Contents Chapter 2 Chapter 3 Chapter 4 Chapter 5 Chapter 6 Chapter 7 Chapter 8 Chapter 9 Chapter 10 Chapter 11 Chapter 12 Chapter 13 Basic Group Theory Group Representations ‘General Properties of Irreducible Vectors and Operators Representatons of the Symmetric Groups ‘One-Dimensional Continuous Groups Rotations in 3-Dimensional Space — The Group $O(3) The Group SU(2) and More About $O(3) Euclidean Groups in Two- and Three-Dimensional Space The Lorentz and Poincaré Groups, and Space-Time Symmetries Space Inversion Invariance Time Reversal Invariance Finite-Dimensional Representations of the Classical Groups There are no problema in Chapter 1. 3 97 103 107Chapter 2 Basic Group Theory Problem 2.1 Prove that the identities (i) e~' = €, (ti) a~ta alla €G follow from the basic azioms of Definition 2.1. ¢, and (iii) ea = a for SoLvution: (i) Since e € G, it follows from Definition 2.1-(iii} that ¢ has an inverse, say e7? € G, auch that ec'=e. Multiply both sides of this equation by ~! from the left: Meet) mente. Using associativity, rewrite this as: (eTeje! = ete. But ae =a, for all a € G. In particular, this is true for e7!; -1 -1 ele tee Furthermore, e7! has an inverse, call it (e7!)-!, such that (e~!}(e7! both sides of the above equation on the right by this inverse yields: (e eNety' =e. ‘Using associativity once more: eee) =e, 2 Group Theory in Physics Multiply both sides on the left by a-' to get a-aa-") Using associativity (a“taja7? = at, Multiply both sides on the right by (a7")~! to get ata=e for alla € G. (iii) Starting with the equation we just proved, multiply both sides on the left by a: aja-'a) =a, Using associativity, rewrite this aa (aa")a = a, and now Definition 2.1-(iit) yields ca=a. Problem 2.2 Show that there is only one group of order three, using a step-by-step procedure to construct the group multiplication table. SOLUTION: Let G = {e,a,6} — with the implicit assumption ¢ ¢ a ~ 6 — and define an operation denoted by juxtaposition such that G equipped with this operation is a group. The trivial part of the multiplication table is: eab a b The Rearrangement Lemma implies that the entries in a row or column of a multipli- cation table are distinct, ic. any element of the group appears only once in any given row or column. [t follows that there is only one possibility for the completion of the second row: a? = 6 which implies ab = e, (The other choice a? = e implies ab = 6 and then 6 would appear twice in column three). The table so far looks like this: ren oe noBasic Group Theory a Requiring that the elements in the second and third columns be distinct leads uniquely to the completed table: e b a € & Therefore, there is a unique way to construct the multiplication table of a group of order three; this implies that, up to isomorphism, the group G of order three is unique. a a 6 ea Problem 2.3 Construct the multiplication table of the permutation group Ss using the cycle structure notation. (The geometrical interpretation represented by Fig, 2.8 should be of great help). SOLUTION: Let me introduce a graphical way of representing permutations of n objects: Draw n evenly spaced dots, much like this: ‘They represent n positions each holding one of n distinguishable objects, which, for obvious reasons, are not shown, A permutation consists of moving the n objects to new positions. This is denoted by drawing a new series of dots under the first one and drawing an arrow pointing from the ald location te the new one. For example, the permutation (12) among three objects would look something like this: x lo To compute the product of two permutations, remember that permutations are oper- ations. The product (12)(123) is a composition: it instructs you to apply (123) first and then apply (12) to the result. Let's compute (31)(12): SK ai To obtain the result, go to the top row, and, starting at position 1, trace the arrows to the bottom row: 1—+3— 3, 4 Group Theory in Physics In —incomplete— cycle structure notation: (13---). Go back to the top row; trace the arrows starting at position 3, in order to figure out where the object that used to be in position 3 ends up. 3— 12 Since these are permutations among three objects, we are done: (12){31) = (132) The above notation has numerous advantages: « It works equally well for arbitrary n, as opposed to the geometrical interpretation such as that of Fig. 2.2 which gets rather tedious as n gets large. « The nature of permutations as operations is made explicit. « Computing long strings of compositions is much easier this way — compare with the cycle structure notation. As practice, establish the following three results that. will prove useful in the following: (3112) = (123) (23)(12) (321) (123)(12) = (31) The trivial part of the multiplication table with the above three results is: e (12) (23) (32) (123) (821) (12) oe (23) (321) (31) (123) e (123) (31) (312) e (321) e (123) it immediately follows that (321)(12) = (23) — no other choices. At this point, it would be fairly easy to explicitly calculate al] the remaining elements. However, it is more instructive to proceed in a deductive manner — we will have an opportunity to see how tight the group structure is, Using these results, we can find all operations whose result is (12). For example: (321) = (23)(12) which implies that (23)(321) (23)(23)(12) e(12) (12).Basic Group Theory 5 Similarly, (123) = (31)(12) implies (81)(123) = (12) (31) = (123)(12) implies (921)(31) = (12) (28) = (321)(12) implies —(123)(23) = (12) The partially completed table is: e (12) (23) (31) (123) (321) (22) (23) (821) (12) (31) (123) e (123) (31) (12) (312) oe (321) (23) (12) e (123) Furthermore, observe: ® (123)(31) cam only be equal to (23) — the only element of the group that does not aheady appear in row five. @ Similarly, we have: (321}(23) = (31). # (23)(123) cannot be equal to (123); so (23)(123) = (31). This, in turn, implies that (23)(31) = (123) and (12)(31) = (321). © Similarly, (12)(123) = (23), which implies (31)(123) = (12). Again, (31)(321) cannot be (31), so it is (23), which implies the following: (31)(23) = (321), (12}(321) = (13) and, finally, (12)(23) = (123) The complete $3 multiplication table is: e (12) (23) (31). (123) (321) (2) e (223) (321) (23) (13) (23) (321) e (123) (31) (12) (31) (123) (a2ty (12) (23) (123) (31) (12) (23) (312) e (921) (23) (31) (12) e (128) Problem 2.4 Show that every element of a group belongs to one and only one class, and the identity element forms @ class by itself. SoLuTion: Let [a] stand for the equivalence class of a, ie. la] = {ge G:g~a} (i) A statement equivalent to the one we are supposed to prove states: 6 Group Theory in Physics ‘Two equivalence classes are either disjoint or identical. If [a] and [6] are disjoint there is nothing to prove. Otherwise, there exists a p € G such that p € [a] and p € [6]; our goal is to show [a] = [8]. Indeed: For all g € [a] we have p ~ g; but p ~ b and transitivity implies that g ~ b, that is, q € [4], for all 7 € [a]. This proves [a] < [B). Similarly, p ~ q' for all q’ € [b]; but p~ a. Transitivity implies that q’ € [a], and this yields (4) [al These two imply that [a] = [b], which completes the proof of the first statement. (ii) Consider the equivalence class of e. Let p € (el; then there exists ag € G such that p = geg™', by the very definition of the equivalence relation. But this implies p=gg"' =e, Therefore, all the p € [e] are identity elements, and uniqueness of the identity implies that [e] = {e}. a Problem 2.5 Enumerate the subgroups and classes of the group Sa. Which of the sub- groups are invariant ones? Find the factor groups of the invariant subgroups. SOLUTION: The group 5, is: Sy = {e,(12), (14), (14), (23), (24), (34), (12)(34), (13)(24), (14)(23), (123), (124), (132), (134), (142), (143), (234), (243), (1234), (1243), (1924), (1342), (1423), (1432)) Tt has five conjugacy classes: te}, {(12), (13), (14), (23), (24), (34)}, {(12)(34), (18)(24), (14)(23)}, {(123), (124), (192), (134), (142), (143), (234), (243)}, {(1234), (1243), (1324), (1342), (1423), (1432)}. It is of order 24, so it can only have proper subgroups of orders 2,3,4,6,8 and 12. In addition to the two trivial ones—S; itself and {e}—some of the subgroups are: @ One invariant subgroup of order twelve, isomorphic to the tetrahedral group. T = {e,(12)(34), (13}(24), (14}(23), (123), (124), (192), (194), (142), (149), (234), (243)}- # The following subgroup of order eight is isomorphic to the dihedral group of Prob- lem 2.8. Dg = {e, (1234), (13)(24), (1492), (13), (12)(34), (24), (14)(23)}.Basic Group Theory 7 @ Four subgroups of order six, all isomorphic to 54. Indeed they are the groups of permutations of any three of the four objects, leaving the fourth invariant. « Three subgroups of order four are isomorphic to Cy, the cyclic group of order four. They are groups of the form {e,9,97,9°} with g any 4-cycle, « However, there is only one invariant subgroup of order four: V4 = fe, (12)(34), (13)(24), (14)(23)} known as Klein's 4-group. Four subgroups of order three, of the form {¢,g,9"} with g any 3-cycle. « Six of the subgroups of order two are of the form {e,(ij)}. Three more are obtained by considering {e, (17 )(41)]. ‘This list of subgroups is not necessarily complete. However, we have found all the invariant ones: T and Vy. ‘The factor groups are: © S,/T = {T,(if)T} with (ij) any 2-eycle. This factor group is isomorphic to Cy © Sy[Va = {Vay (12)V4, (29) Va, (13)Vs, (123)Ve, (321) V4}, isomorphic to