [k
i
, l
i
, m
i
,
i
= 0 . (15.23)
Therefore M
i,f
= 0 [see Eq. (15.16)] unless m 1, 0, 1. The transition
m = 0 is associated with linear polarization in the z direction, whereas
the transitions m = 1 are associated with clockwise and counterclockwise
circular polarizations respectively.
Exercise 15.2.2. Show that the selection rule for the quantum number l is
given by
l = l
f
l
i
1, 1 . (15.24)
Solution 15.2.2. Using Eq. (15.40), which is given by
_
L
2
,
_
L
2
, r
= 2
2
_
rL
2
+L
2
r
_
, (15.25)
together with Eq. (15.18) yield
k
f
, l
f
, m
f
,
f
[
_
L
2
,
_
L
2
, r
[k
i
, l
i
, m
i
,
i
= 2
4
(l
f
(l
f
+ 1) +l
i
(l
i
+ 1)) k
f
, l
f
, m
f
,
f
[ r [k
i
, l
i
, m
i
,
i
=
4
(l
f
(l
f
+ 1) l
i
(l
i
+ 1))
2
k
f
, l
f
, m
f
,
f
[ r [k
i
, l
i
, m
i
,
i
,
(15.26)
thus with the help of the identity
(l
f
(l
f
+ 1) l
i
(l
i
+ 1))
2
2 (l
f
(l
f
+ 1) +l
i
(l
i
+ 1))
= (l
i
+l
f
) (l
i
+l
f
+ 2)
_
(l
i
l
f
)
2
1
_
,
(15.27)
one nds that
(l
i
+l
f
) (l
i
+l
f
+ 2)
_
(l
i
l
f
)
2
1
_
k
f
, l
f
, m
f
,
f
[ r [k
i
, l
i
, m
i
,
i
= 0 . (15.28)
Since both l
i
and l
f
are non negative integers, and consequently l
i
+l
f
+2 > 0,
one nds that k
f
, l
f
, m
f
,
f
[ r [k
i
, l
i
, m
i
,
i
can be nonzero only when l
i
=
l
f
= 0 or [l[ = 1. However, for the rst possibility, for which l
i
= m
i
= l
f
=
m
f
= 0, the wavefunctions of both states [k
i
, l
i
, m
i
,
i
and [k
f
, l
f
, m
f
,
f
is a
function of the radial coordinate r only [see Eq. (6.130)], and consequently
k
f
, l
f
, m
f
,
f
[ r [k
i
, l
i
, m
i
,
i
= 0. Therefore the selection rule is given by l
1, 1.
15.3 Semiclassical Case
Consider the case where one mode of the EM eld, which has angular fre-
quency and polarization vector , is externally driven to a coherent state
[, where [[ 1. In the semiclassical approximation the annihilation op-
erator of the driven mode a is substituted by the complex constant (and
the operator a
by
). Furthermore, all other modes are disregarded. When

is close to a specic transition frequency
a
= (E
+
E
) / between two
atomic states, which are labeled by [+ and [, the atom can be approxi-
mately considered to be a two level system. In the dipole approximation the
matrix element +[ H
p
[ is given by [see Eqs. (15.3), (15.9) and (15.16)]
+[ H
p
[ =

2
_
e
it
e
it
_
, (15.29)
where
= 2ie
a
_
2
V
+[ r [ . (15.30)
It is convenient to express the complex frequency as =
1
e
i
, where
both
1
and are real. The frequency
1
is called the Rabi frequency. Due
to selection rules the diagonal matrix elements of H
p
vanish.
The Schrdinger equation is given by
i
d
dt
[ = H[ , (15.31)
where the matrix representation in the basis [+ , [ of the Hamiltonian
H is given by [see Eq. (15.29)]
15.3. Semiclassical Case
H =
2
_

a

1
_
e
i(t+)
e
i(t+)
_
1
_
e
i(t+)
e
i(t+)
_

a
_
. (15.32)
It is convenient to express the general solution as
[ (t) = b
+
(t) exp
_
it
2
_
[+ +b
(t) exp
_
it
2
_
[ . (15.33)
Substituting into the Schrdinger equation yields [see Eq. (6.230)]
i
d
dt
_
b
+
b
_
=
1
2
_

1
_
e
i
e
i(2t+)
_
1
_
e
i
e
i(2t+)
_

__
b
+
b
_
,
(15.34)
where
=
a
. (15.35)
In the rotating wave approximation the rapidly oscillating terms e
i(2t+)
are disregarded, since their inuence in the long time limit is typically neg-
ligible. This approximation is equivalent to the assumption that the second
term in Eq. (15.29) can be disregarded. Furthermore, the phase factor can
be eliminated by resetting the time zero point accordingly. Thus, the Hamil-
tonian can be taken to be given by
H =
2
_

a

1
e
it
1
e
it
a
_
, (15.36)
and the equation of motion in the rotating frame can be taken to be given
by
i
d
dt
_
b
+
b
_
=
1
2
_

1
1

__
b
+
b
_
. (15.37)
The time evolution is found using Eq. (6.138) [see also Eq. (6.234)]
_
b
+
(t)
b
(t)
_
=
_
_
cos i
sin
2
1
+()
2
i
1 sin
2
1
+()
2
i
1 sin
2
1
+()
2
cos +i
sin
2
1
+()
2
_
_
_
b
+
(0)
b
(0)
_
,
(15.38)
where
=
_
2
1
+ ()
2
t
2
. (15.39)
15.4 Problems
1. Show that
_
L
2
,
_
L
2
, r
= 2
2
_
rL
2
+L
2
r
_
. (15.40)
2. Consider an atom having a set of orthonormal energy eigenstates [
n
.
The oscillator strength f
nm
associated with the transition between state
[
n
to state [
m
is dened by
f
nm
=
2m
e
n,m
3
[
f
[ r [
i
[
2
. (15.41)
Show that
f
n,n
= 1 . (15.42)
3. Calculate the lifetime of all states of hydrogen atom having principle
quantum number n = 2.
4. Consider a hydrogen atom that is initially at time t in its ground
state. An electric eld in the z direction given by
E(t) = E
0
z

2
2
+t
2
, (15.43)
where is a constant having the dimension of time, is externally applied.
2p
to nd the atom in the sub-shell 2p at time
t .
5. Consider a particle having mass m and charge q moving in a one dimen-
sional harmonic oscillator having angular resonance frequency . Cal-
culate using the dipole approximation the rate of spontaneous emission
from the number state [n to the ground state [0.
15.5 Solutions
1. Using the relations [L
x
, z] = iy, [L
y
, z] = ix and [L
z
, z] = 0 one nds
that
_
L
2
, z
=
_
L
2
x
, z
+
_
L
2
y
, z
= i(L
x
y yL
x
+L
y
x +xL
y
)
= iV z ,
(15.44)
where V = r LL r. Thus the following holds
_
L
2
, r
= iV. With
the help of the identities
15.5. Solutions
[L
x
, V
z
] = L
x
[L
x
, y] [L
x
, y] L
x
+ [L
x
, L
y
] x +x[L
x
, L
y
] = iV
y
,
[L
y
, V
z
] = [L
y
, L
x
] y y [L
y
, L
x
] +L
y
[L
y
, x] + [L
y
, x] L
y
= iV
x
,
[L
z
, V
z
] = [L
z
, L
x
y] [L
z
, yL
x
] + [L
z
, L
y
x] + [L
z
, xL
y
] = 0 ,
one nds that
_
L
2
,
_
L
2
, z
= i
_
L
2
, V
z
=
2
(L
x
V
y
+V
y
L
x
L
y
V
x
V
x
L
y
)
=
2
(L VVL) z ,
(15.45)
thus
_
L
2
,
_
L
2
, r
=
2
(L VVL)
=
2
(L(r L) L(L r) (r L) L+(L r) L)
= 2
2
_
rL
2
+L
2
r
_
.
(15.46)
2. Trivial by the Thomas-Reiche-Kuhn sum rule (4.66).
3. The rate of spontaneous emission
(se)
if,
in general is given by Eq. (15.14)
(se)
if,
= 4
fs
3
i,f
2c
2
[M
i,f
[
2
. (15.47)
The transition frequency
i,f
between the levels n = 2 to n = 1 is given
by [see Eq. (7.66)]
i,f
=
m
e
e
4
2
3
_
1
2
2
+
1
1
2
_
, (15.48)
thus in terms of the Bohrs radius a
0
[see Eq. (7.64)] one has
(se)
if,
=
3
3
2
8
5
fs
m
e
c
2
M
i,f
a
0
2
.
The matrix element of z = r cos and of u
= 2
1/2
(x iy) =
2
1/2
r sine
i
are given by [see Eq. (7.92)]
n
, l
, m
[ z [n, l, m
=
_
0
dr r
3
R
n
l
R
nl
1
_
1
d(cos )
2
_
0
d cos
_
Y
m
Y
m
l
,
(15.49)
and
n
, l
, m
[ u
[n, l, m
=
_
1
2
_
0
dr r
3
R
n
l
R
nl
1
_
1
d(cos )
2
_
0
d sine
i
_
Y
m
Y
m
l
,
(15.50)
where n, l and m are the quantum members of hydrogens energy eigen-
vectors. The nal state is taken to be the ground state (n, l, m) = (1, 0, 0).
In the dipole approximation the transition (2, 0, 0) (1, 0, 0) is forbidden
due to the selection rule l 1, 1. Using the identities
R
10
(r) = 2
_
1
a
0
_
3/2
e
r/a
0
, (15.51)
R
20
(r) = (2 r/a
0
)
_
1
2a
0
_
3/2
e
r
2a
0
, (15.52)
R
21
(r) =
_
1
2a
0
_
3/2
r
3a
0
e
r
2a
0
, (15.53)
Y
0
0
(, ) =
_
1
4
, (15.54)
Y
1
1
(, ) =
1
2
_
3
2
sine
i
, (15.55)
Y
0
1
(, ) =
1
2
_
3
cos , (15.56)
Y
1
1
(, ) =
1
2
_
3
2
sine
i
, (15.57)
where a
0
is Bohrs radius [see Eq. (7.64)], one nds for the radial part
that
_
0
dr r
3
R
10
R
21
=
2
7
6
3
5
a
0
, (15.58)
and for the angular part that
1
_
1
d(cos )
2
_
0
d cos
_
Y
0
0
_
Y
0
1
=
1
3
, (15.59)
1
_
1
d(cos )
2
_
0
d cos
_
Y
0
0
_
Y
1
1
= 0 , (15.60)
1
_
1
d(cos )
2
_
0
d sine
i
_
Y
0
0
_
Y
0
1
= 0 , (15.61)
15.5. Solutions
and
1
_
1
d(cos )
2
_
0
d sine
i
_
Y
0
0
_
Y
1
1
= 0 , (15.62)
_
1
2
1
_
1
d(cos )
2
_
0
d sine
i
_
Y
0
0
_
Y
1
1
=
_
1
3
, (15.63)
_
1
2
1
_
1
d(cos )
2
_
0
d sine
i
_
Y
0
0
_
Y
1
1
=
_
1
3
, (15.64)
1
_
1
d(cos )
2
_
0
d sine
i
_
Y
0
0
_
Y
1
1
= 0 . (15.65)
Thus, by combining all these results one nds that the inverse lifetime of
the states (2, 1, 1), (2, 1, 0) and (2, 1, 1) is given by
(se)
=
3
3
2
8
5
fs
m
e
c
2
2
7
3
5
2
, (15.66)
whereas the lifetime of the state (2, 0, 0) is innite (in the dipole approx-
imation).
4. The probability p
2pm
to nd the atom in the state [n = 2, l = 1, m is
calculated using Eq. (10.42) together with Eq. (7.84)
p
2pm
=
e
2
E
2
0
dt
e
i
3E
I
4
t

2
+t
2
2
[2, 1, m[ z [1, 0, 0[
2
. (15.67)
where
E
I
=
e
4
2
2
(15.68)
is the ionization energy. The following holds
dt
e
i
3E
I
4
t

2
+t
2
=
1
dxe
ix
1 +
_
x
_
2
=

_
dxe
ix
(x i) (x +i)
,
(15.69)
where
=
3E
I
4
, (15.70)
thus with the help of the residue theorem one nds that
dt
e
i
3E
I
4
t

2
+t
2
= e
. (15.71)
The matrix element 2, 1, m[ z [1, 0, 0 is calculated with the help of Eq.
(15.49)
2, 1, m[ z [1, 0, 0 =
_
0
dr r
3
R
21
R
10
1
_
1
d(cos )
2
_
0
d cos (Y
m
1
)
Y
0
0
=
2
7
2a
0
3
5

m,0
,
(15.72)
where
a
0
=

2
e
2
(15.73)
is the Bohrs radius, thus
p
2pm
=
2
15
3
10
_
eE
0
a
0

_
2
e
3E
I
2

m,0
. (15.74)
5. The oscillator is assumed to move along the z direction. The rate of
spontaneous emission
(se)
|n|0,
with polarization into solid angle d
is given in the dipole approximation by [see Eqs. (15.14) and (15.16)]
d
(se)
|n|0,
=
q
2
(n)
3
2c
3
[0[ z [n[
2
_
k,
z
_
2
d , (15.75)
where
k,
is the polarization unit vector. With the help of Eqs. (5.11),
(5.28) and (5.29) one nds that
d
(se)
|n|0,
=
q
2
(n)
3
2c
3
n
2m
n,1
_
k,
z
_
2
d . (15.76)
Integrating over d in spherical coordinates and with the help of the
relation
_
d cos
2
=
_

d
_
1
1
d(cos ) cos
2
=
4
3
, (15.77)
and summing over the two orthogonal polarization yields the total rate
of spontaneous emission
(se)
|n|0
=
2q
2
2
3mc
3

n,1
. (15.78)
16. Identical Particles
This chapter reviews the identical particles postulate of quantum mechanics
and second quantization formalism. It is mainly based on the rst chapter of
Ref. [6].
16.1 Basis for the Hilbert Space
Consider a system containing some integer number N of identical particles.
For the single particle case, where N = 1, the state of the system [ can be
expanded using an orthonormal basis [a
i
i
that spans the single particle
Hilbert space. Based on the single particle basis [a
i
i
we wish to construct
a basis for the Hilbert space of the system for the general case, where N can
be any integer. This can be done in two dierent ways, depending on whether
the identical particles are considered to be distinguishable or indistinguishable
(see example in Fig. 16.1).
Suppose that the particles can be labelled by numbers as billiard balls.
In this approach the particles are considered as distinguishable. For this case
a basis for the Hilbert space of the many-particle system can be constructed
from all vectors having the form [1 : i
1
, 2 : i
2
, , N : i
N
. The ket vector
[1 : i
1
, 2 : i
2
, , N : i
N
represents a state having N particles, where the
particle that is labelled by the number m (m = 1, 2, , N) is in the single
particle state [a
i
m
. Each ket vector [1 : i
1
, 2 : i
2
, , N : i
N
can be char-
acterized by a vector of occupation numbers n = (n
1
, n
2
, ), where n
i
is
the number of particles occupying the single particle state [a
i
. Let g
n
be
the number of dierent ket-vectors having the form [1 : i
1
, 2 : i
2
, , N : i
N
that are characterized by the same vector of occupation numbers n. It is easy

to show that
g
n
=
N!
i
n
i
!
, (16.1)
where N =

i
n
i
is the number of particles.
Alternatively, the particles can be considered as indistinguishable. In this
approach all states having the same vector of occupation numbers n rep-
resent the same physical state, and thus should be counted only once. In
Chapter 16. Identical Particles
other words, when the particles are considered as indistinguishable the sub-
space corresponding to any given vector of occupation numbers n is rather
than being g
n
- fold degenerate (as in the approach where the particles are
considered to be distinguishable) is taken to be nondegenerate. The identical
particle postulate of quantum mechanics states that identical particles should
be considered as indistinguishable. Consequently, a basis for the Hilbert space
of the many-particle system can be constructed from the set of ket vectors
[ n
n
. The ket vector [ n represents a state that is characterized by a vector
of occupation numbers n = (n
1
, n
2
, ), where the integer n
i
is the number
of particles that are in the single particle state [a
i
. Such a basis is considered
to be both orthonormal, i.e.
n
1
[ n
2
=
n1, n2
, (16.2)
where
n
1
, n
2
= 1 if n
1
= n
2
and
n
1
, n
2
= 0 otherwise, and complete
[ n
[ = 1 . (16.3)
It it convenient to introduce the creation operators a
i
. With analogy with
the case of a harmonic oscillator [see Eq. (5.32)] and the case of EM eld [see
Eqs. (14.65) and (14.66)] the state [ n is expressed as
[ n =
1
n
1
!n
2
!
_
a
1
_
n1
_
a
2
_
n2
[0 , (16.4)
where [0 represents the state where all occupation numbers are zero. Equa-
tion (16.4) suggests that the creation operators a
i
maps a given state to a
state having additional particle in the single particle quantum state [a
i
. The
operator a
i
is the Hermitian conjugate of the annihilation operator a
i
. The
number operator N
i
is dened by
N
i
= a
i
a
i
. (16.5)
In addition to the above discussed principle of indistinguishability, the
identical particle postulate of quantum mechanics also states that all particles
in nature are divided into two type: Bosons and Fermions. Moreover, while
for the case of Bosons, the creation and annihilation operators satisfy the
following commutation relations
[a
i
, a
j
] =
_
a
i
, a
j
_
= 0 , (16.6)
_
a
i
, a
j
_
=
ij
, (16.7)
for the case of Fermions the following holds
[a
i
, a
j
]
+
=
_
a
i
, a
j
_
+
= 0 , (16.8)
_
a
i
, a
j
_
+
=
ij
, (16.9)
16.1. Basis for the Hilbert Space
state 1
n
1
=0
1
2
3
state 2
n
2
=2
state 3
n
3
=1
state 4
n
4
=0
state 1
n
1
=0
3
1
2
state 2
n
2
=2
state 3
n
3
=1
state 4
n
4
=0
state 1
n
1
=0
2
3
1
state 2
n
2
=2
state 3
n
3
=1
state 4
n
4
=0
state 1
n
1
=0
state 2
n
2
=2
state 3
n
3
=1
state 4
n
4
=0
(a)
(b)
state 1
n
1
=0
1
2
3
state 2
n
2
=2
state 3
n
3
=1
state 4
n
4
=0
state 1
n
1
=0
1
2
3
state 2
n
2
=2
state 3
n
3
=1
state 4
n
4
=0
state 1
n
1
=0
3
1
2
state 2
n
2
=2
state 3
n
3
=1
state 4
n
4
=0
state 1
n
1
=0
3
1
2
state 2
n
2
=2
state 3
n
3
=1
state 4
n
4
=0
state 1
n
1
=0
2
3
1
state 2
n
2
=2
state 3
n
3
=1
state 4
n
4
=0
state 1
n
1
=0
2
3
1
state 2
n
2
=2
state 3
n
3
=1
state 4
n
4
=0
state 1
n
1
=0
state 2
n
2
=2
state 3
n
3
=1
state 4
n
4
=0
state 1
n
1
=0
state 2
n
2
=2
state 3
n
3
=1
state 4
n
4
=0
(a)
(b)
Fig. 16.1. In this example the number of particles is N =

i
ni = 3, where the
occupation numbers are given by n = (n1, n2, n3, n4, ) = (0, 2, 1, 0, ). When
the particles are considered as distinguishable [see panel (a)] the corresponding
subspace is g n degenerate, where g n = N!/
i
ni! = 3. On the other hand, when
the particles are considered as indistinguishable [see panel (b)], the corresponding
subspace is nondegenerate.
where
_
a
i
, a
j
_
+
denotes anti-commutation, i.e.
[A, B]
+
= AB +BA (16.10)
for general operators A and B.
Exercise 16.1.1. Show that for both Bosons and Fermions
[N
i
, N
j
] = 0 . (16.11)
Solution 16.1.1. For Bosons this result is trivial [see Eqs. (16.6) and (16.7)].
It is also trivial for Fermions when i = j. Finally, for Fermions when i ,= j
one has
N
i
N
j
= a
i
a
i
a
j
a
j
= a
i
a
j
a
i
a
j
= a
i
a
j
a
j
a
i
= a
j
a
i
a
j
a
i
= a
j
a
j
a
i
a
i
= N
j
N
i
.
(16.12)
16.2 Bosons
Based on Eqs. (16.2), (16.4), (16.6) and (16.7) a variety of results can be
obtained:
Exercise 16.2.1. Show that for Bosons
_
a
i
,
_
a
i
_
n
_
= n
_
a
i
_
n1
. (16.13)
Solution 16.2.1. Trivial by Eq. (2.174), which states that for any operators
A and B
[A, B
n
] = nB
n1
[A, B] , (16.14)
and by Eq. (16.7).
a
i
[0 = 0 . (16.15)
Solution 16.2.2. The norm of the vector a
i
[0 can be expressed with the
help of Eqs. (16.4) and (16.7)
0[ a
i
a
i
[0 = 0[
_
a
i
, a
i
_
+a
i
a
i
[0
= 0 [0 +0, 0, , n
i
= 1, 0, [0, 0, , n
i
= 1, 0, ,
(16.16)
thus with the help of the normalization condition (16.2) one nds that
0[ a
i
a
i
[0 = 0 and therefore a
i
[0 = 0.
N
i
[ n = n
i
[ n . (16.17)
Solution 16.2.3. With the help of Eqs. (16.4), (16.13) and (16.15) one nds
that
16.3. Fermions
N
i
[ n = a
i
a
i
[ n
=
1
n
1
!n
2
!
_
a
1
_
n1
_
a
2
_
n2
a
i
a
i
_
a
i
_
ni
[0
=
1
n
1
!n
2
!
_
a
1
_
n1
_
a
2
_
n2
a
i
__
a
i
,
_
a
i
_
ni
_
+
_
a
i
_
ni
a
i
_
[0
=
1
n
1
!n
2
!
_
a
1
_
n1
_
a
2
_
n2
a
i
n
i
_
a
i
_
ni1
[0
= n
i
[ n .
(16.18)
a
i
[n
1,
n
2
, , n
i
, =
n
i
[n
1,
n
2
, , n
i
1, , (16.19)
a
i
[n
1,
n
2
, , n
i
, =
n
i
+ 1 [n
1,
n
2
, , n
i
+ 1, . (16.20)
Solution 16.2.4. Equation (16.20) follows immediately from Eqs. (16.4) and
(16.6). Moreover, with the help of Eqs. (16.4), (16.13) and (16.15) one nds
that
a
i
[n
1,
n
2
, , n
i
, =
1
n
1
!n
2
!
_
a
1
_
n1
_
a
2
_
n2
a
i
_
a
i
_
ni
[0
=
1
n
1
!n
2
!
_
a
1
_
n1
_
a
2
_
n2

