Complex Analysis (part 1): Complex Numbers and Derivatives (by Evan Dummit, 2022, v. 1.

00)

Contents
1 Complex Numbers and Derivatives 1

1.1 Complex Arithmetic . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1

1.1.1 The Complex Numbers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1

1.1.2 Polar Form, Complex Exponentials, Powers, and Roots . . . . . . . . . . . . . . . . . . . . . 3

1.1.3 Basic Topology of C . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7

1.2 The Complex Derivative and Holomorphic Functions . . . . . . . . . . . . . . . . . . . . . . . . . . . 9

1.2.1 Limits . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9

1.2.2 The Complex Derivative . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12

1.2.3 Partial Derivatives and the Cauchy-Riemann Equations . . . . . . . . . . . . . . . . . . . . . 14

1.2.4 Holomorphic Functions and Angles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18

1 Complex Numbers and Derivatives

In this chapter, our goal is to discuss functions of a complex variable, and to construct various elementary functions
such as polynomials, rational functions, and the exponential function. We then give various approaches to studying
the notion of dierentiation in the context of functions of a complex variable.

1.1 Complex Arithmetic

• In this section we review the arithmetic of complex numbers.

1.1.1 The Complex Numbers

• Before dening the arithmetic operations for complex numbers, we will give a brief review of their history in
mathematics.

◦ Complex numbers were rst encountered by mathematicians in the 1500s who were trying to write down
general formulas for solving cubic equations (i.e., equations like x3 + x + 1 = 0), in analogy with the
well-known formula for the solutions of a quadratic equation. It turned out that their formulas required
manipulation of complex numbers, even when the cubics they were solving had three real roots.

◦ It took over 100 years before complex numbers were accepted as something mathematically legitimate:
even negative numbers were sometimes suspect, so (as the reader may imagine) their square roots were
even more questionable.

◦ The stigma is still evident even today in the terminology (imaginary numbers); nonetheless, complex
numbers are very real objects (no pun intended), and have a wide range of uses in mathematics, physics,
and engineering.

• Denition: A complex number is a number of the form a + bi, where a and b are real numbers and i is the
so-called imaginary unit, dened so that i2 = −1. The real part of z = a + bi, denoted Re(z), is the real
number a, while the imaginary part of z = a + bi, denoted Im(z), is the real number b. The set of all complex
numbers is denoted C.

1
 √
◦ The notation −1 is also often used to denote the imaginary unit
√i. In certain disciplines (especially
electrical engineering), the letter j may instead be used to denote −1, rather than i (which is instead
used to denote electrical current).

◦ Examples: Some complex numbers are 4 + 3i, 3 − πi, 6i = 0 + 6i, and −5 = −5 + 0i. Their real parts
are 4, 3, 0, and −5 respectively, while their imaginary parts are 3, −π , 6, and 0 respectively.

• Denition: The complex conjugate of z = a + bi, denoted z , is the complex number a − bi. The√modulus (also
called the absolute value, magnitude, or length) of z = a + bi, denoted |z|, is the real number a2 + b2 .

◦ The notation for conjugate varies among disciplines. The notation z∗ is often used in physics and
computer programming to denote the complex conjugate (in place of z) since it is easier to type on a
standard keyboard.
√
◦ Example: For z = 3 + 4i we have z = 3 − 4i and |z| = 32 + 42 = 5.

• Two complex numbers are added (or subtracted) simply by adding (or subtracting) their real and imaginary
parts: (a + bi) + (c + di) = (a + c) + (b + d)i.

◦ Example: The sum of 1 + 2i and 3 − 4i is 4 − 2i . The dierence is (1 + 2i) − (3 − 4i) = −2 + 6i .

• Two complex numbers are multiplied using the distributive law and the fact that i2 = −1: (a + bi)(c + di) =
2
ac + adi + bci + bdi = (ac − bd) + (ad + bc)i.

◦ Example: The product of 1 + 2i and 3 − 4i is (1 + 2i)(3 − 4i) = 3 + 6i − 4i − 8i2 = 11 + 2i .

2
◦ Observe in particular that for z = a + bi, we have |z| = a2 + b2 = z · z .

a + bi (a + bi)(c − di) ac + bd
• For division, we rationalize the denominator using the conjugate: = = 2 +
c + di (c + di)(c − di) c + d2
bc − ad
i.
c2 + d2
2i 2i(1 + i) −2 + 2i
◦ Example: The quotient of 2i by 1−i is = = = −1 + i .
1−i (1 − i)(1 + i) 2

• Here are a few more simple properties of complex number arithmetic:

• Proposition (Complex Arithmetic): Suppose z and w are complex numbers.

