Homework 1 MA20219 Analysis 2B Solutions
Homework 1 MA20219 Analysis 2B Solutions
Homework 1 MA20219 Analysis 2B Solutions
∂f x2 − x21 ∂f −2x1 x2
= 22 + ex1 sin x2 , = 2 + ex1 cos x2 .
∂x1 (x1 + x22 )2 ∂x2 (x1 + x22 )2
Since these exist and are continuous on the open set R2 \{0}, it follows from a theorem in the lecture
notes that f is differentiable on R2 \{0}.
Solution. a) We have
�� ��
∂f 0 f ([ 0t ]) − f ([ 00 ]) t−0
= lim = lim = 1.
∂x1 0 t→0 t t→0 t
∂f
By symmetry we have ∂x2 ([ 00 ]) = 1.
b) Let v = [ vv12 ] ∈ R2 be such that v1 v2 �= 0. Then
�
1 t �= 0,
f (tv) =
0 t = 0.
This function is not continuous at t = 0 and therefore it is not differentiable at t = 0. Hence the
directional derivative at 0 in the direction v does not exist.
Alternatively, by using the definition of directional derivative we have
f (tv) − f (0) 1−0
lim = lim ,
t→0 t t→0 t
which does not exist.
1
Solution. We have
∂f ∂f
= 2x1 + 3, = 1.
∂x1 ∂x2
Since these exist and are continuous on R2 it follows from a theorem in the lecture notes that f is
differentiable on R2 and therefore in particular in 0. The gradient of f at zero equals
� �
3 1 .
Changing bases from the standard basis to the given basis, we have that the matrix respresentation
of f � (0) with respect to the basis {[ 11 ] , [ 32 ]} is
� �
� � 1 3 � �
3 1 = 4 11 .
1 2
and �� ��
3t
� � �� f − f (0)
� 3 2t (3t)2 + 9t + 2t − 0
f 0; = lim = lim = 11,
2 t→0 t t→0 t
� �
which also gives that the asked for matrix representation is 4 11 .
2
Homework 2 MA20219 Analysis 2B solutions
20. Let X be a normed space, let f : D ⊂ X → R2 and let x0 ∈ X . Define for j ∈ {1, 2} the function
fj : D ⊂ X → R by � �
f (x)
f (x) = 1 .
f2 (x)
Show that f is differentiable at x0 if and only for all j ∈ {1, 2} the functions fj are differentiable
at x0 . Moreover, show that the relationship between the derivatives is:
� � �
� f1 (x0 )
f (x0 ) = � .
f2 (x0 )
Since limx→x0 g(x) = 0 we have limx→x0 gj (x) = 0 for j ∈ {1, 2}. Therefore
which are first order Taylor formulas for f1 and f2 . Therefore, f1 is differentiable at x0 with
f1� (x0 ) = A1 and f2 is differentiable at x0 with f2� (x0 ) = A2 .
Conversely, assume that f1 and f2 are differentiable at x0 . Then (8) holds with A1 = f1� (x0 ) and
A2 = f2� (x0 ). Combining these in a vector equation gives (7) which in �turn� gives (6) (where the
A1
condition on g follows from the conditions on g1 and g2 ). Defining A := A2 we have a first order
Taylor formula for f and we read off that A = f � (x0 ).
� �
2x01 x02 x201 + 3
.
− sin x01 ex02
Since all entries of the Jacobian matrix exist and are continuous on R2 , it follows from a result in
the lecture notes that f is differentiable on R2 .
1
� �
f1 (x)
22. A differentiable function f : D ⊂ R2 → R2 is called divergence-free if, writing f (x) = f2 (x)
, we
∂f1 ∂f2
have ∂x1 + = 0 on D. Define g := f ◦ h (with h the “polar coordinates to Cartesian coordinates
∂x2
transformation” from above).
(a) Determine necessary and sufficient conditions on the partial derivatives of g for f to be
divergence-free.