__
a
i
,
_
a
i
_
ni
_
+
_
a
i
_
ni
a
i
_
[0
=
n
i
n
1
!n
2
!
_
a
1
_
n
1
_
a
2
_
n
2

_
a
i
_
n
i
1
[0
=
n
i
[n
1,
n
2
, , n
i
1, .
16.3 Fermions
The anti-commutation relations (16.8) for the case i = j yields
_
a
i
_
2
= 0. As
can be seen from Eq. (16.4), this implies that the only possible occupation
numbers n
i
are 0 and 1. This result is known as the Paulis exclusion principle,
according to which no more than one Fermion can occupy a given single
particle state. For Fermions Eq. (16.4) can be written as (recall that 0! =
1! = 1)
[ n =
_
a
1
_
n
1
_
a
2
_
n
2
[0 , (16.21)
where n
i
0, 1.
Exercise 16.3.1. Show that for Fermions
a
i
[0 = 0 . (16.22)
Solution 16.3.1. The norm of the vector a
i
[0 can be expressed with the
help of Eqs. (16.21) and (16.9)
0[ a
i
a
i
[0 = 0[
_
a
i
, a
i
_
+
a
i
a
i
[0
= 0 [0 0, 0, , n
i
= 1, 0, [0, 0, , n
i
= 1, 0, ,
(16.23)
thus with the help of the normalization condition (16.2) one nds that
0[ a
i
a
i
[0 = 0 and therefore a
i
[0 = 0.
N
i
[ n = n
i
[ n , (16.24)
where N
i
= a
i
a
i
.
Solution 16.3.2. Using Eqs. (16.8), (16.9) and (16.21) one nds that
N
i
[ n = a
i
a
i
_
a
1
_
n1
_
a
2
_
n2
[0
= (1)
2

j<i
nj
_
a
1
_
n1
_
a
2
_
n2
a
i
a
i
_
a
i
_
ni
[0
=
_
a
1
_
n1
_
a
2
_
n2
a
i
a
i
_
a
i
_
ni
[0 .
(16.25)
For the case n
i
= 0 this yields [see Eq. (16.22)] N
i
[ n = 0, whereas for the
case n
i
= 1 one has a
i
_
a
i
_
n
i
=
_
a
i
, a
i
_
+
a
i
a
i
= 1a
i
a
i
, thus N
i
[ n = [ n.
Both cases are in agreement with Eq. (16.24).
a
i
[n
1,
n
2
, , n
i
, = n
i
(1)
j<i
n
j
[n
1,
n
2
, , n
i
1, , (16.26)
a
i
[n
1,
n
2
, , n
i
, = (1 n
i
) (1)
j<i
nj
[n
1,
n
2
, , n
i
+ 1, . (16.27)
Solution 16.3.3. According to Eq. (16.8) a
i
a
j
= a
j
a
i
. For i = j this
yields
_
a
i
_
2
= 0. These relations together with Eq. (16.21) leads to Eq.
(16.27) (note that 1 n
i
= 1 if n
i
= 0 and 1 n
i
= 0 if n
i
= 1). Similarly,
Eq. (16.26) is obtained by using the identity a
i
_
a
i
_
ni
= 1 a
i
a
i
and by
considering both possibilities n
i
= 0 and n
i
= 1.
16.4. Changing the Basis
16.4 Changing the Basis
In the previous section the creation a
i
and annihilation a
i
operators were
dened based on a given single particle orthonormal basis [a
i
i
. Consider
an alternative single particle basis [b
j
j
, which is made of eigenvectors of
the single particle observable B
SP
, i.e. the following holds B
SP
= B
SP
and
B
SP
[b
j
=
j
[b
j
, (16.28)
where
j
is the single particle eigenvalue corresponding to the eigenvector
[b
j
. Moreover, this basis is assumed to be both orthonormal, i.e.
b
j
[b
j
=
j,j
, (16.29)
and complete, i.e.
j
[b
j
b
j
[ = 1 . (16.30)
Exploiting the completeness of the original single particle orthonormal basis
[a
i
i
, i.e. the fact that
i
[a
i
a
i
[ = 1 , (16.31)
allows expressing the eigenvector [b
j
as
[b
j
=

i
a
i
[b
j
[a
i
. (16.32)
The single particle state [a
i
can be expressed in the notation of many particle
states as a
i
[0, thus Eq. (16.32) can be rewritten as
[b
j
= b
j
[0 . (16.33)
where the operator b
j
is given by
b
j
=
i
a
i
[b
j
a
i
. (16.34)
The operator b
j
can be considered as the creation operator of the single
particle state [b
j
. It is the Hermitian conjugate of the annihilation operator
b
j
=
i
b
j
[a
i
a
i
. (16.35)
An important example is the case where the single particle observable is
taken to be the position observable r. For this case Eq. (16.32) becomes
[r
=

i
i
(r
) [a
i
, (16.36)
where [r
is a single particle position eigenvector, and where

i
(r
) = r
[a
i
is the wavefunction of the single particle state [a

i
. Expressing [a
i
as a
i
[0
allows rewriting Eq. (16.36) as
[r
(r
) [0 , (16.37)
where the operator
(r
), which is given by
(r
) =
i
(r
) a
i
, (16.38)
is the Hermitian conjugate of the quantized eld operator
(r
) =

i
i
(r
) a
i
. (16.39)
Note that while
i
(r
) is a wave function, (r
) is an operator on the Hilbert

space of the many particle system.
Exercise 16.4.1. Calculate
_
(r
) ,
(r
, where [A, B]
= AB BA
for general operators A and B, and where the minus sign is used for Bosons
and the plus sign for Fermions.
Solution 16.4.1. With he help of Eqs. (16.7) and (16.9) one nds that
_
(r
) ,
(r
=

i,i
i
(r
i
(r
)
_
a
i
, a
=

i
i
(r
i
(r
)
=

i
r
[a
i
a
i
[r
= r
[r
,
(16.40)
thus [see Eq. (3.66)]
_
(r
) ,
(r
= (r
) . (16.41)
Similarly, one nds that
[ (r
) , (r
)]
= 0 , (16.42)
_
(r
) ,
(r
= 0 . (16.43)
16.5. Many Particle Observables
_
d
3
r
(r
) = N , (16.44)
where
(r
) =
(r
) (r
) , (16.45)
and where
N =
i
N
i
(16.46)
The operator (r
) is called the number density operator, and the operator

N is called the total number of particles operator.
Solution 16.4.2. Using the denition of (r
) one nds that

_
d
3
r
(r
) (r
) =

i,i
i
a
i
_
d
3
r
i
(r
)
i
(r
)
=

i
a
i
a
i
= N .
(16.47)
16.5 Many Particle Observables
Observables of a system of identical particles must be dened and must be
represented by Hermitian operators in a way that is consistent with the prin-
ciple of indistinguishability. Below we consider both, one-particle observables
and two-particle observables, and discuss their representation as operators
on the Hilbert space of the many-particle system.
16.5.1 One-Particle Observables
Consider a single particle observable such as the observable B
SP
, which was
introduced in the previous section [see Eqs. (16.28), (16.29), (16.30)]. It is
convenient to employ the single particle basis [b
j
j
, which is made of single-
particle eigenvectors of B
SP
that satisfy B
SP
[b
j
=
j
[b
j
[see Eq. (16.28)],
in order to construct creation b
j
and annihilation b
j
operators. In the many-
particle case, the same physical variable that B
SP
represents for the single
particle case is represented by the operator B, which is given by
B =

j
j
b
j
b
j
. (16.48)
This can be seen by recalling that the operator b
j
b
j
represents the number
of particles in the single particle state [b
j
and that
j
is the corresponding
eigenvalue. With the help of Eqs. (16.28), (16.29), (16.30) and (16.34) (16.35)
the operator B can be expressed in terms of the operators a
i
and a
i
B =

i,i
a
i
[ B
SP
[a
i
a
i
a
i
. (16.49)
16.5.2 Two-Particle Observables
Consider two-body interaction that is represented by an Hermitian operator
V
TP
on the Hilbert space of two-particle states. A basis for this Hilbert space
can be constructed using a given orthonormal basis for the single particle
Hilbert space [b
j
j
. When the two particles are considered as distinguish-
able the basis of the Hilbert space of the two-particle states can be taken
to be [j, j
j,j
. The ket vector [j, j
represents a state for which the rst

particle is in single particle state [b
j
and the second one is in state [b
j
.
Assume the case where the single particle basis vectors [b
j
are chosen in
such a way that diagonalizes V
TP
, i.e.
V
TP
[j, j
= v
j,j
[j, j
, (16.50)
where the eigenvalue v
j,j
is given by
v
j,j
= j, j
[ V
TP
[j, j
. (16.51)
In the many-particle case, the same physical variable that V
TP
represents
for the two-particle case is represented by the operator V , which is given by
V =
1
2
j,j
v
j,j
b
j
b
j
b
j
b
j
. (16.52)
To see that the above expression indeed represents the two particle interaction
consider the expectation value n[ V [ n with respect to the many body state
[ n = [n
1,
n
2
, . The following holds [see Eqs. (16.6) , (16.7), (16.8) and
(16.9)]
b
j
b
j
b
j
b
j
= b
j
b
j
b
j
b
j
= b
j
_
_
b
j
, b
j
_
b
j
b
_
b
j
= b
j
_
_
b
j
, b
b
j
b
_
b
j
= b
j
_
j,j
b
j
b
_
b
j
= N
j
j,j
+N
j
N
j
,
(16.53)
16.6. Hamiltonian
where the upper sign is used for Bosons and the lower one for Fermions. Thus
V can be rewritten as
V =
1
2
j,j
v
j,j
N
j
(N
j

j,j
) . (16.54)
Separating the terms for which j ,= j
from the terms for which j = j
yields
V =

j<j
v
j,j
N
j
N
j
+
1
2
j
v
j,j
N
j
(N
j
1) , (16.55)
thus the matrix element n[ V [ n is given by
n[ V [ n =

j<j
n
j
n
j
v
j,j
+
j
n
j
(n
j
1)
2
v
j,j
. (16.56)
While the factor n
j
n
j
represents the number of particle pairs occupying
single particle states j and j
for the case j ,= j
, the factor n
j
(n
j
1) /2
represents the number of particle pairs occupying the same single particle
states j. Thus the above expression for V (16.52) properly accounts for the
two-particle interaction.
With the help of Eqs. (16.29), (16.30) and (16.34) (16.35) the operator V
can be expressed in terms of the operators a
i
and a
i
V =
1
2
,i
,i
,i
j,j
a
i
, a
i
[j, j
j, j
[ V
TP
[j, j
, j [a
i
, a
i
a
i
a
i
a
i
a
i
=
1
2
,i
,i
,i
j,j
a
i
, a
i
[j, j
j, j
[ V
TP
[j, j
j, j
[a
i
, a
i
a
i
a
i
a
i
a
i
=
1
2
,i
,i
,i
a
i
, a
i
[
_
_
j,j
[j, j
j, j
[
_
_
V
TP
[a
i
, a
i
a
i
a
i
a
i
a
i
,
(16.57)
thus
V =
1
2
,i
,i
,i
a
i
, a
i
[ V
TP
[a
i
, a
i
a
i
a
i
a
i
a
i
. (16.58)
16.6 Hamiltonian
Consider the case where the single-particle Hamiltonian is given by
H
SP
= T
SP
+U
SP
, (16.59)
where the operator T
SP
, which is given by
T
SP
=
p
2
SP
2m
, (16.60)
where p
SP
is the single-particle momentum vector operator and where m is
the mass of a particle, is the single-particle kinetic energy operator, and
where the operator U
SP
(r
) the the single-particle potential energy. The

many-particle kinetic energy operator is found using Eq. (16.49)
T =
1
2m
i,i
a
i
[ p
2
SP
[a
i
a
i
a
i
. (16.61)
The matrix element a
i
[ p
2
SP
[a
i
can be written using the wavefunctions
i
(r
) = r
[a
i
[recall Eq. (3.29), according to which r
[ p[ = i
for a general state []

a
i
[ p
2
SP
[a
i
=

2
2m
_
d
3
r
i
(r
)) (
i
(r
)) . (16.62)
Thus, in terms of the quantized eld operator (r
) [see Eqs. (16.38) and

(16.39)] the operator T can be expressed as
T =

2
2m
_
d
3
r
(r
) (r
) . (16.63)
Integration by parts yields an alternative expression
T =
2
2m
_
d
3
r
(r
)
2
(r
) . (16.64)
Similarly, the many-particle potential energy operator is found using Eq.
(16.49) [recall Eq. (3.23), according to which r
[ f (r) [ = f (r
(r
) for
a general state [ and for a general function f (r)]
U =

i,i
a
i
[ U
SP
(r
) [a
i
a
i
a
i
=
_
d
3
r
U
SP
(r
(r
) (r
) .
(16.65)
In addition, consider the case where the particles interact with each other
via a two-particle potential V
TP
(r
1
, r
2
). The corresponding many-particle
interaction operator is found using Eq. (16.58). The two-particle matrix ele-
ments of V
TP
are given by
a
i
, a
i
[ V
TP
[a
i
, a
i

=
_
d
3
r
_
d
3
r
i
(r
i
(r
) V
TP
(r
, r
)
i
(r
)
i
(r
) ,
(16.66)
16.6. Hamiltonian
thus
V =
1
2
_
d
3
r
_
d
3
r
V
TP
(r
, r
(r
(r
) (r
) (r
) . (16.67)
Combining all these results yields the total many-particle Hamiltonian
H =

2
2m
_
d
3
r
(r
) (r
)
+
_
d
3
r
U
SP
(r
(r
) (r
)
+
1
2
_
d
3
r
_
d
3
r
V
TP
(r
, r
(r
(r
) (r
) (r
) .
(16.68)
Exercise 16.6.1. Show that the Heisenberg equation of motion for the eld
operator (r
) is given by
i
d
dt
(r
, t)
=
_
2
2m
2
+U
SP
(r
)
_
(r
, t)
+
_
d
3
r
V
TP
(r
, r
(r
, t) (r
, t) (r
, t) .
(16.69)
Note that in the absence of two-particle interaction the above equation for
the eld operator (r
, t) is identical to the single-particle Schrdinger equa-

tion for the single particle wavefunction (r
). Due to this similarity the

many-particle formalism of quantum mechanics is sometimes called second
quantization.
Solution 16.6.1. The Heisenberg equation of motion [see Eq. (4.37)] is given
by
i
d
dt
= [H, ]
. (16.70)
For general operators A, B and C the following holds
[AB, C]
= A[B, C]
[A, C]
B
= A[B, C]
[C, A]
B .
(16.71)
Below we employ this relation for evaluating commutation relations. For
Fermions the upper sign (anti-commutation) is chosen, whereas for Bosons
the lower one is chosen (commutation). With the help of Eqs. (16.41), (16.42)
and (16.43) one nds (for both Bosons and for Fermions) that
[T, (r
)]
2
2m
_
d
3
r
(r
)
2
(r
) , (r
=

2
2m
_
d
3
r
(r
)
2
(r
)
=

2
2m
2
(r
) ,
(16.72)
and
[U, (r
)]
=
_
d
3
r
U
SP
(r
)
_
(r
) (r
) , (r
=
_
d
3
r
U
SP
(r
) (r
) (r
) (16.73)
= U
SP
(r
) (r
) .
(16.74)
Similarly
[V, (r
)]
=
1
2
_
d
3
r
_
d
3
r
V
TP
(r
, r
)
_
(r
(r
) (r
) (r
) , (r
=
1
2
_
d
3
r
_
d
3
r
V
TP
(r
, r
)
_
(r
(r
) , (r
(r
) (r
)
=
1
2
_
d
3
r
_
d
3
r
V
TP
(r
, r
(r
) (r
) (r
) (r
1
2
_
d
3
r
_
d
3
r
V
TP
(r
, r
) (r
(r
) (r
) (r
)
=
_
d
3
r
V
TP
(r
, r
(r
) (r
) (r
)
(16.75)
where in the last step it was assumed that V
TP
(r
, r
) = V
TP
(r
, r
). Com-
bining these results lead to Eq. (16.69).
16.7 Momentum Representation
In the momentum representation the Hamiltonian is constructed using a
single-particle basis made of momentum eigenvectors [p
. The wavefunc-
tions of these single-particle states are proportional to e
ik
[see Eq. (3.75)],

where
k
=
p
. (16.76)
16.7. Momentum Representation
These wavefunctions can be normalized when the volume of the system is
taken to be nite. For simplicity, consider the case where the particles are
conned within a volume 1 = L
3
having a cubic shape. The normalized wave
functions are taken to be given by
r
[k
=
k
(r
) =
1
1
e
ik
, (16.77)
where [k
labels a momentum eigenvector having an eigenvalue k
. The
requirement that the wavefunctions
k
(r
) satisfy periodic boundary con-

ditions, i.e.
k
(r
) =
k
(r
+L x) =
k
(r
+L y) =
k
(r
+Lz), yields a
discrete set of allowed values of the wave vector k
k =
2
L
(n
x
, n
y
, n
z
) , (16.78)
where n
x
, n
y
and n
z
are all integers. The orthonormality condition reads
_
V
d
3
r
k
(r
)
k
(r
) =
1
1
_
V
d
3
r
e
i(k
)r
=
k
,k
.
(16.79)
In the momentum representation the many-particle kinetic energy T is
given by [see Eq. (16.61)]
T =
1
2m
,k
[ p
2
SP
[k
a
k
=

2
2m
k
2
a
k
a
k
,
(16.80)
the many-particle potential energy U is given by [see Eq. (16.65)]
U =

k
,k
U
k
k
a
k
a
k
, (16.81)
where
U
k
k
= k
[ U
SP
(r
) [k
=
1
1
_
V
d
3
r
U
SP
(r
) e
i(k
)r
,
(16.82)
and the many-particle interaction operator is given by [see Eq. (16.58)]
V =
1
2
,k
,k
,k
, k
[ V
TP
[k
, k
a
k
a
k
, (16.83)
where
k
, k
[ V
TP
[k
, k
=
1
1
2
_
V
d
3
r
_
V
d
3
r
V
TP
(r
, r
) e
i(k
)r
e
i(k
)r
.
(16.84)
The assumption that V
TP
(r
, r
) is a function of the relative coordinate r =

r
only, together with the coordinates transformation

r
0
=
r
+r
2
, (16.85)
r = r
, (16.86)
yields (note that r
= r
0
+r/2 and r
= r
0
r/2)
k
, k
[ V
TP
[k
, k
=
1
1
2
_
V
d
3
r
0
e
i(k
+k
)r0
_
V
d
3
r V
TP
(r
0
+r/2, r
0
r/2) e
i(k
+k
)r
2
=
k
+k
,k
+k
1
1
_
V
d
3
r v
TP
(r) e
i(k
+k
)r
2
,
(16.87)
where
v
TP
(r) = V
TP
(r
0
+r/2, r
0
r/2) . (16.88)
Thus the only allowed processes for this case are those for which the total
momentum is conserved, i.e. k
+k
= k
+k
. Using the notation

q = k
= k
, (16.89)
one can express V as
V =
1
2
,k
,q
v
q
a
k
a
k
a
k
q
a
k
+q
, (16.90)
where
v
q
=
1
1
_
V
d
3
r v
TP
(r) e
iqr
2
. (16.91)
16.8 Spin
In addition to spatial (orbital) degrees of freedom, the particles may have
spin. We demonstrate below the inclusion of spin for the case of momentum
representation. The basis for single-particle states is taken to be [k
,
16.9. The Electron Gas
where the quantum number indicates the spin state. The single-particle
orthonormality condition reads
k
[k
=
k
,k

,
. (16.92)
The commutation (for Bosons) and anti-commutation (for Fermions) rela-
tions [see Eqs. (16.6), (16.7), (16.8) and (16.9)] become
[a
k
,
, a
k
,
]
=
_
a
,
, a
= 0 , (16.93)
_
a
k
,
, a
=
k
,k

,
, (16.94)
For the example above, the Hamiltonian becomes
H =

2
2m
k
2
a
,
a
k
,
+

k
,k
U
k
k
a
,
a
k
+
1
2
,k
,q,
v
q
a
a
k
q,
a
k
+q,
.
(16.95)
16.9 The Electron Gas
Consider a free (i.e. noninteracting) gas of N 1 electrons occupying volume
1. The Hamiltonian is given by [see Eq. (16.95)]
H =

2
2m
k
2
a
,
a
k
,
. (16.96)
In the momentum representation the single particle state [k
, has a wave-
function given by [see Eq. (16.77)]
r
[k
, =
1
1
e
ik
, (16.97)
and thus the quantized eld operator
(r
) is given by [see Eq. (16.39)]
(r
) =
1
e
ik
a
k
,
. (16.98)
The single particle state [k
, has energy given by
k
=

2
k
2
2m
, (16.99)
where m is the electron mass [see Eq. (16.96)].
The allowed values of k
are determined by boundary conditions. Consider

for simplicity the case where the gas is conned in a cube (having edge length
of 1
1/3
). Imposing periodic boundary conditions on the wavefunction of the
single particle states [k
, leads to the requirement that

k
=
2
1
1/3
(n
x
, n
y
, n
z
) , (16.100)
where n
x
, n
y
and n
z
are all integers. Thus, the density of states per spin in
k
space is 1/8
3
.
In the ground state [
0
all single particle states for which [k
[ k
F
are
singly occupied, whereas all single particle states for which [k
[ > k
F
remain
empty, i.e.
[
0
=

|k
|kF,
,
[0 . (16.101)
The Fermi wave vector is chosen such that the number of single particle states
for which [k
[ k
F
is N. Since the density of states per spin in k
space is
1/8
3
one nds that
2
1
8
3
4
3
k
3
F
= N , (16.102)
thus
k
3
F
=
3
2
N
1
. (16.103)
The Fermi energy
F
is the corresponding energy
F
=

2
k
2
F
2m
. (16.104)
The density of states D() per spin and per unit volume is given by
D() =
1
1
(
k
) . (16.105)
where
k
is given by Eq. (16.99). By replacing the sum by an integral one
nds that
D() =
1
1
_

2
k
2
2m
_
=
1
1
1
8
3
4
_
0
dk
k
2
_

2
k
2
2m
_
=
1
4
2
_
2m
2
_
3/2

_
0
d
)
=
m
2
2
2m .
(16.106)
16.10. Problems
The ground state energy is given by
E
0
= 1
F
_
0
d
D(
=
m
2m1
5/2
F
5
2
3
. (16.107)
16.10 Problems
1. Find the many-particle interaction operator V for the case where the
two-particle potential is a constant V
TP
(r
1
, r
2
) = V
0
.
2. The same for the Coulomb interaction
V
TP
(r
1
, r
2
) =
e
2
[r
1
r
2
[
. (16.108)
3. Show that
d
dt
+J = 0 , (16.109)
where (r
) =
(r
) (r
) is the the number density operator [see Eq.