1. We have Re(z) = (z + z)/2 and Im(z) = (z − z)/(2i).

2. We have z + w = z + w, zw = z · w, and z = z.
3. We have |z| = |z| and |zw| = |z| · |w|.
4. We have z=z if and only if z is real, while z = −z if and only if z is purely imaginary (of the form ri
where r is real).

5. We have Re(z) ≤ |z| and Im(z) ≤ |z|.

6. (Triangle Inequality) We have |z + w| ≤ |z| + |w|.
◦ Proofs: (1)-(5) are easy algebraic calculations.

◦ For (6), use (1) and (2) to observe zw + wz = 2Re(zw), and (5) and (3) to observe 2Re(zw) ≤
2 |zw| = 2 |z| |w|.
2 2 2 2 2
◦ Then |z + w| = (z +w)(z + w) = zz +zw+wz +ww = |z| +|w| +2Re(zw) ≤ |z| +|w| +2 |z| |w| =
(|z|+|w|)2 . Since both |z + w| and |z|+|w| are nonnegative, taking the square root yields the desired
|z + w| ≤ |z| + |w|.

• We emphasize that (2) above shows that the conjugate is both additive and multiplicative.

◦ Example: If z = 1 + 2i and w = 3 − i, then z = 1 − 2i and w = 3 + i. We compute z + w = 4 + i,

z + w = 4 − i , zw = 5 + 5i and z · w = 5 − 5i, so indeed z + w = z + w and zw = z · w.

2
 z
◦ The multiplicativity of the conjugate explains the procedure for performing division: we write =
w
z·w z·w 2
= 2 , where the denominator is now the real number |w| .
w·w |w|

• The real numbers and the complex numbers are both examples of elds: sets of numbers that can be added,
subtracted, multiplied, and divided (except by zero) and possess various algebraic relations involving these
operations. More formally:

• Denition: A eld is an ordered triple (F, +, ·) consisting of a set of numbers F together with two binary
operations ,
1 + (addition) and · (multiplication), satisfying the following axioms:

[F1] Addition is associative: a + (b + c) = (a + b) + c for any elements a, b, c in F.

[F2] Addition is commutative: a+b=b+a for any elements a, b in F.
[F3] There is an additive identity 0 satisfying a+0=a for all a in F.
[F4] Every element a in F has an additive inverse −a satisfying a + (−a) = 0.
[F5] Multiplication is associative: a · (b · c) = (a · b) · c for any elements a, b, c in F.
[F6] Multiplication is commutative: a·b=b·a for any elements a, b in F.
[F7] There is a multiplicative identity 1 6= 0, satisfying 1·a=a for all a in F.
−1 −1
[F8] Every nonzero a in F has a multiplicative inverse a satisfying a·a = 1.
[F9] Multiplication distributes over addition: a · (b + c) = a · b + a · c for any elements a, b, c in F.

1.1.2 Polar Form, Complex Exponentials, Powers, and Roots

• We often think of the real numbers geometrically, as a line. The natural way to think of the complex numbers
is as a plane, with the x-coordinate denoting the real part and the y -coordinate denoting the imaginary part.

◦ Once we do this, there is a natural connection to polar coordinates: namely, if z = x + yi is apcomplex

number which we identify with the point (x, y) in the complex plane, then the modulus |z| = x2 + y 2
is simply the coordinate r when we convert (x, y) from Cartesian to polar coordinates.

◦ Furthermore, if we are given that |z| = r, we can uniquely identify z given the angle θ that the line
connecting z to the origin makes with the positive real axis. (This is the same θ from polar coordinates.)

• From polar coordinates (or simple trigonometry), we see that we can write z in the form z = r [cos(θ) + i sin(θ)],
which is called the polar form of z.
1 The result of applying these operations to elements a and b is denoted by a+b and a·b (or simply ab), respectively. The denition
of binary operation means that a+b and a·b are also numbers in F.

3
 ◦ The length r is simply the modulus of z, while the angle θ is called the argument of z and sometimes
denoted θ = arg(z).
◦ We will emphasize that although r is unique, θ is not: since the sine and cosine are periodic with period
2π , any θ that diers by an integral multiple of 2π yields an equivalent polar form. We will implicitly
identify polar forms yielding the same complex number.

◦ Example: For r = 2 and θ = 0 we obtain z = 2[cos 0 + i sin 0] = 2. Taking r=2 and θ = 2π also yields
z = 2[cos 2π + i sin 2π] = 2.
◦ z = x + iy then we can compute the (r, θ) form fairly easily by solving x = r cos θ
Conversely, if we know
and y = r sin θ θ.
for r and
y
−1 y
p  
◦ Explicitly, we have r = x2 + y 2 = |z| and we can take θ = tan−1 if x > 0 and θ = tan +π
x x
if x < 0. (The extra +π is needed when x < 0 because of the fact that the principal arctangent function
only has range (−π/2, π/2), so we would otherwise get the wrong value for θ if z lies in the second or
third quadrants.)
√
◦ Example: If z = 1 + i, then the corresponding values of r
2 and θ =
and θ above are r = |z| =
π √ h π π i
tan−1 (1) = , so we can write z in polar form as z = 2 cos + i sin . Indeed, we may check that
4 4 4
√ h π π i √ h √ √ i
2 cos + i sin = 2 22 + i 22 = 1 + i, as it should be.
4 4
√
◦ Example: If z = −1 + i 3, then the corresponding values of r and θ above are r = |z| = 2 and
√
 
2π 2π
θ = π + tan−1 (− 3) = 2π/3, so we can write z in polar form as z = 2 cos + i sin .
3 3

• We can also repackage the polar form using complex exponentials.