(b) Show that g : (0, ∞) × (−π, π) → R2 defined by
�� �� � − sin(ϕ) �
r r
g = cos(ϕ)
ϕ
r
corresponds to a divergence-free f .
so that the Jacobian in zero equals J := [ 01 00 ]. The Hessian matrix of the first component of f is
� �
0 cos x2
,
cos x2 −x1 sin x2
2
�0 1
�
which in zero equals H2 := 1 −1 . For the first component of the second order Taylor polynomial
we then have
�� �� � � �� �� � � ��
� � 1 1 0 1 x1 x1 1 x2 x
f1 (0) + 0 0 x + �H1 x, x� = , = , 1 = x1 x 2 .
2 2 1 0 x2 x2 2 x1 x2
For the second component of the second order Taylor polynomial we have
�� � � � � ��
� � 1 1 0 1 x1 x
f2 (0) + 1 0 x + �H2 x, x� = 1 + x1 + , 1
2 2 1 −1 x2 x2
�� � � ��
1 x2 x 1
= 1 + x1 + , 1 = 1 + x1 + x1 x2 − x22 .
2 x1 − x2 x2 2
3
Homework 3 MA20219 Analysis 2B solutions
Solution. We first note that f is polynomial and therefore twice differentiable so that the results
from the lecture notes apply.
The gradient equals � �
6x21 − 150 + 6x22 12x1 x2 − 9x22 .
Solving 12x1 x2 − 9x22 = 0 shows that either x2 = 0 or x2 = 43 x1 . Substituting the first possibility
into 6x21 − 150 + 6x22 = 0 gives x1 = ±5. Substituting the second possibility into� 6x21�− 150 + 6x�22 = �0
−3
gives x1 = ±3. Therefore the points where the gradient equals zero are [ 50 ], −5 3
0 , [ 4 ] and −4 .
The Hessian matrix equals � �
12x1 12x2
.
12x2 12x1 − 18x2
At the point [ 50 ] this equals � �
60 0
,
0 60
which is positive definite by Problem 24 (since the determinant is 3600, which is positive and �60 is�
positive), so that [ 50 ] is a local minimum by a theorem from the lecture notes. At the point −5
0
the Hessian matrix equals � �
−60 0
,
0 −60
� �by Problem 24 (since the determinant is 3600, which is positive and −60
which is negative definite
is negative), so that −5 0 is a local maximum by a theorem from the lecture notes. At the point
[ 34 ] the Hessian matrix equals � �
36 48
,
48 −36
which is neither positive nor negative semi-definite by Problem 24 (since its determinant is negative).
Therefore
� −3 � [ 34 ] is not a local extremum by a theorem from the lecture notes. Similarly
� −3 � at the point
−4 the Hessian matrix is neither positive nor negative semi-definite. Hence −4 is not a local
extremum.
34. Let X and Y be normed spaces and let B : X × X → Y be a continuous bilinear function. Define
f : X → Y by
f (x) := B(x, x).
Show that f is differentiable on X and that for x0 ∈ X
1
B(x−x0 ,x−x0 )
Hence, defining g(x) := �x−x0 � , we have
Since limx→x0 √x−x0 = 0 we have by continuity of B that indeed limx→x0 g(x) = B(0, 0) = 0.
�x−x0 �
We therefore see that the conditions in the definition of derivative are fulfilled. Hence f � (x0 ) = v �→
B(x0 , v) + B(v, x0 ).
Solution. We have f (x) = B(x, x) for (for example) the bilinear function B : R2 × R2 → R given
by �� � � ��
a1 b
B , 1 = a1 b1 + a 1 b2 + a 2 b 1 .
a2 b2
Therefore � �
v
f (x0 ) 1 = x01 v1 + x01 v2 + x02 v1 + v1 x01 + v1 x02 + v2 x01 .