(16.45)] and where the current density operator J is given by
J(r
) =

2im
_
(r
) (r
)
_
(r
)
_
(r
. (16.110)
4. Consider two identical Bosons having mass m in a one dimensional po-
tential U (x) well given by
U (x) =
_
0 if 0 x L
else
. (16.111)
The particles interact with each other via a two-particle interaction given
by V
TP
= V
0
L (x
1
x
2
), where V
0
is a constant. Calculate the ground
state energy to lowest nonvanishing order in V
0
.
5. By denition, an ideal gas is an ensemble of non-interacting identical
particles. The set of single particle eigenenergies is denoted by
i
. Cal-
culate the average energy H and the average number of particles N in
thermal equilibrium as a function of the temperature T and the chemical
potential for the case of
a) Fermions.
b) Bosons.
6. Bogoliubov transformation - Consider the transformation
b
k
= u
k
a
k
+v
k
a
k
, (16.112)
where a
k
(a
k
) is the annihilation (creation) operator corresponding to
the single particle state [k ([k), and where u
k
and v
k
are real coe-
cients. The state [V
b
is dened by the condition
b
k
[V
b
= 0 . (16.113)
a) For the case of Fermions, under what conditions the operators b
k
and b
k
can be considered as annihilation and creation operators?
Evaluate the expectation value V
b
[ a
k
a
k
[V
b
.
b) The same for Bosons.
7. Find the eigenenergies of the Hamiltonian
H =
_
a
k
a
k
+
_
a
k
+a
__
, (16.114)
where a
k
and a
k
are Boson annihilation and creation operators corre-
sponding to the single particle state [k
, and where
k
and are real
coecients.
8. Find eigenvectors and eigenvalues of the quantized eld operator (r)
for the case of Bosons. Evaluate the expectation values with respect to
the number operator N and with respect with the Hamiltonian of the
many body system (with one-particle and two-particle interactions).
9. Consider a free (i.e. noninteracting) gas of N 1 electrons occupying
volume 1. Calculate the correlation function
C
(r
) =
0
[
(r
(r
) [
0
, (16.115)
where [
0
is the ground state of the N electrons gas,
(r) is the quan-

tized eld operator and stands for a spin state.
16.11 Solutions
1. In general V is given by Eq. (16.90) where for this case
v
q
= V
0
q,0
, (16.116)
thus
V =
V
0
2
,k
a
k
a
k
=
V
0
2
,k
_
_
a
a
k
, a
k
+a
k
a
a
k
_
.
(16.117)
16.11. Solutions
With the help of Eq. (16.71) one nds that [see also Eqs. (16.6), (16.7),
(16.8) and (16.9)]
_
a
a
k
, a
k
= a
[a
k
, a
k
]
_
a
k
, a
a
k
=
k
,k
a
k
,
(16.118)
[for Fermions the upper sign (anti-commutation) is taken, whereas for
Bosons the lower one is taken (commutation)], thus
V = V
0
N (N 1)
2
, (16.119)
where N is the total number of particles operator. Note that N (N 1) /2
is the number of interacting pairs in the system.
2. For this case the Fourier transform f (q) of the function 1/ [r[ is needed
1
[r[
=
_
d
3
q f (q) e
iqr
. (16.120)
Applying the Laplace operator
2
and using the identity
2
1
[r[
= 4 (r) (16.121)
yield
4 (r) =
_
d
3
q f (q) [q[
2
e
iqr
, (16.122)
thus with the help of the identity
dk e
ikx
= 2 (x) , (16.123)
one nds that
f (q) =
1
2
2
q
2
, (16.124)
where q = [q[, and therefore
1
[r[
=
1
2
2
_
d
3
q
1
q
2
e
iqr
(16.125)
With the help of this result one nds that V is given by [see Eqs. (16.90)
and (16.91)]
V =
1
21
,k
,q
4e
2
q
2
a
k
a
k
a
k
q
a
k
+q
. (16.126)
3. With the help of Eq. (16.69) and its Hermitian conjugate one nds that
d
dt
=
d
(r
)
dt
(r
) +
(r
)
d (r
)
dt
=
1
i

2
2m
_
(r
)
2
(r
, t)
_
(r
, t)
_
(r
,
(16.127)
where the assumptions U
SP
(r
) = U
SP
(r
) and V
TP
(r
, r
) = V
TP
(r
, r
)
have been made, thus
d
dt
+J = 0 . (16.128)
Note the similarity between this result and the continuity equation that
is satised by a single-particle wavefunction [see Eq. (4.70)].
4. For the unperturbed case, i.e. when V
0
= 0, the single-particle wavefunc-
tions of the normalized eigenstates are given by
j
(x) =
_
2
L
sin
jx
L
, (16.129)
where j = 1, 2, , and the corresponding single-particle eigenenergies
are
j
=

2
2
j
2
2mL
2
. (16.130)
For this case the ground state is the many-particle state [GS = [n
1
= 2, n
2
= 0, n
3
= 0, ,
i.e. the state for which both particles are in the j = 1 single-particle state.
In perturbation theory to rst order in V
0
the energy of this state is given
by [see Eq. (9.32)]
E = 2
1
+GS[ V [GS +O
_
V
2
0
_
, (16.131)
where the many-particle interaction operator V is given by Eq. (16.58).
The matrix element GS[ V [GS is given by
GS[ V [GS =
1
2
1, 1[ V
TP
[1, 1 GS[ a
1
a
1
a
1
a
1
[GS
=
1
2
1, 1[ V
TP
[1, 1 GS[ a
1
_
a
1
a
_
a
1
, a
1
__
a
1
[GS
= 1, 1[ V
TP
[1, 1 GS[
N
1
(N
1
1)
2
[GS
= 1, 1[ V
TP
[1, 1 ,
(16.132)
where the two-particle matrix element 1, 1[ V
TP
[1, 1 is given by
16.11. Solutions
1, 1[ V
TP
[1, 1 =
_
L
0
dx
1
_
L
0
dx
2

1
(x
1
)
1
(x
2
) V
TP
(x
1
, x
2
)
1
(x
1
)
1
(x
2
)
= V
0
L
_
L
0
dx
1

4
1
(x
1
)
=
3
2
V
0
,
(16.133)
thus
E =

2
2
mL
2

3
2
V
0
+O
_
V
2
0
_
. (16.134)
5. The grandcanonical partition function [see Eq. (8.334)] is evaluated by
summing over all many-particle states
Z
gc
= Tr
_
e
H+N
_
=

n1,n2,
n
1,
n
2
, , n
i
, [ e
H+N
[n
1,
n
2
, , n
i
, ,
(16.135)
where
H =

i
i
a
i
a
i
, (16.136)
N =

i
a
i
a
i
, (16.137)
and = 1/k
B
T, thus one nds that
Z
gc
=

i
ni
e
ni(i)
. (16.138)
and
log Z
gc
=

i
log
_
ni
e
ni(i)
_
. (16.139)
a) In this case the summation over n
i
includes only two terms n
i
= 0
and n
i
= 1, thus
log Z
gc
=
i
log
_
1 +e
(i)
_
. (16.140)
The average energy is found using Eq. (8.335)
H =
_
log Z
gc
_
log Z
gc
=

i
i
e
(
i
)
1 +e
(i)
,
(16.141)
whereas the average number of particles is found using Eq. (8.338)
N =
log Z
gc
i
e
(i)
1 +e
(
i
)
, (16.142)
In terms of the Fermi-Dirac function f
FD
(), which is given by
f
FD
() =
1
exp[ ( )] + 1
, (16.143)
these results can be rewritten as
H =
i
f
FD
(
i
) , (16.144)
and
N =
i
f
FD
(
i
) . (16.145)
b) In this case the summation over n
i
includes all integers n
i
=
0, 1, 2, , thus
log Z
gc
=
i
log
_
1
1 e
(i)
_
. (16.146)
The average energy is found using Eq. (8.335)
H =
i
f
BE
(
i
) ,
whereas the average number of particles is found using Eq. (8.338)
N =
i
f
BE
(
i
) , (16.147)
where
f
BE
() =
1
exp[ ( )] 1
(16.148)
is the Bose-Einstein function .
6. The operators a
k
and a
k
satisfy [see Eqs. (16.6), (16.7), (16.8) and (16.9)]
[a
k
, a
k
]
=
_
a
k
, a
= 0 , (16.149)
_
a
k
, a
=
k
,k
. (16.150)
Similarly, The operators b
k
and b
k
can be considered as annihilation and
creation operators provided that they satisfy
16.11. Solutions
[b
k
, b
k
]
=
_
b
, b
= 0 , (16.151)
_
b
k
, b
=
k
,k
. (16.152)
Using the denition (16.112) together with Eqs. (16.6) and (16.8) these
conditions become
v
k
u
k
_
a
, a
k
+u
k
v
k
_
a
k
, a
= 0 , (16.153)
v
k
u
k
_
a
k
, a
+u
k
v
k
_
a
, a
k
= 0 , (16.154)
u
k
u
k
_
a
k
, a
+v
k
v
k
_
a
k
, a
k
=
k
,k
. (16.155)
Note that by inverting the transformation between the operators a
k
, a
k
,
a
k
and a
k
and the operators b
k
, b
k
, b
k
and b
k
, which can be expressed
in matrix form as [see Eq. (16.112)]
_
_
_
_
b
k
b
k
b
k
b
k
_
_
_
_
=
_
_
_
_
u
k
0 0 v
k
0 u
k
v
k
0
0 v
k
u
k
0
v
k
0 0 u
k
_
_
_
_
_
_
_
_
a
k
a
k
a
k
a
k
_
_
_
_
, (16.156)
one nds that
_
_
_
_
a
k
a
k
a
k
a
k
_
_
_
_
=
1
u
k
u
k
v
k
v
k
_
_
_
_
u
k
0 0 v
k
0 u
k
v
k
0
0 v
k
u
k
0
v
k
0 0 u
k
_
_
_
_
_
_
_
_
b
k
b
k
b
k
b
k
_
_
_
_
.
(16.157)
This result together with Eq. (16.113) imply that the expectation value
V
b
[ a
k
a
k
[V
b
is given by
V
b
[ a
k
a
k
[V
b
=
_
v
k
u
k
u
k
v
k
v
k
_
2
V
b
[ b
k
b
k
[V
b
, (16.158)
thus for both Bosons and Fermions [see Eq. (16.152)]
V
b
[ a
k
a
k
[V
b
=
_
v
k
u
k
u
k
v
k
v
k
_
2
. (16.159)
a) For the case of Fermions one nds using Eq. (16.9) that the condi-
tions (16.153), (16.154) and (16.155) become (recall that [A, B]
+
=
[B, A]
+
)
(v
k
u
k
+u
k
v
k
)
k
,k
= 0 , (16.160)
(v
k
u
k
+u
k
v
k
)
k
,k
= 0 , (16.161)
(u
k
u
k
+v
k
v
k
)
k
,k
=
k
,k
, (16.162)
thus
v
k
u
k
+u
k
v
k
= 0 , (16.163)
u
2
k
+v
2
k
= 1 . (16.164)
These conditions are guarantied to be satised provided u
k
and v
k
are expressed using a single real parameter
k
as
u
k
= cos
k
, v
k
= sin
k
, (16.165)
u
k
= cos
k
, v
k
= sin
k
. (16.166)
For this case Eq. (16.157) becomes
_
_
_
_
a
k
a
k
a
k
a
k
_
_
_
_
=
_
_
_
_
cos
k
0 0 sin
k
0 cos
k
sin
k
0
0 sin
k
cos
k
0
sin
k
0 0 cos
k
_
_
_
_
_
_
_
_
b
k
b
k
b
k
b
k
_
_
_
_
, (16.167)
and Eq. (16.159) becomes
V
b
[ a
k
a
k
[V
b
= sin
2
k
. (16.168)
b) For the case of Bosons one nds using Eq. (16.7) that the condi-
tions (16.153), (16.154) and (16.155) become (recall that [A, B]
=
[B, A]
)
v
k
u
k
+u
k
v
k
= 0 , (16.169)
u
2
k
v
2
k
= 1 . (16.170)
These conditions are guarantied to be satised provided that u
k
and
v
k
are expressed using a single real parameter
k
as
u
k
= cosh
k
, v
k
= sinh
k
, (16.171)
u
k
= cosh
k
, v
k
= sinh
k
. (16.172)
For this case Eq. (16.159) thus becomes
V
b
[ a
k
a
k
[V
b
= sinh
2
k
. (16.173)
7. Consider the unitary transformation [see for comparison Eq. (9.49)]
H
k
= e
L
k
H
k
e
L
k
, (16.174)
where
H
k
=
k
_
a
a
k
+
_
a
k
+a
__
, (16.175)
and where
L
k
=
_
a
k
a
_
. (16.176)
e
L
Ae
L
= A+[L, A] +
1
2!
[L, [L, A]] +
1
3!
[L, [L, [L, A]]] + , (16.177)
16.11. Solutions
and the identities
_
a
k
, a
a
k
_
= a
k
, (16.178)
_
a
k
, a
k
a
k
_
= a
k
, (16.179)
_
a
k
a
k
, a
k
+a
_
= 2 , (16.180)
one nds that
H
k
= H
k

k
_
a
k
+a
_
2
k

2
+
k

2
=
k
_
a
a
k

2
_
.
(16.181)
Thus, the unitary transformation
H = U
HU , (16.182)
where
U = exp
_
L
k
_
, (16.183)
which yields
H =
_
a
a
k

2
_
, (16.184)
can be employed for diagonalization of H. Let [ n be a number state,
which satisfy
a
a
k
[ n = n
k
[ n , (16.185)
where n
k
is the number of particles in single-particle state [k
. The
following holds
HU [ n = UU
HU [ n
= U

H[ n
=

k
_
n
k

2
_
U [ n ,
(16.186)
thus the eigenvectors of H are the vectors U [ n and the corresponding
eigenenergies are given by
E
n
=

k
_
n
k

2
_
. (16.187)
8. Consider the state [(r
), which is dened by
[(r
) = D
(r
)
[0 , (16.188)
where (r
) ( and where the operator D

(r
)
is given by [see for com-
parison Eq. (5.36)]
D
(r
)
= e
_
dr
((r
(r
(r
)(r
))
. (16.189)
For general operators A and B the following holds [see Eq. (2.175)]
e
A+B
= e
A
e
B
e
1
2
[A,B]
= e
B
e
A
e
1
2
[A,B]
, (16.190)
provided that
[A, [A, B]] = [B, [A, B]] = 0 . (16.191)
Moreover, with the help of Eq. (16.41) one nds that
__
dr
(r
(r
) ,
_
dr
(r
) (r
)
_
=
_
dr
_
dr
(r
(r
)
_
(r
) ,
(r
=
_
dr
[(r
)[
2
,
(16.192)
thus [see for comparison Eq. (5.39)]
D
(r
)
= e
_
dr
(r
(r
)
e
_
dr
(r
)(r
)
e
1
2
_
dr
[(r
)[
2
= e
_
dr
(r
)(r
)
e
_
dr
(r
(r
)
e
1
2
_
dr
[(r
)[
2
.
(16.193)
Using the last result (16.193) it is easy to show that D
(r
)
is unitary
D
(r
)
D
(r
)
= D
(r
)
D
(r
)
= 1 , (16.194)
and thus [(r
) is normalized. With the help of Eq. (16.193) together

with the relation (r) [0 = 0 one nds that
[(r
) = e
1
2
_
dr
[(r
)[
2
e
_
dr
(r
(r
)
[0 . (16.195)
To show that [(r
) is an eigenvector of the quantized eld operator (r)

the commutation relation
_
exp
__
dr
(r
(r
)
_
, (r)
is evaluated
below. For general operators A and B and for a smooth function f (A)
the following holds
16.11. Solutions
[f (A) , B] =
df
dA
[A, B] , (16.196)
provided that [[A, B] , A] = 0 [see Eq. (2.174)]. Using this general result
[with f (A) = e
A
, A =
_
dr
(r
(r
) and B = (r)] together with

Eq. (16.41) yields
_
e
_
dr
(r
(r
)
, (r)
_
= e
_
dr
(r
(r
)
__
dr
(r
(r
) , (r)
_
= e
_
dr
(r
(r
)
_
dr
(r
) (r r
)
= e
_
dr
(r
(r
)
(r) ,
(16.197)
The last result together with the relation (r) [0 = 0 can be used to
show that the state [(r
) is an eigenvector of (r) with eigenvalue (r)

(r) [(r
)
= (r) e
1
2
_
dr
[(r
)[
2
e
_
dr
(r
(r
)
[0
= e
1
2
_
dr
[(r
)[
2
_
e
_
dr
(r
(r
)
(r) +
_
(r) , e
_
dr
(r
(r
)
__
[0
= (r) e
1
2
_
dr
[(r
)[
2
e
_
dr
(r
(r
)
[0 ,
(16.198)
that is
(r) [(r
) = (r) [(r
) . (16.199)
The expectation value with respect to the number operator N [see Eqs.
(16.44) and (16.45)] is given by
(r
)[ N [(r
) =
_
d
3
r
(r
)[
(r
) (r
) [(r
)
=
_
d
3
r
[(r)[
2
,
whereas the expectation value with respect to the Hamiltonian H [see
Eq. (16.68)] is given by
(r
)[ H[(r
) =

2
2m
_
d
3
r
(r) (r)
+
_
d
3
r
U
SP
(r
) [(r)[
2
+
1
2
_
d
3
r
_
d
3
r
V
TP
(r
, r
) [(r
)[
2
[(r
)[
2
.
(16.200)
9. The correlation function C
(r
) is given by
C
(r
) =
1
1
,k
e
i(k
0
[ a
,
a
k
,
[
0
, (16.201)
where [
0
is the ground state of the free electron gas [see Eq. (16.101)],
thus
C
(r
) =
1
1
|k
|k
F
e
ik
(r
)
. (16.202)
For N 1 the summation can be approximately substituted by inte-
gration over the Fermi sphere having radius k
F
[see Eq. (16.103)]. In
spherical coordinates in which the z axis is taken to be in the direction
of the vector r
one has
C
(r
) =
1
8
3
2
1
_
kF
0
dk
k
2
_
1
1
d(cos ) e
ik
cos [r
[
, (16.203)
thus
C
(r
) =
1
2
2
sin(k
F
[r
[) k
F
[r
[ cos (k
F
[r
[)
[r
[
3
.
(16.204)
With the help of Eq. (16.103) the result can be expressed as
C
(r
) =
3N
21
sin(k
F
[r
[) k
F
[r
[ cos (k
F
[r
[)
(k
F
[r
[)
3
.
(16.205)
17. Superconductivity
In this chapter two models are discussed, the Londons model, in which a
macroscopic wavefunction is introduced to describe the state of a supercon-
ductor, and the model by Bardeen, Cooper and Schrieer (BCS), which pro-
vides an insight on the underlying microscopic mechanisms that are respon-
sible for superconductivity.
17.1 Macroscopic Wavefunction
In this chapter the Londons equations are derived from the assumption that
the state of a superconductor can be describe using a macroscopic wavefunc-
tion.
17.1.1 Single Particle in Electromagnetic Field
Consider a single particle having charge q and mass m in electromagnetic
eld characterized by the scalar potential and the vector potential A. The
electric eld E and the magnetic eld B are given by (in Gaussian units) [see
Eqs. (1.41) and (1.42)]
E =
1
c
A
t
, (17.1)
and
B = A , (17.2)
where c = 2.99 10
8
ms
1
is the speed of light in vacuum. Let r = (x, y, z)
be the position vector of the particle in Cartesian coordinates. The variable
vector canonically conjugate to the position vector r is given by [see Eq.
(1.61)]
p = m r+
q
c
A . (17.3)
The classical equation of motion is given by [see Eq. (1.60)]
mr = q
_
E+
1
c
r B
_
. (17.4)
Chapter 17. Superconductivity
The Schrdinger Equation. The Hamiltonian of the system is given by
[see Eq. (1.62)]
H =
_
p
q
c
A
_
2
2m
+q . (17.5)
The Schrdinger equation for the wavefunction (r
, t
) is given by [see Eq.

