18.

04 Practice problems exam 2, Spring 2018 Solutions

Problem 1. Harmonic functions

(a) Show u(x, y) = x3 − 3xy 2 + 3x2 − 3y 2 is harmonic and find a harmonic conjugate.
It’s easy to compute:

ux = 3x2 − 3y 2 + 6x, uxx = 6x + 6

uy = −6xy − 6y, uyy = −6x − 6

It’s clear that ∇2 u = uxx + uyy = 0, so u is harmonic.

If v is a conjugate harmonic function to u, then u + iv is analytic and the Cauchy-Riemann
equations tell us that vx = −uy and vy = ux . Therefore, we can integrate ux and uy to find
v.

vx = −uy = 6xy + 6y ⇒ v = 3x2 y + 6xy + g(y)

vy = ux = 3x2 − 3y 2 + 6x ⇒ v = 3x2 y − y 3 + 6xy + h(x)

Comparing the two expressions for v we see that g(y) = −y 3 + C and h(x) = C. So
v = 3x2 y − y 3 + 6xy + C .
(b) Find all harmonic functions u on the unit disk such that u(1/2) = 2 and u(z) ≥ 2 for
all z in the disk.
answer: The only possibility is the constant function u(z) ≡ 2. The maximum principle
for harmonic functions says that if u takes a relative maximum or minimum at an interior
point then it is constant. (This is a consequence of the mean value theorem.)
(c) The temperature of the boundary of the unit disk is maintained at T = 1 in the first
quadrant, T = 2 in the second quadrant, T = 3 in the third quadrant and T = 4 in the
fourth quadrant. What is the temperature at the center of the disk
answer: The mean value theorem says that f (0) is the average over any circle centered at
0. This is clearly the average of the (constant) values in each quadrant. So f (0) = 2.5.
(d) Show that if u and v are conjugate harmonic functions then uv is harmonic.
answer: Easy method. We know f = u + iv is analytic. Therefore f 2 = u2 − y 2 + 2iuv
is also analytic. So, Im(f 2 ) = 2uv is harmonic. QED
Calculation method.

(uv)xx = uxx v + 2ux vx + uvxx

(uv)yy = uyy v + 2uy vy + uvyy

We know u, v are harmonic and satisfy the Cauchy-Riemann equations ux = vy , uy = −vx .

So adding the above equations we get

(uv)xx + (uv)yy = (uxx + uyy )v + 2(−ux uy + ux uy ) + u(vxx + vyy = 0.

We have shown that uv is harmonic.

1
18.04 Practice problems exam 2, Spring 2018 Solutions 2

(e) Show that if u is harmonic then ux is harmonic.

answer: Easy method. For some conjugate v, f = u+iv is harmonic. Since f 0 = ux +ivx ,
we know Re(f 0 ) = ux is harmonic.
Direct calculation (ux )xx + (ux )yy = (uxx + uyy )x = 0.
(f ) Show that if u is harmonic and and u2 is harmonic the u is constant.
(We always assume harmonic functions are real valued.)
answer: We calculate this directly.
(u2 )xx = 2(ux )2 + 2uuxx , (u2 )yy = 2(uy )2 + 2uuyy .
Assume that u and u2 are harmonic, then
0 = (u2 )xx + (u2 )yy = 2((ux )2 + (uy )2 ) + 2u(uxx + uyy ) = 2((ux )2 + (uy )2 ).
As a sum of squares, (ux )2 + (uy )2 = 0 implies ux = uy = 0. This implies u is constant.
QED.

Problem 2.
1
Let f (z) = . Find Laurent series for f on each of the 3 annular regions
(z − 1)(z − 3)
centered at z = 0 where f is analytic.
answer: The poles are at z = 1 and z = 3. This divides the plane into 3 annular regions,
with f analytic on each region:
A1 : |z| < 1, A2 : 1 < |z| < 3, A3 : 3 < |z|.
Im(z)

A3

A2
A1
Re(z)
1 3

Using partial fractions we get f (z) = − 12 · z−11

+ 12 · z−3
1
. We write each of these terms as
geometric series in each region.
1 1
=− = −(1 + z + z 2 + . . .) (converges for |z| < 1)
z−1 1−z  
1 1 1 1 1
= = 1 + + 2 + ... (converges for |z| > 1)
z−1 z(1 − 1/z z z z
1 1
=− = −(1 + (z/3) + (z/3)2 + . . .) (converges for |z| < 3)
z−3 3(1 − z/3)
3 32
 
1 1 1
= = 1 + + 2 + ... (converges for |z| > 3)
z−3 z(1 − 3/z z z z
18.04 Practice problems exam 2, Spring 2018 Solutions 3

On each region we can add the appropriate form of these series.