�
v2
This can alternatively be written as
� � � �
� v1 � � v1
f (x0 ) = 2x01 + 2x02 2x01 .
v2 v2
36. Let f : D ⊂ C → C and let z0 ∈ D. Show that f is complex differentiable at z0 if and only if z0 is
an interior point of D and there exists a a ∈ C and a g : D → C such that for all z ∈ D
Then the desired equation relating f and g is clearly satisfied. We have for z �= z0
� �
f (z) − f (z0 ) � z − z0
g(z) = − f (z0 ) .
z − z0 |z − z0 |
2
Now assume the first order Taylor formula condition. By assumption z0 is an interior point of D.
Rewriting the given relation between f and g gives for z �= z0
f (z) − f (z0 ) |z − z0 |
−a= g(z).
z − z0 z − z0
The right-hand side converges to zero as z → z0 by Algebra of Limits since the first term has
absolute value one and the second term by assumption converges to zero. Hence the left-hand side
converges to zero as z → z0 . This implies that f is complex differentiable at z0 with derivative a.
In particular this shows that a is unique and equals f � (z0 ).
37. Define the function f : C → C as the complex function corresponding to the real function
�� �� � �
x1 cos x1 cosh x2
fR = .
x2 − sin x1 sinh x2
Prove that f is complex differentiable on C. This function is called the complex cosine function
and will be denoted by z �→ cos z.
From this we see that all partial derivatives exist and are continuous on R2 . Hence by a result from
the lecture notes fR is differentiable on R2 . Moreover, the Jacobian matrix has the form required
by the Cauchy–Riemann Theorem at every x ∈ R2 . Therefore, by the Cauchy–Riemann Theorem,
f is complex differentiable on C.
3
Homework 4 MA20219 Analysis 2B solutions
47. Let Γ be the boundary of the triangle with vertices 0, 1 and i (traversed in that order). Calculate
(using the convention from the lecture notes about how an oriented polygon is seen as a circuit)
�
z dz,
Γ
Therefore �
z dz = i.
Γ
(b) We have u(x, y) = x and v(x, y) = −y and that S (the set enclosed by Γ) is the triangle with
vertices 0, 1 and i. It follows that the assumptions from Problem 41 (b) are satisfied and therefore
� �� �� ��
∂v ∂u ∂u ∂v
z dz = − + dxdy + i − dxdy = i 2 dxdy = i,
γ S ∂x ∂y S ∂x ∂y S
��
where we have used that S 1 dxdy is the area of S, which equals 21 .
48. Let γ : [a, b] → C be continuously differentiable and let φ : [c, d] → [a, b] be a continuously
differentiable bijection with φ� (t) > 0 for all t ∈ [c, d]. Define γ1 : [c, d] → C by γ1 := γ ◦ φ. Prove
that
Solution. Let z ∈ Γ. Then there exists a t ∈ [a, b] such that z = γ(t). Since φ is onto, there exists
a θ ∈ [c, d] with t = φ(θ). Then z = γ(φ(θ)) = γ1 (θ). Hence z ∈ Γ1 .
Let z ∈ Γ1 . Then there exists a θ ∈ [c, d] such that z = γ1 (θ). Define t := φ(θ). Then z = γ(t), so
that z ∈ Γ. We conclude that Γ = Γ1 .
Since γ1 is the composition of two differentiable functions it is differentiable and γ1� (t) = γ � (φ(t)) φ� (t).
Since γ � and φ are continuous, so is their composition γ � ◦ φ. Combined with the fact that φ� is
continuous, it follows that the product of γ � ◦ φ and φ� is continuous. Hence γ1� is continuous.
