ASSIGNMENT – 2

C-linear maps versus R-linear maps;

C-differentiable mappings versus R-differentiable mappings:

(1) As C is an R-vector space as well as a C-vector space, we have to distinguish be-

tween R-linear mappings and C-linear mappings of C into itself; recall for instance,
that z → z is an example of an R-linear map which is not C-linear. However, we
often identify C ' R2 as R-vector spaces and when we do so, we identify z = x+iy
as (x, y) as well. With this understanding, suppose T is an R-linear map from C
into itself and we write
T (z) = λz + µz
Show that the constants λ, µ are given in terms of the values of T by
1
λ = (T (1) − iT (i))
2
and
1
µ = (T (1) + iT (i)) .
2
Show that the mapping T is C-linear iff T (1) = iT (1); in such a case, show that T
has the form T (z) = T (1)z and T that is the composition of a the rotation about
the origin by Arg(T (1)) followed by a dilation by the factor |T (1)|.
Next, let us view the map T which as stated above is only known to be R-linear,
also as a map from R2 into itself. We know from basic linear algebra that every
such map is represented by a (2 × 2)-matrix; say T is represented by
 
a b
c d
Verify then that T (1) = a + ic and T (i) = b + id. Show that T is C-linear if and
only if the entries satisfy the following:
(
a = d, and,
c = −b
and show that this happens if and only if T commutes with
 
0 −1
.
1 0
What does all this mean for C-differentiability versus R-differentiability?

(2) ‘Wirtinger derivatives’ or partial-derivatives expressed in terms of z, z – complex notation!:

First recall/read about the Cauchy – Riemann equations (for instance, just about
page-32 from the first chapter of Lang’s book would be enough).
Suppose f is a complex valued function defined on an open set U ⊂ C with
the property that its partial derivatives exist i.e., writing f in terms of its real
1
2 ASSIGNMENT – 2

and imaginary parts as usual as f = u + iv, the partial derivatives of u, v exist

everywhere throughout U . Then, define
∂f 1  ∂f ∂f 
= −i
∂z 2 ∂x ∂y
∂f 1 ∂f
 ∂f 
= +i .
∂z 2 ∂x ∂y
The LHS of the above pair of equations are referred to as ‘Wirtinger derivatives’.
Suppose now that f is a (totally) differentiable map – in the sense of Calculus of
several real variables – between a pair of open sets in C identified with R2 as real
vector spaces.
Verify that f is C-differentiable on U if and only if ∂f /∂z ≡ 0 on U and in such
a case, its C-derivative is given by ∂f /∂z.
(Remark: the point of introducing ‘Wirtinger derivatives’ is to write the Cauchy–
Riemann equations as a single equation ∂f /∂z = 0. The partial differential oper-
ator ∂/∂z is known as the Cauchy – Riemann operator).

(3) Show that the following analogue of the mean value theorem is false for C-differentiable
functions: Suppose f is C-differentiable on an open subset U ⊂ C and l is a
line-segment which lies fully within U . Call the end-points of l as a, b; then, in
particular a, b are a pair of points of U . Show that the assertion/statement that
there exists a point c on the line-segment l such that

f (b) − f (a) = f 0 (c)(b − a),

is false.

(4) On the complex-logarithm and its derivatives:

(i) Define
∞
X wn
f (w) = (−1)n−1
n=1
n

Show that the radius of convergence of this series is 1 and recall that taking
w = x real, just gives the usual Taylor series of log(1 + x) studied in one-
variable Calculus or Real Analysis. Prove then that

exp(f (z − 1)) = z

for all z in D1 (1), the disc centered at the point 1 with radius 1. Conclude
therefore that
∞ n
n−1 (z − 1)
X
f (z − 1) = (−1)
n=1
n

is a power series representation of Log(z) centered at the point 1, valid on the

disc D1 (1).
 ASSIGNMENT – 2 3

(ii) Let z0 ∈ C∗ . Let ζ be any complex number such that exp(ζ) = z0 . For z in
the disc D|z0 | (z0 ) define
   
z z − z0
g(z) = f −1 +ζ =f + ζ.
z0 z0
Prove that exp ◦ g is the identity function on D|z0 | (z0 ). Remark: f (z) and
g(z) may well differ by a constant (indeed, such a constant would be an in-
teger multiple of 2πi and that’s the most that can go ‘wrong’), if the discs
D1 (1) and D|z0 | (z0 ) intersect and z is a point in their intersection. Finally
prove that g and thereby every right inverse of the exponential map which is
given by a power series, centered at some point in C∗ is infinitely many times
C-differentiable.