On A1 : |z| < 1:
∞  
1 2

1 1 2

1X 1
f (z) = 1 + z + z + ... − · 1 + z/3 + (z/3) + . . . = 1 − n+1 z n .
2 2 3 2 3
n=0

On A2 : 1 < |z| < 3:

  ∞ ∞
1 1 1 1 1 2 1X 1 1X
(z/3)n .


f (z) = − · 1 + + 2 + . . . − 1 + z/3 + (z/3) + . . . = − −
2 z z z 6 2 zn 6
n=1 n=0

On A3 : |z| > 3:
∞
3 33 1 X (−1 + 3n−1 )
   
1 1 1 1 1
f (z) = − · 1 + + 2 + ... + 1 + + 2 + ... = .
2 z z z 2z z z 2 zn
n=1

Problem 3.
Find the first few terms of the Laurent series around 0 for the following.
(a) f (z) = z 2 cos(1/3z) for 0 < |z|.
answer: Using the known series for cos(z) we get
 
1 1 1 1
f (z) = z 2 1 − + − . . . = z2 − + − ...
2! · 32 z 2 4! · 34 z 4 2! · 32 4! · 34 z 2

1
(b) f (z) = for 0 < |z| < R. What is R?
ez
−1
answer: Writing out ez as a power series we have
1 1 1
f (z) = = ·
z+ z 2 /2! 3
+ z /3! + . . . z 1 + z/2! + z 2 /3! + . . .

For z near 0 the expression z/2! + z 2 /3! + . . . is small so we can use the geometric series:
1
1 − (z/2! + z 2 /3! + . . .) + (z/2! + z 2 /3! + . . .)2 − (z/2! + z 2 /3! + . . .)3 + . . . .


f (z) =
z
It is hard to get a general expression for the terms of this series, but we can compute the
first few explicitly.
 1 z2

1 z 2 3 z 3
f (z) = 1 − + z (−1/3! + 1/4) + z (−1/4! + 2/(2!3!) − 1/8) = 1− + − 0 · z + ...
z 2 z 2 12

Problem 4.
∞
X zn
What is the annulus of convergence for .
n=−∞
2|n|
18.04 Practice problems exam 2, Spring 2018 Solutions 4

answer: We find region for singular and regular parts seperately.

∞
X 1
Singular part: . Either by recognizing this as a geometric series or using the ratio
2 zn
n
n=1
test we see it converges if |z| > 1/2.
∞
X zn
Regular part: . Either by recognizing this as a geometric series or using the ratio test
2n
n=0
we see it converges if |z| < 2.
The annulus of convergence is 1/2 < |z| < 2 .

Problem 5.
Find and classify the isolated singularities of each of the following. Compute the residue at
each such singularity.
z3 + 1
(a) f1 (z) =
z 2 (z + 1)
answer: f1 has a pole of order 2 at z = 0 and a apparently a simple pole at z = −1. (In
fact we will see that z = −1 is a removable singularity.)
z 3 +1
Res(f1 , 0): Let g(z) = z 2 f1 (z) = z+1 . Clearly we want the coefficient of z in the Taylor
0
series for g. That is Res(f1 , 0) = g (0) = −1 . (Alternatively we could have written 1/(z+1)
as a geometric series and found the coefficient of z from that.)
3
Res(f1 , −1): Let g(z) = (z + 1)f1 (z) = z z+1
2 . Res(f1 , −1) = g(−1) = 0. So the singularity
is removable. In retrospect we could have seen this because z + 1 = (z + 1)(z 2 − z + 1).
1
(b) f2 (z) = z
e −1
answer: f2 has poles whenever ez − 1 = 0, i.e. when z = 2nπi for any integer n. We’ll
show the poles are simple and compute their residues all at once by computing lim (z −
z→2nπi
2nπi)fz (z).
z − 2nπi 1
Res(f, 2nπi) = lim (z − 2nπi)fz (z) = lim z
= 2nπi = 1.
z→2nπi e −1
z→2nπi e
(The limit was computed using L’Hospital’s rule.) Since the limit exists the pole is simple
and the limit is the residue.
(c) f3 (z) = cos(1 − 1/z)
answer: f3 has exactly one singularity, which is at z = 0. We’ll find the residue by
computing the first few terms of the Laurent expansion.
ei(1−1/z) + e−i(1−1/z) ei e−i/z + e−i ei/z
cos(1 − 1/z) = = .
2 2
Using the power series for ew we have
 
i −i/z i i 1
ee = e 1 − − 2 + ...
z 2z
 
−i i/z −i i 1
e e =e 1 + − 2 + ...
z 2z
18.04 Practice problems exam 2, Spring 2018 Solutions 5

Looking at just the 1/z terms we have

−iei + ie−i
Res(f3 , 0) = = sin(1).
2

Alternatively we could have used the trig identity cos(1−1/z) = cos(1) cos(1/z)+sin(1) sin(1/z).