We have, using the change of variables θ := φ(t) and the change of variables theorem for real
integrals,
� d � d � d � b
� � � � �
|γ1 (t)| dt = |γ (φ(t)) φ (t)| dt = |γ (φ(t))| φ (t) dt = |γ � (θ)| dθ.
c c c a
1
Similarly we have [we could have split the integrals below into real and imaginary parts so as to
apply the change of variables theorem for real integrals rather than directly use it for complex
integrals]
� � d � d
f (z) dz = f (γ1 (t)) γ1� (t) dt = f (γ(φ(t))) γ � (φ(t)) φ� (t) dt
γ1 c c
� b �
= f (γ(θ)) γ � (θ) dθ = f (z) dz.
a γ
2
Homework 5 MA20219 Analysis 2B solutions
54. Define a function which is complex differentiable on C\(−∞, 0] and coincides with the usual cube
root function on (0, ∞). [Hint: use polar coordinates. Formally applying the exponential rules will
tell you what the correct formula should be.]
� �1/3
Solution. We want f (reiϕ ) = reiϕ . Formally applying the exponential rules gives that this
√
equals 3 reiϕ/3 . So this is how we will define f (though we first write the complex exponential in
terms of real functions so as to be able to use Problem 49 later).
Define the function f : C\(−∞, 0] → C through f (reiϕ ) = u ([ ϕr ]) + iv ([ ϕr ]) with u, v : (0, ∞) ×
√ √
(−π, π) → R defined by u(r, ϕ) := 3 r cos ϕ3 and v(r, ϕ) := 3 r sin ϕ3 . We have
∂u r−2/3 ϕ ∂u 1 ϕ ∂v r−2/3 ϕ ∂v 1 ϕ
= cos , = −r1/3 sin , = sin , = r1/3 cos .
∂r 3 3 ∂ϕ 3 3 ∂r 3 3 ∂ϕ 3 3
We see that these partial derivatives exist and are continuous on (0, ∞) × (−π, π). Therefore, by
a theorem from the lecture notes, u and v are differentiable on (0, ∞) × (−π, π). We also see
that they satisfy the polar Cauchy–Riemann equations from Problem 49. Hence f is complex
differentiable on C\(−∞, 0]. The defined function coincides with the usual cube root on (0, ∞)
√
since f (r) = u(r, 0) + v(r, 0) = 3 r for all r > 0.
�
55. Let γ : [0, π] → C be defined by γ(t) = eit . Calculate γ ez dz.
� wk
Solution. We have ew = ∞ , and this series converges uniformly on B R (0) for any R > 0.
k=0 k! �
Substituting w = 1/z gives e = ∞1/z 1 −k
k=0 k! z , where this series converges uniformly on the exterior
of Br (0) for any r > 0. In particular, we obtain uniform convergence on ∂B1 (0). By a theorem
from the lecture notes we can then interchange the infinite sum and the integral to obtain
� � ∞
� �∞ �
1 −k 1
1/z
e dz = z dz = z −k dz.
∂B1 (0) ∂B1 (0) k! k! ∂B1 (0)
k=0 k=0
57. Given an example of a function f : W ⊂ C → C with W open and non-empty which is holomorphic
on W with derivative zero, but which is not constant.
1
Homework 6 MA20219 Analysis 2B solutions
(i) B1 (0);
(ii) C\R.
Solution. (i) B1 (0) is convex, so star-shaped with respect to any z0 ∈ B1 (0). To show that this set
is convex, consider z0 , z1 ∈ B1 (0). Then for t ∈ [0, 1]
Solution. By the chain rule, z �→ ecos(z) is holomorphic on C. Since C is star-shaped, it follows from
the local version of Cauchy’s theorem that z �→ ecos(z) has an anti-derivative on C. The fundamental
� cos(z)
theorem of complex integration then shows that γ e dz = 0 since γ is a circuit.
Solution. Denote this star-shaped set by W . Let z0 ∈ W be a point with respect to which W is
star-shaped. Let z1 , z2 ∈ W . By star-shapedness we have that the line segments [z0 , z1 ] and [z0 , z2 ]
are contained in W . It follows that the contour −[z0 , z1 ] + [z0 , z2 ] which has initial point z1 and
terminal point z2 is contained in W . Therefore W is contour-connected.
�∞ 1 k
67. Consider the power series k=1 k2 z .
1
(b) By the Weierstrass M -test with majorizing sequence Mk := k2
we have that the power series
does converge uniformly on B 1 (0).