(5) Partial fraction expansion – a simple case of decomposition of functions: While we’ve
been discussing ways of constructing new-functions out-of-old, systematically, more
and more complex(!), it’s important in applications to perform the reverse i.e.,
break down complex functions into simple ones. The FTA (fundamental theorem
of algebra) already does this for an arbitrary non-constant polynomials by leading
to their expression as a product of powers of linear polynomials. Derive out of
this, the following decomposition result for rational functions. Let R = P1 /P2 be
any rational function which we may well assume to be in reduced form i.e., P1 , P2
are polynomials with P2 6≡ 0 and with gcd(P1 , P2 ) = 1. Prove that there exists a
polynomial Q(z) and constants ck,j ∈ C, integers nj for j = 1, . . . , N := deg(R2 ),
such that
N X nj
X ck,j
R(z) = Q(z) +
j=1 k=1
(z − zj )k
Thus: every rational fraction can be decomposed into a (finite) sum of powers-of-
simple-rational functions, upto a polynomial Q; here, by a simple non-polynomial
rational function we mean a function of the form r(z) = 1/(z − ζ) for some fixed
ζ ∈ C. Such partial fraction decompositions (PFDs) have many applications:
within mathematics, for instance in computing integrals in basic calculus of a
single real variable; while this itself is of use in real world applications of basic
calculus, PFDs also help solving certain differential equations that arise in engi-
neering problems using the Laplace transform and its inverse.

Applications of analytic functions to differential equations (these applications in-

volve rather lengthy proofs and you may consult Lang’s book for complete solutions
if you don’t succeed in your attempts):

(6) Linear Differential Equations. Let a0 (z), . . . , ak (z) be C-analytic functions in a

neighbourhood of the origin in C. Assume that a0 (0) 6= 0. Given numbers
c0 , . . . , ck−1 , there exists a unique analytic function f at the origin (i.e., in a neigh-
bourhood of the origin) with (Dn f )(0) = cn for n = 0, 1, 2, . . . , k − 1, where D
denotes the differential operator ∂/∂z and such that
a0 (z)Dk f (z) + a1 (z)Dk−1 f (z) + . . . + ak (z)f (z) = 0
4 ASSIGNMENT – 2

(Hint: First argue that you may assume a0 (z) = 1; then solve for f by a formal
power series. Then, show that this formal power series converges).

(7) General autonomous Ordinary Differential Equations. Let g be a complex ana-

lytic function in a neighbourhood of the origin in C. Show that there exists a
unique analytic function f at the origin (i.e., in a neighbourhood of the origin),
with f (0) = 0 and satisfying f 0 (z) = g(f (z)).

Further examples of C-differentiable functions.

(8) Let h(t) be any continuous complex valued function on the standard unit interval
[0, 1] in R. Define:
Z 1
h(t)
H(z) = dt
0 t−z
for all z ∈ C\[0, 1]. Show that H(z) is C-differentiable and compute its derivative.

Basic properties of C-differentiable functions.

(9) Show that if f is a C-differentiable function on a domain D ⊂ C, whose deriv-

ative vanishes everywhere on D, then f must be a constant function. Next, let
f, g be a pair of C-differentiable functions on a domain D ⊂ C whose derivatives
agree everywhere on D: f 0 (z) = g 0 (z) for all z ∈ D. Show that f and g differ at-
most by a constant i.e., for some c ∈ C, we have for all z ∈ C, that g(z) = f (z)+c.

(10) Suppose that f is a C-differentiable function on a domain D ⊂ C. Prove that in

each of the following cases, we can conclude that f is a constant function.
(i) <(f ) is constant,
(ii) =(f ) is constant,
(iii) |f | is constant,
(iv) Arg(f ) is constant.
What can you say about a pair of C-differentiable functions on a domain D ⊂ C
whose real parts agree?

(11) Suppose f is a C-differentiable function with nowhere vanishing derivative on a

domain D ⊂ C. Write f in terms of it real and imaginary parts as usual as:
f (z) = u(x, y) + iv(x, y) and consider the level sets of the curves of u and v i.e.,
the sets
{z = x + iy : u(x, y) = u0 }, {z = x + iy : v(x, y) = v0 }
for fixed real numbers u0 , v0 . Prove that these level curves are orthogonal to each
other wherever they intersect.

(12) Show that the real and imaginary parts u, v of a C-differentiable function f = u+iv
on a domain D ⊂ C satisfy the Laplace equation, assuming that they are C 2 -
smooth – recall that, we say that a f satisfies the Laplace equation if
∂ 2f ∂ 2f
∆f := + ≡0
∂x2 ∂y 2
 ASSIGNMENT – 2 5

It’s implicitly assumed here that the partial derivatives appearing in the above all
exist on the domain (open + connected subset of R2 ) where f is defined.
Next, for each of the following functions, show that either a C-differentiable func-
tion f exists with the given function being its real part (in which case, determine
f ) on some domain or prove that none exists:

(i):
x2 + y 2
u1 (x, y) =
1 + 2x + x2 + y 2
(ii):
x(1 + x2 + y 2 )
u2 (x, y) = .
1 − 2x2 − 2y 2 + (x2 + y 2 )2