◦ We will later give a more precise approach to complex exponentials using power series, but for now, we
will simply give a denition using the familiar real exponential and real sine and cosine functions.

• Denition: If z = x + iy is a complex number, we dene the complex exponential ez = ex (cos y + i sin y).

◦ Examples: We have eiπ/2 = e0 (cos π2 + i sin π2 ) = i and e1+iπ = e1 (cos π + i sin π) = −e .

◦ The motivation here is that we want the complex exponential to obey the familiar rules for the real
exponential function, and so in particular we want ex+iy = ex eiy .
◦ It therefore suces just to dene eiy , which we do via Euler's identity eiθ = cos(θ) + i sin(θ).
◦ Euler's identity encodes a lot of information: for example, we claim that eiθ · eiϕ = ei(θ+ϕ) . Expanding
both sides with Euler's identity yields

eiθ · eiϕ = [cos θ + i sin θ] [cos ϕ + i sin ϕ]

= [cos θ cos ϕ − sin θ sin ϕ] + i [sin θ cos ϕ + cos θ sin ϕ]
= cos(θ + ϕ) + i sin(θ + ϕ) = ei(θ+ϕ)

where the key step in the middle uses the usual addition identities cos(θ + ϕ) = cos θ cos ϕ − sin θ sin ϕ
and sin(θ + ϕ) = sin θ cos ϕ + cos θ sin ϕ for sine and cosine.

◦ Another way of interpreting this calculation is that the (initially rather arbitrary-seeming) sine and cosine
addition formulas actually just reect the natural structure of the multiplication of complex numbers.

• Another convenient result follows by applying Euler's identity to the simple relation ei(nθ) = (eiθ )n , which
n
when written out yields De Moivre's identity cos(nθ) + i sin(nθ) = [cos θ + i sin θ] .

◦ By expanding the right-hand side using the binomial theorem we can obtain identities for sin(nθ) and
cos(nθ) in terms of sin θ and cos θ.
2
◦ Example: Setting n = 2 produces cos(2θ) + i sin(2θ) = [cos θ + i sin θ] = (cos2 θ − sin2 θ) + i(2 sin θ cos θ),
2 2
and so we recover the double-angle formulas cos 2θ = cos θ − sin θ and sin 2θ = 2 sin θ cos θ .

4
 −1 cos θ − i sin θ
◦ Example: Setting n = −1 produces cos(−θ) + i sin(−θ) = [cos θ + i sin θ] = , and so we
cos2 θ + sin2 θ
recover the standard identities cos2 θ + sin2 θ = 1, cos(−θ) = cos θ, and sin(−θ) = − sin θ.

• Using Euler's identity and the polar form of complex numbers above, we see that every complex number can
be written in exponential form as z = r · eiθ for the same r and θ we described above.
√
◦ Example: We can draw 1+i in the complex plane, or use the formulas, to see that |1 + i| = 2 and
π √ iπ/4
arg(1 + i) = , and so we see that 1+i= 2·e .
4
√ √ π
◦ Example: Either by geometry or trigonometry, we see that 1−i 3 =2 and arg(1 − i 3) = − , hence
3
√
1 − i 3 = 2 · e−iπ/3 .
√
◦ Example: Using the formulas for r and θ above, we have 3 + 2i = 13 · ei·arctan(2/3) .

• The rectangular a + bi form of a complex number is more convenient for addition, while the polar reiθ form is
iθ iϕ i(θ+ϕ)
more convenient for multiplication, since we may easily multiply (re )(se ) = (rs)e . (This calculation
is often summarized as lengths multiply, angles add.)

◦ In particular, it is very easy to take powers of complex numbers when they are in exponential form: we
have (r · eiθ )n = rn · ei(nθ) .

• Example: Compute (1 + i)8 .

√ √
8 √
◦ From above we have 1+i = 2 · eiπ/4 , so (1 + i)8 = 2 · eiπ/4 = ( 2)8 · e8iπ/4 = 24 · e2iπ = 16 .
(Note how much easier this is compared to multiplying 1 + i by itself eight times.)
√
• Example: Compute (1 − i 3)9 .
√ −iπ/3
√
◦ From above we have 1 − i 3 = 2 · e , so (1 − i 3)9 = 29 · e−9iπ/3 = 512 · e−3iπ = −512 .