�∞
68. Consider the power series k=0 (2 + (−1)k )k z k .
(a) What happens if we try to apply the ratio test for the radius of convergence?
(b) Determine the radius of convergence.
1
Solution. We have �
3k k even
ak =
1 k odd.
It follows that �
ak 3k k even
=
ak+1 3−k−1
k odd,
� � � �
� ak � � ak �
which implies that lim inf k→∞ � ak+1 � = 0 and lim supk→∞ � ak+1 � = ∞. It follows that the conditions
of the ratio test are not satisfied.
We have �
�
k 3 k even
|ak | =
1 k odd,
�
so that lim supk→∞ k
|ak | = 3. It follows that the radius of convergence is 13 .
2
Homework 7 MA20219 Analysis 2B solutions
1
79. Determine the radius of convergence of the power series of z 2 −1
centred at z0 = 3 without calculating
the series.
Solution. The given function is complex differentiable on C\{−1, 1}. In particular, it is holomorphic
on B2 (3). Therefore the radius of convergence of its power series centred at z0 = 3 is at least 2
by the Radius of Convergence Corollary. Since the singularity 1 is on the boundary of B2 (3), the
radius of convergence cannot be larger than 2. Therefore the radius of convergence equals 2.
lim |p(z)| = ∞.
|z|→∞
We conclude that
|an |
|p(z)| ≥ |z n |
.
2
Since the right-hand side converges to infinity as |z| → ∞, so does the left-hand side.
(b) We prove this by contradiction. Assume that the non-constant complex polynomial p does not
have a complex zero. Then f := 1/p is entire. Additionally, lim|z|→∞ f (z) = 0 since lim|z|→∞ |p(z)| =
∞. It follows that f is bounded on C: since lim|z|→∞ f (z) = 0 there exists an R > 0 such that
outside of BR (0) we have |f (z)| ≤ 1; by Weierstrass’ Extreme Value Theorem, there exists a M > 0
such that on B R (0) we have |f (z)| ≤ M ; so |f (z)| ≤ max{1, M } for all z ∈ C. So f is a bounded
entire function, so by Liouville’s theorem it is constant. This is the desired contradiction.
81. Let f be an entire function with the property that there exists a M > 0 such that |f (z)| ≤ M eRe(z)
for all z ∈ C. Prove that f (z) = Cez for some constant C.
Solution. The given inequality is equivalent to |e−z f (z)| ≤ M for all z ∈ C. Since as the product
of two entire functions, z �→ e−z f (z) is entire, we have by Liouville’s Theorem that it is constant.
Hence there exists a C ∈ C such that e−z f (z) = C. This gives the desired conclusion.
1
Homework 8 MA20219 Analysis 2B solutions
Solution. (a) By the definition of order of a zero with g ≡ 1 it follows that the order is 2.
(b) We use Problem 82. We have f (0) = 0, f � (z) = 2 sin(z) cos(z) so that f � (0) = 0 and f �� (z) =
2 cos2 (z) − 2 sin2 (z) so that f �� (0) = 2 �= 0. It follows that the order is 2.
93. Use the identity theorem to show that there is precisely one function which is holomorphic on C
and coincides with z 2 on (0, 1).
[You may use the fact that the equality is true for z, w ∈ R and that the sine and the cosine are
holomorphic on C.]
Solution. Fix w ∈ R. Both the left-hand side and the right-hand side of the equality are holomorphic
on C as functions of z. By assumption they are equal for z ∈ R. Since R contains an accumulation
point it follows that the equality holds for all z ∈ C. We conclude that the equality holds for all
z ∈ C and w ∈ R.
Now fix z ∈ C. Both the left-hand side and the right-hand side of the equality are holomorphic on
C as functions of w. By the first part of the proof, they are equal for w ∈ R. Since R contains an
accumulation point it follows that the equality holds for all w ∈ C.
Therefore the inequality holds for all z, w ∈ C.