Problem 6.
(a) Find a function f that has a pole of order 2 at z = 1 + i and essential singularies at
z = 0 and z = 1.
answer: It’s easiest to write this as a sum.
1
f (z) = e1/z + e1/(z−1) + .
(z − 1 − i)2

The term e1/z has an essential singularty at z = 0. Since the other two terms are analytic
at z = 1, f has an essential singurity at z = 0.
The singularities at 1 and 1 + i can be analyzed in the same manner.
(b) Find a function f that has a removable singularity at z = 0, a pole of order 6 at z = 1
and an essential singularity at z = i.
answer: We’ll do this in the same way as part (a).
z 2 + 8z 1
f (z) = + + e1/(z−i) .
sin(z) (z − 1)6

Problem 7.
True or false. If true give an argument. If false give a counterexample
(a) If f and g have a pole at z0 then f + g has a pole at z0 .
(b) If f and g have a pole at z0 and both have nonzero residues the f g has a pole at z0 with
a nonzero residue.
(c) If f has an essential singularity at z = 0 and g has a pole of finite order at z = 0 the
f + g has an essential singularity at z = 0.
(d) If f (z) has a pole of order m at z = 0 then f (z 2 ) has a pole of order 2m
Answers.(a) False. Counterexample: f (z) = 1/z, g(z) = −1/z.
(b) False. Counterexample: f (z) = 1/z, g(z) = 1/z.
(c) True. When you add Laurent series you simply add the coefficients. The singular part
of the series for f has infinitely many nonzero coefficients. After a certain point, the singular
part of g has all zero coefficients. So after that point, the singular part of f + g has the
same coefficients as f . That is, it has infinitely many nonzero coefficients, so the singularity
is essential.
(d) True. We know f (z) = (z − z0 )m g(z), where g(z0 ) 6= 0. So, f (z)2 = (z − z0 )2m g(z)2 ,
where g(z0 )2 6= 0. This shows, f 2 has a 0 of order 2m.
18.04 Practice problems exam 2, Spring 2018 Solutions 6

Problem 8.
Find the Laurent series for each of the following.
(a) 1/e(1−z) for 1 < |z|.
answer: f (z) = 1/e(1−z) = ez−1 is analytic on the entire plane. So,

z2 z3
 
−1 z −1
f (z) = e e = e 1+z+ + + ...
2! 3!

is the Taylor series for all z. Hence it is the Laurent series on |z| > 1.

Problem 9.
1 1 2z
Let h(z) = − + 2 in the disk |z| < 2π.
sin(z) z z − π 2
(a) Show that all the apparent singularities are removable.
(b) Find the first 4 terms of the Taylor series around z = 0.
Answers.(a) The apparent singularities of h are at 0, ±π. There might be a slicker way
to do this part, but here’s one that’s not too painful. Write
z − sin(z) 2z
At z = 0: h(z) = + 2 . The second term is analytic, so doesn’t contribute
z sin(z) z − π2
to the singularity at 0. Writing out the first term in terms of Taylor series we have

z − sin(z) z 3 /3! − z 5 /5! + . . .

=
z sin(z) z 2 − z 4 /3! + . . .

The numerator has a zero of order 3 and the denominator one of order 2, so the entire term
has a 0 of order 1, i.e. the singularity is removable.
We can play the same game at z = π. To make things easier we use partial fractions

1 1 1 1 (z − π) − sin(z) 1 1
h(z) = − + + = − + .
sin(z) z z − π z + π (z − π) sin(z) z z+π

The second and third terms are analytic at z = π, so don’t contribute to the singularity.
The first term can be written as
(z − π) + −(z − π) + (z − π)3 /3! − (z − π)5 /5! + . . .


(z − π) − sin(z) (z − π)3 /3!
= =
(z − π) sin(z) (z − π) (−(z − π) + (z − π)3 /3! − (z − π)5 /5! + . . .) (z − π)2 (−1 + (z − π)2 /2 + . . .)