(4.175)]
i
d
dt
=
1
2m
_
i
q
c
A
_
2
+q . (17.6)
The continuity Equation. The continuity equation expresses the proba-
bility conservation law [see Eq. (4.70)]
d
dt
+J = 0 , (17.7)
where
=
(17.8)
is the probability density and
J =

m
Im(
)
q
mc
A (17.9)
is the current density [see Eq. (4.181)]. For a wavefunction having the form
=
1/2
e
i
, (17.10)
where is real, one has [see Eq. (6.340)]
J =

m
_
q
c
A
_
. (17.11)
Gauge Invariance. Consider the following gauge transformation [see Eqs.
(12.49) and (12.50)]
A

A = A+ , (17.12)
= , (17.13)
where = (r) is an arbitrary smooth and continuous function of r, which
is assumed to be time independent. This transformation leaves E and B
unchanged [see Eqs. (12.1) and (12.2)], however, the wavefunction is trans-
formed according to the following rule. Given that the wavefunction (r
, t
)
solves the Schrdinger equation with vector A and scalar potentials, the
transformed Schrdinger equation with vector

A = A+ and scalar =
potentials is solved by the transformed wavefunction

(r
, t
), which is given
by [see Eq. (12.53)]
(r
, t
) = exp
_
iq(r
)
c
_
(r
, t
) . (17.14)
17.1. Macroscopic Wavefunction
17.1.2 Drude Model
Consider a conductor containing charge carriers having charge q and mass m
in electromagnetic eld. The density of charge carriers (i.e. number per unit
volume) is n. Scattering is taken into account in the Drude model by adding
a damping term to the classical equation of motion (17.4)
m
_
r+
1
tr
r
_
= q
_
E+
1
c
r B
_
, (17.15)
where
tr
is the so-called scattering time. For simplicity the applied magnetic
eld is assumed to vanish, i.e. B = 0. In terms of the current density vector
J, which is related to the the velocity vector v = r by the relation
v =
J
qn
, (17.16)
Eq. (17.15) yields
m
q
2
n
_
J
t
+
1
tr
J
_
= E . (17.17)
Maxwells Equations. The macroscopic Maxwells equations (in Gaussian
units) for the electric eld E, electric displacement D, magnetic induction
B and magnetic eld H in the presence of external charge density
ext
and
external current density J
ext
are given by
H =
4
c
J
ext
+
1
c
D
t
, (17.18)
E =
1
c
B
t
, (17.19)
D = 4
ext
, (17.20)
B = 0 . (17.21)
For an isotropic and linear medium the following relations hold
D = E+4P , (17.22)
D = E , (17.23)
P =
e
E , (17.24)
B = H+ 4M , (17.25)
B = H , (17.26)
M =
m
H , (17.27)
where P is the electric polarization, = 1 + 4
e
is the permittivity (di-
electric constant of the medium),
e
is the electric susceptibility, M is the
magnetization, = 1 + 4
m
is the permeability and
m
is the magnetic
susceptibility.
Dielectric Constant. While the external charge density
ext
is related to
D by the relation D = 4
ext
[see Eq. (17.20)], the induced charge
density
ind
, which is dened as the change in charge density with respect to
the unperturbed case, is related to the electric polarization by the relation
P =
ind
, and the total charge density
ind
+
ext
is related to the electric
eld E by the relation
E = 4 (
ind
+
ext
) . (17.28)
Thus, the dielectric constant of the medium is the ratio between
ext
and
the total charge density
ind
+
ext
[see Eq. (17.23)]
ind
+
ext
=
1
ext
. (17.29)
For the case where the induced charge is carried by electrons having mass
m
e
, charge q
e
, density n
e
and scattering time
tr,e
Eq. (17.17) becomes
m
e
q
2
e
n
e
_
J
e
t
+
1
tr,e
J
e
_
= E , (17.30)
where the electron current density J
e
is related to the induced electron charge
density
ind
=
e
by the continuity equation (17.7)
d
e
dt
+J
e
= 0 . (17.31)
Applying to Eq. (17.30) and using Eqs. (17.31) and (17.28) lead to
d
2
e
dt
2
+
1
tr,e
d
e
dt
=
2
p,e
(
e
+
ext
) , (17.32)
where
p,e
, which is given by
2
p,e
=
4q
2
e
n
e
m
e
, (17.33)
is the so-called electron plasma frequency. Consider the case where
ext
is
taken to oscillate according to
ext
= Re
_
ext,0
e
it
_
, where the angular
frequency is a real constant. Seeking a solution to Eq. (17.32) having the
form
e
= Re
_
e,0
e
it
_
yields
e,0
=

2
p,e
2
+i
1
tr,e

2
p,e
ext,0
, (17.34)
thus the frequency dependent dielectric constant
e
() is given by [see Eq.
(17.29)]
e
() = 1

2
p,e
2
i
tr,e
i
tr,e
1
. (17.35)
Alternatively, in terms of the skin depth
e
, which is given by
e
=
c
p,e
_
2
tr,e
, (17.36)
the dielectric constant
e
() can be expressed as
e
() = 1 +
2ic
2
2
e
2
1
1 i
tr,e
. (17.37)
Complex Conductivity. Consider the case where the applied electric eld
E is taken to be given by E = Re
_
E
0
e
it
_
. Seeking a solution of Eq. (17.30)
having the form J
e
=Re
_
J
e,0
e
it
_
yields
J
e,0
=
e
() E
0
, (17.38)
where
e
() =

e,0
1 i
tr,e
, (17.39)
is the complex conductivity () and where
e,0
=
q
2
e
n
e
tr,e
m
e
. (17.40)
In terms of the complex conductivity
e
() the dielectric constant
e
() can
be expressed as [see Eq. (17.35)]
e
() = 1 +i
4
e
()
. (17.41)
17.1.3 The Macroscopic Quantum Model
The macroscopic quantum model is based on the hypothesis that some prop-
erties of a superconducting media can be described by a single wavefunction
s
(r
, t
). It is assumed that the local density of superconducting charge car-

riers n
s
is related to the wavefunction by
n
s
= [
s
(r
, t
)[
2
. (17.42)
In the presence of in an electromagnetic eld the time evolution of
s
(r
, t
)
is governed by the Schrdinger equation [see Eq. (17.6)]
i
d
s
dt
=
1
2m
s
_
i
q
s
c
A
_
2
s
+q
s
. (17.43)
where m
s
and q
s
are the mass and charge respectively of a superconducting
charge carrier. Furthermore, it is assumed that the current density carried
by a superconductor having a macroscopic wavefunction given by [see Eq.
(17.10)]
s
(r
, t
) =
_
n
s
(r
, t
)e
i(r
,t
)
, (17.44)
is given by
J
s
=
q
s
n
s
(r
, t
)
m
s
_
s
c
A
_
. (17.45)
Note that while J in Eq. (17.11) represents probability current density, J
s
in
a superconductor represents charge current density.
17.1.4 London Equations
London equations can be derived from the macroscopic quantum model by
assuming that the superconducting charge carriers density n
s
is constant.
2nd London Equation. By taking the curl of Eq. (17.45) and employing
Eq. (17.2) one obtains the second London equation, which reads
J
s
=
q
2
s
n
s
m
s
c
B . (17.46)
Taking the curl of Eq. (17.18), assuming = = 1 (valid provided that
frequency is suciently low) and employing Eqs. (17.46), (17.19) and (17.21)
together with the general vector identity
B = ( B)
2
B (17.47)
lead to
2
B =
1
2
L
B+
1
c
2
2
B
t
2
, (17.48)
where
L
=
_
m
s
c
2
4n
s
q
2
s
(17.49)
is the London penetration depth in Gaussian units (
L
=
_
m
s
/
0
n
s
q
2
s
in
SI units). In terms of the superconducting plasma frequency
p,s
, which is
given by
2
p,s
=
4n
s
q
2
s
m
s
, (17.50)
the London penetration depth can be expressed as
L
=
c
p,s
. (17.51)
For time independent B the solution of Eq. (17.48) yields an exponential
decay of B with characteristic decay length given by the London penetration
depth
L
. Thus, except of a region having characteristic width
L
near the
surfaces the magnetic eld inside a superconductor vanishes (even in the
presence of an externally applied magnetic led). This expulsion of a magnetic
eld froma superconductor, which is called the Meissner eect , represents the
perfect diamagnetism of superconductors. As can be seen from Eq. (17.18), in
the absence of time dependent electric eld (i.e. when displacement currents
vanish) the expulsion of a magnetic eld also implies that the supercurrent
density J
s
also vanishes deep inside a superconductor.
1st London Equation. When the superconducting density of charge carri-
ers n
s
is assumed to be a constant Eq. (17.43) becomes [see Eq. (17.44)]
d
dt
=
1
2m
s
_
s
c
A
_
2
+q
s
, (17.52)
or [see Eq. (17.45)]
d
dt
=
m
s
2q
2
s
n
2
s
J
2
s
+q
s
. (17.53)
Taking the time derivative of Eq. (17.45) and employing Eq. (17.1) together
with the last result yield the rst London equation
m
s
q
2
s
n
s
_
J
s
t
+
1
2q
s
n
s
J
2
s
_
= E . (17.54)
Typically in superconductors the electric eld E on the right hand side
of Eq. (17.54) can be neglected in comparison with the term proportional to
J
2
s
on the left hand side of Eq. (17.54). The factor J
2
s
can be estimated by
the relation

J
2
s
2 [J
s
[
2
/l
0
, where l
0
is a length scale that characterizes
the spacial variations of the current density J
s
. Moreover, the ratio [J
s
[ /l
0
can be estimated from the second London equation (17.46)
[J
s
[
l
0

q
2
s
n
s
m
s
c
[B[ . (17.55)
Combining these results allows estimating the term proportional to J
2
s
on
the left hand side of Eq. (17.54)
m
s
q
2
s
n
s
1
2q
s
n
J
2
s
1
c
[v
s
[ [B[ , (17.56)
v
s
=
J
s
q
s
n
s
(17.57)
is the velocity of superconducting charge carriers. In view of the classical
equation of motion (17.4) the above estimate shows that the ratio between
[E[ and the the term proportional to J
2
s
in Eq. (17.54) represents the ratio
between electric and magnetic forces acting on the superconducting charges.
Typically in metals electric forces are strongly suppressed due to screening,
and consequently can be neglected in comparison with magnetic forces. Ne-
glecting the E term in Eq. (17.54) leads to
J
s
t
+
1
2q
s
n
s
J
2
s
= 0 . (17.58)
Homogeneous solutions (i.e. position independent solutions) of the rst
London equation (17.54) satisfy
m
s
q
2
s
n
s
J
s
t
= E . (17.59)
Comparing the above relation with Eq. (17.17) of the Drude model yields
agreement provided that the scattering time
tr
is assumed to be innite.
From this point of view the rst London equation represents the nullication
of resistance in superconductors. Equation (17.59) implies that the supercon-
ducting conductivity
s
() for the homogeneous case is given by [see Eq.
(17.39)]
s
() = i
q
2
s
n
s
m
s
. (17.60)
Flux Quantization. Consider a close curve ( inside a superconductor. In-
tegrating Eq. (17.45), which is given by
J
s
=
q
s
n
s
m
s
_
s
c
A
_
, (17.61)
along the curve yields
_
C
dr J
s
=
q
s
n
s
m
s
_
_
C
dr
q
s
c
_
C
dr A
_
. (17.62)
The assumption that the superconducting wavefunction
s
=
s
e
i
is con-
tinuous implies that
_
dr = 2n, where n is integer. The integral over
A can be calculated using Stokes theorem [see Eqs. (12.2) and (12.47)]
_
C
dr A =
C
, (17.63)
where
C
=
_
ds B is the magnetic ux threaded through the area enclosed
by the closed path (. With these results Eq. (17.62) becomes
_
C
dr J
s
=
hq
s
n
s
m
s
_
n
s
_
, (17.64)
where
s
=
hc
q
s
(17.65)
is the so called superconducting ux quantum (in Gaussian units).
As was shown above, the second London equation implies that the super-
current density J
s
vanishes deep inside a superconductor. Consider a close
curve ( inside a superconductor and assume that the distance between any
point on ( and the nearest surface is much larger than the London penetra-
tion depth
L
. For such a curve the left hand side of Eq. (17.64) vanishes,
and consequently
C
= n
s
, (17.66)
i.e. the magnetic ux is quantized in units of the superconducting ux quan-
tum.
Two Fluid Model. In the limit of vanishing temperature only supercon-
ducting charge carriers are present in a superconductor. However, at nite
temperature also normally conducting charge carriers may be present. Con-
sequently, the total complex conductivity () [see Eq. (17.38)] is taken to
be given by
() =
n
() +
s
() , (17.67)
where the normal conductivity
n
() is assumed to be given by the Drude
formula (17.39)
n
() =
q
2
e
n
n
tr,e
m
e
1
1 i
tr,e
, (17.68)
where n
n
is the density of normal charge carriers, and the super conductivity
s
() is assumed to be given by Eq. (17.60)
s
() = i
q
2
s
n
s
m
s
. (17.69)
The dielectric constant () [see Eq. (17.41)] in the two uid model is
thus given by
() = 1 +i
4
n
()
+i
4
s
()
, (17.70)
or in terms of the skin depth [see Eq. (17.36)] and the London penetration
depth
L
[see Eq. (17.49)]
() = 1 +
2i
(
e
k)
2
1
1 i
tr,e

1
(
L
k)
2
, (17.71)
where k = /c. Note that 1/
2
n
n
[see Eq. (17.36)], whereas 1/
2
L
n
s
[see Eq. (17.49)]. Note also that the ratio between these characteristic length
scales is given by
L
=
_
m
s
m
n
n
n
s
q
2
q
2
s
tr
2
. (17.72)
17.2 The Josephson Eect
A Josephson junction is formed between two superconductors that are weakly
connected to each other. Electrons can ow between the superconductors by
crossing a barrier. In this chapter the Josephson relations are derived from
the model that was discussed in the previous chapter.
17.2.1 The First Josephson Relation
In terms of the superconducting ux quantum
s
[see Eq. (17.65)] Eq. (17.61)
can be rewritten as
J
s
=
q
s
n
s

GI
, (17.73)
where
GI
=
2
s
A . (17.74)
The phase factor
GI
is commonly called the gauge invariant phase.
Consider an integral over J
s
(17.73) along a path going through a Joseph-
son junction from point r
1
on the interface between the rst superconductor
and the barrier to point r
2
on the interface between the second superconduc-
tor and the barrier. The path is assumed to be parallel to the direction of
the current density J
s
. The total current owing through the junction I is
expected to be a function of the phase dierence , which is obtained from
integrating
GI
=
_
r2
r
1
dr
GI
= (r
2
) (r
1
)
2
s
_
r2
r
1
dr A . (17.75)
17.2. The Josephson Eect
Furthermore, by symmetry, I is expected to be an odd function of . In
addition, since the macroscopic wavefunction
s
=
s
e
i
is unaected when
is replaced by +2 [see Eq. (17.44)], I is expected to be a periodic function
of with a period of 2, i.e. I can be expressed as
I =

n=1
I
n
sin(n) , (17.76)
where I
n
are all constants. As will be shown below [see Eq. (17.298)], for the
case where the superconductors are weakly coupled all terms except for the
rst one can be neglected. For that case the general current -phase relation
(17.76) becomes the so-called rst Josephson relation, which is given by
I = I
c
sin , (17.77)
where I
c
is the so-called critical current of the junction.
17.2.2 The Second Josephson Relation
The time derivative of (17.75), which is denoted by

, can be expressed
with the help of Eq. (17.53) as
=
1
_
m
s
2q
2
s
n
2
s
_
J
2
s
(r
2
) J
2
s
(r
1
)
_
+q
s
((r
2
) (r
1
))
_
s
_
r2
r
1
dr
A
t
.
(17.78)
The assumption J
s
(r
2
) = J
s
(r
1
) leads to
=
2c
s
_
r2
r1
dr
_

1
c
A
t
_
. (17.79)
With the help of Eq. (17.1) this becomes
=
2c
s
V , (17.80)
where V , which is given by
V =
_
r2
r1
dr E , (17.81)
is the voltage across the Josephson junction. As will be shown below, the
elementary superconducting charge carrier is a pair of electrons, i.e. q
s
= 2e,
and consequently Eq. (17.80) becomes
=
2eV
. (17.82)
The above result is the second Josephson relation.
17.2.3 The Energy of a Josephson Junction
Let I (t) and V (t) be the current through and voltage across a Josephson
junction, respectively, at time t. The energy U
J
of the junction can be eval-
uated by integrating the work done by the source
U
J
=
_
t
dt
I (t
) V (t
) . (17.83)
With the help of the rst (17.77) and second (17.82) Josephson relation this
becomes
U
J
=
I
c
2e
_

d
sin
, (17.84)
thus up to a constant U
J
is given by
U
J
= E
J
cos , (17.85)
where
E
J
=
I
c
2e
=

s
I
c
2c
. (17.86)
The energy U
J
(17.85) can be expressed as [see Eq. (17.298)]
U
J
= E
J
_
1
_
I
I
c
_
2
. (17.87)
To second order in I this becomes [see Eq. (17.299)]
U
J
= E
J
+
L
J
I
2
2
+O
_
I
4
_
, (17.88)
where
L
J
=

s
2cI
c
(17.89)
is the so-called Josephson inductance. Note, however, that an inductor-like
behavior of a Josephson junction is expected only when I I
c
.
17.3 RF SQUID
A radio frequency (RF) superconducting quantuminterference device (SQUID)
is made of a superconducting loop interrupted by a Josephson junction (see
Fig. 17.1). Consider a close curve ( going around the loop. The requirement
that the phase of the macroscopic wavefunction is continues reads
17.3. RF SQUID
2n =
_
C
dr , (17.90)
where n is integer. The section of the close curve ( inside the superconductor
is denoted by (
and the integral through the junction is denoted as an

integral from point r
1
to point r
2
. With the help of Eq. (17.61) the above
condition becomes
2n =
m
s
q
s
n
_
r
2
r1
dr J
s
+
m
s
q
s
n
_
C
dr J
s
+
2
s
_
C
dr A . (17.91)
Consider the case where the curve is chosen such that the supercurrent density
J
s
vanishes everywhere on the curve (
(i.e. inside the superconductor the

distance between any point on (
and the nearest surface is much larger than

the London penetration depth
L
). For this case Eq. (17.91) becomes
2n = +
2
s
, (17.92)
where
=
m
s
q
s
n
_
r2
r1
dr J
s
=
_
r2
r1
dr
GI
(17.93)
is the gauge invariant phase dierence across the junction [see Eqs. (17.73)
and (17.75)] and where
=
_
C
dr A (17.94)
is the magnetic ux threaded through the area enclosed by the closed path
( [see Eq. (17.63)].
The junctions critical current is labeled by I
c
. It is assumed that the
junction has capacitance, which is denoted by C
J
. Consider the case where a
magnetic ux that is denoted by
e
is externally applied by a magnetic eld,
which is applied perpendicularly to the plane of the loop. The total magnetic
ux threading the loop is given by
=
e
+I
s
, (17.95)
where I
s
is the circulating current owing in the loop and is the self induc-
tance of the loop.
17.3.1 Lagrangian
The Lagrangian of the system [see Eq. (1.16)] can be expressed as a function
of the dimensionless ux coordinate , which is dened by
Fig. 17.1. RF SQUID is made of a superconducting loop interrupted by a Joseph-
son junction.
=
2
s
, (17.96)
and its time derivative

. According to Faradays law of induction the voltage
across the capacitor (in Gaussian units) is
V =
c
, (17.97)
and therefore the kinetic energy of the system T is the capacitance energy
T =
C
J
2
2c
2
=
C
J
2
s

2
8
2
c
2
. (17.98)
The potential energy U has two contributions, the inductive energy (in
Gaussian units)
I
2
s
2c
=
(
e
)
2
2c
=

2
s
(
e
)
2
8
2
c
, (17.99)
where
e
=
2
e
s
(17.100)
17.3. RF SQUID
is the normalized external ux, and the Josephson energy U
J
[see Eqs. (17.85)
and (17.92)]
U
J
=
s
I
c
2c
cos . (17.101)
Thus the Lagrangian / = T U is given by
/ =
C
J
2
s

2
8
2
c
2

2
s
(
e
)
2
8
2
c
+

s
I
c
2c
cos , (17.102)
or in a dimensionless form by
/ = E
0
_

L
J
2
p
u(;
e
)
_
, (17.103)
where the energy constant E
0
is given by
E
0
=

2
s
8
2
c
, (17.104)
the junctions plasma frequency
p
is given by
p
=
_
c
L
J
C
J
=
_
2ecI
c
C
J
, (17.105)
where L
J
=
s
/2cI
c
is the Josephson inductance [see Eq. (17.89)], the di-
mensionless potential u(;
e
) is given by
u(;
e
) = (
e
)
2
2
L
cos , (17.106)
and the dimensionless parameter
L
is given by
L
=
2I
c
s
. (17.107)
The resulting Euler - Lagrange equation of motion (1.8) is given by
d
dt
_
/
_
=
/
, (17.108)
thus
L
J
2
p
+
e
+
L
sin = 0 . (17.109)
With the help of Eqs. (17.92), (17.95) and (17.97) the equation of motion can
be rewritten as
I
s
= I
c
sin +C
J

V . (17.110)
The above equation states that the circulating current I
s
equals the sum of
the current I
c
sin through the Josephson junction and the current C
J