Solution. By the maximum modulus theorem for a compact set (Problem 83) we have that the
maximum is reached on the unit circle B 1 (0) (and by the maximum modulus theorem, it is reached
only there). Writing z = x + iy with x, y ∈ R, we have
This is a real-valued function of one real variable which we are maximizing over [−1, 1]. The
maximum is reached at x = 0. We then have y = ±1 and therefore z = ±i.
96. Use the Minimum Modulus Theorem to prove the Fundamental Theorem of Algebra.
1
Solution. Let p be a non-constant complex polynomial. Since lim|z|→∞ |p(z)| = ∞ by Problem 80,
by the Weierstrass Extreme Value Theorem, |p| has a minimum on C. The details of that are as
follows. That lim|z|→∞ |p(z)| = ∞ by definition means that for all M ≥ 0 there exists a R > 0 such
that for all z with |z| ≥ R we have |p(z)| ≥ M . Letting M := |p(0)| we obtain a R > 0 such that for
all z with |z| ≥ R we have |p(z)| ≥ |p(0)|. By the Weierstrass Extreme Value Theorem, there exists
a z0 ∈ B R (0) such that |p(z)| ≥ |p(z0 )| for all z ∈ B R (0). In particular, |p(0)| ≥ |p(z0 )|. Combined
with the earlier inequality, this shows that for all z with |z| ≥ R we have |p(z)| ≥ |p(0)| ≥ |p(z0 )|.
Therefore, we have |p(z)| ≥ |p(z0 )| for all z ∈ C, i.e. |p| has a minimum on C.
Let z0 be a point where this minimum is reached. Since p is not constant, it then follows from the
Minimum Modulus Theorem applied with W = C that p(z0 ) = 0.
97. Let z0 , z1 , z2 ∈ C and let r > 0 be such that |z1 − z0 |, |z2 − z0 | �= r. Let W := C\{z1 , z2 }. Determine
when ∂Br (z0 ) is homologous to zero in W .
Solution. We have C\W = {z1 , z2 }. Therefore, the condition that ∂Br (z0 ) is homologous to zero
in W is n∂Br (z0 ) (z1 ) = 0 = n∂Br (z0 ) (z2 ). This condition is satisfied if and only if |z1 − z0 |, |z2 − z0 | >
r.
98.∗ Let f be a continuous function on the unit circle. Show that f is the uniform limit of a sequence of
polynomials if and only if f has a continuous extension to the closed unit disc which is holomorphic
on the open unit disc.
[Hint: Use Problems 71, 83 and an idea from the proof of Problem Problem 59]
Solution. First assume that f is the uniform limit (on the unit circle) of a sequence of polynomials
(pn )n∈N . This sequence of polynomials is well-defined on the closed unit disc. By the Maximum
Modulus Theorem for compact sets (Problem 83), and the fact that (pn )n∈N is convergent and
therefore Cauchy in the space of continuous functions on the unit circle, we have that (pn )n∈N is
Cauchy in the space of continuous functions on the closed unit disc. Since that space is complete,
the (pn )n∈N converges to a limit g which is a continuous function on the closed unit disc. By
uniqueness of limits, it follows that g coincides with f on the unit circle. Since the uniform limit on
an open set of holomorphic functions is holomorphic (by Problem 71), we have that g is holomorphic
on the open unit disc.
Now assume that f has a continuous extension g to the closed unit disc which is holomorphic on
the open unit disc. For r ∈ (0, 1), define gr : B 1/r (0) → C by gr (z) = g(rz). Then gr is holomorphic
on B1/r (0) and as in Problem 59, we have that gr converges uniformly to g on ∂B1 (0) as r ↑ 1.
Since gr is holomorphic on B1/r (0), it is analytic on that set and therefore it has a power series
expansion which converges uniformly on B 1 (0). The partial sums of this power series expansion
gives a sequence of polynomials (pn )n∈N which converges uniformly to gr on B 1 (0).