As before, the numerator has a zero of order 3 and the denominator one of order 2, so the
singularity is removable.
The singularity at z = −π is handled identically to z = π.
18.04 Practice problems exam 2, Spring 2018 Solutions 7

(b) For this part let’s work on each term in the original expression of h.
1 1 1
=
sin(z) z 1 − (z /3! − z 4 /5! + . . .)
2

1
1 + (z 2 /3! − z 4 /5! + . . .) + (z 2 /3! − z 4 /5! + . . .)2 + . . .


=
z
1
1 + z 2 /3! + z 4 (−1/5! + 1/(3!)2 ) + . . .


=
z
z2 7z 4

1
= 1+ + + ...
z 6 360

2z 2z 2z
=− 2 = − 2 (1 + z 2 /π 2 + z 4 /π 4 + . . .)
z2 −π 2 2 2
π (1 − z /π ) π
Combining all the parts we get
1 z 7z 3 1 2z 2z 3
h(z) = + + + ... − − 2 − 4 + ...
z 6 360  z π π
1 2 3 7 2
=0+z − +z − + ...
6 π2 360 π 4

Problem 10.
Find the residue at ∞ of each of the following.
(a) f (z) = ez
z−1
(b) f (z) = .
z+1
Z
Answers.(a) Easy method: f (z) is entire so f (z) dz = 0 for all closed C. Since the
C
residue at infinity is minus the integral over a closed curve containing all the singularities
we must have Res(f, ∞) = 0.
1 1/w
Method 2. Let g(w) = w2
e . The Laurent series for g is
 
1 1
g(w) = 2 1 + + . . . .
w w
So Res(f, ∞) = − Res(g, 0) = 0.
1 1 1/w − 1
(b) Since Let g(w) = 2
f (/1/w) = 2 . Writing 1/(w + 1) as a geometric series
w w 1/w + 1
we get
1 1 1 2
g(w) = 2
(1 − w)(1 − w + w2 − w3 + . . .) = 2 (1 − 2w + 2w2 − . . .) = 2 − + 2 − . . .
w w w w
Therefore Res(f, ∞) = − Res(g, 0) = 2.

Problem 11.
Use the following steps to sketch the stream lines for the flow with complex potential Φ(z) =
z + log(z − i) + log(z + i)
18.04 Practice problems exam 2, Spring 2018 Solutions 8

(i) Identify the components, i.e. sources, sinks, etc of the flow.
(ii) Find the stagnation points.
(iii) Sketch the flow near each of the sources.
(iv) Sketch the flow far from the sources.
(v) Tie the picture together.
answer: (i) The log terms with positive coefficients represent sources. The term z represents
a steady stream. So this is two sources in a steady stream.
(ii) Stagnation poihts are places where Φ0 (z) = 0. Computing:
2z (z + 1)2
Φ0 (z) = 1 + = 2 .
z2+1 z +1
So there is a single stagnation point at z = −1.
(iii-v) Near the sources the flow looks like a source. Far away it looks like uniform flow to
the right. By symmetry (or direct calculation) there are streamlines on the x-axis. We get
the following picture. (I used Octave to draw draw the underlying vector field.)

Problem 12.
Compute the following definite integrals
√
Z π
1
(a) 2 dθ. (answer: π 2)
−π 1 + sin (θ)
Z ∞
x
(b) 2 + 4x + 13)2
dx. (answer: −π/27)
−∞ (x
Z ∞
x sin(x)
(c) p.v. 2
dx.
−∞ 1 + x
Z ∞
cos(x)
(d) p.v. dx.
−∞ x + i
Z ∞
xe2ix
(e) I = p.v. 2
dx.
−∞ x − 1
18.04 Practice problems exam 2, Spring 2018 Solutions 9

z − 1/z z2 − 1
(a) answer: On the unit circle z = eiθ , sin(θ) = = . So the integral becomes
2i 21
1 dz −4z
Z Z
= dz
|z|=1 1 + ((z 2 − 1)/2iz)2 iz |z|=1 i(z 4 − 6z 2 + 1)

−4z
Let f (z) = . So the integral is
i(z 4 − 6z 2 + 1)
Z X
f (z) dz = 2πi residues of f inside the unit disk.
|z|=1
√ √
The poles of f are at z 2 = 3 ± 8. Of these, only 2
pz =√3 − 8 is inside
p the √
unit circle. So
there are two poles inside the unit circle at z1 = 3 − 8 and z2 = − 3 − 8. These are
simple poles and we can compute the residue using L’Hospital’s rule.