Sy. a Problem 2.8 Let H be any subgroup of G, which is not necessarily invariant. Is it possible to define products of left cosets directly by the equation pH gH = (pq)H, hence obtain a “factor group” consisting of left cosels? Apply this definition to the special case of H = {e,(12)} for Ss, and point out logical difficulties if there are any. SOLUTION: The definition of a group requires the existence of a well defined operation that associates an ordered pair of elements of the underlying set G with another one in the same set. This “association” is a mapping between two sets: The set of all ordered pairs of elements of G, denoted by G x G and called the Cartesian product set, and ad. The shorthand notation for all this is O:GxG—-+G There is also a notation to explicitly display the action of a mapping by showing its result on a pair of elements: ©: (a,8)—+e 8 Group Theory in Physics for a,b,c € G. For a mapping to be well-defined, each element of the domain must have a unigue image. To be more specific, suppose a = 4 € G, Then, given an clement ¢ € G, it must be true that a @¢— ©, where we have used the more familiar infix notation for the group operation. This is exactly the requirement that our “definition” fails to meet, as we shall presently show. First some more notation: Let Ly denote the set of all left cosets of H, i.e. Ly ={X €G:X =9H,¥oe G} Define a binary operation that combines two elements of Ly and produces another one, or, in symbols: ©:lyx Ly — Lr by pH © qH +— (pq)H We must investigate whether this operation is well defined. Let’s suppose that Af is not an invariant subgroup. Then, there exists a p € G@ such that pHp-' ¢ H, which implies that there exists an A € Hf such that php7! does not belong to H. For these special p,h we have: (pH) © [(h#) 0 @"H)] (pH) © (hp ")H (php™)H. i Since php"! does not belong to A, it follows that: (php) eH. Furthermore, A ¢ H, which implies that 4M = eH, by the rearrangement lemma. So, (pH) @ [(hH) © (p7".H)] must be equal to (pH) @ [(eH) @ (p"'H)|. However, (PH) o [(eH)@ (A) = po (ep) pHOp'H eH H. a ‘We have shown that: (pif) @ [(AH1) @ (pH) # (pi) 0 [leH) oH) despite the fact that eff = hi. Therefore, “* is indeed poorly defined. If H were invariant, (php~")M would be equal to Hf and this problem would not arise. Z Problem 2.7 Prove thai G = H, ® Hz implies G/M, ~ Hy and G/Hz ~ H, where ~ means “isomorphic to”.Basic Group Theory 9 SOLUTION: Since H), HM are invariant subgroups, the sets G/M, and G/H, equipped with the usual multiplication of cosets are groups. If it is also true that G = Ay @ Ho, we have: Gilly = {gly:g eG) {AyhgH, + hy © Hy, hg © Aa} {(hi Ay hay): hy © Hi, fe © Hz) {(eH)(hoffy) + hz © Ha} {hall} a i This states that the coscis of JJ; generated by the elements of H, are the only elements of the factor group Gj Ff. The above equation suggests 2 natural correspondence ha € Hy 74 aM € G/T which is trivially one-to-one and onto. This identification is a homomorphism since T(kh') = (hh’)H = (hH)(A'H) = TAT) for all A, h' € G. Therefore T is an isomorphism and G/H, = Hy. Similarly for G/ Hz ~ My. a Problem 2.8 Consider the dihedral group Dg which is the symmetry group of the square consisting of rotations around the center and reflections aboul the vertical, horizontal, and diagonal ares. Enumerate the group elements, the classes, the subyroups, and the invariant subgroups. Identify the factor groups, Is the full group the direct product of some of its subgroups? SOLUTION: Let a rotation by 1/2 about the centre be denoted by g and let a reflection about the (24) diagonal be denoted by h. Then Dy = {e,.9.97,9°, h, gh, gth, oh} subject to ¢ = g' = h? = (gh)*, One says that D, is generated by g and h. ' The group D, has five conjugacy classes: febs{o7} {o.97}. Ga. g7h}, (9h, aA}. It is of order 8 and therefore has non-trivial subgroups of orders 2 and 4 only: Tin cycle notation: g = (1234), g? = (13)(24), g* = (1432), A = (13), ph = (1234), g™A = (24), gh = (14)(23) 10 Group Theory in Physics Order 2: ferh}, feghh, M = {e9"), {e,g*A}, fe,g*h} Order 4: NY = f{eg.97,97), Ni, NY feo? gh, g?A}, {e,g?,h, 97h}. ‘The indicated ones are the invariant subgroups. The factor groups are: D4/Na {Nay gNa, ha, gh}, Daf N's {NANG}, DiylN 26 = {N%9N7}, DyfN3, = (NAG, AN). It is now easy to see that D, is not the direct product of any of its subgroups, since the factor group 04/2 is not isomorphic to any of the invariant subgroups of order 4. (Another way to prove this is to observe that if a group is the direct product of any of its invariant subgroups, then the intersection of these invariant subgroups must contain exactly one element: the identity [Prove!].) For reference, here is the multiplication table for D,: e g gf g& bh gh gth gh go gf ge gh gth gh h Foe g Ph @h hgh e g gt gh A gh gh A gh gth gh ¢ g® gg gh oh gh Gh g e¢ gf gf? gh gh kh ph P@ gi e GF Ph gh gh hog gg og eChapter 3 Group Representations Problem 3.1 Consider the siz transformations associated with the dihedral group Dy defined in Chap.2 (cf. Fig. 2.2). Let V be the 2-dimensional Euclidean space spanned by 4, and é, as shown. Write down the matriz representation of elements of Dy on V with respect to this Cartesian bases. So.ution: The group D; = 55 consists of: 53 = {e, (12), (23), (41), (123), (421)}. Let T be the representation. It is obvious that the matrix which corresponds to € is: mo=(91) The two cycles are rotations by 24/3 and 47/3; it therefore follows from Eq. (3.1-5) that the matrix realizations for the two d-cycles are: T(123) = -3( ts ey; T(321) a(y ): 2\V4 1 It is clear from Fig, 3.1-c, that T(23)|é,] = -é, and T(23)[é,] = éy, which imply Ten=( 4) aE Recalling that T is a homomorphism: T(12) = F((123)(23)) = T(123)T (23). Carrying out the matrix multiplication: 1f-t 4 T(12) = => (1) =—3 ( vo} ) IL 12 Group Theory in Physics man) =3( Js ae Similarly: Problem 3.2 Show that the 2-dimensional representation of rotations in the plane given in Ezample § of See. 9.1 con be decomposed into two {-dimensional representations, SoLution: Let {é,} be a basis for the plane, as given in the above mentioned Exam- ple. These basis vectors transform as: & = U(d)ér = dr cong + érsing and é, = U(d)éa = —é) sind + &, cos ¢. We will explicitly construct an invariant subspace, To this end, let: = ~gylte +i). ‘Under the action of L'(¢), these transform to: & = Ube, I s as = (8) |Jq-a = a [(-é cos 6 — égin ¢) —1(—€, sin 6 + €g cos f)] is Fw [-41 (cas ¢ — isin) — i8 (cos ¢— isin G}] = (5 (a ie ev, Collecting, we get: mgt a, =a". We conclude that, for all Jz) € span {é,}: U(é}l2) = [x)e™ € span {24} and so span {é,} is an invariant subspace, Moreover, since the dimension of this invariant subspace is one, it is minimal. Ina similar fashion, we obtain: U(d)é_ = eel#, which proves that span {é_} is another invariant subspace. This completes the decomposition R@ R= span {é,) @ span {é_}.Group Representations 13 Problem 3.3 In example 7 of Sec. 9.1, show thal the mapping in function space defined by Eq. (9.1-10) forms a representation of the relevant group G if x! = DT(g)iz!, where {DT (a),0 € G) are the transposes of the representation matrices of the group G (such as these described in Examples 4,5,6 in the same section), SOLUTION: According to Eq. (3.1-10), the action of the group on the function space is given by: cg Lf by fle) = fle’) In order to establish that this mapping provides us with a representation of the group G, we must show that it is a homomorphism. Rename the two variables z' and x? so that we can take advantage of the summation convention to rewrite Eq, (3.1-9) as f(z) = fiz' with t= 1,2 The image of f under the action of g, € G is: fz) 44 fie = fiz" = AD" (gy) Under the further action of another clement g2 € G: f(z) ®% fiz! = fT (or)\j2" = ADT (9:)'; (DT (a:)'a2*) =f (BT (9:)';D7 (o2)'s) x = iD" (gogi)iz! which shows that the homomorphism property is indeed satisfied: F(z) B% FD" (gag ja? ‘Transpose and inverse are not the only operations that will effect a representation of a group in a function space. In fact, any operation that reverace the otder of multiplication of the representation matrices will work equally well. An important example is Hermitian conjugation. a Problem 3.4 Prove that if D(G) is any representation of a fintte group G on an inner product space V, and x,y €V, then (z,¥) = > (Dly)=|D(g)y) sea defines o new scalar preduct on V. (Verify that the azioma of Appendiz /1.5 are satisfied.) “4 (Group Theory in Physics SOLUTION: Assuming that (-|-} is an inner product, let’s check to see if (-,-) satisfies the axioms for an inner product: (i) The first axiom requires that (z,y) must be equal to (y, 2)’. (zy) = 2 (D(g)21D(9)y) eG YE (Pla)ylD(g)z)" eG (5 (tsi) ee (y.z)". tt (ii) Check linearity ti (2, 0191 + o2ya) ¥ (Y(g)z1P(9) lary + aayal) reo = ¥ [a(D(e)z|D(g)y:) + a2(D(9)21D(9)y2)| ea = a [= (oto +a, eG = 4i(2,yr) + aa(z, yo). x (D104) peg and; finally, (iii) Positivity of the norm follows from (z,2) ¥ (Dig}z|Dig)z) Pr = 0, since each (D(g)2|D(g)r) 2 0. a Problem 3.5 Find one set of (unitary) representation matrices D*(p),p € Sy, for the # = 3, 2-dimensional irreducible representation of the group 5y. fHint: Examine the solution to Problem 1, check the irreducibility, unitarity, ... etc.} SOLUTION: From Problem 3.1, we know that the set of matrices: Te) = ( ): wa = (4%) me) = ({ {):(Group Representations 16 nay = (44). 