Let ε > 0. Consider a sequence (rk )k∈N with rk ∈ (0, 1) and rk → 1. Then there exists a K ∈ N
such that for all k ≥ K we have (the norm here is the maximum norm on the closed unit disc)
�g − grk �∞ ≤ ε/2.
Let (pn,k )n∈N be the partial sum sequence for grk . Then there exists a Nk ∈ N such that for all
n ≥ Nk we have
�grk − pn,k �∞ ≤ ε/2.
Define the sequence of polynomials (qn )n∈N by qn := pn,K and define N := NK . We then have for
all n ≥ N
�g − qn �∞ ≤ �g − grK �∞ + �grK − pn,K �∞ ≤ ε.
This shows that (qn )n∈N converges to g uniformly on the unit circle.
2
Homework 9 MA20219 Analysis 2B solutions
� cos z �
106. Calculate Res sin z , 0 .
(c) Calculate
�
eiz
dz,
ΓR cosh z
where ΓR is the boundary of the rectangle with vertices −R, R, R + iπ and −R + iπ (in that
order). Denote the top, bottom, left and right of this rectangle by ΓTR , ΓB L R
R , ΓR , ΓR .
eiz
(d) Show that the function f (z) := cosh z satisfies
(e) Use 107d to write the integral of f over ΓTR as a constant times the integral over ΓB
R.
(f) Show that |eiz | ≤ 1 for z ∈ {x + iy : x ∈ R, y ≥ 0}.
(g) Show that | cosh z| ≥ | sinh x|, where z = x + iy and x, y ∈ R.
1
(h) Conclude that we have |f (z)| ≤ | sinh x| for z ∈ {x + iy : x ∈ R, y ≥ 0, x �= 0}.
(i) Use the ML-inequality and the result from (107h) to show that
� �
lim f (z) dz = 0 = lim f (z) dz.
R→∞ ΓL R→∞ ΓR
R R
�∞ cos x
and the above established results to determine −∞ cosh x dx.
Solution. (a) Multiplying by ez , we see that ez + e−z = 0 if and only if e2z + 1 = 0. So we want
to solve e2z = −1. Writing −1 in polar coordinates this gives e2z = eiπ+2iπk with k ∈ Z, i.e.
2z = iπ + 2iπk, which gives
π
z = i + iπk.
2
(b) Using a result from the lecture notes, we have
� iz �
e π e−π/2 e−π/2
Res ,i = = ,
cosh z 2 sinh(i π2 ) i
1
(c) i π2 is the only singularity inside the contour, so by the Residue Theorem (using that nγ (i π2 ) = 1,
which is geometrically clear):
� � iz �
eiz e π
dz = i2πRes ,i = 2πe−π/2 .
ΓR cosh z cosh z 2
(d) We have
(e) A parametrization of ΓB B T T
R is γR (θ) = −R + 2θR and a parametrization of ΓR is γR (t) =
R − 2tR + iπ. We have (using part 107d for the second equality and using the change of
variables θ = 1 − t):
� � 1 � 1 � 1
−π
f (z) dz = −2Rf (R−2tR+iπ)dt = 2Re f (R−2tR)dt = 2Re−π f (−R+2θR)dθ.
ΓT
R 0 0 0
So we see that � �
−π
f (z) dz = e f (z) dz.
ΓT
R ΓB
R
(f) We have
|eiz | = |ei(x+iy) | = |eix−y | = |eix e−y | = |eix | |e−y | = |e−y | ≤ 1,
where in the last step we have used that y ≥ 0.
(g) We have by the reverse triangle inequality
� � � �
|ez + e−z | ≥ �|ez | − |e−z |� = �ex − e−x � .
(h) Since we bounded the numerator from above by 1 and the denominator from below by the
positive number | sinh x| the desired inequality between fractions follows.
(i) The ML-inequality gives (using that the length of ΓR R is π)
�� �
� � 1
� �
� f (z) dz � ≤ π max |f (z)| ≤ π .
