You are on page 1of 7

Assignment 7

Arfken 6.4.1
Taking the contour C to be a circle of radius r centered at z0 , we can use rei = z z0 to calculuate
I

rn ein irei d
2


n+1

r
i(n+1)

n 6= 1
n+1 e

0
=i
2


n = 1
0

0 n 6= 1
=
2i n = 1

(z z0 ) dz =

Note that though we have taken C to be a circle, because (z z0 )n is analytic at all other points in the
plane, C could be deformed to be anything in the complex plane.
Arfken 6.4.2
Evaluate the contour integral with a circular contour, C of radius 1. By the argument from the previous
problem, we can always deform this contour to any in the plane encircling z = 0 (the single point of nonanalyticity). We have
1
2i

Z 2
1
rmn1 ei(mn1) rei id
2i 0
2

r mn i(mn)

e
n 6= m

i(mn)

1
0
=
2

2


n=m
0

0 n 6= m
=
1 n=m

z mn1 dz =

= mn
which is the Kronecker delta.
Arfken 6.4.6
The function f (z) is analytic on and in C. From the Cauchy integral formula, we have
f (z) =

1
2i

I
C

f (z 0 ) 0
dz
z0 z

Taking the derivative of this with respect to z, we get


f 0 (z) =

1
2i

I
C

f (z 0 )
dz 0
(z 0 z)2

We will assume that a similar result holds for the nth derivative:
I
n!
f (z 0 )
f (n) (z) =
dz 0
0
2i C (z z)n+1
1

Taking another derivative with respect to z (not z 0 !), we can establish this for n + 1
I
f (z 0 )
(n + 1)!
dz 0
f (n+1) (z) =
0
n+2
2i
(z

z)
C
Therefore since it is true for n = 1, assumed for n and shown for n + 1, by induction, this is true.
Arfken 6.4.8
We can use the Cauchy integral formula for the nth derivative to convert Rodrigues formulae into
Schlaefli integrals.
For the Legendre polynomials, we have
Pn (x) =
=
=
=
=

1 dn 2
(x 1)n
n! dxn
1
f (n) (x)
n
2 n!
I
1 n!
f (z 0 )
dz 0
n
0
2 n! 2i C (z x)n+1
I
2
(z 0 1)n
1
dz 0
2n 2i C (z 0 x)n+1
I
2
(1)n 1
(1 z 0 )n

dz 0
2n
2i C (z 0 x)n+1
2n

For the Hermite polynomials, we get


dn x2
e
dxn
2
= (1)n ex f (n) (x)
I
f (z 0 )
n x2 n!
= (1) e
dz 0
2i C (z 0 x)n+1
I
02
2 n!
ez
dz 0
= (1)n ex
2i C (z 0 x)n+1
I (t+x)2
e
n x2 n!
= (1) e
(dt)
2i C (t)n+1
I
2
n!
=
tn1 et +2tx dt
2i C
2

Hn (x) = (1)n ex

where

t = x z0

For the Laguerre polynomials, we get


ex
n!
ex
=
n!
ex
=
n!

Ln (x) =

dn n x 
x e
dxn

f (n) (x)
I
n!
f (z 0 )
dz 0
2i C (z 0 x)n+1
I
0
n
ex
z 0 ez
=
dz 0
2i C (z 0 x)n+1
which, if we use t = 1 x/z 0 , we can convert into a slightly more standard form
I
1
ext/(1t)
Ln (x) =
dt
2i C (1 t) tn+1
2

Arfken 6.5.3
Our assumptions are that f (z) is analytic and |f (z)| 1 for |z| 1 and that f (0) = 0. First, we have
from the Cauchy integral formula for C the unit circle
I
f (z)
dz = 2if (0) = 0
z
C
Because this will be true for all contours, C, inside the unit circle and encircling the origin z = 0, we conclude
from Moreras theorem that f (z)/z is analytic on |z| 1. Using the Cauchy integral formula again, we have

n
n
I 
f (z)
1
f (z 0 )
1
=
dz 0
z
2i C
z0
z0 z
where we again take C to be the unit circle. Taking magnitudes, we can finish our proof



n
I 

f (z) n 1
f (z 0 )
1
0
=

dz



0
0
z
2i C
z
z z



I

1 1
f
(z 0 )n 0

dz
2 z 0n max C z 0 z




1 1
2if (z)n max



0n
2 z max
n

|f (z)|max
1
where in the second line we have used a variant of the Darboux inequality (problem 6.3.2) and in the third
line we have used the fact that the power of an analytic function will also be analytic. Taking the nth root
and rearranging, we get our answer
|f (z)| |z|
Arfken 6.5.7
The function f (z) is analytic in a region that includes the real axis and is purely imaginary if z is real.
Because f (z) is analytic, we can express it in terms of a Taylor series around some point on the real axis, x0
f (z) =

an (z x0 )n

n=0

Being purely imaginary on the x-axis yields


f (x) = f (x) =

an (x x0 )n =

an (x x0 )n

n=0

n=0

which implies that the coefficients an are purely imaginary. For a general point off the axis, we have
f (z ) =
=

X
n=0

an (z x0 )n
an (z x0 )n

n=0

an (z x0 )

n=0

= [f (z)]

(b) For f (z) = iz = ix y, it is straightforward to see that f (z ) = iz = ix + y and f (z) = iz = ix y.