7(123) = “(45 7); Ta) = “H(i =e) forms a 2-dimensional representation of D5 ~ 53. This representation is unitary, as can be shown by explicit matrix multiplication of each of these matrices with their transposes. The quickest way to check irreducibility is to employ Theorem 3.9. The equivalence classes of Ss are given in Ex. 1 after Definition 2.7. Taking the traces of the above matrices, we get: x:(20 -1) Then: 1 Be ES ri eo = x na lx 1 = gli 493-042-917) 1 which shows that the representation we obtained in Problem 3.1 is irreducible. o Problem 3.6 Find the similarity transformation which reduces the 2-dimenstonal rep- resentation of Cy given in the Example following Theorem 3.8 into diagonal form. SoLution: The 2-dimensional representation is: mo = (0) ma = (fo) It suffices to diagonalize T(a). The characteristic polynomial is: det [D{a) — Al] = 0 or M-1s0. The eigenvalues are: Ay = +1 and the corresponding eigenvectors are: w=(4). 16 ‘Group Theory in Physics The similarity transformation is generated by the matrix: J1fil s= ali -1) which is unitary. Then, the diagonalized T(a) is: Ta) = ST(a)s—" - (34). Problem 3.7 Consider the §-dimensional function space V consisting of polynomials of degree 2 in two real variables (2, y): S(t.y) = 27 + bry + ey? 4+ de tey th where a,b,...h are complex constants. If (z,y) transfarm under the dihedral group Ds as the coordinates of a 2-vector as described in Example 6 of Sec. 3.1 [cf. Problem i], then we obtain a 6-dimensional representation of Dy on V. Jdentify the various invari- ant subspaces of V under Dy, and the corresponding irreducible representations that are contained in this 6-dimensional representation. SOLUTION: Rename the two variables (z",27). According to Eq. (3.1-11) fos by fey=s luo} In order to compute the representation matrices, we need to know how 1, z', x, (2")?, a'z? and (z*) transform under the action of Dy. Using the 2-dimensional representation matrices we obtained in Problem 3.1: 141 fe Dig je? (2)? Le (Df)? ast 2s (Dig ir") (Dlg) Notice once more that the transformation induced by gon the function space requires the use of the representation matrices for g~' on the components of the original 2-vector. For example, let's see exactly how z'z? transforms under the action of g = (31). We do not need to compensate for the g“", since (31) happens to be its own inverse. Then ats? 2, (D(31)}2') (D132) 4 14 vi, va, la = git Se) (Be 2") v3 aya Liu vi ar =) +3 ~7e)Group Representations li which produces the fifth row of U'(31) below. Proceeding in a similar manner, we get the following matrices: 1 l ue) = am : 1 1 1 1, =foa uz) = * , al woi & 3 i 4 2 4&4 1 -1 u(23) = ts ‘ -1 1 1 4 ug) = TP , a 4 “oi a8 3-8 1 1 I 1 a =a A usa) = a 8 lw a. |, a2 4 3 oA ft 4 2 ¢4 1 4} u(321) = Bow “8a ok a ee a = 1 Sed Once more, one does not have to explicitly compute all of the above matrices. Using the homomorphism property and carrying out the matrix multiplications—as we did for 18 Group Theory in Physics the homomorphism property and carrying out the matrix multiplications —as we did for Problem 3.1—saves a considerable amount of time. The representation matrices are already in block diagonal form. Therefore, it is easy to identify the invariant subspaces. The obvious reduction of the representation is: U(Ss) = DON S3) @ DO(S3) & RUS) with R(S3) a 3-dimensional representation, However, Ds does not have a 3-dimensional irreducible representation; so R(53) must be reducible. Indeed, rewriting our polynomial as: rtf fp? oy? Kew) = (049 SY or) will decompose R(5) into an 1-dimensional minimal invariant subspace—generated by (a + e}(z? + y?)/2—and the the 2-dimensional orthogonal complement—generated by bey + (a ~e)(z? ~ y?)/2. The complete reduction of the 6-dimensional representation is; U(S3) = 2D"(S)) @ 2D"(3,) + bry + (a= c} +dr+ey+h Problem 9.8 Let (£},4;) end (x3,y2) be coordinates of two 2-vectors which transform independently under D3 tranaformations as in Problems I and 7. Consider the function space V spanned by the monomials 2,2, 71y2, ¥1%2, Yuva. Show that the realization of the group D3 on this 4-dimensional space is the direct product representation of that of Protlem I with itself. SOLUTION: We anticipate this result since the vector space V is, by construction: Vo = span {|z.25), |tvya)s lina), lta} } = span {{r1),|y1)} @ span {|23), [y2}]- Let's compute the representation matrices in this 4-dimensional vector space. In a way similar to the previous problem: ule) " win =) ve GO | U(23)Group Representations 19 1 vBoVeo8 vo = 3/48 4 aL 3 -Vi -v3 1 1 -v¥3 -V¥3 3 uqaa) = 4 4 a = a F 3°43 V1 1 Vv VB 3 ven) = 1) 1 y a -vi -V9 1 Explicit calculation of the direct product of the six matrices D@(G) with themselves establishes the identity: U( $2) = DO" Ss) @ (53). a Problem 3.9 Reduce the {-dimensional representation of the group Dy (hence 53) ob- tained in the previous problem into its irreducible components. Evaluate the Clebsch- Gordan coefficients. SOLUTION: The representation constructed in the previous problem is 4-dimensional, therefore reducible. [t is easily seen that: W, := span {]rut2) + |y.y2)} is an I-dimensional invariant subspace, since, under the action of any of the U(g), the laiy,} pieces generated by the transformation cancel. Similarly, the 1-dimensional sub- space W, = span {|ziy2) — (yur2)} is also invariant, and finally: Wy := span {|ry22) — lyye) . lewye) + lvze)) is a 2-dimensional invariant subspace. Therefore, changing basis to: ben) bi) ba) = Flin) - nw), bots): glzven) + tse), glzise) neal) lieu) + Ine] 20 Group Theory in Physics reduces this representation inte block diagonal form: UM(S) = DOS.) @ DPMS.) & DONS) or, in the notation of page 50, D™? = @BpD* with 4=1,2,3. 4a Note that each irreducible representation appears only once. ‘The elemeats of the transformation that effects this change of basis are, by definition, the Clebach-Gordan coefficients. In our case « =» = 3, and we can write the coefficients down by inspection: (1,1(8,3)1,1,1) =Jg = (2,2(8,3)1,1,1) (1, 2(8,3)1, 1,1) (2,1(3,3)1,1, 1) (1, L(3,3)1, 2,1) (2, 2(3,3)L, 2,1) (1, 2(3,3)1,2,1) = dg = —(2,1(3,3)1,2, 1) (1, 1(3,9)1,3,1) = y= —(2,2(3,3)1,3, 1} (1, 2(3,3)1, 3,1) (2, 1(3,3)1,3, 1} (1,1(3,3)1,9,2) = (2, 2(3, 3)L,3, 2) (1, 2(3,3)1,3,2) = (2, 1(3,3)1,3, 2). Problem 3.10 The tetrahedral group consists of all rotations which leave the regular tetrahedron invariant. This group has four ¥-fold ares, and three 2-fold ares. Enumer- ate the group elements, the distinct classes, and any invariant subgroups. Describe the irreducible representations and construct the character table, SOLUTION: The tetrahedral group T is the subgroup of 5, that consists of: T = {e,(123), (321), (124), (421), (144), (431), (234), (432) (12)(34), (1)(24), (14)(28)}. It has four equivalence classes: E— = {ej Cy = {(UZH(94), (13)(24), (14)(29))}, Cy = { (123), (421), (134), (432)}, Co = {(821), (124), (431), (284)}. There is one invariant subgroup: N = {e, (1234), (13)(24), (14)(23)}Group Representations 21 with factor group: TIN = {N, (123)N, (321)N} which is isomorphic to C5. The tetrahedral group has four equivalence classes; it, therefore, has only four nonequiv- alent, irreducible representations, whose dimensions satisfy: This equation has the unique solution: fy =M2=ny=1 and ny =F. The homomorphism - T*S TiN 4, will provide us with all the 1-dimensional irreducible representations of J’. To be more specific, construct a mapping T’ —+ T//N such that it maps @ The identity and all 9 € C; to N. « Allg EC, ta (129)N, e Allg EC, to (321)N. It is easy to check that this indeed forms a homomorphism. We can now use the character table of C3 to obtain the characters of all the 1- dimensional representations of T: EG G& i]t or oT ot ()1 1 —e G))1 1 a oo ¢ (4) [3 xf? yf yt? with = tris, Using the orthonormality condition (cf. Eq. 3.6-2), keeping » = 4 and letting v run through 1,2,3 we get x=-1 xf) <0 = xf! This completes the character table: 22 Group Theory in Physics Problem 3.11 Construct the character table of Sy. [Hint: Make use of the irreducible representations of any factor groups that may ezist. Then complete the table by using orthonormalily and completeness relations.) SOLUTION: The group 5; is of order 4! and has the following five equivalence classes (cf. Problem 2.5): € = {e} Co = {(12),(13), (14),(23), (24), (34), CE = {(22)(34), (15)(24), (14)(23)}, Cy = (123), (124), (194), (422), (431), (234), (432)}, Cy = 4 (1234), (1243), (1324), (1942), (4231), (4921)} Therefore, it has five inequivalent, irreducible representations whose dimensions satisfy 5. Eilat = 4! =u 1 with the unique solution: m=leny , ng=2 and ny =3= ng. In order to construct the character table, note: «© The trivial 1-dimensional representation has character x8:(1 21.11). « Sq has an invariant subgroup of order 12: the tetrahedral group T of the previous Problem. The homomorphism: SPS S/T = {T,12)T} Bs Cy gives us the second 1-dimensional representation, with character: x: (1 1-1 -1 1). © There is an invariant subgroup of order 4: V4 (cf. Problem 2.5}. Observe that Sul Va {¥a, (12)¥4, (23)Va, (15) Wa, (123)Ma, (321)Va) = Ds The group Ds has a 2-dimensional representation; the homomorphism S23 Saf¥a = Dy TE (Ds) should provide us with the 2-dimensional representation of SyGroup Representations 23 Let’s construct the first homomorphism—the others are natural. Observe that Ds has three equivalence classes: é& = {e}, Ky = {(12),(23),(31)}, Ky = {(123),(321)}. Let our map take the equivalence classes of S, into the equivalence classes of Ds in the following