V
through the capacitor.
17.3.2 Flux Quantum Bit
Consider the case where the externally applied ux
e
is chosen to be close
to a half integer value in units of the superconducting ux quantum
s
. To
study this case the normalized externally applied ux
e
(17.100) is taken to
be given by
e
= +
e,r
, (17.111)
where
e,r
1. Using the notation
= +
r
, (17.112)
the dimensionless potential u (17.106) becomes
u = (
r
e,r
)
2
+ 2
L
cos
r
. (17.113)
Consider that case where
e,r
= 0 (i.e. when
e
= ). As can be seen
from Eq. (17.95), for this case the relative coordinate
r
is proportional to
the circulating current I
s
r
=
2
s
I
s
. (17.114)
Moreover, when
e,r
= 0 one nds to second order in
r
that the potential u
is given by
u = 2
L
+ (1
L
)
2
r
+O
_
4
r
_
. (17.115)
Thus if
L
> 1 the point
r
= 0 becomes a local maxima point of u. The
corresponding potential barrier centered at
r
= 0 (i.e. at = ) separates
two symmetric potential wells on the right and on the left. At suciently low
temperatures only the two lowest energy levels are expected to be occupied.
In this limit the Hamiltonian of the system can be expressed in the basis of
the states [ and [, that represent localized states in the left and right
well, respectively, having opposite circulating currents. Such a device can be
used as an articial two-level system, i.e. as a quantum bit (qubit in short).
17.3.3 Qubit Readout
Readout of the state of an RF SQUID can be performed by inductively cou-
pling the superconducting loop to an LC resonator (see Fig. 17.2), which is
made of an inductor and a capacitor in parallel having inductance L and
capacitance C respectively. The mutual inductance between the RF SQUID
and the resonator is denoted by M. Detection is performed by injecting a
monochromatic input current I
in
into the LC resonator at a frequency close
17.3. RF SQUID
Fig. 17.2. The LC resonator that is coupled to the RF SQUID allows readout.
to the resonance frequency and measuring the output voltage V
out
(see Fig.
17.2).
The total magnetic ux threading the SQUID loop is given by [compare
with Eq. (17.95)]
=
e
+
i
, (17.116)
where the term
i
represents the ux generated by both, the circulating
current in the RF SQUID I
s
and by the current in the inductor of the LC
resonator I
L
i
= I
s
+MI
L
, (17.117)
where is the self inductance of the loop. Similarly, the magnetic ux in
the inductor of the LC resonator is given by
= I
L
L +MI
s
. (17.118)
Eqs. (17.117) and (17.118) can be rewritten in a matrix form
_
_
=
_
M
M L
__
I
s
I
L
_
. (17.119)
Inverting the above relation allows expressing the currents I
s
and I
L
in terms
of
i
=
e
and
I
s
=

i
(1 K
2
)

M
L(1 K
2
)
, (17.120)
I
L
=

L(1 K
2
)

M
i
L(1 K
2
)
, (17.121)
where the dimensionless constant K is given by
K =
M
L
. (17.122)
Lagrangian. The Lagrangian of the system / = T U [see Eq. (1.16)] is
expressed below as a function of the coordinates and and their time
derivatives

and . The contributions to the total kinetic energy T are the
capacitance energy of the Josephson junction that is given by Eq. (17.98) and
the capacitance energy of the capacitor in the LC resonator, thus T is given
by
T =
C
J
2
s

2
8
2
c
2
+
C
2
2c
2
. (17.123)
The inductive energy U
I
stored in the RF SQUID loop and the lumped in-
ductor L is calculated using Eqs. (17.120) and (17.121)
U
I
=
1
2c
_
I
s
I
L
_
_
M
M L
__
I
s
I
L
_
=
1
2c (1 K
2
)
_
i

_
_
1

M
L
M
L
1
L
__
_
=

2
i

2
i
M
L
+

2
L
2c (1 K
2
)
=

2
2cL
+
_
M
L
_
2
2c(1 K
2
)
=
C
2
e
2
2c
+

2
s
_

e
2M
s
L
_
2
8
2
c(1 K
2
)
,
(17.124)
where
e
=
1
LC
(17.125)
is the LC angular resonance frequency. The total potential energy U is given
by
U = U
I
I
in
c

s
I
c
2c
cos , (17.126)
where the term I
in
/c is the potential energy of the current source and
(
s
I
c
/2c) cos is the Josephson energy [see Eq. (17.101)].
With the help of the above relations one nds that the Lagrangian of the
system / = T U can be expressed as
17.3. RF SQUID
/ = /
0
+/
1
, (17.127)
where /
0
, which is given by
/
0
=
C
2
2c
2

C
2
e
2
2c
+
I
in
c
, (17.128)
is the Lagrangian of the driven LC resonator. The Lagrangian of the super-
conducting loop /
1
is given by [see Eqs. (17.104), (17.105) and (17.107)]
/
1
=
C
J
2
s

2
8
2
c
2

2
s
_

e
2M
s
L
_
2
8
2
c(1 K
2
)
+

s
I
c
2c
cos
= E
0
_

L
J
2
p
u
K
(;
e,e
)
_
,
(17.129)
where the dimensionless potential u
K
(;
e,e
) is given by [compare with Eq.
(17.106)]
u
K
(;
e,e
) =
(
e,e
)
2
1 K
2
2
L
cos , (17.130)
and where the eective external ux
e,e
is given by
e,e
=
e
+
2M
s
L
. (17.131)
Note that /
1
depends on the eective external ux
e,e
, which, in turn,
depends on the coordinate of the LC resonator [see Eq. (17.131)]. This
dependence gives rise to the coupling between the LC resonator and the RF
SQUID.
The Euler - Lagrange equations (1.8) are given by
d
dt
_
/
_
=
/
, (17.132)
d
dt
_
/

_
=
/
, (17.133)
thus
C
J
2c
=
s
(
e
)
2

M
L
(1 K
2
)
I
c
sin , (17.134)
and
C
c
=

M

s
(e)
2
L(1 K
2
)
+I
in
. (17.135)
With the help of Eqs. (17.120) and (17.121) the equations of motion can be
rewritten as
I
s
= I
c
sin
C
J
c
, (17.136)
and
I
in
=
C
c
+I
L
. (17.137)
While Eq. (17.136) expresses the law of current conservation in the SQUID
loop, Eq. (17.137) expresses the same law in the LC resonator.
Hamiltonian. The variables canonically conjugate to and are given by
[see Eq. (1.20)]
Q =
/
=
2E
0
L
J
2
p
. (17.138)
q =
/

=
C
c
2
. (17.139)
The Hamiltonian is given by [see Eq. (1.22)]
H = Q

+q / = H
0
+H
1
, (17.140)
where
H
0
=
c
2
q
2
2C
+
C
2
e
2
2c

I
in
c
, (17.141)
and where
H
1
=
L
J
2
p
Q
2
4E
0
+E
0
u
K
(;
e,e
) . (17.142)
Quantization is achieved by regarding the variables , Q, , q as Her-
mitian operators satisfying the following commutation relations [see Eqs.
(3.6), (3.7) and (3.8)]
[, Q] = [, q] = i , (17.143)
and
[, ] = [, Q] = [q, ] = [q, Q] = 0 . (17.144)
17.3. RF SQUID
Adiabatic Approximation. The dynamics of the superconducting loop is
typically much faster than the one of the LC resonator. This can be exploited
for simplifying the systems equations of motion by employing the adiabatic
approximation (see section 13.6). Similarly to /
1
, the Hamiltonian H
1
of the
superconducting loop depends on the eective external ux
e,e
, which in
turn depends on the coordinate of the LC resonator [see Eq. (17.131)].
As a basis for expanding the general solution we use the eigenvectors of
the following Schrdinger equation
H
1
[n() =
n
() [n() , (17.145)
where is treated here as a parameter (rather than a degree of freedom).
The local eigenvectors are assumed to be orthonormal
m() [n() =
nm
. (17.146)
The eigenenergies
n
() and the associated wavefunctions
n
(, ) are
found by solving the following Schrdinger equation [see Eq. (4.50)]
J
d
2
n
d
2
+u
K
(;
e,e
)
n
=

n
E
0
n
, (17.147)
where
J
=
L
J
p
2E
0
_
2
. (17.148)
The state of the entire system (t) at time t is expanded at any point
using the local basis [n() [see Eq. (13.71)]
(t) =

n
n
(, t) [n() . (17.149)
In the adiabatic approximation the time evolution of the coecients
n
is gov-
erned by the following set of decoupled equations of motion [see Eq. (13.83)]
[H
0
+
n
()]
n
= i
n
. (17.150)
Note that in the present case the geometrical vector potential [given by Eq.
(13.76) for the case m = m] vanishes since the wavefunctions
n
(, ) can
be chosen to be real [see Eq. (13.84)].
Two level approximation. Consider again the case where the externally
applied ux
e
is chosen to be close to a half integer value in units of the
superconducting ux quantum
s
. The potential u
K
(17.130) can be expressed
as
u
K
=
(
r
e,e,r
)
2
1 K
2
+ 2
L
cos
r
, (17.151)
where
e,e,r
and
r
are dened by [see Eqs. (17.111) and (17.112)]
e,e
=
e
+
2M
s
L
= +
e,e,r
, (17.152)
= +
r
. (17.153)
In what follows we focus on the case where [
e,e,r
[ 1 and
L
_
1 K
2
_
>
1. In this case the potential u
K
given by Eq. (17.151) contains two wells sepa-
rated by a barrier near
r
= 0 (see Fig. 17.3). As was discussed above, at low
temperatures only the two lowest energy levels are expected to contribute. In
this limit the Hamiltonian H
1
can be expressed in the basis of the states [
and [, representing localized states in the left and right well respectively
having opposite circulating currents. In this basis, H
1
is represented by the
2 2 matrix
H
1
=
_
e,e,r

e,e,r
_
. (17.154)
The real parameters and (having units of energy) can be determined
by solving numerically the Schrdinger equation (17.147). It is convenient to
express in terms of the parameter I
cc
=

s
I
cc
2c
. (17.155)
The physical meaning of the parameter I
cc
, which has units of current, will
be discussed below.
Using the notation
tan =

e,e,r
=
2c
s
I
cc
e,e,r
, (17.156)
H
1
can be rewritten as
H
1
=
_
_
s
I
cc
e,e,r
2c
_
2
+
2
_
cos sin
sin cos
_
. (17.157)
The eigenvectors and eigenenergies are denoted as
H
1
[ =
[ , (17.158)
where [compare with Eqs. (6.199) and (6.200)]
[+ =
_
cos

2
sin

2
_
; [ =
_
sin

2
cos

2
_
, (17.159)
and where
17.3. RF SQUID
=
_
_
s
I
cc
e,e,r
2c
_
2
+
2
=
_
1 +
1
2
_
s
I
cc
e,e,r
2c
_
2
_
+O
_
4
e,e,r
_
.
(17.160)
Exercise 17.3.1. Show that the expectation value of the circulating current
I
s
when the qubit is in state n is given by
n[ I
s
[n = c
e
. (17.161)
Solution 17.3.1. With the help of Eqs. (17.95), (17.117), (17.120), (17.130)
and (17.142) one nds that
c
H
1
e
= I
s
, (17.162)
thus
n[ I
s
[n = c n[
H
1
e
[n = c
e
. (17.163)
With the help of Eqs. (17.160) and (17.161) one nds that the expectation
value of the circulating current I
s
corresponding to the eigenvectors [+ and
[ is given by
[ I
s
[ = I
cc
s
I
cc
e,eff,r
2c
_
1 +
_
s
Icc
e,eff,r
2c
_
2
. (17.164)
In the limit where
s
I
cc
[
e,e,r
[ /2c (i.e. when
e,e
is far from the
point
e,e
= ) the eigenvectors of H
1
become the states [ and [ [see
Eqs. (17.156) and (17.159)]. Thus, in view of the above result (17.164), it is
evident the the constant I
cc
represents the absolute value of the circulating
current associated with the states [ and [.
In the adiabatic approximation the dynamics of the LC resonator is gov-
erned by the Hamiltonian H
, which is given by
H
= H
0
+
() , (17.165)
when the qubit is the state [ [see Eq. (17.150)]. With the help of Eqs.
(17.141), (17.152) and (17.160) one nds that
H
=
c
2
q
2
2C
+U
()
I
in
c
, (17.166)
Fig. 17.3. Eigenstates of H1. (a)-(c) The rst 3 lowest energy states for the case
e,e,r
= 0. (d) The energy of the two lowest states vs.
e,e,r
.
17.4. BCS Model
where
U
() =
C
2
e
2
2c

_
_
s
I
cc
e,e,r
2c
_
2
+
2
, (17.167)
and where
e,e,r
=
e
+
2M
s
L
. (17.168)
For simplicity, consider the case where
e
= . For that case one nds to
second order in M that [see Eqs. (17.125) and (17.160)]
U
() =
C
2c
2
+O
_
M
4
_
, (17.169)
where
=
e
_
1
M
L
MI
2
cc
c
. (17.170)
Thus the eective angular resonance frequency of the LC resonator
de-
pends on the state being occupied by the qubit. This dependence allows
reading out the qubit state by measuring the resonance frequency of the LC
resonator.
17.4 BCS Model
This chapter briey discusses the BCS microscopic model of superconductiv-
ity.
17.4.1 Phonon Mediated Electron-Electron Interaction
Let (r
) be the electron density in a medium having volume 1. Classically,

The two-particle Coulomb interaction V
TP
(r
1
, r
2
) = e
2
/ [r
1
r
2
[ [see Eq.
(16.108)] gives rise to energy V given by [see Eqs. (16.67) and (16.45) for
comparison with the analogous second-quantization expression]
V =
1
2
_
d
3
r
_
d
3
r
V
TP
(r
, r
) (r
) (r
) . (17.171)
With the help of the Fourier expansion
(r
) =
1
(2)
3/2
1
_
d
3
q
(q
) e
iq
(17.172)
and Eqs. (4.47) and (16.125) one nds that [see Eq. (16.126) for comparison
with the analogous second-quantization expression]
V =
1
21
_
d
3
q
4e
2
q
2
(q) (q) . (17.173)
The eect of induced charges in the medium (i.e. screening) can be taken
into account by dividing by the dielectric constant of the medium [see Eq.
(17.29)]
V =
1
21
_
d
3
q
4e
2
q
2
(q) (q) . (17.174)

The expression for the dielectric constant
e
() given by Eq. (17.35) is
evaluated by calculating the induced electron charge density
e
due to exter-
nal charge density
ext
. However, the induced ion charge density
i
associated
with lattice vibrations was disregarded. To account for the eect of lattice
vibrations on electron-electron interaction both contributions to the total in-
duced charge density
e
+
i
have to be taken into account. With analogy to
Eq. (17.32), the equation of motion for the ion induced charge density
i
is
given by
d
2
i
dt
2
+
1
tr,i
d
i
dt
=
2
p,i
(
e
+
i
+
ext
) , (17.175)
where
p,i
, which is given by
2
p,i
=
4q
2
i
n
i
m
i
, (17.176)
is the ion plasma frequency and where m
i
, q
i
and n
i
are the ionic mass, charge
and density, respectively. Note that
p,i

p,e
since the ion mass m
i
is much
larger than electron mass m
e
.
The electron density n
e
is assumed to be a constant provided that the
media is homogeneous. It can be evaluated by summing the Fermi-Dirac
function f
FD
(
i
) over all electronic states having energies
i
[see Eq. (16.145)]
n
e
() =
1
1
i
f
FD
(
i
) =
1
1
i
1
exp[ (
i
)] + 1
, (17.177)
where 1 is the volume,
1
= k
B
T is the thermal energy and where is
the chemical potential. The external charge density
ext
together with the
induced charge density
e
+
i
give rise to a spatially varying scalar potential
(r), which is related to the total charge density by the Poissons equation
2
= 4 (
e
+
i
+
ext
) . (17.178)
In the Thomas-Fermi approximation, which is valid provided that (r) is a
slowly varying function of position on the length scale of electron wavelength,
the local value of electron density is taken to be given by n
e
( q
e
(r)),
17.4. BCS Model
and consequently the electron induced charge density
e
(r) is taken to be
given by
e
(r) = q
e
[n
e
( q
e
(r)) n
e
()] . (17.179)
For suciently small this becomes
e
(r) = q
2
e
n
e
(r) . (17.180)
When the thermal energy k
B
T is much smaller than the Fermi energy
F
the
factor n
e
/ is approximately the density of states at the Fermi energy
F
,
which is given by [see Eq. (16.106)]
n
e

m
2
e
v
F
3
, (17.181)
where v
F
is the so-called Fermi velocity (which is dened by the relation
k
/k
= v
F
, where k
is the wave number and where the derivative is

taken at
k
=
F
). For this case Eq. (17.180) becomes
e
(r) = q
2
e
m
2
e
v
F
3
(r) . (17.182)
Combing Eqs. (17.178) and (17.182) yields the following relation between the
electron induced charge density
e
and the total charge density
e
+
i
+
ext
e
= k
2
TF
(
e
+
i
+
ext
) , (17.183)
where
k
2
TF
=
4q
2
e
m
2
e
v
F
3
. (17.184)
ext
is taken to oscillate in time and space accord-
ing to Re
_
ext,0
e
i(qrt)
_
, where both the real angular frequency and the
real vector q are constants. In steady state the induced charges oscillate ac-
cording to
e
= Re
_
e,0
e
i(qrt)
_
and
i
= Re
_
i,0
e
i(qrt)
_
. Substituting
into Eqs. (17.175) and (17.183) yields
i,0
tr,i
i,0
=
2
p,i
_
e,0
+
i,0
+
ext,0
_
, (17.185)
and
q
2
e,0
= k
2
TF
_
e,0
+
i,0
+
ext,0
_
. (17.186)
e,0
=
k
2
TF
2
_
1+
i
tr,i
_
2
p,i

ext,0
(q
2
+k
2
TF
)
_
2
_
1+
i
tr,i
_
2
p,i
1
_
+k
2
TF
, (17.187)
and
i,0
=
q
2
ext,0
(q
2
+k
2
TF
)
_
2
_
1+
i
tr,i
_
2
p,i
1
_
+k
2
TF
, (17.188)
thus the dielectric constant (q, ) is given by [see Eq. (17.29)]
1
(q, )
=

e,0
+
i,0
+
ext,0
ext,0
=
q
2
2
_
1+
i
tr,i
_
2
p,i
(q
2
+k
2
TF
)
_
2
_
1+
i
tr,i
_
2
p,i
1
_
+k
2
TF
.
(17.189)
For the case where
tr,i
1 this becomes
1
(q, )
=
q
2
q
2
+k
2
TF
2
p,i
, (17.190)
where
2
p,i
=
2
p,i
q
2
q
2
+k
2
TF
. (17.191)
The expression for the Coulomb energy (17.174) together with the result
for the dielectric constant (q, ) (17.190) lead to the eective interaction
coecient for a pair of electrons having wave vectors k
and k
and energies
k
and
k
respectively
v
k
,k
=
4e
2
q
2
(q, )
=
4e
2
q
2
+k
2
TF
2
p,i
, (17.192)
where q = k
and where = (
k

k
) /.
The fact that
1
(q, ) becomes negative when <
p,i
indicates that
the eective (i.e. phonon mediated) electron-electron interaction becomes at-
tractive in the limit of low frequencies. The characteristic energy interval
p,i
in which the interaction becomes attractive is of the order of the so-
called Debye energy
D
, which represents the largest energy of an acoustic
phonons in the lattice.
17.4. BCS Model
17.4.2 The Hamiltonian
In the BCS model the Hamiltonian of electrons in a superconducting metal
H =

k
(
k

F
)
_
a
,
a
k
,
+a
,
a
k
,
_
g
1
FD<
k
,
k
<F+D
,k
,
a
,
a
k
,
a
k
,
.
(17.193)
where labels spin up state, labels spin down state,
k
is the energy of
both single particle states [k
, and [k
, and
F
is the Fermi energy [see
Eq. (16.104)]. The coupling constant g > 0 gives rise for an eective electron-
electron attracting interaction. The interaction is assume to couple pairs of
electrons whose energies are inside an energy interval of width 2
D
around
the Fermi energy
F
.
As can be seen from the comparison with the more general many-particle
interaction operator V given by Eq. (16.95), the BCS Hamiltonian contains
only interaction terms that represents annihilation (the factor a
k
,
a
k
,
)
and creation (the factor a
,
a
,
) of electrons pairs having zero total an-
gular momentum. Moreover, the summation is restricted only to the energy
interval of width 2
D
in which attractive interaction is expected, and the eec-
tive interaction coecients are all assumed to be identical [see for comparison
Eq. (17.192)].
17.4.3 Bogoliubov Transformation
As will be seen below, the Hamiltonian can be approximately diagonalized
by employing the Bogoliubov transformation [see Eqs. (16.112), (16.151) and
(16.152)]
_
_
_
_
b
k,
b
k,
b
k,
b
k,
_
_
_
_
=
_
_
_
_
e
i
k
cos
k
0 0 e
i
k
sin
k
0 e
i
k
cos
k
e
i
k
sin
k
0
0 e
i
k
sin
k
e
i
k
cos
k
0
e
i
k
sin
k
0 0 e
i
k
cos
k
_
_
_
_
_
_
_
_
a
k,
a
k,
a
k,
a
k,
_
_
_
_
,
(17.194)
where both
k
and
k
are real. The inverse transformation is given by
_
_
_
_
a
k,
a
k,
a
k,
a
k,
_
_
_
_
=
_
_
_
_
e
i
k
cos
k
0 0 e
i
k
sin
k
0 e
i
k
cos
k
e
i
k
sin
k
0
0 e
i
k
sin
k
e
i
k
cos
k
0
e
i
k
sin
k
0 0 e
i
k
cos
k
_
_
_
_
_
_
_
_
b
k,
b
k,
b
k,
b
k,
_
_
_
_
.
(17.195)
The following holds
a
k,
a
k,
= sin
2
k
(1 N
k,
) + cos
2
k
N
k,
+
sin(2
k
)
2
_
b
k,
b
k,
+b
k,
b
k,
_
,
(17.196)
and
a
k,
a
k,
= sin
2
k
(1 N
k,
) + cos
2
k
N
k,
+
sin(2
k
)
2
_
b
k,
b
k,
+b
k,
b
k,
_
,
(17.197)
where
N
k,
= b
k,
b
k,
(17.198)
is number operator (with respect to the b
k,
and b
k,
operators ).
Using the above results together with Eqs. (16.151) and (16.152) one nds
that the Hamiltonian can be rewritten as
H = 2
(
k

F
) sin
2
(
k

F
) cos (2
k
) (N
k
,
+N
k
,
)
+
(
k

F
) sin(2
k
)
_
b
k
,
b
k
,
+b
,
b
,
_
g
1
FD<
k
,
k
<F+D
,k
k
B
k
,
(17.199)
where
B
k
= a
k
,
a
k
,
=
e
2i
k
sin(2
k
)
2
(N
k
,
+N
k
,
1)
+e
2i
k
_
sin
2
k
b
,
b
,
cos
2
k
b
k
,
b
k
,
_
.
(17.200)
Formally, the operator B
k
can be written as
B
k
= B
k
+B
k
, (17.201)
where B
k
is the expectation value of B
k
in thermal equilibrium and
where
17.4. BCS Model
B
k
= B
k
B
k
. (17.202)
In the mean eld approximation the coupling term B
k
B
k
is approximated
by
B
k
B
k