3

Arfken 6.5.8
The Laurent series for f (z) = (ez 1)1 about the origin is given by

f (z) =

an (z z0 )n

n=

1
2i

an =

I
C

f (z 0 ) dz 0
(z 0 z0 )n+1

Calculating the coefficients, an using the unit circle as C with z0 = 0, we have


an =

1
2i

1
2i

dz 0
z 0 n+1 (ez0

dz

1)

k1

Bk z 0
k!

dz 0
z 0 n+1 k=0
I

1 X Bk
kn11
=
z0
dz 0
2i
k! C

k=0

=
=

X
Bk

k!

k,n+1

k=0
 Bn+1
(n+1)!

n 1
n < 1

where we have used the series definition of the Bernoulli numbers and the result of problem 6.4.2 to define
the Kronecker delta in the last line.
The Laurent expansion is now

X
Bn+1 n
1
=
z
ez 1 n=1 (n + 1)!

B0
B2
+ B1 +
z +
z
2
z
1 1
+
= +
z
2 12
=

taking the first three terms.


Arfken 6.5.9
We want to prove that a Laurent expansion about a point z0 is unique. We can do this most easily by
induction. Assuming that

X
X
an (z z0 )n =
bn (z z0 )n
f (z) =
n=N

n=N

multiply both sides by (z z0 )N and set z = z0 . The result is


aN = bN
Continue this process by multiplying the sums by (z z0 )N , taking a derivative and setting z = z0 . This
results in
aN +1 = bN +1
4

Assume now that ak = bk for all k N with k > 0, we will show it for ak+1 = bk+1 . Because of this
assumption, we can cancel the first terms up to k + 1. The series then read

an (z z0 )n =

n=k+1

bn (z z0 )n

n=k+1

Taking k + 1 derivatives and setting z = z0 , we get

(k + 1)! ak+1 = (k + 1)! bk+1

and our proof via induction is done and the Laurent expansions are unique.

Arfken 6.5.10
For the function
f (z) = [z(z 1)]1 the Laurnet expansion about z = 1 for small values of |z 1|
P
is given by f (z) = n= an (z 1)n where an is found via a contour integral using a circle of radius r
centered at z = 1 (i.e. z 1 = rei ) as our contour C.
I
1
f (z 0 ) dz 0
2i C (z 0 z0 )n+1
I
1
dz 0
=
2i C (z 0 1)n+2 z 0
I
1
1
1
dz 0
=
2i C (z 0 1)n+2 1 (1 z 0 )
I
1 X
(1 z 0 )k
=
2i
(z 0 1)n+2
k=0 C
Z 2

1 X
(1)k rkn2
ei(kn2) riei d
=
2i
0

an =

k=0

(1)k rkn1 k,n+1

k=0


=

(1)n+1 n 1
0
n < 1

Thus the Laurent expansion is

X
1
=
(1 z)n
z (z 1)
n=1

This expansion will hold up to the next singularity at z = 0. Hence, the radius of convergence of this Laurent
expansion is |z| < 1.
(b) The Laurent expansion for this function around z = 1 but for |z 1| very large can be found in a similar
5

way,
an =
=
=
=
=
=

I
f (z 0 ) dz 0
1
2i C (z 0 z0 )n+1
I
1
dz 0
0
2i C (z 1)n+2 z 0
I
1
1
1
dz 0
0
n+2
0
2i C (z 1)
(z 1) + 1
I
1
1
1
dz 0
2i C (z 0 1)n+3 1 + (z 0 1)1
I
1 X
(1 z 0 )k
dz 0
2i
(z 0 1)n+3
C
k=0
Z 2

1 X
k kn3
(1) r
ei(kn3) riei d
2i
0
k=0

(1)k rkn2 k,n+2

k=0


=

(1)n n 2
0 n > 2

Thus the Laurent expansion is


2
X
1
(1 z)n
=
z (z 1) n=

X
n=2

1
(1 z)n

Arfken 6.6.1 A function has a pole of order m at z = z0 . Find the coefficient of (z z0 )1 , a1 .


The Laurent expansion for f (z) near z0 will be given by
f (z) =

an (z z0 )n

n=m

am
am+1
a1
+
+ +
+ a0 +
m
m1
(z z0 )
(z z0 )
z z0

To read off any given constant coefficient we need to eliminate all of the z dependance. For instance to get
am , multiply by (z z0 )m and set everything to z = z0 :


(z z0 )f (z)

= am
z=z0

To get higher order coefficients, in addition to the multiplication, we need to take derivatives, e.g.
am+1



d
=
(z z0 )f (z)
dz
z=z0

Finally, to get the residue, we take m 1 derivatives


a1



1
dm1
=
(z z0 )f (z)
(n 1)! dz m1
z=z0
6

Arfken 6.6.2 A function f (z) is a quotient of two everywhere analytic functions, f1 (z) and f2 (z). Further
there is a pole in f (z) where f2 (z0 ) = 0. f1 (z0 ) 6= 0 and f20 (z0 ) 6= 0. Find the residue.
Since f2 (z0 ) = 0, we can write f2 (z) = (z z0 )g(z) where g(z) is some function without a zero at z0 .
To verify this, take the derivative: f20 (z) = g(z) + (z z0 )g 0 (z). At z = z0 , this becomes f20 (z0 ) = g(z0 )which
by assumption is not zero. Thus f2 (z) has a single (simple) zero at z0 and thus f (z) has a simple pole at
z0 . We can then write f (z) as
f1 (z)
f (z) =
(z z0 )g(z)
The residue of this function is just f (z)(z z0 ) evaluated at z = z0 . Thus we get
f1 (z0 )
g(z0 )
f1 (z0 )
= 0
f2 (z0 )

a1 =

You might also like