way: EUG i &, GUC, + Ks, Cy ot Kp It is easy to check that this map is a homomorphism. All we have to do now is to look up the character of the 2-dimensional representation of D3, which leads to: x :(220 0-1). Check irreducibility by showing that xO a1. Finally, use the orthonormality and completeness relations to compute the characters for the 3-dimensional representations, The complete character table is: Problem 3.12 Analyze the irreducible representations of the dihedral group Dj, /cf. Prob- lem 2.8): (i) Enumerate the irreducible representations, and (ii) Construct the character table for these representations. So.vTion: In the case of the group Dy, it is actually easier to construct the character table before enumerating the irreducible representations. The reason for this is that Dy ‘has many 1-dimensional representations. It has five equivalence classes (cf. Problem 2.8); therefore, it has five inequivalent, irreducible representations, whose dimensions satisfy: 5 Lm)? =8 rr’ Group Theory in Physica This equation has the unique solution: m=nj=nj=ng=1 and ny =2. In a way similar to the previous two problems, the homomorphisms from Dy, to its various normal subgroups of order 4 produce all the inequivalent, 1-dimensional representations: 1 1 1 1 (3)) 1 - obo aoe (y} 2 ob a o-oo. ()| 2 x ¥ xt x Completeness of the second column implies that x? = —2 and orthogonality demands that the rest be zero. This completes the character table: ‘We have already enumerated the 1-dimensional representations, since they are identi- cal to their characters. Construct the 2-dimensional ones by superimposing the diagram of Problem 2.8 on an orthonormal set of basis vectors for the plane. The identity and the three rotations are easily seen to produce the following matrices: ue = (4 t). oy = (2 4). vio - (7 @) va = (5 5)- The reflection (13) takes é, to é, and has the following matrix realization: vw=(4 a ‘The other three matrices are obtained by matrix multiplication of one of the above four matrices with U(A}, due to the homomorphism property: vor) = (79).Group Representations 25 vy = (4), U(g’h) = ci &).Chapter 4 General Properties of Irreducible Vectors and Operators Problem 4.1 Let G = T? be the discrete translational symmetry group, V be the vector apace of one particle on the one-dimensional lattice of Chap. 1, and P, be the prajection operator to the irreducible representation described in this chapter. If|z) = |nb+y),n an integer and —b/2 < y < 6/2, prove that Py|z) = Pyly)em!** so that Pylz)e"* = Pyly)e!*™. Compare with Eg. (1.9-6). SOLUTION: Suppose that there exists an integer n such that |x) = [nb +y), Then Plt) = Prlnb+y) YT (m)|nb + yee LP(M|T mje’ TMU yp) neo il Ere*inet"| ecthab P, | ye, where the second and last equality follow from the definition of P, (cf. Definition 4.2), the third from the definition of T(n), and the fourth from the identity T(m)T(n)} = Tim +m) and a change in summation variable from m to m' = m 4 n, Multiplying both sides by e'"*, with x = n+ y yields: Pelz)e™ = Pely)e'™™. a Problem 4.2 Let G= S5 and V = Vz = Vz where V3 ts the two-dimensional vector space of Problem 9.1. Starting with basis vectors tsé,, Ex@y, éyés, éyé,, construct four new basis vectors which transform irreducibly under Ss. Use the projection operator technique. 27 28. Group Theory im Physics SoLvuTion: We will construct the irreducible basis by computing the appropriate projection operators and applying them to an arbitrary [z) € V. The projection operators are defined by Eq. (4.2-1) and Definition 4.2, which, put together, read: fi Ma 1 i, Fat = ng a Polat Ug). The superscript j = i has been retained to signify that no summation on t or 7 is to be performed. The projection operators are seen to be linear combinations of the representation operators /(G); they can, therefore, be realized as 4 x 4 matrices, In order to proceed with the calculation, we need the following: # A realization of the ['(G) over the 4-dimensional vector space V2 * V;. In Problem 3.8 we obtained one over a function space which is closely related to this. In fact, the transpeses of the six matrices listed in the above mentioned Problem can serve our purposes. « A complete list of the irreducible representations of $5. This information is con- tained in the character table of 53 (cf. Table 3.3). « Matrix realizations of all the irreducible representations. The character table will provide realizations of all the 1-dimensional ones—since these realizations are iden- tical to their characters. A matrix realization of the 2-dimensional representation of S; was obtained in Problem 3.1. ‘We will explicitly display the construction of only one of the four projection operators, since the computation is trivial but somewhat tedious. Choose p = 3. This is a 2 dimensional representation and therefore has two projection operators associated with it. Let's construct the i = 1 = 7 one. We have ng = 6, n3 = 2 and the D(g-')} are given by the the realization obtained in Problem 3.1. Putting all this together we have: 1 2 1 Pit = § 1 1 1 -v5 -v3 3 ,Uf-vi a 3 Vi 24) -V3 3 =-1 V3 3°48 WGGeneral Properties of Inreducible Vectors and Operators 29 1 va va a -V5 -vI 1 vi Va 3 Lily 1 -3 “ga| -vV3 -3 1 VG 30 -W3 -vio1 1 -V§ -vW3o3 Vi} vBo1) a v5 24[ v8 -2 1 -V8 3043 V1 Adding these matrices gives: Poyt = 5 coos Tn a similar fashion we obtain: nope Puy = Hsoou Pay = 5 | | | coos Pal = mye 1 a 0 1 ‘We now construct the irreducible basis by taking an arbitrary |x) € V: lz) = a€.é, + B82, + yee. + 66,8, and operating on it with each of the projection operators. The results are: Poole) = SES) with (1)) = fete + fe) Paylz) B49) with |(2)) := [sé ~ @é,], 30 Group Theory in Physics i Pinlz) P+79),1) with [(3)y1) := (ed) + &y2els Paylt) = t= 415), 2) with [(3},2) = (éeée — & 2]. ‘The basis vectors {\(1)},((2)), ((3),1),|(3),2)} transform irreducibly; indeed, operat ing on them with the U/(G) one gets the matrix elements of the x“ representation. Notice that these irreducible vectors span the invariant subspaces generated in Problem 3.9. For reference, we give the two operators with p = 3 but 1 # 7. We will need these for Problem 4.4: rope Pon = sooe 0 0 0 0 o 0 1 0-1 ool a) o 60 0-1 0 1 1 te Oo Po = ‘ oo Notice that these matrices are transposea of one another. a Problem 4.3 Prove that the operators P,' have the following property: Pit = Py}. SOLUTION: Tey 1 (hat = [oe vow] Ty agi na Dug ura] “| -E o ED, tohu")] 7 = [5 “yuit ie yale yu «| = Pi where the first and last equalities are definitions; the second involves taking the complex conjugate of the scalar factors D,(g1)} and the Hermitian conjugate of the operator U; the third takes advantage of the unitarity of the matrix D(g); and the fourth involves a change of the summation variable from g to g~! and an application of the Rearrangement Lemma to show that these two summations are equivalent. As an example, observe that t (Pi) = Pint for the projection operators of the previous Problem. |General Properties of Irreducible Vectors and Operators 31 Problem 4.4 Prove that: a G) PAB D = Pass and (ii) PP Pe; = SueboSiiuPe. Use (ii) to interpret Pel ag the “transfer operator” from the vectors of type |ol} to the type |ok). SoLuTion: (i) Using the identity established in the previous Problem and Eq. (4,2-3) in turn, we get: (PuliPd = Pai Pol P65 = Pay tL An example of this can be provided by the matrix realizations of the projection operators obtained in Problem 4.2. (ii) Using Eq. (4.2-3) twice: nininl Pui [beudl Pel Sob ov8\54 Pol = Sober Sl6i Pal, In order to facilitate the interpretation of the above equation, let’s use the represen- tation apace V of Problem 4.2 as a conerete example. Let |e) = adsde + Best, + yey + 6,8, be an arbitrary element of V. We will compute Fis Pent Pisyal=) (no summation) where the necessary matrix realizations of these operators and vector can be found in Problem 4.2. Notice that we have taken all the 6's on the left-hand-side of (ii) into account, 90 the result is guaranteed to be non-zero. In Problem 4.2 we computed: Pople) = *5*K3),2) BD an ee = tae al This operator, as expected, picked off the piece of |z) that is parallel to [(3),2). Operate with Fis)? on this result: a-é PotPanlt) = —— oeeo eess eoco \ 32 Group Theory in Physics o 1 1 0 xt, + 2,2.) a-& = Fa The result is a vector parallel to the other irreducible basis vector of this invariant subspace, while, of course Poayn|(3), 1) = 18). 1) The essence of all these lies in the observation that Pisyfl(3), 2) = |(3),1) The operators that pre- and post-multiply P,* serve as label carriers to ensure that one starts with the right vector and ends up with the right vector. This is made explicit by the string of 6's on the right-hand-side of (ii). The reader can now explicitly calculate the result of the other choices of labels on these operators, and, in particular, the"reverse” Ploy Peay Pia) lz) (no summation) of what we just computed. a Problem 4.5 Suppressing the irreducible label » and matriz indices (i,j), Definition 4.9 for irreducible operators can be written as U(g)OU(g)* = OD{g) where {D(g);g € G} form a representation of the group G. Consider replacing the right- Aand side of this equation by the following alternatives, in turn: (i) QD(g)" Gi)OD(g"*) (iii)O.D(g)* (iv)D(g)7O (v) D(g")O (vi)D(g)tO, where the order of appearance of the two factors is important because of the implied matrix multiplication. In which of the six cases do we get true representations of the group G on the space of linear operators {O!