_
B
_
B
k
+
_
B
_
B
k
+B
k
B
k
, (17.203)
i.e. the term B
k
, which represents the deviation from the expectation value,
is considered as small. Employing this approximation, using Eq. (17.200)
and disregarding all constant terms [the rst line in Eq. (17.199) and terms
containing only expectation values, e.g.
_
B
_
B
k
] result in the mean eld
Hamiltonian H
MF
, which is give by
H
MF
=
FD<
k
<F+D
_
(
k

F
) cos (2
k
)
_
e
2i
k
e
2i
k
_
sin(2
k
)
2
_
(N
k
,
+N
k
,
)
+
D
<
k
<
F
+
D
_
(
k

F
) sin(2
k
) +
e
2i
k
cos
2
k
e
2i
k
sin
2
b
k
,
b
k
,
+
D
<
k
<
F
+
D
_
(
k

F
) sin(2
k
) +e
2i
k
cos
2
e
2i
k
sin
2
,
b
,
.
(17.204)
where
=
g
1
FD<
k
<F+D
B
k
. (17.205)
Note that all terms outside the energy interval
F
D
<
k
<
F
+
D
were
disregarded in the above expression for H
MF
.
Diagonalization is achieved when the last two sums in Eq. (17.204) vanish,
i.e. when the terms e
2i
k
are real and positive and when the parameters
k
are chosen such that the following condition is satised
(
k

F
) sin(2
k
) +[[ cos (2
k
) = 0 . (17.206)
For this case
k
=
1
2
tan
1
_
[[
k

F
_
, (17.207)
i.e.
sin(2
k
) =
[[
_
(
k

F
)
2
+[[
2
, (17.208)
cos (2
k
) =

k

F
_
(
k

F
)
2
+[[
2
, (17.209)
and
sin(
k
) =
_
1

k

F
(
k
F)
2
+||
2
2
, (17.210)
cos (
k
) =
_
1 +

k
F
(
k

F
)
2
+||
2
2
, (17.211)
and thus H
MF
becomes diagonal
H
MF
=

k
k
N
k
,
, (17.212)
where
k
=
_
(
k

F
)
2
+[[
2
. (17.213)
17.4.4 The Energy Gap
The value of the energy gap [[ can be determined from Eq. (17.205). Let
n
k
,
denotes the expectation value of the operator N
k
,
, i.e.
N
k
,
= n
k
,
. (17.214)
In thermal equilibrium at temperature T the following holds [see Eqs.
(16.142), (16.143), (17.212) and (17.213)]
n
k
,
=
1
e
+ 1
, (17.215)
where = 1/k
B
T, and where k
B
is Boltzmanns constant. Moreover, in ther-
mal equilibrium
_
b
,
b
,
_
= b
k
,
b
k
,
= 0 and thus B
k
is given by
[see Eq. (17.200) and recall that [[ = e
2i
k
]
B
k
=
(1 n
k
,
n
k
,
)
2
_
(
k

F
)
2
+[[
2
, (17.216)
and thus Eq. (17.205) can be expressed as
17.4. BCS Model
1 =
g
21
FD<
k
<F+D
1 n
k
,
n
k
,
_
(
k

F
)
2
+[[
2
, (17.217)
or [see Eqs. (17.213) and (17.215)]
1 =
g
21
FD<
k
<F+D
1 2
_
e
+ 1
_
1
=
g
21
FD<
k
<F+D
tanh
_
2
_
.
(17.218)
Replacing the sum by an integral leads to
1 =
gD
0
2
_
D
D
d
tanh

2
+||
2
2
_
2
+[[
2
, (17.219)
where D
0
is the density of states per unit volume.
Zero Temperature. For the case of zero temperature, where all occupation
numbers vanish, i.e. n
k
,
= 0, Eq. (17.219) becomes
1 =
gD
0
2
_
D
D
d
2
+
2
0
=
gD
0
2
log

D
+
_
2
D
+
2
0
D
+
_
2
D
+
2
0
,
(17.220)
where
0
stands for the value of [[ at zero temperature. The assumption
0

D
leads to
1 =
gD
0
2
log
4
2
D
2
0
, (17.221)
thus
0
= 2
D
exp
_
1
gD
0
_
. (17.222)
Critical Temperature. The energy gap [[ vanishes when T = T
c
, where
T
c
is the critical temperature. For this case Eq. (17.219) becomes
1 =
gD
0
2
_
D
D
d
tanh

c
= gD
0
_

c
D
2
0
dx
tanhx
x
,
(17.223)
where
c
= 1/k
B
T
c
. Integration by parts (note that lim
x0
tanhxlog x = 0)
yields
1 = gD
0
_
tanh

c
D
2
log

c
D
2

_

c
D
2
0
dx
log x
cosh
2
x
_
. (17.224)
For the case of weak coupling, for which
D
2
1 , (17.225)
one has
1 gD
0
_
log

c
D
2

_

0
dx
log x
cosh
2
x
_
. (17.226)
Using the identity
_

0
dx
log x
cosh
2
x
= log
4
+C
E
, (17.227)
where C
E
0.577 is Eulers constant, one nds that [see Eq. (17.222)]
k
B
T
c
=
e
C
E

0
0.566
0
. (17.228)
General Temperature. The energy gap [[ at temperature T can be nu-
merically evaluated from Eq. (17.219). To a good approximation the solution
can be expressed by the following analytical relation
[[
0
_
1
_
T
T
c
_
3
. (17.229)
17.4.5 The Ground State
The ground state [
0
of the mean eld Hamiltonian H
MF
=

k
k
N
k
,
(17.212) is a state for which all occupation numbers vanish, i.e. N
k
,
[
0
=
b
k,
b
k,
[
0
= 0, and therefore b
k,
[
0
= 0 for all k and . Moreover, [
0
is required to be normalized, i.e.

0
[
0
= 1.
Claim. The ground state [
0
is given by
[
0
=

k
/
k
[0 , (17.230)
where
/
k
=
_
e
i
k
cos
k
e
i
k
sin
k
a
,
a
,
_
. (17.231)
17.4. BCS Model
Proof. By employing the fact that
_
/
, /
k
_
= 0 provided that k
,= k
and
the relation
/
/
k
= cos
2
k
+ sin
2
k
a
k
,
a
k
,
a
,
a
,
sin
k
cos
k
_
e
2i
k
,
a
,
+e
2i
k
a
k
,
a
k
,
_
,
(17.232)
one nds that [
0
is indeed normalized as required
0
[
0
= 0[
/
k
[0
= 0[
/
k
[0
= 0[
_
cos
2
k
+ sin
2
_
1 a
,
a
k
,
__
1 a
,
a
k
,
__
[0
= 1 .
(17.233)
Moreover, using the relations [see Eq. (17.194)]
b
k
,
/
k
=
_
e
i
k
cos
k
a
k
,
e
i
k
sin
k
a
,
_
_
e
i
k
cos
k
e
i
k
sin
k
a
,
a
,
_
,
(17.234)
and
b
k,
/
k
=
_
e
i
k
cos
k
a
k
,
e
i
k
sin
k
a
,
_
_
e
i
k
cos
k
e
i
k
sin
k
a
,
a
,
_
,
(17.235)
one nds that
b
k
,
[
0
= b
k
/
k
[0
=

k
=k
/
k
b
k
,
/
k
[0
=

k
=k
/
k
sin
k
cos
k
_
a
k
,
a
,
+ 1
_
a
,
[0
=

k
=k
/
k
sin
k
cos
k
a
,
a
k
,
a
,
[0
= 0 ,
(17.236)
and similarly
b
k,
[
0
= 0 .
Alternatively, the ground state [
0
, which is given by Eq. (17.230), can
be expressed as
[
0
= C
0
_
1
k
a
,
a
,
_
[0 , (17.237)
where C
0
is a normalization constant, which is given by
C
0
=
e
i
k
cos
k
, (17.238)
and where [see Eqs. (17.210) and (17.211)]
k
= e
2i
k
tan
k
= e
2i
k
_
1

k

F
(
k
F)
2
+||
2
1 +

k
F
(
k
F)
2
+||
2
.
(17.239)
Furthermore, since
_
a
,
a
,
_
2
= 0 the following holds
[
0
= C
0
exp
_
k
a
,
a
,
_
[0 . (17.240)
17.4.6 Pairing Wavefunction
For a general function of position (r
) the following holds

_
dr
_
dr
(r
(r
(r
)
=
1
1
,k
,
a
,
_
dr
_
dr
(r
) e
ik
ik
=

k
,k
,
a
,
1
1
_
dr
e
i(k
)r
. .
=
k
,k
_
dr
(r
) e
ik
=

k
,
a
,
_
dr
(r
) e
ik
,
(17.241)
where
(r
) is quantized eld operators [see Eq. (16.98)]. In view of the

above result the ground state [
0
, which is given by Eq. (17.240), can be
expressed as
[
0
= C
0
exp
__
dr
_
dr
(r
(r
(r
)
_
[0 , (17.242)
where the function (r
), which is called the pairing wavefunction, sat-

isfy
_
dr
(r
) e
ik
=
k
, (17.243)
where
k
is given by Eq. (17.239).
The energy region near
F
in which
k
changes signicantly has a char-
acteristic width given by the energy gap
0
[see Eq. (17.239)]. The corre-
sponding region in k
space has thus a characteristic size given by

0
/v
F
,
where v
F
is the so-called Fermi velocity (which is dened by the relation
k
/k
= v
F
, where the derivative is taken at
k
=
F
). Consequently the
paring wavefunction (r
) is expected to have a characteristic size given

by , where
=
v
F
[[
, (17.244)
is the so-called BCS coherence length.
17.5 The Josephson Eect
Consider the global transformation a
k
,
a
k
,
e
i/2
and a
,
a
,
e
i/2
,
where is real. Such a transformation leaved the Hamiltonian (17.193) un-
changed, however, the factor B
k
is transformed according to B
k
B
k
e
i
[see Eq. (17.200)] and the energy gap is transformed according to [see Eq.
(17.205)]
e
i
. (17.245)
Moreover, the ground state [
0
is modied [see Eq. (17.237)] and becomes
[
0
[ (), where
[ () =

k
/
k
() [0 , (17.246)
where the operator /
k
() is given by
/
k
() =
_
e
i
k
cos
k
e
i
e
i
k
sin
k
B
_
. (17.247)
As can be seen from Eq. (17.246), the vector state [ () becomes identical
to the ground state [
0
(17.237) when = 2n, where n is integer. In view
of the fact that the pair creation operator B
k
= a
,
a
,
(17.200) in Eq.
(17.247) is multiplied by the factor e
i
one may argue that the phase can
be considered as the phase of Cooper pairs.
Claim. The state [ () (17.248) can be alternatively expressed as
[ () = e
in
P
[
0
, (17.248)
where [
0
is the BCS ground state (17.237) and where
n
P
=
1
2
,
a
,
a
k
,
(17.249)
is the so-called pair number operator.
Proof. On one hand
_
a
,
a
k
,
+a
,
a
k
,
_
/
k
() [0
= e
i
e
i
k
sin
k
_
a
,
a
k
,
+a
,
a
k
,
_
a
,
a
,
[0
= e
i
e
i
k
sin
k
_
a
,
a
,
_
1 a
,
a
k
,
_
+a
,
a
,
_
1 a
,
a
k
,
__
[0
= e
i
e
i
k
sin
k
_
a
,
a
,
+a
,
a
,
_
[0 ,
(17.250)
and thus
n
P
/
k
() [0 =
1
2
_
a
,
a
k
,
+a
,
a
k
,
_
/
k
() [0
=
=k
/
k
() e
i
e
i
k
sin
k
a
,
a
,
[0 .
(17.251)
On the other hand
i

/
k
() [0 = i
=k
/
k
()
/
k
[0 , (17.252)
and therefore [see Eq. (17.247)]
n
P
[ () = i

[ () , (17.253)
where [ () =

k
/
k
() [0 [see Eq. (17.246)]. The above result together
with the Taylor expansion formula for the exponential function [see Eq.
(3.31)] lead to
e
inP
[ (0) = [ () , (17.254)
which proofs the claim (17.248).
17.5.1 The Second Josephson Relation
Consider the case where a voltage V is applied to a superconductor. The
added energy of = eV per electron, where e is the electron charge, can be
taken into account by adding a term to the Hamiltonian of the system, which
becomes
H(V ) = H
MF
+ 2n
P
, (17.255)
where H
MF
is given by Eq. (17.212) and where the pair number operator n
P
is given by Eq. (17.249). As will be shown below, the added term 2n
P
gives
rise to time dependence of the complex energy gap [see Eq. (17.205)].
Claim. The following holds
i
d
dt
= 2 . (17.256)
Proof. With the help of the Heisenberg equation of motion (4.38) one nds
that
i
dB
k

dt
= [B
k
, H(V )] . (17.257)
thus [see Eqs. (17.196), (17.197), (17.200) and (17.212)]
i
dB
k

dt
=
k
e
2i
k
__
sin
2
k
b
,
b
,
cos
2
k
b
k
,
b
k
,
, N
k
,
+N
k
,
__
+
e
2i
k
sin
2
(2
k
)
2
__
N
k
,
+N
k
,
, b
k
,
b
k
,
+b
,
b
,
__
+e
2i
k
cos (2
k
)
__
sin
2
k
b
,
b
,
cos
2
k
b
k
,
b
k
,
, N
k
,
+N
k
,
__
+e
2i
k
sin(2
k
)
__
b
,
b
,
, b
k
,
b
k
,
__
.
(17.258)
With the help of the commutation relations
_
b
,
b
,
, N
k
,
+N
k
,
_
= 2b
,
b
,
, (17.259)
_
b
,
b
,
, b
k
,
b
k
,
_
= N
k
,
+N
k
,
1 , (17.260)
one nds that
i
dB
k

dt
= e
2i
k
sin(2
k
) N
k
,
+N
k
,
1 , (17.261)
and therefore [see Eqs. (17.208) and (17.216)]
i
dB
k

dt
= 2B
k
. (17.262)
Thus, the complex energy gap , which is given by [see Eq. (17.205)]
=
g
1
B
k
, (17.263)
satises Eq. (17.256).
For a xed the solution of Eq. (17.256) reads
(t) = (0) e
i(t)
, (17.264)
where the phase factor (t) is given by
(t) =
2t
=
2eV t
. (17.265)
Taking the time derivative (which is denoted by overdot) yields, in agreement
with Eq. (17.82), the second Josephson relation
=
2eV
. (17.266)
17.5.2 The Energy of a Josephson Junction
Consider a system composed of two superconductors that are separated one
from the other by a thin insulating layer, which serves as a tunneling barrier.
The Hamiltonian of the system is assumed to be given by
H = H
1
+H
2
+H
T
, (17.267)
where H
1
and H
2
are the Hamiltonians of the two decoupled superconductors
and where the tunneling Hamiltonian H
T
H
T
=

k
,k
t
k
,k
_
a
1,k
,
a
2,k
,
+a
1,k
,
a
2,k
,
_
+t
,k
_
a
2,k
,
a
1,k
,
+a
2,k
,
a
1,k
,
_
,
(17.268)
where the annihilation operators of the rst (second) superconductor are
labeled by a
1,k
,
(a
2,k
,
). With the help of Eq. (17.195) one nds that H
T
can be expressed as
H
T
=

k
,k
H
k
,k
, (17.269)
where
H
k
,k
= /
k
,k
_
b
1,k
,
b
2,k
,
b
1,k
,
b
2,k
,
_
+/
,k
(b
2,k
,
b
1,k
,
b
2,k
,
b
1,k
,
)
+E
k
,k
_
b
1,k
,
b
2,k
,
+b
1,k
,
b
2,k
,
_
+E
,k
_
b
2,k
,
b
1,k
,
+b
2,k
,
b
1,k
,
_
,
(17.270)
the coecients /
k
,k
and E
k
,k
are given by
/
k
,k
=
k
,k
cos
1,k
sin
2,k
+
,k
sin
1,k
cos
2,k
, (17.271)
E
k
,k
=
k
,k
sin
1,k
sin
2,k

,k
cos
1,k
cos
2,k
, (17.272)
and where
,k
= e
i(
1,k

2,k
)
t
k
,k
. (17.273)
We employ below time independent perturbation theory to calculate the
correction E to the systems energy to lowest nonvanishing order in the tun-
neling coecients [t
k
,k
[. The averaged total energy change E is evaluated
by summing over all basis states of the combined system and multiplying the
energy change of each state by its thermal occupation probability. As can be
seen from Eq. (9.32) E vanishes to rst order in [t
k
,k
[. To second order in
[t
k
,k
[ the correction E is found to be given by
E = 2

k
,k
[/
k
,k
[
2
_
n
k
n
k
k
+
k
(1 n
k
) (1 n
k
)
k
+
k
_
2

k
,k
[E
k
,k
[
2
_
n
k
(1 n
k
)
k

k
+
(1 n
k
) n
k
k

k
_
,
(17.274)
where
k
is given by Eq. (17.213) and n
k
is given by Eq. (17.215). With the
help of Eqs. (17.208), (17.271), (17.272) and (17.273) one nds that
[/
k
,k
[
2
= [
k
,k
[
2
_
cos
2
1,k
sin
2
2,k
+ sin
2
1,k
cos
2
2,k
_
+
1
2
Re
_
t
2
k
,k

1,k

2,k
_
,
(17.275)
and
[E
k
,k
[
2
= [
k
,k
[
2
_
sin
2
1,k
sin
2
2,k
+ cos
2
1,k
cos
2
2,k
1
2
Re
_
t
2
k
,k

1,k

2,k
_
.
(17.276)
In what follows it will be assumed, for simplicity, that all tunneling ampli-
tudes t
k
,k
are identical. Moreover, the two superconductors will be assumed
to be of the same type, i.e. [
1
[ = [
2
[ [[. For this case all the terms
t
2
k
,k

2
can be expressed as
t
2
k
,k

2
= T [[
2
e
i
, (17.277)
where T = [t
k
,k
[
2
and where is the relative phase dierence between
the two superconductors. The energy correction E can be expressed as a
function of as
E = (E)
0
E
J
cos , (17.278)
where (E)
0
is independent on and where
E
J
=

k
,k
T [[
2
1,k

2,k
_
1 n
k
n
k
k
+
k
+
n
k
n
k
k

k
_
=

k
,k
T [[
2
1,k

2,k
(1 2n
k
)
k
(1 2n
k
)
k
2
k

2
k
.
(17.279)
Replacing the sum by an integral leads to
E
J
=
[[
2
e
2
R
N
_

0
_

0
d
1
d
2
2
tanh

2
2

1
tanh

1
2

2
2
1
2
2
, (17.280)
where 1D
0
is the density of states,
n
=
_
2
n
+[[
2
, (17.281)
[see Eq. (17.213)] and where R
N
, which is given by
R
N
=

4e
2
1
2
D
2
0
T
, (17.282)
is the so-called normal state resistance.
The variable transformation
n
= [[ cosh
n
, (17.283)
n
= [[ sinh
n
, (17.284)
leads to
E
J
=
[[
e
2
R
N
I
_
[[
2
_
, (17.285)
where the function I (x) is given by
I (x) =
_

0
_

0
d
1
d
2
tanh(xcosh
2
) cosh
1
tanh(xcosh
1
) cosh
2
cosh
2
1
cosh
2
2
.
(17.286)
In the limit of zero temperature the integral can be evaluated using the
variable transformation
p
=

1
+
2
2
, (17.287)
m
=

1
2
2
, (17.288)
which together with the identities
cosh
1
+ cosh
2
= 2 cosh
p
cosh
m
, (17.289)
cosh
1
cosh
2
= 2 sinh
p
sinh
m
, (17.290)
_

d
cosh
= , (17.291)
lead to
E
J
=
[[
e
2
R
N
_

0
_

0
d
1
d
2
cosh
1
+ cosh
2
=
[[
4e
2
R
N
_

d
1
d
2
cosh
1
+ cosh
2
=
[[
4e
2
R
N
_

d
p
cosh
p
_

d
m
cosh
m
,
(17.292)
thus
E
J
=
[[
4e
2
R
N
. (17.293)
For arbitrary temperature the result is
E
J
=
[[
4e
2
R
N
tanh
[[
2
. (17.294)
17.5.3 The First Josephson Relation
As was shown above [see Eq. (17.278)], the energy of a Josephson junction
U
J
having phase relative to the energy when the phase vanishes is given
by [compare with Eq. (17.85)]
U
J
= E
J
cos . (17.295)
Let I (t) and V (t) be the current through and voltage across a Josephson
junction, respectively, at time t. Assume that initially at time t = 0 the
phase vanishes. Energy conservation leads to the requirement that
U
J
=
_
t
0
dt
I (t
) V (t
) . (17.296)
With the help of the second Josephson relation