(z − z1 )(−4z) −4z1 −1 1
Res(f, z1 ) = lim 4 2
= 3 = 2 = √ .
z→z1 i(z − 6z + 1) i(4z1 − 12z1 ) i(z1 − 3) i 8
The residue at z2 has the same value. So,
Z
4π √
f (z) dz = 2πi(Res(f, z1 ) + Res(f, z2 )) = √ = π 2.
|z|=1 8

(b) answer: Call the integral in question I. Let f (z) = z/(z 2 + 4z + 13)2 . This decays
faster than 1/z 2 so we can use path

Im(z)

CR

−2 + 3i

Re(z)
−R C1 R
Z Z
We know lim f (z) dz = 0, so, letting R go to infinity in f (z) dz we get
R→∞ CR C1 +CR
X
I = 2πi residues of f in the upper half-plane.

The poles of f are at −2 ± 3i. Only z1 = −2 + 3i is in the upper half-plane. All we have
z
to do is compute the residue. Let g(z) = (z − z1 )2 f (z) = (z−(−2−3i)) 2 . Since g is analytic

at z1 we have
Res(f, z1 ) = g 0 (z1 ) = some algebra = i/54.
So I == −π/27.

(c) answer: Call the integral in question I. Replace sin(x) by eix and let
Z ∞
˜ xeix ˜
I = p.v. 2
dx, so, I = Im(I)
−∞ 1 + x
18.04 Practice problems exam 2, Spring 2018 Solutions 10

zeiz
Let f (z) = and use the contour C1 + CR .
1 + z2
Im(z)

2Ri CR
i
C1
Re(z)
−R R

The only pole of f in the upper half-plane is at z = i. It is easy to compute Res(f, i) =

ie−1 2i = e−1 /2. So, Z
f (z) dz = 2πi Res(f, i) = πie−1 .
C1 +CR

Since |z/(1 + z 2 )|
< M/|z| for Large z and the coefficient of ix in the exponent of f is
positive, we know
lim f (z) dz = 0.
R→∞
RR R∞
Also, limR→∞ ˜
−R f (z) dz = p.v. −∞ f (x) dx = I.
In conclusion we have
I˜ = 2πi Res(f, i) = πie−1 .
˜ = πe−1 .
So I = Im(I)

eix + e−ix eiz e−iz

(d) answer: Write cos(x) = . So, Let f1 (z) = z+i and f2 (z) = z+i .
2
∞ Z ∞ ∞
cos(x) 1 1
Z Z
p.v. dx = p.v. f1 (x) dx + p.v. f2 (x) dx.
−∞ x+i −∞ 2 −∞ 2
We compute these integrals using two different contours
Im(z) Im(z)

2R1 i C R1

C1 −R2 R2
Re(z) Re(z)
−R1 R1 C2

−i

−2R2 i −CR2

The reasoning is the same as in part (b). Both f1 and f2 have a single pole at z − i. So,
using the contour C1 + CR1 we find
Z ∞
1 1
p.v. f1 (x) dx = 2πi Res f1 in the upper half plane. = 0.
−∞ 2 2
Likewise, using the contour C1 − CR2 we find
Z ∞
1 1
p.v. f2 (x) dx = 2πi Res f2 in the lower half plane. = −2πi Res(f2 /2, −i) = −πie−1 .
−∞ 2 2
18.04 Practice problems exam 2, Spring 2018 Solutions 11

(The minus sign is because C1 − CR2 is oriented in the clockwise direction.)

Answer to problem: the integral is −πie−1 .

ze2iz
(e) answer: Since our integrand f (z) = has poles on the real axis we will need to
z2 − 1
use an indented contour.
Im(z)

2Ri
CR

−C2 −C4
C1 −1 C3 1 C5
Re(z)
−R −1−r1 −1+r1 1−r2 1+r2 R

As usual, we chose the contour so that the integral over CR goes to 0 as R goes to infinity.
Since f has no poles inside the contour we have
Z
f (z) dz = 0.
C1 −C2 +C3 −C4 +C5 +CR

The poles of f at ±1 are simple. So, letting R → ∞ and r1 , r2 → 0 we get

I = πi(Res(f, −1) + Res(f, 1)).

The residues are straightforward to compute.

Res(f, −1) = e−2i /2, Res(f, 1) = e2i /2.

So, I = πi(e2i + e−2i )/2 = πi cos(2) .