}; thereby obtain viable allernatives fo the original definition? Among the original definition and the allernatives, which ones are equivalent to each other if fa) D(G) is unitary, (b) D(G) is equivalent to D(G)*, fe) D(G) is both unilery and equivalent to D(G)"? Sotution: In order to see why Definition 4.3 provides us with a representation of the relevant group, let's show that the homomorphism property is satisfied.General Properties of Inreducible Vectors and Operators 33 Let g = igo. Then U(g) =U(q)U(ga) and UO) = (dla) O [Vl UT) = U(g:) [U(:)0.U¢@5")] U(r") = Ula) [O;D(s2)i] Vig)" = [P@0,tiay"| Diet = Dig)! D(9s)f OD(g19:)F AD(9) The crucial step is the recombination of D(g,) with D(g;) te form D(g). From this we can immediately see that (ii) and (iii) are not viable alternatives—since they reverse the order of multiplication—but (i) does provide a representation. However, if D(g) appears to the left of O, we do need to reverse the order of multiplication—which all of (iv), (¥) and (vi) accomplish. Te demonstrate this, consider (vi): Let g = gigz, as before. Then: U(g}Oui(g) 0 [OC )O(o)] % [Oto U@r")] = Ula) [U(o2)0W (a5")] Uae") = U(a)[D(g}0,] Ula M(x)! [U(g OU (94")] DY(9q); D*(g )$Ox [D(g1)D(g2)I" #Os D(g)tOr Collecting, the viable alternatives are (i), (iv), (v) and (vi). (a) Suppose that D(G) is unitary. Then: Dig) = Dig)" = Dg)" ou and so (v) and (vi) are equivalent. Similarly, (b) Our original definition is equivalent te (1), and (iv) is equivalent to (vi), {c) The original definition is equivalent Lo (i) and (iv), (v) and (vi) are all equivalent to one another. aChapter 5 Representations of the Symmetric Groups Problem 5.1 Display all the standard Young tableauz of the group Sy. From the result, enumerate the inequivalent irreducible representations of 5, and specify their dimensions. SOLUTION: The partitions of 4 are: {4},{3,1),12,2), {2,1,1}. {1 U1,0) and the corresponding standard Young tableaux are: e.-7 217) _Oy2]3) on [2/4 an [1 ]3]4 Ta] ob il LE ay [14 ®=BT 8" = Pla 1(2 aE L[4 a =(3 of? = [2] of =13 [a] 3 6 Group Theory in Physics According to Theorem 5.2, there are five inequivalent irreducible representations; one for each normal tableau. The dimensions of these representations are equal to the number of standard tableaux associated with each normal tableau (Miller, Theorem 4.2]. Therefore: ® There are two I-dimensional irreducible representations, corresponding to @, and Os. «* A 2-dimensional representation is associated with O5. © Finally, there are two 3-dimensional inequivalent representations, corresponding to ©, and O. Notice that this serves as an example to Theorem 5.8. The group algebra of 5, is of dimension 4!; we have completely reduced this algebra into a direct sum of linearly independent left ideals: * One |-dimensional ideal, corresponding to ©. © Three 3-dimensional ones, each associated with one of the @2, 8S" and Of), All of these generate equivalent representations. « Two 2-dimensional ideals are generated by 03 and ef), ® Another three 3-dimensional ideals are generated by the standard tableaux aasoci- ated with @,. « Finally, @ generates an 1-dimensional left ideal. Adding the dimensions of the left ideals we get: a= 14(3x3)4 (2% 2)4 (3 x8) 41 which is an identity. a Problem 5.2 Repeat the above for Sg. SOLUTION: The partitions of 5 and the corresponding standard tableaux are:Representations of the Symmetric Groups 3T Partition [| Normal Tableau Standard Tableaux 6, =D [2737475] {4,1} = est ef”), ef), of) tay | o=fFP2T8) | of, 09,05", 02 ah) 6, {3,11} a= QR) girs) T {2.2\1} @.=[3 of), of" ef", ape 5 Tl2 {2,1,,1} = oF ef, 09,0) [5] 1 2 {1,1,1,1,1} 6, =|3 4 Similarly to the previous Problem, there are seven inequivalent irreducible represen- tations; one for each normal tableau: ® There are two 1-dimensional irreducible representations, corresponding to @, and Or. ® Two 4-dimensional representations are associated with ; and Oy. ® Two §-dimensional representations generated by G3 and Gs. 38 Group Theory in Physica « Finally, there is one 6-dimensional representation, corresponding to Oy. Furthermore, the dimensions of the generated left ideals sum to the dimension of the group algebra: Sl=2+(2x4*)4 (2x57) +6? as in the previous problem. 2 Problem 5.3 Verify by explictt calculation that e, of the group algebra Ss (given after Definition 5.4) is a primitive idempotent. fef. Appendix IIS]. SOLUTION: According to Theorem 3 in Appendix III, an idempotent ¢; is primitive if and only if it satisfies ejre; = A,e;, for all r in the group algebra. The Young symmetrizer ¢, of 5 is: es =e + (12) — (31) - (321) We must now take an arbitrary element r € 3, and show that the above relation is satisfied. However, all such r can be written aa linear combinations of the natural basis elements for the algebra; 90, it suffices to check that enpez = Aper for all p € S5. This is a rather trivial calculation, so we will only display a sample: ep(123}e, = fe + (12) — (31) — (321)] (123) [e + (12) — (31) — (321)] fe + (12) — (31) — (321}] ((123) + (31) — (23) — e] —3[e + (12) — (31) - (821)] = —dep. Proceeding in a like manner, we can establish the required result. a Problem 5.4 Prove that: if \ > y, eet = 0 for all pg € Sq. [Hfint: use Theorem 5.6 and Theorem ili.4.) SoLution: If dm, we know that e,ge, =0, for all g¢ $,. Then ee, = 0 equi = 0 peveh = 0 fe = 0 Problem 5.5 Prove that D(p), p€ 5, as defined by Eq. (5.5-13) forms a representation of Sn.Representations of the Symmetric Groups 39 SOLUTION: The defining Eq. 5.5-13 reads: dik i in Dip) = 6 fi, = Ati We must show that the representation matrices so defined satisfy D(pq) = D(p)D{q) for all p,q € 5, Indeed: DipDta) = (858k) (58, --48,) = a ar ee = arth. gtrn = Depa) This calculation is fairly simple, deapite the horrible looks. All one has to do is to not get lost in the jungle of indices. We have freely permuted the 6, as allowed by the two equivalent forms of the definition of the representation matrices. a Problem 5.6 Prove that there is no nth-rank totally anti-symmetric tensor in m- dimensional space if n> m. SOLUTION: Let a be the total anti-symmetrizer; it obviously satisfies (ij)a = a(ij) = —a 48 a consequence of the rearrangement lemma and the fact that multiplication by a transposition changes the parity of a permutation. Consider an element |i, ---i,) of the natural basis of V® such that i, = i, ie. it has 4 duplication in the ’ and | positions. We can show that a annihilates all such basis elements of Vii: (ea) fi, ati, a [(RE) Ita ---in)] aliy --i,) vin) in) which implies that aliy--+¢,) =0 This is an example of a general property of the irreducible symmetrizers which can be stated in rough terms as: If the tensor on which the syrmmmetrizer is to operate cannot “accommodate” the symmetry, it vanishes under the action of the symmetrizer. The required result now follows from the observation that, ifn > m™ all natural basis elements of V@ contain at least one such duplication, and, therefore are all annihilated by a. o 40 Group Theory in Physics Problem 5.7 Verify that the mized symmetry tensora {|m,a,a),a = 1,2} span an in- variant subspace under Ss in the tensor space V;. fef. Sec. 5.5, Example §.] SOLUTION: The irreducible symmetrizer ey, is given by fe + (12)l[e - (31)] e + (12) — (31) — (321) om th It is easy to show that re, with r € 5; is proportional either to e, or to el), For example, (31)em = (31) + (123) — e— (23) = = (em is ef2°)) Therefore, re,|m, a, 2) can only be a linear combination of the |m,e, a}, a= 1,2, wh: shows that the subspace spanned by these two tensors is invariant. a Problem 5.8 (i) Show that in Example § of Sec, 5.5 the two vectors {|m,i, 1},i = 1,2} span the subspace {e,,)c), |) € V2}. (ti) Show that this subspace is irreducible under Gy. SOLUTION: Part (i) is trivial: Notice that e_ annihilates }+ ++}, |+ — +)},|-——) and |— + —), which follows from either explicit calculation or by observing that ¢, anti- symmetrizes the first and third positions, while these tensors have duplications there [cf Problem 5.6]. The other two basis tensors give: em|— + +) —|ra, 1,1) enlt——} = —|r2,2,1). (ii) It is instructive—although not the shortest way to obtain the result—to actually construct the representation of G; on T(1). Then we will be able to look at the repre- sentation matrices and decide whether they have any common cigenvectors. (Recall that the representation is 2-dimensional; if it is reducible, the irreducible subspaces will have to be of dimension one). Let's simplify the notation by renaming the two tensors: (4) = Im 11) W(-)) = [mm 2.1) ‘We wish to investigate the action of elements of G2 on these tensors. According to Eqs. (8.5-B) and (5.5-6) which define the action of Gm elements on Vi", we have: 7ellt+—-)- 14 = +)—1- + 4] = Sliuish)ehetatt — Fliniaks)o oto — Tiaiskade ot a? gl(+))Representations of the Symmetric Groups 41 with summation over repeated indices implied. Carrying out these summations and collecting the various terms, we obtain: gl(+)) = (det g) [4st - K-95] In a similar fashion: al(-)) = (deta) [-I+))9% + (-)9r] Therefore, the representation matrices are [ef. Eq. (3.1-2)]: = st fc vis) = (ders) ( # a These matrices are themselves elements of Ga—check this!