= (2e/) V (17.266) and
Eq. (17.295) this becomes
E
J
cos =

2e
_

0
d
I (t
) . (17.297)
Taking the derivative with respect to leads, in agreement with Eq. (17.77),
to the rst Josephson relation
I = I
c
sin , (17.298)
where the so-called critical current I
c
is given by
I
c
=
2eE
J
=
2cE
J
s
, (17.299)
where
s
=
hc
2e
(17.300)
is the superconducting ux quantum, which is identical to the superconduct-
ing ux quantum given by Eq. (17.65) provided that the charge q
s
is taken to
be 2e. Note also that for the normal ux quantum
0
given by Eq. (12.48)
the charge of elementary carrier is e.
17.6 Problems
1. Rotating Superconductor - Consider a superconductor rotating at
angular frequency around the z axis. In the presence of an externally
applied magnetic eld B calculate the magnetic eld deep inside the
superconductor.
2. Consider a conductor containing charge carriers having charge q and
mass m. The density of charge carriers at point r is n(r) and the current
density is J(r). Contrary to the case of a normal metal, it is assumed
that all charge carriers at point r move at the same velocity v, which is
related to J by the relation [see Eq. (17.16)]
v =
J
qn
. (17.301)
Show that in steady state this assumption leads to the 2nd London equa-
tion [see Eq. (17.48)]
17.7. Solutions
2
H =
1
2
L
H , (17.302)
where H is the magnetic eld and where
L
=
_
mc
2
4nq
2
. (17.303)
3. Calculate n
P
and
_
(n
P
)
2
_
with respect to the BCS ground state [
0
,
where n
P
is the pairs number operator (17.249).
17.7 Solutions
1. In classical mechanics a mass particle in a rotating frame experiences
a force perpendicular to its velocity called the Coriolis force. For the
present case the Coriolis force F
F
= 2m
s
v , (17.304)
where = z is the rotation vector and where v = r is the velocity
vector. Additional force perpendicular to the velocity, which is acting in
the presence of a magnetic eld B, is the Lorentz force F
L
=
q
s
c
v B
[see Eq. (17.4)]. From this point of view the eect of rotation can be taken
into account by replacing the magnetic eld B by an eective magnetic
eld B
e
given by
B
e
= B+
2m
s
c
q
s
. (17.305)
With this approach Eq. (17.48) (for time independent B) becomes
2
B =
1
2
L
_
B+
2m
s
c
q
s

_
. (17.306)
Thus the magnetic eld deep inside the superconductor is given by
(2m
s
c/q
s
) .
2. The total energy of the system in steady state is given by E = T +U
H
,
where
T =
_
V
nmv
2
2
dV
is the kinetic energy and where
U
H
=
1
8
_
V
H
2
dV (17.307)
is the magnetic energy [see Eq. (14.38)]. With the help of the Maxwells
equation (17.18) and Eq. (17.16) E can be expressed in terms of H as
E =
1
8
_
V
_
2
L
(H)
2
+H
2
_
dV . (17.308)
Let H be an innitesimally small change in H, and let E be the corre-
sponding change in the energy. The requirement that E obtains a mini-
mum value leads to
0 = E =
1
4
_
V
_
2
L
(H) (H) +H H
dV . (17.309)
With the help of the general vector identity [see Eq. (14.41)]
(F
1
F
2
) = (F
1
) F
2
F
1
(F
2
) , (17.310)
one nds (for the case where F
1
and F
2
are taken to be given by F
1
=
H and F
2
= H) that
(H) (H) = ((H)) H (HH) .
(17.311)
The vector identity (H) = ( H)
2
H together with the
Maxwells equation (17.21) lead to
(H) (H) =
_
2
H
_
H (HH) . (17.312)
The volume integral over the second term on the right hand side can be
expressed as a surface integral using the divergence theorem. However,
when boundary conditions of H = 0 on the surfaces are applied the
surface integral vanishes. Thus Eq. (17.309) becomes
0 = E =
1
4
_
V
_
2
L
2
H+H
_
H dV . (17.313)
The requirement that E vanishes for arbitrary (small) H leads to Eq.
(17.302). The assumption that B = H [see Eq. (17.26)], i.e. the assump-
tion that the medium is isotropic and linear, implies that in steady state
Eq. (17.302) is equivalent to the 2nd London equation (17.48).
3. The following holds [see Eqs. (17.230) and (17.232)]
0
[ a
,
a
k
,
[
0
= 0[

k
=k
/
k
/
,
a
k
,
/
k
[0
= 0[
_
e
i
k
cos
k
e
i
k
sin
k
a
k
,
a
k
,
_
a
,
a
k
,
_
e
i
k
cos
k
e
i
k
sin
k
a
,
a
,
_
[0
= sin
2
k
0[ a
k
,
a
k
,
a
,
a
k
,
a
,
a
,
[0
= sin
2
k
,
(17.314)
17.7. Solutions
thus
0
[ n
P
[
0
=
sin
2
k
. (17.315)
Similarly, since
_
a
,
a
k
,
_
2
= a
,
a
k
,
one nds that
0
[ n
2
P
[
0
=
1
4
,k
0
[
_
a
,
a
k
,
+a
,
a
k
,
__
a
,
a
k
,
+a
,
a
k
,
_
[
0
=
1
4
0
[ a
,
a
k
,
+a
,
a
k
,
+ 2a
,
a
k
,
a
,
a
k
,
[
0
+
1
4
=k
0
[
_
a
,
a
k
,
+a
,
a
k
,
__
a
,
a
k
,
+a
,
a
k
,
_
[
0
=
1
2
sin
2
k
+ sin
4
k
+

k
=k
sin
2
k
sin
2
k
,
(17.316)
thus
0
[ (n
P
)
2
[
0
=
0
[ n
2
P
[
0
(
0
[ n
P
[
0
)
2
=
1
2
sin
2
k
+ sin
4
k
+

k
=k
sin
2
k
sin
2
k

k
,k
sin
2
k
sin
2
=
1
2
sin
2
_
1 sin
2
_
=
1
2
sin
2
k
cos
2
k
.
18. Open Quantum Systems
This chapter is mainly based on the book [7].
18.1 Classical Resonator
Consider a classical mechanical resonator having mass m and resonance fre-
quency
0
. The resonator is driven by an external force F
ex
that is given
by
F
ex
= F
0
cos (
p
t) = F
0
Re
_
e
ipt
_
, (18.1)
where F
0
is a real constant. The equation of motion is given by
m x +m
2
0
x = F
ex
. (18.2)
In steady state we seek a solution having the form
x = Re
_
Ae
ipt
_
, (18.3)
where A is a complex constant. Substituting such a solution into the equation
of motion (18.2) yields
A =
1
m
F
0
2
0
2
p
. (18.4)
This result is clearly nonphysical since it diverges at resonance
p
=
0
. This
can be xed by introducing a damping term in the equation of motion
m x +m x +m
2
0
x = F
ex
, (18.5)
where is the damping rate. For this case the steady state amplitude becomes
nite for any driving frequency
A =
1
m
F
0
2
0
2
p
i
p
. (18.6)
However, also (18.5) is a nonphysical equation of motion. The equipartition
theorem of classical statistical mechanics predicts that in equilibrium at tem-
perature T the following holds
Chapter 18. Open Quantum Systems
x
2
_
=
k
B
T
m
2
0
. (18.7)
However, as can be seen from Eq. (18.5), when F
0
= 0 the steady state
solution is given by x(t) = 0, contradicting thus the equipartition theorem.
This can be xes by introducing yet another term f (t) in the equation of
motion representing uctuating force
m x +m x +m
2
0
x = f (t) +F
ex
. (18.8)
The uctuating force has vanishing mean f (t) = 0, however its variance
is nite

f
2
(t)
_
> 0. In exercise 1 below the autocorrelation function of the
uctuating force f (t) is found to be given by (18.152)
f (t) f (t +t
) = 2mk
B
T (t
) . (18.9)
Similarly to the classical case, also in the quantum case nonphysical be-
havior is obtained when damping is disregarded. This happens not only for
the above discussed example of a driven resonator. For example, recall that
for a general quantum system driven by a periodic perturbation the time
dependent perturbation theory predicts in the long time limit constant rates
of transition between states [e.g., see Eq. (10.38)]. Such a prediction can
yield correct steady state population of quantum states only when damping
is taken into account.
Damping and uctuation in a quantum system can be taken into account
by introducing a thermal bath, which is assumed to be weakly coupled to the
system under study. Below this technique is demonstrated for two cases. In
the rst one, the system under study (also referred to as the closed system)
is a mechanical resonator, and in the second one it is taken to be a two level
system. In both cases the open system is modeled by assuming that the closed
system is coupled to a thermal bath in thermal equilibrium.
18.2 Quantum Resonator Coupled to Thermal Bath
Consider a mechanical resonator having mass m and resonance frequency
0
.
The resonator is coupled to a thermal bath, which is modeled as an ensemble
of harmonic oscillators.
18.2.1 The closed System
First, we consider the isolated resonator. The Hamiltonian is given by [see
Eqs. (5.9), (5.10), (5.11), (5.12), (5.13) and (5.16)]
18.2. Quantum Resonator Coupled to Thermal Bath
H
0
=
p
2
2m
+
1
2
m
2
0
x
2
=
0
_
a
a +
1
2
_
,
(18.10)
where
a =
_
m
0
2
_
x +
ip
m
0
_
, (18.11)
a
=
_
m
0
2
_
x
ip
m
0
_
, (18.12)
and where
_
a, a
= 1 . (18.13)
The inverse transformation is
x =
_

2m
0
_
a
+a
_
, (18.14)
p = i
_
m
0
2
_
a
a
_
. (18.15)
18.2.2 Coupling to Thermal Bath
Damping is taken into account using a model containing a reservoir of har-
monic oscillators interacting with the resonator. The total Hamiltonian is
given by:
H
t
= H
0
+H
r
+1 , (18.16)
where H
0
is given by Eq. (18.10), H
r
is the Hamiltonian of the thermal bath,
which is assumed to be a dense ensemble of harmonic oscillators
H
r
=

k
k
_
b
k
b
k
+
1
2
_
, (18.17)
and 1 is a coupling term
1 = a
k
b
k
+a
k
b
k
, (18.18)
where
k
are coupling constants. The bath operators satisfy regular harmonic
oscillator commutation relations
[a, b
k
] =
_
a, b
k
_
=
_
a
, b
k
=
_
a
, b
k
_
= 0 , (18.19)
_
b
k
, b
l
_
=
k,l
, (18.20)
and
[b
k
, b
l
] =
_
b
k
, b
l
_
= 0 . (18.21)
a + (i
0
+) a = F (t) , (18.22)
where is a constant and where the uctuating term F (t) is given by
F (t) = i
k
exp(i
k
t) b
k
(0) . (18.23)
Solution 18.2.1. In general, the Heisenberg equation of motion of an oper-
ator O is given by Eq. (4.37)
O =
i
[O, H] +
O
t
. (18.24)
a = i
0
a i
k
b
k
, (18.25)
a
= i
0
a
+i
k
b
k
, (18.26)
b
k
= i
k
b
k
i
k
a , (18.27)
and
k
= i
k
b
k
+i
k
a
. (18.28)
The solution of Eq. (18.27) is given by
b
k
(t) = exp[i
k
(t t
0
)] b
k
(t
0
)
i
k
_
t
t
0
dt
exp[i
k
(t t
)] a(t
) .
(18.29)
Choosing the initial time to be given by t
0
= 0 and substituting Eq. (18.29)
into Eq. (18.25) yield
a +i
0
a +
_
t
0
dt
a (t
k
[
k
[
2
exp[i
k
(t t
)]
= i
k
exp(i
k
t) b
k
(0) .
(18.30)
The states of the thermal bath are assumed to be very dense, thus one can
replace the sum over k with an integral
k
[
k
[
2
exp[i
k
(t t
)]
d[()[
2
exp[i(t t
)] ,
(18.31)
where () is the density of states. Assuming () is a smooth function
near =
0
one nds that
_
t
0
dt
a (t
k
[
k
[
2
exp[i
k
(t t
)]
_
t
0
dt
a (t
) [(
0
)[
2
_

dexp[i(t t
)]
. .
2(tt
)
= [(
0
)[
2
a(t) .
(18.32)
Thus using the notation
= [(
0
)[
2
, (18.33)
one has
a + (i
0
+) a = F (t) , (18.34)
a
+ (i
0
+) a
= F
(t) , (18.35)
where
F (t) = i
k
exp(i
k
t) b
k
(0) , (18.36)
F
(t) = i
k
exp(i
k
t) b
k
(0) . (18.37)
The uctuation terms F (t) and F
(t) represent noisy force acting on the

resonator.
From Eqs. (18.34), (18.35), (18.14), and (18.15) one nds that
p +p +m
2
0
x = f (t) , (18.38)
where
f (t) = i
_
m
0
2
_
F
(t) F (t)
. (18.39)
In classical mechanics the momentum p is given by p = m x. Using this
substitution the equation of motion for the quantum operator p (18.38) takes
a form analogues to the classical equation of motion of a mechanical resonator
having damping rate and inuenced by a force f (t)
m x +m x +m
2
0
x = f (t) . (18.40)
18.2.3 Thermal Equilibrium
(t) F (t +t
)
_
= 2 n
0
(t
) , (18.41)
F (t) F
(t +t
)
_
= 2 ( n
0
+ 1) (t
) , (18.42)
and
F (t) F (t +t
) =
(t) F
(t +t
)
_
= 0 , (18.43)
where
n
0
=
1
e
0
1
, (18.44)
and where = 1/k
B
T.
Solution 18.2.2. The modes of the thermal bath are assumed to be in ther-
mal equilibrium. In general, thermal averaging of an operator O
k
, associated
with mode #k in the thermal bath, is given by [see Eqs. (8.8) and (8.42)]
O
k
= Tr (
k
O
k
) , (18.45)
where the density operator
k
is given by
k
=
1
Z
e
Hr,k
, (18.46)
where
Z = Tr
_
e
H
r,k
_
, (18.47)
H
r,k
=
k
_
b
k
b
k
+
1
2
_
, (18.48)
and = 1/k
B
T. Using these expressions one nds that [see Eq. (8.134)]
_
b
k
(t) b
k
(t)
_
=
1
e
k
1
n
k
. (18.49)
Using Eq. (18.20) one nds that
_
b
k
(t) b
k
(t)
_
= n
k
+ 1 . (18.50)
In a similar way one also nds that
b
k
=
_
b
k
_
=
b
2
k
_
=
_
b
2
k
_
= 0 . (18.51)
Moreover, using the full bath Hamiltonian H
r
one can easily show that
b
k
b
l
=
_
b
k
b
l
_
= 0 , (18.52)
_
b
k
(t) b
l
(t)
_
=
kl
n
k
, (18.53)
and
_
b
k
(t) b
l
(t)
_
=
kl
( n
k
+ 1) . (18.54)
The uctuating forces are given by Eqs. (18.36) and (18.37). We calculate
below some correlation functions of these forces. Using Eq. (18.51) one nds
F (t) =

F
(t)
_
= 0 . (18.55)
Using Eq. (18.53) one nds that
(t) F (t +t
)
_
=

k
[
k
[
2
exp(i
k
t
) n
k
. (18.56)
Replacing the sum over k with an integral, as in Eq. (18.31), and taking into
account only modes that are nearly resonant with the cavity mode one nds
(t) F (t +t
)
_
= 2 n
0
(t
) , (18.57)
where
n
0
=
1
e
0
1
. (18.58)
Similarly
F (t) F
(t +t
)
_
= 2 ( n
0
+ 1) (t
) , (18.59)
and
F (t) F (t +t
) =
(t) F
(t +t
)
_
= 0 . (18.60)
Exercise 18.2.3. Show that the expectation value

a
a
_
in steady state is
given by
a
_
= n
0
. (18.61)
Solution 18.2.3. Multiplying Eq. (18.34) by the integration factor e
(i0+)t
yields
d
dt
_
ae
(i
0
+)t
_
= F (t) e
(i
0
+)t
. (18.62)
a (t) = a (t
0
) e
(i0+)(t0t)
+
_
t
t0
dt
F (t
) e
(i
0
+)(t
t)
. (18.63)
Steady state is established when (t t
0
) 1. In this limit the rst term
becomes exponentially small (recall that is positive), i.e. eect of initial
condition on the value of a at time t
0
becomes negligible. Thus in steady
state the solution becomes
a (t) =
_
t
t0
dt
F (t
) e
(i
0
+)(t
t)
, (18.64)
and the Hermitian conjugate is given by
a
(t) =
_
t
t0
dt
(t
) e
(i0+)(t
t)
. (18.65)
a
_
=
_
t
t
0
dt
_
t
t
0
dt
(t
) F (t
)
_
e
(i0+)(t
t)
e
(i0+)(t
t)
= 2 n
0
_
t
t0
dt
e
2(t
t)
= n
0
_
1 e
2(tt0)
_
.
(18.66)
The assumption (t t
0
) 1 allows writing this result as
a
_
= n
0
. (18.67)
The last result

a
a
_
= n
0
veries that the resonator reached thermal
equilibrium in steady state. Similarly, the next exercise shows that in the
classical limit the equipartition theorem of classical statistical mechanics is
satised.
Exercise 18.2.4. Calculate

x
2
_
in steady state.
Solution 18.2.4. According to Eq. (18.14) and
_
a, a
= 1 the following
holds
18.3. Two Level System Coupled to Thermal Bath
x
2
_
=

2m
0
_
a
+a
_ _
a
+a
__
=

2m
0
a
2
+a
2
+a
a +aa
_
=

2m
0
a
2
+a
2
+ 2a
a + 1
_
.
(18.68)
As can be seen from Eq. (18.60),

a
2
_
=

a
2
_
= 0. Thus, with the help of
Eq. (18.67) one has
x
2
_
=

2m
0
(2 n
0
+ 1)
=

2m
0
coth

0
2
,
(18.69)
in agreement with Eq. (8.142). In the classical limit where k
B
T
0
one
has
x
2
_
=
k
B
T
m
2
0
, (18.70)
in agreement with the classical equipartition theorem.
18.3 Two Level System Coupled to Thermal Bath
In this section we discuss a two level system (TLS) coupled to thermal baths,
and obtain the Bloch equations.
18.3.1 The Closed System
The Hamiltonian H
q
of the closed system can be represented by a 22 matrix
[see Eq. (15.36)], which in turn can be expressed in terms of Pauli matrices
(6.136)
H
q
=
2
(t) , (18.71)
where (t) is a 3D real vector, and where the components of the Pauli matrix
vector are given by
x
=
_
0 1
1 0
_
,
y
=
_
0 i
i 0
_
,
z
=
_
1 0
0 1
_
. (18.72)
Let P = be the vector of expectation values P = (
x
,
y
,
z
). We
refer to this vector as the polarization vector. With the help of Eq. (4.38),
which is given by
dA
dt
=
1
i
[A, H] +
_
A
t
_
, (18.73)
dP
z
dt
=
1
i
[
z
, H
q
]
=
1
2i
(
x
[
z
,
x
] +
y
[
z
,
y
])
= (
x
y

y
x
)
= ( P) z .
(18.74)
Similar expressions are obtained for P
x
and P
y
that together can be written
in a vector form as [see also Eq. (6.153)]
dP
dt
= (t) P . (18.75)
The time varying eective magnetic eld (t) is taken to be given by
(t) =
0
z +
1
(t) . (18.76)
While
0
, which is related to the energy gap separating the TLS states by
0
= /, is assumed to be stationary, the vector
1
(t) is allowed to vary
in time, however, it is assumed that [
1
(t)[
0
.
18.3.2 Coupling to Thermal Baths
As we did in the previous section, damping is taken into account using a model
containing reservoirs having dense spectrum of oscillator modes interacting
with the TLS. Furthermore, since the ensembles are assumed to be dense,
summation over modes is done with continuos integrals. The Hamiltonian H
of the entire system is taken to be given by
H = H
q
+
_
d a
1
() a
1
()
+
_
d a
2
() a
2
()
+
_
d
_
1
2
_
e
i
1
+
a
1
() +e
i
1
a
1
()
_
+
_
d
_
4
_
e
i
2
z
a
2
() +e
i
2
a
2
()
z
_
,
(18.77)
where H
q
is the Hamiltonian for the closed TLS,
+
=
_
0 1
0 0
_
,
=
_
0 0
1 0
_
, (18.78)
and the real coupling parameters
1
,
,
1
and
2
are assumed to be fre-
quency independent. The bath modes are boson modes satisfying the usual
Bose commutation relations
_
a
i
() , a
i
(
)
_
= (
) , (18.79)
[a
i
() , a
i
(
)] = 0 , (18.80)
where i = 1, 2. While the coupling to the rst bath (with coupling constant
1
) gives rise to TLS decay through spin ips, the coupling to the second
bath (with coupling constant
) gives rise to pure dephasing.

d
z
dt
=
1
i
[
z
, H
q
]
1
(1 +
z
)
+
2
_
i
+
1
1
+i1
_
,
(18.81)
and
d
+
dt
=
1
i
[
+
, H
q
]
_
1
2
+
+
+
i
_
1
z
+ 2
+
_
1
+1
_
_
,
(18.82)
where
1
1
=
_
1
2
e
i
1
_
de
i(tt0)
a
1
(t
0
, ) , (18.83)
and
1
=
_
4
e
i
2
_
de
i(tt0)
a
2
(t
0
, ) . (18.84)
Solution 18.3.1. With the help of the identities
[
z
,
+
] = 2
+
, (18.85)
[
z
,
] = 2
, (18.86)
[
+
,
] =
z
, (18.87)
one nds that the Heisenberg equation of motion (4.38) for
z
is given by
d
z
dt
=
1
i
[
z
, H
q
]
2i
_
1
2
_
d e
i
1
+
a
1
()
+ 2i
_
1
2
_
d e
i
1
a
1
()
,
(18.88)
for
+
by
d
+
dt
=
1
i
[
+
, H
q
]
i
_
1
2
_
d e
i
1
a
1
()
z
+ 2i
_
4
_
d e
i
2
+
a
2
()
+ 2i
_
4
_
d e
i
2
a
2
()
+
,
(18.89)
for a
1
() by
da
1
()
dt
= ia
1
() i
_
1
2
e
i
1
, (18.90)
and for a
2
() by
da
2
()
dt
= ia
2
() i
_
4
e
i
2
z
. (18.91)
Integrating the equations of motion for the bath operators a
1
() and a
2
()
yields
a
1
() = e
i(tt
0
)
a
1
(t
0
, )
i
_
1
2
e
i
1
_
t
t0
dt
e
i(tt
(t
) ,
(18.92)
and
a
2
() = e
i(tt
0
)
a
2
(t
0
, )
i
_
4
e
i
2
_
t
t0
dt
e
i(tt
z
(t
) .
(18.93)
We now substitute these results into the Eqs. (18.88) and (18.89) and make
use of the following relations
_
d e
i(tt
)
= 2 (t t
) , (18.94)
_
t
t0
(t t
) f (t
) dt
=
1
2
sgn(t t
0
) f (t) . (18.95)
where sgn(x) is the sign function
sgn(x) =
_
+1 if x > 0
1 if x < 0.
, (18.96)
to obtain
d
z
dt
=
1
i
[
z
, H
q
]
2i
_
1
2
_
de
i
1
+
e
i(tt0)
a
1
(t
0
, )
+ 2i
_
1
2
_
de
i
1
e
i(tt0)
a
1
(t
0
, )
,
(18.97)
and
d
+
dt
=
1
i
[
+
, H
q
]
i
_
1
2
_
de
i
1
e
i(tt
0
)
a
1
(t
0
, )
z
+

