Complex Analysis Solutions: Polynomials, Singularities, and Laurent Series, Assignments of Mathematics

Solutions to problems related to complex analysis, including the holomorphicity of polynomials, the calculation of residues at singularities, and the laurent series representation of functions. The problems involve finding the maximum value of a function on the unit disk, evaluating limits, and determining the nature of singularities.

Typology: Assignments

Pre 2010

Uploaded on 10/01/2009

koofers-user-net
koofers-user-net 🇺🇸

3

(1)

10 documents

1 / 2

Toggle sidebar

This page cannot be seen from the preview

Don't miss anything!

bg1
Patrick Corn
Math 185, 7/26/05
Solution Set 9
Problem 1: Let f(z) = Qn
i=1(zai). Then fis a polynomial, and so it’s holomorphic every-
where. Note that |f(0)|= 1, so since the maximum of |f|over the unit disk is attained on the
boundary, there must be a point pwith |p|= 1 such that |f(p)| 1. And this is what we wanted.
Problem 2: Let Rbe a large enough real number, i.e. large enough that all the roots of qare
enclosed by the circle CR. Note that
ZCR
p(z)
q(z)dz
2πRMR
where MRis the maximum value of
p(z)
q(z)
on CR. Now let n= deg(q)deg(p), and note that
limz→∞
znp(z)
q(z)is a finite number. So lim supR→∞ RnMRis finite. Since n > 1, it follows that
limR→∞ RMR= 0, so that
lim
R→∞ ZCR
p(z)
q(z)dz = 0
But
ZCR
p(z)
q(z)dz =ZC
p(z)
q(z)dz
for large enough R, by Cauchy’s theorem for an annulus. So then RC
p(z)
q(z)dz = 0.
Problem 3: (a) The polynomial in the denominator has four distinct roots, equal to e2πik/5for
1k4. These are all simple poles. Also, it is not hard to see that is a simple pole:
f1
z=1
z·1
z4+z3+z2+z+ 1,
and the latter fraction is holomorphic near 0.
(b) It is not hard to see that the zeroes of si n zare precisely the integer multiples of π. They are
all simple zeroes, because the derivative of sin zdoesn’t vanish at any of them. Therefore z=
is a pole of order 2 (a double pole) for all kZ.
The case of is more interesting. Be careful–in fact is not an isolated singularity at all!
There is no punctured disk around contained in the region Gon which 1
sin2zis holomorphic,
because /Gfor all kZ.
(c) First of all, is a removable singularity, because sin zis holomorphic at 0. The only other
possible problem is at z= 0. We can write the Laurent series for sin 1
zin powers of 1
z:
sin 1
z=
X
n=0
(1)nz2n1
(2n+ 1)!
so 0 must be an essential singularity.
Problem 4: For the finite set of points F={z1,...,zn}, we can find a nonnegative integer pk,
1kn, such that (zzk)pkf(z) can be extended to a holomorphic function in a neighborhood
of zk(because each zkis either removable, so pk= 0, or a pole of order pk). Then consider
g(z) = f(z)
n
Y
k=1
(zzk)pk
1
pf2

Partial preview of the text

Download Complex Analysis Solutions: Polynomials, Singularities, and Laurent Series and more Assignments Mathematics in PDF only on Docsity!

Patrick Corn Math 185, 7/26/ Solution Set 9

Problem 1: Let f (z) =

∏n i=1(z^ −^ ai). Then^ f^ is a polynomial, and so it’s holomorphic every- where. Note that |f (0)| = 1, so since the maximum of |f | over the unit disk is attained on the boundary, there must be a point p with |p| = 1 such that |f (p)| ≥ 1. And this is what we wanted.

Problem 2: Let R be a large enough real number, i.e. large enough that all the roots of q are enclosed by the circle CR. Note that ∣ ∣∣ ∣

CR

p(z) q(z)

dz

∣ ≤^2 πRMR

where MR is the maximum value of

∣ p q((zz))

∣ on^ CR.^ Now let^ n^ = deg(q)^ −^ deg(p), and note that

limz→∞ z

n (^) p(z) q(z) is a finite number.^ So lim supR→∞^ R

nMR is finite. Since n > 1, it follows that

limR→∞ RMR = 0, so that

lim R→∞

CR

p(z) q(z)

dz = 0

But (^) ∫

CR

p(z) q(z)

dz =

C

p(z) q(z)

dz

for large enough R, by Cauchy’s theorem for an annulus. So then

C

p(z) q(z) dz^ = 0. Problem 3: (a) The polynomial in the denominator has four distinct roots, equal to e^2 πik/^5 for 1 ≤ k ≤ 4. These are all simple poles. Also, it is not hard to see that ∞ is a simple pole:

f

z

z

z^4 + z^3 + z^2 + z + 1

and the latter fraction is holomorphic near 0. (b) It is not hard to see that the zeroes of sin z are precisely the integer multiples of π. They are all simple zeroes, because the derivative of sin z doesn’t vanish at any of them. Therefore z = kπ is a pole of order 2 (a double pole) for all k ∈ Z. The case of ∞ is more interesting. Be careful–in fact ∞ is not an isolated singularity at all! There is no punctured disk around ∞ contained in the region G on which (^) sin^12 z is holomorphic, because kπ /∈ G for all k ∈ Z. (c) First of all, ∞ is a removable singularity, because sin z is holomorphic at 0. The only other possible problem is at z = 0. We can write the Laurent series for sin (^1) z in powers of (^1) z :

sin

z

∑^ ∞

n=

(−1)nz−^2 n−^1 (2n + 1)!

so 0 must be an essential singularity.

Problem 4: For the finite set of points F = {z 1 ,... , zn}, we can find a nonnegative integer pk, 1 ≤ k ≤ n, such that (z − zk )pk^ f (z) can be extended to a holomorphic function in a neighborhood of zk (because each zk is either removable, so pk = 0, or a pole of order pk). Then consider

g(z) = f (z)

∏^ n

k=

(z − zk)pk

1

Then g(z) is an entire function, and by assumption it has a non-essential singularity at ∞. As we have seen in class, this implies that g(z) is a polynomial. (It is equal to its Taylor series centered at 0, which can only have finitely many nonzero terms.) Thus f is a rational function.

Problem 5: We can write h(z) = (z − z 0 )h 1 (z) with h 1 holomorphic in a neighborhood of z 0 , and h 1 (z 0 ) 6 = 0. Then g(z) h(z)

= (z − z 0 )−^1

g(z) h 1 (z) and g/h 1 is holomorphic in a neighborhood of z 0 , hence representable by a Taylor series

n=0 an(z− z 0 )n^ there. Then the coefficient of (z − z 0 )−^1 in the Laurent series for g/h is equal to a 0 = (^) hg 1 ((zz^00 ) ).

But, as we have seen many times already, h 1 (z 0 ) = h′(z 0 ).

Problem 6: (a) There are q singularities of the form ζk = e^2 πik/q, 0 ≤ k ≤ q − 1. By the previous exercise, the residue at ζk is equal to z

p −qzq−^1 evaluated at^ ζk. This equals ζkp −qζ kq−^1

ζ kp+ q

e^2 πik(p+1)/q q (b) There are double poles at ±1, so we cannot use the previous exercise. Consider the function f 1 (z) = z^5 /(z + 1)^2. It has a Taylor series

n=0 an(z^ −^ 1)

n, and the Laurent series of our function

around 1 is equal to this Taylor series divided by (z − 1)^2. Then the coefficient of (z − 1)−^1 in that Laurent series will be equal to a 1. This equals f 1 ′(1), and now we must do a little computation to get f 1 ′(1) = 1, so the residue at z = 1 is 1. Similarly let f− 1 (z) = z^5 /(z − 1)^2. The residue at z = −1 will equal f (^) −′ 1 (−1) = 1. So both residues are 1. (c) There are two simple poles, ζ and ζ^2 , where ζ = e^2 πi/^3. We can use the previous exercise to find that the residue at ζk^ is the function 2 cosz+1^ z evaluated at ζk. Let’s just leave it like that:

resz=ζk

cos z 1 + z + z^2

cos ζk 2 ζk^ + 1

, k = 1, 2.

(d) There are simple poles at z = kπ, k ∈ Z, so we can use the above exercise to see that

resz=kπ

sin z

cos kπ

= (−1)k.

Problem 7: Let f (z) = (z − z 0 )mf 1 (z), where f 1 (z) is holomorphic in a neighborhood of z 0 and f 1 (z 0 ) 6 = 0. Then

f ′(z) f (z)

m(z − z 0 )m−^1 f 1 (z) + (z − z 0 )mf 1 ′(z) (z − z 0 )mf 1 (z)

m z − z 0

f 1 ′(z) f 1 (z)

and the latter fraction is a holomorphic function in a neighborhood of z 0 , so the principal part of the Laurent series of f ′/f is exactly (^) z−mz 0. So the residue is m, as desired.

Problem 8: First, if g(z) is identically zero, there’s nothing to prove. Otherwise, the zeroes of g(z) are isolated, as we’ve seen, so h(z) = f (z)/g(z) is holomorphic everywhere except for isolated singularities. Let z 0 be one such singularity; we see immediately that

lim sup z→z 0

|h(z)| < ∞

since |h(z)| ≤ 1 on a punctured neighborhood of z 0. Thus z 0 is a removable singularity. So all the (non-∞) singularities of h(z) are removable. So h(z) is actually an entire function. But it’s bounded! So by Liouville’s theorem it is constant.

2