—and they cannot all commute with one another, since G is not abelian. Therefore, they do not ail have a common eigenvector and so the representation is irreducible (cf, Problem 3.5]. aChapter 6 One-Dimensional Continuous Groups Problem 6.1 Show that the rotation matriz R(¢), Eg. (6.1-8), is an orthogonal matriz and prove that every SO(2) matrix represents a rotation in the plane. SOLUTION: Check orthogonality by explicit multi Rayer ay= (BE ar) ( sae etie(s ‘a sing cosp =sing cos¢ ion: and similarly for A7(¢)A(¢). Conversely, suppose S is a real 2 x 2 matrix satisfying SST = 1. The orthogonality condition imposes conatraints on the matrix elements. More explicitly: Su Siz a sn) _ 8,433, susn + Si28a Su 3 so sa) \snsutsese | sh +sh _ f10 ae 01 which is equivalent to the three equations: A4e, = 1 Sy182y + 812832 tntam = 1 i] ‘The first of these relations allows us to make the identifications s;, = coed and #43 = sind. Similarly 492 = cosy) and sy, = sing from the third, while the middle one then implies: coa @sin d +ain dcos Py = sin(y + ¢) = 0. This in turn implies that » = —¢ and 20 We recover: o=(o8 a) (sind cosd 43 44 Group Theory in Physics Problem 6.2 Show that 24 = (#1 — ié2)/ 2 are eigenvectors of J with eigenvalues +1 respectively [ef. Eg. (6.9-8)]. SOLUTION: We will use the explicit form of J to obtain the eigenvalues and eigen- vectors: a 7 _ —t a ( i 0 ) The characteristic polynomial is 47 — 1 = 0 and so the eigenvalues are Ay = +1. The normalized eigenvectors are: é = ~ syle + i€3)Chapter 7 Rotations in 3-Dimensional Space —The Group S$0(3) Problem 7.1 Derive the general expression for the 3x 3 mairiz R(a, 3,4). SoLuTION: From Eq. (7.1-12) we get: Fle, 8.) = Bala) Ral?) Fal) Explicit expressions for these rotations are provided by Eqs. (7.1-13). Performing the matrix multiplications, the result is: Rina cofeottemasiny -sinacosfsinytess sen ain cain commcogcay=sinasiny coma corfainy—sinacosy con sing Aladin) = = sin Gcony in Bain coal Problem 7.2 Derive the relation between the Euler angle variables (a, 8,7) and the an- gle and azis parameters (y,0,¢) for a general rotation. [Hint: use (i) the trace condition, and (ii) the fact that f is left invariant by the rotation R(a,8,7).] So.ution: Let A,() and Ala, §,+) be two different ways to represent the same rotation. We wish to find expressions relating the two parametrizations. Qne such expression can be obtained by considering the traces of the two matrices, which, of course, satisfy: TrA(e, 8,7) = Tria (e) We can immediately compute the trace of A{a,8,7) by referring to the result of the previous problem: TrR(a, 4.7) = coad + (1+ cos 8) cos(a +) too (SFr 45 46 Group Theory in Physics In view of Eq, (7.1-9), we observe: TeRa(d) = Tr (RRa(p) A) = TrR(p) = 2osp+l Equating the two, we get: Qeosy+1 = feos? (227) coat 8 — 1 ary 3 omg = heal ooo which is one of the Eqs. (7.1-4). Now we turn our attention to the fact that: Ria, 6,7) = Rap) = Write # in terms of spherical variables as cos paind nm: singsind cos @ Multiply it on the left by A(a,,7) and set the result equal to #2. This gives us three equations; one for each component. The condition for the third component reads: cos @ = — sin Acosy sind cos ¢ + sin Psin-y ain O sing + cos Pcosh which gives: cos 8 —1 ain B cos(p +7) Equating the first components gives: tanf = (+) sinfcosd? = cosesin Scos? + sin 0 cos ¢ (cos. @cos J cosy — sin rsin 7) — sin @ sin ¢ (cos a cos § ain + — sin aw cos 7) Dividing both sides by sin @, regrouping and using (+) gives: cos @ = —cos(a — 6— +) as a result of several manipulations, This, in turn, implies: t+a- gis nig which is another of the Eqs. (7.1-4). Finally, inserting the above result in (+), we obtain: tang = nisRotations in 3-Dimensional Space—The Group SO(3) 47 Problem 7.3 From geometrical considerations, derive the following result whieh de- scribes the effect of the rotation Ra (i) on an arbitrary sector #: Ral)? = # cos w + fa(1 — cos )(*- A) + (4 x F) sing. SOLUTION: Starting with the given vectors * and f, define an orthonormal set of vectors by with cos@ := 1% -#, The components of # in this basis are: "35 #6 =—sind Ta = F< cos8 Tish) = 0 Rotate # to # = Ra (y)#. Hs components are: cosh 0 sing’ —sind —coadsingd Cd oO 1 0 cos F = cond =sings 0 cosy 0 ain psin# or, rewriting this in vector notation: # = 0050 — Gcosw sind + (fx #)sin Expressing ¢ in terms of # and #1, we arrive at the result: # = Bos 4 AA(1 — conw)(m-#) + (a x #) sing a Problem 7.4 An alternative way of writing the Lie Algebra for SO(3), Eq. (7.2-10), ean be obtained by defining JY = chy, (i.e, 1? = Jy,...efe. ) as the generator for rotations in the k-I plane. Show that; [art am] = i (gt ath — gh gi — gn ptm. gin yh) | Although thia form may appear a little less compact than Eq. (7.2-10), it is more readily generalized to higher dimensions. 48 Group Theory in Physics SOLUTION: We list some of the properties of the totally antisymmetric tensor in three dimensions that will prove useful in the following. ope = 6 open? = 25H Cope? = 6869 — 8L5h a oe 5, 8 By We make no distinction between lower and upper indices; they are placed wherever convenience with the summation convention dictates. The defining equation for J* can be inverted: fos ee ent = canes, = 25 Sou = WWJ, which gives: A= etm We are now ready to compute the commutators: [ee] = ce (0, Jal = ieee ay Jy = ieMa(gmgn — ams") J, F (ems S etn) 5 (eens i etme _ J?) Carrying out the contractions, using the antisymmetry of J* and performing the remain- ing summations, we obtain: (a, i] zy (aman a ae saps) & Problem 7.5 Using the group algebra of §0(3), verify that (J?) =0 for? =1,2,3. SoLutTion: The Casimir operator J? can be written as J? = J'A,, with again no distinction between upper and lower indices. The commutator of J? with the three generators of the algebra becomes: Pd) = Ue] = JM A+ (dil = J (tex Jn) + ie’? Jj) de = teu” (Jn + Ind")Rotations in 3- Dimensional Space—The Group $0(3) 49 where the second step follows by adding and subtracting J*JjJ,, and the final step involves renaming the dummy indices i, j. Now interchange & and m. The second factor is symmetric in these two indices, while the antisymmetric tensor gives a minus sign. However, these are dummy indices; we can call them anything we want, as long as they are summed over the same range. Therefore: en” (J*Jn + Jn d*) = -cu" (dnt* + JJ) from which follows that (PJ =0 a Problem 7.6 Show that the three matrices of By, (7.9-28) are related to those of Bg. (7.29) by a similarity transformation. Ezpress the “canonical” basis veetors (on which the first set of matrices are based) in terms of the “Cartesian” basis vectors é;, i = 1,2,3 on which the second set of matrices are defined. SOLUTION: Observe that the canonical realization of Ja is diagonal. Therefore if the two sets of matrices are related by « similarity transformation, it must be the one that diagonalizes Jy. The matrix realization of Jy given in Eq. (7.2-3) has characteristic polynomial MM-1)=0 and three eigenvalues: 1,0,—1. The corresponding eigenvectors are realized as: wali) =()--ald) The required transformation is: l -1 001 1 -li @ Sa=—]-i 0 =i with = S7=—/] 0 0 v2 Valo vi 0 va\1 io ‘We can now obtain the matrices in the canonical realization by forming the products $71.5, which indeed yield Eq. (7-3.22). The relation between the two sets of basis vectors is given by the expressions for the eigenvectors of Jy: a ik) = J (Fa - 2) and 0) = 50 (Group Theory in Physics Problem 7.7 (i) From the definition of the canonical basis vectors and the Lie Algebra of SO(3) show that U[Ra(m)|Lim) = i — my, where [nF = 1, and tar = — 145 fii) Using '/2 = 1 which follows from Eg. (7.9-20), prove that ni = 1 fer all j by mathemai induction; (iti) Combine the above results to derive the useful formula: Di[Raz)In = (-1) "os (iv) Use this formula to derive the explicit expression for D*|R,(x)]. SoLution: (i) By Theorem 7.2 we have: U[Ba(m)]Jol!"[Ra(e)] = Js (+) Multiplying both sides on the right by U[Ry(r)], recalling that U' is unitary, and applying the result on |jm), we get: Js (U[Ra(x)]Lim)) = U[Ra()] (JsLim)) or, since |jm) is an eigenvector of Js Jy (U{Ra()}lgrm)) = —m (U[Ra(e)\Lm)) ‘We recognize this as a statement that the vector U/[.R2(x)]|jm) is an cigenvector of J5 with eigenvalue —m. The only such vectors are multiples of the eigenvector |j —m); therefore: U[Ra(e)lim) = bi — mjnt, ‘We can constrain the modulus of n!, by computing the matrix elements of (+), as follows: Gjm|Jsljn} = — (ml [Ra( x) JgU'[Ra(x)] Lim) m(jmlim) = -G — mili — (na) "om 512 —(-m)(j — my — m) |nf,] Assuming that the eigenvectors of J; are properly normalized, this leads to: bat By Theorem 7.2 again: U[Rale]}J2U'[Ralx)] = —Je and thus: U[Ri(r\Jelim) = —J,U (Fale )}iim) URAL mz VY +1)—m(m £1) = —Jgli— mpi,Rotations in 3-Dimensional Space—The Group $0(3) $1 which, in turn, implies: Gi — mF Ita = —li —m = 1), Cancelling the common factors, leaves us with: Teast = (ii) The condition that |y!