1
2

+
z
+ 2i
_
4
_
de
i
2
+
e
i(tt0)
a
2
(t
0
, )
+

2

+
z
+ 2i
_
4
_
de
i
2
e
i(tt
0
)
a
2
(t
0
, )
+
2

z
+
.
(18.98)
Thus, by making use of the following relations
=
1
2
(1 +
z
) , (18.99)
+
=
1
2
(1
z
) , (18.100)
+
=
+
z
=
+
, (18.101)
one nds that
d
z
dt
=
1
i
[
z
, H
q
]
1
(1 +
z
)
+
2
_
i
+
1
1
+i1
_
,
(18.102)
and
d
+
dt
=
1
i
[
+
, H
q
]
_
1
2
+
+
+
i
_
1
z
+ 2
_
+
1
+1
+
_
_
.
(18.103)
18.3.3 Thermal Equilibrium
1
1
=
_
1
1
_
= 1
=

1
_
= 0 . (18.104)
Using Eqs. (18.53) and (18.54), the relation
_
d e
i(tt
)
= 2 (t t
) , (18.105)
and assuming the case where the dominant contribution to the TLS dynamics
comes from the bath modes near frequency
0
(recall that
0
= /, where
is the energy gap separating the TLS states), one nds that
_
1
1
(t
) 1
1
(t)
_
=
2
1
2
_
d
_
d
e
i
(tt
)
_
a
1
(t
0
, ) a
1
(t
0
,
)
_
=
2
1
2
_
de
i(tt
)
n()

2
1
n
0
(t t
) ,
(18.106)
where n
0
is given by [see Eq. (18.58)]
n
0
=
1
e
0
1
. (18.107)
Similarly
_
1
1
(t) 1
1
(t
)
_
=
2
1
( n
0
+ 1) (t t
) , (18.108)
(t
) 1
(t)
_
=
2
2
n
0
(t t
) , (18.109)
(t) 1
(t
)
_
=
2
2
( n
0
+ 1) (t t
) , (18.110)
and
1
1
(t
) 1
1
(t) =
_
1
1
(t
) 1
1
(t)
_
1
(t
) 1
(t) =

1
(t
) 1
(t)
_
= 0 .
(18.111)
18.3.4 Correlation Functions
Equation of motion for the polarization vector P can be obtained by tak-
ing the expectation value of Eqs. (18.102) and (18.103). Using the iden-
tity
= (1/2) (
x
i
y
) and the notation P
= (1/2) (P
x
iP
y
) and
u
= (1/2) ( x i y) one nds that
P
z
= ((t) P)
z
1
(1 +P
z
)
+
2
_
i
+
1
1
+i
_
1
__
,
(18.112)
and
P
+
= ((t) P)
+
1
2
+
_
P
+
+
i
_
1
z
_
+ 2
_
+
1
+
__
_
,
(18.113)
where the subscripts z and + denote the components of the vector (t) P
in the z and u
+
directions respectively. However, Eqs. (18.112) and (18.113)
contain product terms between bath operators and TLS operators (e.g. the
term
+
1
1
in Eq. (18.112). To lowest order such terms can be evaluated
by assuming that these operators are independent, e.g.
+
1
1

+
1
1
.
However, this approach, which yields vanishing contribution of all such terms
is too crude. Below we employ a better approximation to evaluate the expec-
tation value of such terms. In the rst step Eqs. (18.102) and (18.103) are
formally integrated. This yields the following results
z
(t) = 1 + (1 +
z
(0)) e
1t
+
_
t
0
dt
_
1
i
[
z
, H
q
] +
2
_
i
+
1
1
+i1
_
_
e
1(t
)
,
(18.114)
and
+
(t) =
+
(0) e
(
1
2
+
)t
+
_
t
0
dt
_
1
i
[
+
, H
q
] +
i
_
1
z
+ 2
_
+
1
+1
+
_
_
_
e
(
1
2
+)(t
t)
.
(18.115)
In the second step these expressions for the TLS operators are substituted
into Eqs. (18.112) and (18.113). In this nal step, correlations are disregarded
(e.g. the expectation value of a term having the form
+
1
1
1
1
is evaluated
using the approximation
_
+
1
1
1
1
_

+
_
1
1
1
1
_
). The expectation values
of bath operators are calculated with the help of the results of the previous
section. This approach yields the following results
+
1
1
=
1
i
_
t
0
dt
e
(
1
2
+
)(t
t)
_
1
1
(t
) 1
1
(t)
_
z
(t
)
=
i
1
n
0
2
P
z
,
(18.116)
_
1
_
=
i
1
n
0
2
P
z
, (18.117)
_
1
z
_
= i
1
n
0
P
+
, (18.118)
and
+
1
+
_
= i
n
0
P
+
, (18.119)
thus
P
z
= ((t) P)
z
1
[1 + (2 n
0
+ 1) P
z
] , (18.120)
and
P
+
= ((t) P)
+
1
2
+
_
(2 n
0
+ 1) P
+
. (18.121)
A similar equation can be obtained for

P
, which together with Eq. (18.121)

can be written as
P
x
= ((t) P)
x
1
2
+
_
(2 n
0
+ 1) P
x
, (18.122)
P
y
= ((t) P)
y
1
2
+
_
(2 n
0
+ 1) P
y
. (18.123)
18.4. Problems
18.3.5 The Bloch Equations
1
(t) = 0, i.e. (t) =
0
z [see Eq. (18.76)]. For
this case Eqs. (18.120) and (18.121) become
P
z
=
1
[1 + (2 n
0
+ 1) P
z
] , (18.124)
=
_
i
0
1
2
+
_
(2 n
0
+ 1)
_
P
. (18.125)
In the long time limit the solution is given by P
(t ) = 0 and
P
z
(t ) = P
z0
, where [see Eq. (18.58)]
P
z0
=
1
(2 n
0
+ 1)
= tanh
_
0
2
_
. (18.126)
Note that Eq. (18.126) is in agreement with the Boltzmann distribution law
of statistical mechanics, according to which in thermal equilibrium the prob-
ability to occupy a state having energy is proportional to exp() (recall
that P
z
is the probability to occupy the upper state of the TLS minus the
probability to occupy the lower one). In terms of the decay times T
1
and T
2
,
which are dened by
T
1
=
1
1
(2 n
0
+ 1)
1
, (18.127)
T
2
=
_
1
2
+
_
1
(2 n
0
+ 1)
1
, (18.128)
the equations of motion for the general case, which are known as optical
Bloch equations, are given by
P
x
= ((t) P)
x
P
x
T
2
, (18.129)
P
y
= ((t) P)
y
P
y
T
2
, (18.130)
P
z
= ((t) P)
z
P
z
P
z0
T
1
. (18.131)
18.4 Problems
1. Calculate the autocorrelation function f (t) f (t +t
) of the classical
uctuating force f (t), which was introduced into the classical equation
of motion (18.8) of a mechanical resonator. The autocorrelation function
should yield a result consisting with the equipartition theorem.
2. Calculate the autocorrelation function f (t) f (t +t
), where the quan-

tum operator f (t) is given by Eq. (18.39).
3. Consider a TLS having energy gap . A perturbation, which is externally
applied, induces transitions between the states having rate
T
. Calculate
the polarization vector P in steady state.
4. Magnetic resonance - A time dependent magnetic eld given by
B(t) = B
0
z +B
1
( xcos (t) + ysin(t)) (18.132)
is applied to a spin 1/2 particle.
a) Use the Bloch equations to determine the polarization P
z
in steady
state.
b) The polarization P
x
in steady state can be expressed as
P
x
= 2
1
[cos (t)
() + sin(t)
()] , (18.133)
where
() and
() are respectively the real and imaginary parts

of the magnetic susceptibility () (i.e. () =
() + i
()).
Note that the term proportional to
() is in phase with respect

to the driving magnetic led in the x direction [recall that B
x
=
B
1
cos (t)], whereas the second term, which is proportional to
()
is out of phase [i.e. proportional to sin(t)] with respect to B
x
.
Calculate ().
18.5 Solutions
1. In the absence of any externally applied driving force, i.e. when F
ex
= 0,
the classical equation of motion is given by (18.8)
m x +m x +m
2
0
x = f (t) , (18.134)
where f (t) represents a random force acting on the resonator due to the
coupling with the thermal bath at temperature T. Bellow we consider
statistical properties of the uctuating function x(t). However, since some
of the quantities we dene may diverge, we consider a sampling of the
function x(t) in the nite time interval (/2, /2), namely
x
(t) =
_
x(t) /2 < x < /2
0 else
. (18.135)
The equipartition theorem requires that
1
2
m
2
0
x
2
_
=
1
2
k
B
T , (18.136)
where
x
2
_
lim
_
+
dt x
2
(t) . (18.137)
18.5. Solutions
Introducing the Fourier transform (FT)
x
(t) =
1
2
_

d x
()e
it
, (18.138)
one nds that
x
2
_
=
1
2
_

d x
()
_

)
lim
_
+
dte
i(+
)t
. .
2(+
)
= lim
d x
()x
() .
(18.139)
Since x(t) is real x
() = x
(). In terms of the power spectrum S

x
(),
which is dened as
S
x
() = lim
[x
()[
2
, (18.140)
one has
x
2
_
=
_

d S
x
() . (18.141)
Next, we take the FT of Eq. (18.134)
_
m
2
im +m
2
0
_
x() = f() , (18.142)
where
f(t) =
1
2
_

df()e
it
. (18.143)
Taking the absolute value squared yields
S
x
() =
S
f
()
m
2
_
()
2
+ (
2
0
2
)
2
_ . (18.144)
Integrating Eq. (18.144) leads to
_

d S
x
() =
1
m
2
_

d
S
f
()
()
2
+ (
2
0
2
)
2
. (18.145)
Assuming that in the vicinity of
0
, i.e. near the peak of the integrand,
the spectral density S
f
() is a smooth function on the scale of the width
of the peak , and also assuming that
0
, one approximately nds
that
_

d S
x
() S
f
(
0
)
1
m
2
_

d
()
2
+ (
2
0
2
)
2
=
S
f
(
0
)
m
2
3
0
_

d
(/
0
)
2
+ (1
2
)
2
S
f
(
0
)
m
2
3
0
_

d
(/
0
)
2
+ 1
. .
0
/
=

m
2
2
0
S
f
(
0
) .
(18.146)
This together with Eqs. (18.136) and (18.141) yields
S
f
(
0
) =
mk
B
T
, (18.147)
thus Eq. (18.144) becomes
S
x
() =
k
B
T
m
1
()
2
+ (
2
0
2
)
2
. (18.148)
At the peak =
0
one nds
S
x
(
0
) =
k
B
T
m
2
0
. (18.149)
The correlation function C (t
) of the uctuating force f is dened as:

C (t
) = f (t) f (t +t
) lim
_
+
dt f (t) f (t +t
) . (18.150)
18.5. Solutions
C (t
) =
1
2
lim
_
+
dt
_

df()e
it
f(
)e
i
(t+t
)
=
1
2
lim
de
it
f()
f(
)
_
+
dte
i(+
)t
. .
2(+
)
= lim
de
it
f()f()
=
_

de
it
S
f
() .
(18.151)
Using Eq. (18.147) and assuming as before that S
f
() is smooth function
near =
0
allow determining the coecient C (t
)
C (t
) =
mk
B
T
de
it
. .
2(t
)
= 2mk
B
T (t
) . (18.152)
2. Using the denition (18.39) and Eqs. (18.57), (18.59) and (18.60) one has
f (t) f (t +t
) =
m
0
2
_
F
(t) F (t)
_
F
(t +t
) F (t +t
)
_
= m
0
(2 n
0
+ 1) (t
)
= m
0
e
0
+ 1
e
0
1
(t
)
= m
0
coth
_
0
2
_
(t
) .
(18.153)
In the classical limit where k
B
T
0
one nds that
f (t) f (t +t
) = 2mk
B
T (t
) , (18.154)
in agreement with Eq. (18.152).
3. The Bloch equation (18.131) for this case becomes
P
z
=
T
P
z
P
z
P
z0
T
1
, (18.155)
thus in steady state
P
z
=
P
z0
1 +
T
T
1
. (18.156)
Clearly, by symmetry, P
x
= P
y
= 0 in steady state.
4. The Hamiltonian of the closed system is given by
H
q
=
2
(t) , (18.157)
where
(t) =
0
z+
1
(cos (t) x + sin(t) y) , (18.158)
and where [see Eq. (4.22)]
0
=
[e[ B
0
m
e
c
, (18.159)
1
=
[e[ B
1
m
e
c
. (18.160)
In terms of the vectors u
= (1/2) ( x i y) the vector (t) is given by

(t) =
0
z+
1
_
e
it
u
+
+e
it
u
_
. (18.161)
In terms of T
1
, T
2
and P
z0
Eqs. (18.120) and (18.121) become
P
z
= ((t) P)
z
P
z
P
z0
T
1
, (18.162)
and
P
+
= ((t) P)
+
P
+
T
2
. (18.163)
With the help of the identities
z u
= i u
, (18.164)
u
+
u
+
= u
= 0 , (18.165)
u
+
u
= i (1/2) z , (18.166)
Eqs. (18.162) and (18.163) become
P
z
=
i
1
_
e
it
P
+
e
it
P
_
2

P
z
P
z0
T
1
, (18.167)
and
P
+
= i
0
P
+
+i
1
e
it
P
z
P
+
T
2
. (18.168)
18.5. Solutions
By employing the transformation into the rotating frame [see for com-
parison Eq. (6.226)]
P
+
(t) = e
it
P
R+
(t) , (18.169)
P
(t) = e
it
P
R
(t) , (18.170)
Eqs. (18.167) and (18.168) become
P
z
=
i
1
(P
R+
P
R
)
2

P
z
P
z0
T
1
, (18.171)
and
P
R+
= i (
0
) P
R+
+i
1
P
z
P
R+
T
2
. (18.172)
a) In steady state, i.e. when

P
z
=

P
R+
= 0, one has
i
1
(P
R+
P
R
)
2
=
P
z
P
z0
T
1
, (18.173)
i (
0
) P
R+
+i
1
P
z
=
P
R+
T
2
, (18.174)
thus (recall that P
R+
= P
R
)
P
z
=
1 +T
2
2
(
0
)
2
1 +T
2
2
(
0
)
2
+
2
1
T
1
T
2
P
z0
. (18.175)
b) In steady state one has
P
R+
=
iT
2
1
[1 +iT
2
(
0
)]
1 +T
2
2
(
0
)
2
+
2
1
T
1
T
2
P
z0
, (18.176)
thus
P
+
=
iT
2
1
[1 +iT
2
(
0
)]
1 +T
2
2
(
0
)
2
+
2
1
T
1
T
2
P
z0
e
it
. (18.177)
Using the relations
P
x
= 2
1
[cos (t)
() + sin(t)
()]
=
1
_
e
it
() +e
it
()
_
,
(18.178)
and
P
x
= P
+
+P
, (18.179)
one nds that
() =
iT
2
[1 +iT
2
(
0
)]
1 +T
2
2
(
0
)
2
+
2
1
T
1
T
2
P
z0
. (18.180)
References
1. Claude Cohen-Tannoudji, Bernard Diu, Franck Laloe, Quantum Mechanics, Wi-
ley, New York (1977).
2. J.J.Sakurai, Modern Quantum Mechanics, Addison-Wesley, New York (1994).
3. L. D. Landau and L. M. Lifshitz, Quantum Mechanics Non-Relativistic Theory,
Oxford: Pergamon Press (1977).
4. L. S. Schulman, Techniques and Applications of Path Integration, , Wiley, New
York (1981).
5. M. V. Berry, Quantal phase-factors accompanying adiabatic changes, Proc. Roy.
Soc. London A 392, 45-57 (1984).
6. Alexander L. Fetter and John Dirk Walecka, Quantum Theory of Many-Particle
Systems, Mcgraw-Hill (1971).
7. Howard Carmichael, An open systems approach to quantum optics, Springer
(August 1993).
Index
action, 1
adiabatic approximation, 351
Aharonov-Bohm eect, 328
angular momentum, 139
Bogoliubov transformation, 411, 451
Bohrs magneton, 30, 72
Bohrs radius, 194
Bohr-Sommerfeld quantization rule,
312
Bose-Einstein function, 416
Boson, 394, 396
bra-vector, 17
canonically conjugate, 5
central potential, 187
chemical potential, 252
closure relation, 18
coherence length, 459
collapse postulate, 29
commutation relation, 34
commuting operators, 35
conservative system, 4
continuity equation, 307
Coulomb gauge, 365, 381
current density, 307
degeneracy, 24
density operator, 207
Diracs notation, 17
Drude model, 425
dual correspondence, 19
Ehrenfests theorem, 76
eigenvalue, 23
eigenvector, 23
equipartition theorem, 478
Euler-Lagrange equations, 2
expectation value, 29
Fermis golden rule, 300
Fermi-Dirac function, 416
Fermion, 394
Feynmans path integral, 327
ne-structure constant, 384
ux quantum, 331, 431, 466
fugacity, 253
gauge invariance, 331, 424
gauge transformation, 353
generalized force, 4
geometrical phase, 352
gyromagnetic ratio, 30
Hamiltons formalism, 1
Hamilton-Jacobi equations, 5
Heisenberg representation, 73
Hermitian adjoint, 22
Hydrogen atom, 192
ideal gas, 411
identical particles, 393
inner product, 15
Josephson eect, 432
Josephson inductance, 434
ket-vector, 17
kinetic energy, 4
Lagrangian, 1
Larmor frequency, 72
linear vector space, 15
London Equations, 428
London penetration depth, 428
macroscopic quantum model, 427
magnetic moment, 29
matrix representation, 20
Maxwells equations, 365, 425
Meissner eect, 429
momentum representation, 54
momentum wavefunction, 55
Index
norm, 16
normal ordering, 109
number density operator, 401
number operator, 98
observable, 22
operator, 17
optical Bloch equations, 487
orbital angular momentum, 139, 147
orthogonal, 16
orthonormal basis, 16
outer product, 18
path integration, 323
Paulis exclusion principle, 397
Plancks constant, 3
plasma frequency, 426, 437
Poissons brackets, 7
position representation, 50
position wavefunction, 51
positive-denite, 38
potential energy, 4
principle of least action, 2
projector, 25
pure ensemble, 210
quantized eld operator, 400
quantum bit, 438
quantum measurement, 28
quantum statistical mechanics, 211
radial equation, 190
reduced mass, 192
rotation, 140
scattering time, 425
Schr dinger equation, 3, 69
Schwartz inequality, 37, 38
second quantization, 405
semiclassical limit, 341
shell, 197
sperical harmonics, 151
spin, 29
spin 1/2, 71, 147
SQUID, 434
state vector, 15
stationary state, 71
Stern-Gerlach, 30
superconductivity, 423
symmetric ordering, 74
thermal bath, 472
thermal equilibrium, 476
Thomas-Fermi approximation, 448
Thomas-Reiche-Kuhn sum rule, 78
time evolution operator, 69
time-dependent perturbation theory,
293
trace, 33
transformation function, 55
translation operator, 50
tunneling, 314
turning point, 308
uncertainty principle, 36
unitary, 32
vector potential, 365, 381
Weyl transformation, 75
WKB approximation, 305

QM Lecture Notes

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

QM Lecture Notes

Uploaded by

Copyright:

Available Formats

Eyal Buks

Quantum Mechanics - Lecture

(t). The trajectory Q

(t). As the following theorem shows, the principle of

(t), for which the action obtains

(t) from point Q

Fig. 1.2. The classical trajectory Q (t) and the trajectory Q

is clearly an operator since for any [ T the object

is linear since for every [

. What is the dual of the ket-vector [ =

is a linear functional that

is given by Eq. (2.29).

and is dened by the following duality relation

Proof. Applying XY on an arbitrary ket-vector [ T and employing the

. Thus, there exists a unitary transformation

= [ [, thus clearly the operator A is Hermitian.

A[u 0. Moreover, using a complete orthonormal basis [n

be a complete orthonormal basis made of eigenvectors of B

). Using this basis for evaluating the trace one has

= A, thus the two eigenvalues

is the modulus of n divided by 6 (e.g., 1

be an eigenvalue of the observable x, and let [x

since x is Hermitian. As we will see below transformation

can be performed using the translation

be an eigenvalue of the observable x, and let [x

is an eigenvector of x with an eigenvalue x

is normalized since J is unitary.

) of a state vector [ is dened as:

) of state vector [, what is the wavefunction

be an eigenvalue of the observable p, and let [p

since p is Hermitian. Similarly to the case of the position

) of a given state [ is dened as

) and its mo-

) are the direct and inverse

(x) x (x) , (3.78)

commutes with the position operator x) and (3)

) is normalizable, however, the inte-

), which are dene as

is an eigenvector of p with eigenvalue p

. Thus by using the notation

is Hermitian implies that

and p have the

[ = [ p [. On the other hand,

commutes with the position operator

[ = [ p [ one nds that

) : the Schwartz in-

Eyal Buks Quantum Mechanics - Lecture Notes 67

is the Hamiltonian of the system.

, the dual of the Schrdinger equation (4.1) is given by

u = 1 for any time, namely u is unitary.

(t) = [; x[ u(t, 0) [(0)[

) in the coordinate representation satises

[ is the momentum wave function and

[ > a the potential V (x

[ > a is taken to be given by

[ yields [see Eqs. (3.23) and (3.29)]

[ and inserting the

) satises the following equation

(x)) /2). We have

, and subtracting the complex conjugate of

is time independent, the continuity

. To that end we begin by evaluating the matrix element

), which appears on the left hand side of the above