� ΓR � R
z∈ΓR | sinh R|
R
The right-hand side converges to zero as R → ∞ and therefore (by the squeezing principle) so
does the left-hand side.
Similarly we have
�� �
� � 1 1
� �
� f (z) dz � ≤ π max |f (z)| ≤ π =π ,
� ΓL � L
z∈ΓR | sinh(−R)| | sinh R|
R
2
(j) Taking limits in the mentioned equality gives (using the above results):
� ∞
eix
(1 + e−π ) dx = 2πe−π/2 ,
−∞ cosh x
3
Homework 10 MA20219 Analysis 2B solutions
114. Calculate the Laurent series of the following functions with the given centre and determine its
punctured disc of convergence.
z
(a) (z+1)(z+2) with centre z0 = −1,
2
(b) (z−1)(z+1) with centre z0 = 1,
1
(c) z 2 (z+1)
with centre z0 = −1.
Solution. (a) The function has poles at z = −1 and z = −2, so the punctured disc of convergence
is {z : 0 < |z + 1| < 1}. The partial fraction expansion is
−1 2
+ .
z+1 z+2
By the geometric series
� ∞
2 2
= =2 (−z − 1)k ,
z+2 1 − (−z − 1)
k=0
for |z + 1| < 1. It follows that
� ∞
z −1
= +2 (−1)k (z + 1)k .
(z + 1)(z + 2) z+1
k=0
(b) The function has poles at z = −1 and z = 1, so the punctured disc of convergence is {z : 0 <
|z − 1| < 2}. The partial fraction expansion is
1 1
− .
z−1 z+1
By the geometric series
∞ � �
1 1 1 1� z−1 k
= = −
z+1 2 1 + z−1
2
2 2
k=0
for |z − 1| < 2. It follows that
∞ � �k ∞ � �
2 1 1� −1 k
� −1 k+1
= − (z − 1) = (z − 1)k .
(z − 1)(z + 1) z−1 2 2 2
k=0 k=−1
(c) The function has poles at z = 0 and z = −1, so the punctured disc of convergence is {z : 0 <
|z + 1| < 1}. The partial fraction expansion is
1 1 1
− + .
z2 z z + 1
By the geometric series
� ∞
1 −1
= =− (z + 1)k ,
z 1 − (z + 1)
k=0
1 �∞ k
which converges for |z + 1| < 1. Differentiating the geometric series equality 1−w = k=0 w gives
for |w| < 1
�∞
1
= kwk−1 .
(1 − w)2
k=1
Using this gives
� ∞ � ∞
1 1 k−1
= = k(z + 1) = (n + 1)(z + 1)n .
z2 (1 − [z + 1])2
k=1 n=0
1
It follows that
∞
� ∞
�
1 1 k
= + (k + 2)(z + 1) = (k + 2)(z + 1)k .
z 2 (z + 1) z+1
k=0 k=−1
115. Classify zero as a removable singularity, pole or essential singularity of the following functions
1
(a) z3
− cos z,
(b) ze1/z ,
sin z
(c) z ,
cos z
(d) z .
Solution. (a) Since cos z is analytic in zero we can ignore it and only consider the non-analytic part
1
z3
. It follows that zero is a pole.
� (−1)n 2n
In more detail, using that cos z = ∞ n=0 (2n)! z we have that the Laurent series with centre zero
�∞ k
is k=−∞ ak z with
0 k < 0 and k �= −3
1 k = −3
ak = (−1)k/2
k ≥ 0 and k even
k!
0 k ≥ 0 and k odd.
� �∞ 1 −n+1 �1
(b) The power series expansion of ew is ∞ 1 n
n=0 n! w so that ze
1/z =
n=0 n! z = j=−∞ 1 j
(1−j)! z .
It follows that zero is an essential singularity.
�
In more detail, we have that the Laurent series with centre zero is ∞ k
k=−∞ ak z with
�
1
(1−k)! k≤1
ak =
0 k > 1.