/2| = 1 follows directly from Eq. (7,3-20) and can be thought of as another statement of the Condon-Shortley convention. We wish to show that the same condition holds for arbitrary 7. Assume that |yj| = 1, or, equivalently: URN) = Li 3) In order to proceed inductively, consider the direct product state: weft 5 lin@ls5) =(j+Li4+1) an equality that follows from Eq, (7.7-6). The action of U[.A,()] on this state is given by: UlPal i +15 +1) U T d 1 si = ecu ml t el with no overall phase factor, Therefore Vi (iii) Combining the above result and the fact that 71.4, = —n#, which we established earlier, we get: mat ‘The defining relation for D/|Ft(x)] is: U[Ra(x)]lim) = [irw') D)[Ra(m)] which implies . . (= mom = Lien)? Ral) Tn view of the orthonormality property of the |jrn): DPURa =) = nit or, finally ; . ; Dialer yin = (-1P°" 8%, (iv) Observe that ; | | ULRi(n)] = oP eine 48-5 52 Group Theaery in Physics and so the representation matrix for this rotation is given by the following matrix element: dim UR (e lfm) = Gime ULRa(r)eF |jm) ef (im![L[Ra(x)|imjen™ afin yiom gn " DLR («IN Problem 7.8 Verify that the vector in By, (7.7-6) is an eigenvector of J? with eigenvalue G+ FG +77 +1). SOLUTION: The generators of the Lie algebra on the direct product space are defined by Eq. (7.7-4) aa: J = igy +195! ‘These generators are guaranteed to satisfy the same Lie algebra as the original ones, but it in instructive to actually compute the commutators and see exactly how the above definition accomplishes this. We wish to verify that UF gp Fm eg? Indeed: (ae? PY] = (eel + leaner + tea) — (AOI + 18I]( 1 + 185) = (hOr ASI) + (AO IS} + (1BIj)(@1') + (18 J))(184) — (A814 @1’) - (@1 18) — (1BS)(eO1') — (1B5, (1B; = (Jedi) @1' + 1@(Sid}) - (d)@U - LA(IR) = (Jn, J@1' + 100, 3] = teu™ (lm @1' + 18J,,) = jews where we have used standard properties of the direct product, i.e. linearity, the distribu- tive property, etc. The interpretation of J? on the direct product representation becomes simple. Using Eq. (7.3-6), we get (sii? = (ae 4 + ee psRotations in 3-Dimensional Space—The Group 50(3) 53 Let J? operate on [j7) 1/7"): PUSS) (J@1' +1@Jj)(O1' + LOI") + (s@1' +1045) 57) L's") EF RO Helis) » = biel’) (0 +57+ G47) WBVTNGING+7 +) which shows that |jj)@|j’j') is an eigenvector of J? with eigenvalue (j + 3")(7 +7’ + 1). a Problem 7.9 If {T;;:i,j = 1,2,3} are components of a second rank tensor, show that (i) TT = &T, is invariant under $0(3); fii) Ty = (Ty — Tya)/2 remains antisymmet- ric after an SO(3) transformation, and T, = ¢"T,;/2 transform as a vector; and (it) Ty = (Ty +Tye)/2 remains symmetric under an 50(3) transformation, the 5 independent components of T transform as D? under rotations. SoLution: Let é,, @9 and és be a Cartesian basis for the 3-dimensional vector space Vs. Second rank tensors over this space form @ 9-dimensional vector space, denoted by V4@V5 or V3, and spanned by the vectors ¢;@4;; an arbitrary element of V2 can be written as: T=T!smé; One must be careful to distinguish between T, a geometrical object, and its components T? which depend on the choice of basis for the tensor space. Once again, we make no distinction between upper and lower indices. Let T be a second-rank tensor. Under an 50(3) transformation: T+ rerte Recall that all our transformations are active, in accord with the footnote on page 3 of the book. (i) Compute the trace of T’: Thr ‘Tr(RTR™} Tr(TRR) = DT i where the second equality follows from the cyclic property of the trace, (ii) We anticipate this result in view of Lemma 5.1, which states that GL(n}—and, in particular, SO(n) C GL(n}—transformations are "symmetry preserving”. All RB € SO(3) obey R-? = RT, or, in terms of components: (Ray = Bi; 54 Group Theory in Physics Define the antisymmetric tensor T by ty Under the action of an $0(3) transformation, the image of T has components: Ty = (RTRs ; (Rit Tu Ft; — Rs"Tn Ri) = 3(%-%) which is indeed antisymmetric. A rank two antisymmetric tensor has three independent components. We must now show that these transform under rotations as the components of a vector. This is most conveniently done in terms of the “dual” of T, denoted by *T, and defined by: 7 (eT) Se Tye We must compute the image of ¢T under an $O(3) transformation : agp OM, = gee’ = SEHR Ti FO; Multiply the right-hand side by 6", = R°,.R*,, which follows from the orthogonality of Ai ef. Eq. (7.1-3)]. Rename the first subscript of the antisymmetric tensor to n and sum over it: 1 CM = oo Rei Ti Rj zee FT "RE He Using Eq. (7.1-6) we get: Th = seein = (+T Ae which is the transformation law for a vector. ' (iii) First we show that T, defined by: ‘eT is netually 8 paeudo-vector. This distinction will be made clear when we consider O(3) ‘transformetions.Rotations in 3-Dimensional Space—The Group 50(3) 55 remains symmetric under an SO(3) transformation. The image of T has components: i, = ; (RAT WR + Ry"Tank") = 3 (%+%) which is symmetric. Observe that another way of stating this result is to say that the subspace of symmetric 2-tensors is invariant under SO(3), We must now show that a tracelesa symmetric tensor transforms under SO(3) trana- formations according to the spin 2 representation. Our approach to this will consist of the following steps: « We will display a basis for this invariant subspace. © Observe that the full tensor space is constructed by taking the direct product Va@V5. It therefore serves aa a representation space for the 1 x | representation, which is reducible according to: D“'= Papen We will construct a basis for the J = 2 irreducible subspace. © And finally, we will show that our basis for the invariant subspace of traceless sym- metric tensors is a linear combination of the basis elements of the J = 2 irreducible subspace. ‘We will then conclude that the two subspaces generate equivalent representations, which will complete the proof. STEP 1: It easy to show that the following tensors form a basis for the subspace of traceless symmetric tensors [Prove!]: 4. = $(é1@4 ~ 468) ey os Hés94 - 408) én = 7léi@4 + 204) dua a= 7l41@4 +604) gay SS He@és + é58%2) STEP 2: We now construct the J = 2 sector of the 1 x 1 representation. We know that the eigenvector of J; with eigenvalue Mf = 2 is given by: (22) = [Ly 1) @|L, 2) as follows from Eq. (7.7-6). We could proceed by operating with J_ repeatedly—as described in Sec. 7.7—and thus obtain the other four eigenstates. However, this is a 56 Group Theory in Physics lot more work than necessary, since Appendix V contains a table of the Clebsch-Gordan coefficients for this reduction. The results are: (2,2) = [1,1)@/[1,1) Bt) = g(ien.o) + 1,0)eH1,1)) 2,0) = Je qustyot,—1) + ih -a)@0,1))+ 2 0,0)00,0 y-1) = Zp(lto@lt,-1) + 1,—1)@H1,0) 2) = [1,—1)@f1,-1) STEF 3: We must now establish the connection between the |2,m) and our basis for the subspace of traceless symmetric tensors. To this end, recall that in Problem 7.6 we showed that the Cartesian and canonical bases are related by: Ni £1) = Jp (Fé —ies) and 1,0) = 8s All that is left to do is to compute the five |2,m} by performing the indicated tensor products. As an example, consider: (2,2) (1,1) @1, 1) 1 ox lo. = ala — ie Bla — i83) Dig pei coded a dnd snd = [ér@8i — 2242 + 1(21:@e: + 2.@2;,)) = ba +83} It ia an interesting exercise to go back to part (ii), carry out the construction of the J = 1 sector of the direct product representation and identify the eigenstates of J that correspond to the three linearly independent antisymmetric tensors. The latter ones are in 1-1 correspondence with their duals and allow us to construct the duals of the eigenvectors of Jy. There is an interesting pattern there, well worth the effort of identifying it. a Problem 7.10 Prove that on the space of multicomponent functions {"; m= —j,...,j} the mapping given by Eg. (7.6-9) (in Definition 7.9) forms a representation of the sym- metry group SO(3). [Hint: pattern the proof after that given in Example 7, Sec. 9.1.] SoLuTion: Under the action of R € 5O(3), the function y transforms according to Eq. (7.69): ey with ws) := DR eZ) ‘We must check that this map is a homomorphism. Let p By yRotations in 3-Dimensional Space—The Group SO(3) 57 and of A ye Then, under the composition of maps > —+ p", we have: we) = DLR R5'2) = DLR)" [Ry] (Re Rg*2) = DLR) ey 2] and indeed this map is a homomorphism. aChapter 8 The Group SU(2) and More About §0(3) Problem 8.1 Verify that U(¢,0,0)|pz,o) is an cigenstate of the lincar momentum op- erators P = (P,P, P,) with eigenvalues p which, when converted into spherical coordi- nates, give (p,@, 4). SOLUTION: Operate on [p,c) with P: Filp.2) FU(d,8,0)|pz, 0) = U(9,0,0) [0%(6, 0,0) (0(4,0,0))'] lp=,0) = U(d,0,0)Pj|pz,0)R'(d, 6,0); = U(9,0,0)Palpz, 2) R'(o, 4,0)? = |p,o)pht'(¢,0,0}} where the first equality is a definition; the second follows from the unitarity of the group operators; the third follows from the fact that the P; are Cartesian components of a vector operator; the fourth follows from the fact that the momentum of the state |pz,c) is entirely along the z-axis, therefore P, and P, annihilate it; and the fifth follows from the fact that |[pz,o) is an eigenstate of P, with the indicated eigenvalue. Using the expression for A{a, 8,7) obtained in Problem 7.1, we get: i Pip, a) = |p, o)sin@ cos ¢, sin @sin ¢, cod)” which is the required result. a Problem 8.2 Identify the conventionally defined properties of the spherical harmonics that lead to the precise relation between Yim(0,¢) and £M(¢,0,0) given by Ey. (8.5-10). SOLUTION: Eq. (8.5-10) reads: Yan(®¢) = f+ [ota 0,0)8" 59 60 Group Theory in Physics The constant ia chosen such that Eq. (8.69) is satisfied. In order to see the reason for the complex conjugation, observe that Eq. (7.3-16) implies: D6,0,0)% oc em (In the Condon-Shortley convention the d-matrix is real, therefore complex conjugation does not affect it), However, the conventional definition for the spherical harmonics (see any Quantum Mechanics text) is such that: Fim (8, 6) ox ef which makes complex conjugation necessary for the identification. i Problem 8.3 (i) Prove Eq. (8.6-4) explicitly, (ii) Can this equation be “inverted” to yield a formula which expresses ¥(8,¢) as a linear combination of terms such as Fim (8, d)Fiem(@, 4)? Why? SoLutrow: (i) From Eq. (7.7-16) we have: DUR) DI RY = pas (rum'(j7") JM) DR) (IN(G7)an') The summations over M and N are trivial due to Eq. (7.7-12). Setting n = 0 =n! we “ DARED Rg" = =2 (ramn’(33")J,m + mn!) DR)" (JC 5")00) Taking the complex conjugate of both sides, keeping in mind that in the Condon-Shortley convention the Clebsch-Gordan coefficients are all real: LET asT = SS (am'(gi")J,mm +m’) [DCRYS+™ | (J0(53")00) JMN Renaming the 7's to I's, using Eq. (8.5-10) and setting a = @, 8 = ¢, 7 = 0: ar an aati apy = (ram! (U}lyre +m’) 5 (8,4) = TE pee (# #)(L0( 0'}00) Fi, #)Viem (8, #) = a ea mnt + mY mans ® 8 OUUE 0} which is—the correct form of—Eq. (8-6-4).The Group SU(2) and More About $0(3) 61 (ii) Observe that Eq. (7.7-17) is of no use for our purposes, since the indices n and n! are summed over, whereas we would like to set them equal to 0 if we are to recover the spherical harmonics. However, a partial inverse will do. So start with Eq. (7.7-16) again: D (Ryn D? (RB) = ie {ram'(3j'}.M).D" (RYN (J N(Gs"Inn'} Premultiply both sides by (J'M'(jj")mm’) and sum over m and m': DY VM jam!) DI (RED? UAE = SM" ij!)mm’} (mm'(5j")J M) DD! (RYM (IN(js" nn’) JMNmm!' Carry out the summation over m and m! on the right-hand side using the orthogonality relation, Eq. (7.7-13): MG mm) DRED AY = 3 Sh6i.D! (RM (IN (;)nn’) JIMN LORIN ING nm) Set n= 0 = n', which forces N = 0 due to Eq, (7.7-12): X (J'M'i" nm!) D'(R)B DD” (AS = Be" (YM (J0(i")00) mam! Rename the free indices J’ and M‘ to J and M to obtain: JM (37")mm’) pee DY Ry =o Gane ws DUR Take the complex conjugate of both sides, let (jj") + (Il') and multiply by the normal- ization constants to get: 4n(2L +1) (JM(I)}mm') ¥ea(8, 3) =x WE Ar+H (OCHO) inl #¥rw (A, #) Problem 8.4 Derive the recursion formula for the spherical harmonics Yim(8,¢) which increments both | and m by +1. SoLvtion: From Eq, (8.6-7) we have: VOF 1 cos 6¥;,.(8, 4) = Co TAY snl 0,6) + ER mt Dy 0,4) 62 ‘Group Theory in Physics From Eq, (8:6-8): Yl(t—1)—m(m—D¥4gim(8.¢) = l* (5 — (m= I) cot ‘) Fierm—1(9, 4) while Eq. (8.6-6) gives: OF DTFE = mime Missle) = 4 = (F + (m4 1) 0k) Yr) Putting all these together we obtain the following result: VI +1 008 OYin( 8,4) = Agme”* (3 -(m- 1yeoed) ‘i-1,m—1(8, 3) wf — Bue * (& + (m+ 1) cot ‘) Vist mer(#, 4) where: _ t+ m)(i — m) Alum = (2 — DIG = 1) — m(m — 1] and (+m41}U-m+) Bum = \ (Oy ayll+ NL +2) ~ mim + Problem 8.5 Define the action of Jy on the symmetric tensors of Vy", and verify that {é'")) as defined by Fg. (8.1-23) do transform as canonical components of the j = n/2 irreducible representation of SU (8). SoLuTion: Let ve denote the subspace of Vf" that consists of symmetric tensors of rank n in two dimensions. We will generalize Js and J, to act on tensors of arbitrary rank, explicitly construct the canonical basis of the subspace of symmetric tensora, and then we will express the tensor defined by Eq. (8.1-21) as a linear combination of the canonical basis tensors. At that point, the required result will almost be self-evident. First, turn Jj and J, into tensor operators. The way to do this is a straight-forward generalization of Eq. (7.7-4). Define: Jg:= @18-+-O14+190---B14+...41919-- By nm terms n terms = terms and similarly for Jx: Jy := 19-91 @---@1+...+191@---OJs = terme mn terms 8 termeThe Group SU(2) and More About $0(3) 63 where we use a tilde over a symbol to denote its symmetric nature. This definition pro- duces tensor operators that satisfy the same Lie algebra as the original vector operators. It easy but somewhat tedious to verify that: [sis Ja) = 444 and [4d] = 2h which is the SU(2) Lie algebra. Observe that n = 2 is a special case of Problem 7.8 for appropriate choices of j and j’. An arbitrary element of SU(2) ia realized in our tensor space by the operator: O(a,8,7) = etal ith ads which ia a generalization of Eq. (7.2-9), and in particular: #(8) := 0(0, 8,0) = «#* Observe that for n = 1 this reduces to Eq. (8.1-18). We now turn to the construction of the canonical basis of the subspace of symmetric tensora. Before tackling the general case, let's look at n = 2. The subspace of symmetric tensors of rank 2 is given by: V7 = spar {el Jp (Hel-) + els). |-re1-)} Define j :=n/2 = 1, and let m run from —j to j in integer steps. Furthermore, identify tensors [jm) as follows: fi) = inel+) [t0) = valve) + ely] [1 =1) = [-)@l-+) Tt should be clear that the number of times a |+) (respectively |—)) appears is j +m (resp. j—m) and this observation constitutes the definition of the label m for a particular tensor. The normalization factor is: 1 na jtm ayo k (me — kylet is a binomial coefficient. [t arises from the combinatorics of the problem, i.e. (j2~)=(j20) 64 Group Theory in Phyaics is the number of basis tensors in V;* with |+} appearing j +m times and |—) appearing j —m times that must be combined to form a basia tensor for ¥>°. In order te show that our notation is more than suggestive, let’s compute the effect Jy on our basis tensors: Ji) (Jg@1 + Jo@1) |+) 8/4) (4@1)H) B14) + (18J5)]4) 014) (+3) em i} i which proves that [11) is an eigenvector of Jy with eigenvalue 1. In a manner similar to the above we can establish that all our chosen basis tensors are canonical, therefore we have constructed a basis for the subspace of rank-2 symmetric tensors which at the same time is aj = 1 representation of SU(2). The case of arbitrary n is not any more complicated. In general, then, define the 27 + 1 tensors: [jm) := 1 | Ss |w1e--e1ee1-20--21-) Gin) bo where S isa aymmetrizer (cf. Chapter 5), ie. it performs 4 sum over all the possible permutations of the symbols shown. These obviously form a basis for our subspace, and, furthermore, one can now show that: Salim) = |jm)m Jatin) = Lim) Vii +1) ~ mim £1) [Verify!] which proves that our construction yields the canonical basis of Vj’, (An easy way to verify the top equation is to observe that J; merely counts the difference between the number of times a |+) appeara and the number of times a |—) appears—which is always 2m—and divides by 2). Since these tensors form a basis, it follows that any rank-2 symmetric tensor can be written as a linear combination of our canonical tensors, In particular, let: € = £*|+) + €-|-) € Wj Then, the—obviously symmetric—tensor £@---@€ must be a linear combi- SL nation of our canonical basis tensors. Indeed: fe @E= TeV S ||H@-@l4)8|-}8---@L} n ™ itm i Tn order to write this in terms of our canonical basis, define: et yeepThe Group SU(2) and More About $0(3) 65 Then our symmetric tensor can be written in the much neater form: §8---BE = |jm) E" with summation over m from —j to j implied. We have shown that the ") as defined by Eq. (8.1-23) are the canonical components of the j = n/'2 irreducible representation of the SU(2) Lie algebra. Only our normalization is different—we have an extra /nl—but as this js common to all the canonical components we could have absorbed it in the definition of the canonical basis tensors, In any case, it does not affect the expression for the representation matrices, as we now proceed to verify. Applying #(8) on our symmetric tensor we get: (£@---@EY == F(8) [f@---@E] or, in our particular basis: Ljrm et = FB) mm) eo In order to obtain a relationship between the components, take the matrix element: (irm!|P( 8) arm) eo (ime Lim) eo") en It of, using the analogue of Eq. (7.3-17)}: em) = ware”) There are many reasons for approaching this problem in this particular way. For example, it will serve as a concrete example of how tensors of arbitrary rank behave under group transformations, something that will be more fully explored in Chapter 13. Another reason is that our construction is the heart of Schwinger’s uncoupled oscillator model for angular momentum. (For a reference to the original article, as well as a discussion of the model see J.J. Sakurai, Modern Quantum Mechanics, pp. 217-221). Problem 8.6 Evaluate the integral [e088 [a6 Vins, O¥in(A, 81¥rn(B, 8) SoLuTion: Let Tim [dcon8 [ab Vingl@, #¥in( 8, 8)¥om (84) Using Eq. (86-4) to eliminate the last two spherical harmonics we obtain: r= fOEMOP SY facoeo [do ¥ire0, 9) 9 (arn (I)! ma +) ¥ es ms mer( 8, 4)(L/0(10")00) u 1 V2 +1 66 Group Theory in Physics where we have already taken into account the selection rule Eq. (7.7-12). Interchanging the summation and integration: I = OE com rm! (L)L' m + ma") 1 4 rai! ott" )00) x [cosa [49 Vil, 4)¥irmim (6,8) Imposing the orthonormality condition, Eq. (8.6-9) eisnerey mml(UVE! 1 4 L I= Bob (I) mnt mn) ee EOI 00) 85, and finally: (21+ 1)(2F + 1) = (mm'(lN")L'm +m") ('0(tt")00) Problem 8.7 Derive the invariant integration measure in the angle-and-azis parametriza- tion of the 50/3) group. SOLUTION: First we must express an arbitrary SU(2) element in this parametrization, The operator that performs a rotation about the vector #—characterized by two angles @ and ¢—by an angle wp is given by Eq. (7.2-8) as: Ral) = This equation is true for both $0(3) and SU(2) since these two groups are locally iso- morphic. The difference lies in the range of the variables. Then all elements of SU(2) are of the form: DMM o)R = (m'je“44 my ‘We need to compute this matrix element. To this end, observe in turn: z ot et eS = cos Ody + [Fs += -|sino 2 2 therefore: uy a-J|4) = +i) + <-sin 8) Applying this operator once more we. get: (ii P14) = G4)The Group SU(2) and More About 50(3) 67 Using the definition of the exponential of an operator and the previous two equations: =ip 2 1 =ip gil) le) ++ (