• Taking roots of complex numbers is also fairly straightforward using the polar form, since we may interpret
roots as fractional powers. We do need to be slightly careful, since there are in general n dierent nth roots
of any nonzero complex number when n is a positive integer.

◦ Explicitly, suppose we wish to solve z n = reiθ . If we write z in polar form as z = seiϕ , then z n = sn einϕ .
n iθ
By the uniqueness of the modulus we have s = r and also e = einϕ , which is equivalent to having
θ = nϕ + 2kπ for some integer k.
θ 2kπ
◦ s and ϕ yields s = r1/n (the real nth root of the nonnegative real number r), and ϕ = +
Solving for
n n
for some integer k . We can see that for k = 0, 1, . . . , n−1, these values of ϕ yield distinct complex numbers
but that any other value of k will simply repeat one of these n values, since its argument will dier from
one of these by an integer multiple of 2π .
√
◦ Thus, we see that the n possible nth roots of z = reiθ are n re(θ+2πk)i/n for integers k = 0, 1, . . . , n − 1.

• Example: Find all complex square roots of 2i.

√
◦ We are looking for square roots of 2i = 2eiπ/2 . By the formula, the two square roots are 2ei[π/4+kπ]
√ √ 5iπ/4
for k = 0, 1, which are 2eiπ/4 and 2e .

◦ Converting from exponential to rectangular form using Euler's formula gives the two square roots in
a + bi form as 1+i and −1−i .

◦ Indeed, we can easily multiply out to verify that (1 + i)2 = (−1 − i)2 = 2i, as it should be.

• Example: Find all complex numbers z = a + bi with z 3 = 1.

◦ We are looking for cube roots of 1 = 1 · e0 . By the formula, the three cube roots of 1 are 1 · e2kiπ/3 , for
0 2πi/3 4πi/3
k = 0, 1, 2, which are e ,e ,e .

5
 √ √
1 3 1 3
◦ Converting to rectangular form using Euler's formula gives the roots as 1, − + i, − − i in
2 2 2 2
a + bi form.

• The n solutions to the equation z n = 1 are called the nth roots of unity, and are given explicitly by
1, ζn , ζn2 , . . . , ζnn−1 where ζn = e2πi/n = cos(2π/n) + i sin(2π/n).

◦ Since their arguments are the multiples of 2π/n, the nth roots of unity form the vertices of a regular
n-gon on the unit circle.

◦ Additionally, we can use the nth roots of unity to give the factorization of the polynomial z n − 1:
specically, we have z − 1 = (z − 1)(z − ζn ) · · · (z − ζnn−1 ).
n

◦ The set G = {1, ζn , . . . , ζnn−1 } is closed under multiplication hence is a nite multiplicative group; indeed,
G is a cyclic group of order n that is generated by ζn since all of the elements are powers of ζn .

◦ The roots of unity have many interesting properties and show up often in algebra, number theory, and
analysis (particularly Fourier analysis).

• Although we can always extract roots by converting to exponential form, if the argument is irrational then it
can be dicult to simplify the resulting expressions back into a + bi form, even when the results turn out to
be pleasant.

◦ For example, if we wish to nd the complex square roots of

√15 + 8i, converting to exponential form
yields 15 − 8i = 17ei·arctan(8/15) , which yields square roots ± 17ei·arctan(8/15)/2 ; this expression is not
particularly easy to evaluate.

◦ We can give an explicit formula for evaluating square roots directly in a + bi form: explicitly, the two
√
a+ a2 + b2
square roots of z = a + bi are given by ±(c + di) where c is the positive square root of and
2
b
d= .
2c
√
2 2 a+ a2 + b2 b2
◦ To verify this formula we simply observe that c −d = + √ =a and 2cd = b,
2 a + a2 + b2
2 2 2
so that (c + di) = (c − d ) + 2cdi = a + bi as claimed.

• Example: Find the complex square roots of 15 + 8i.

r √
15 + 152 + 82 8
◦ Per the formula with a = 15 and b = 8, we obtain c= =4 and d= = 1, so the
2 2·4
square roots are ±(4 + i) .

• Many textbooks introduce complex numbers as a tool for giving meaning to the formal symbols obtained
when using the quadratic formula to solve quadratic equations that do not have real solutions.

◦ Explicitly, if a, b, c are real numbers and a 6= 0, then we may complete the square in the expression
2 b 2 4ac−b2
az + bz + c and write it as a(z + 2a ) + 4a .
◦ We may then obtain the usual quadratic formula for the roots of the polynomial
√ az 2 + bz + c = 0; namely,
−b ± b2 − 4ac
z= .
2a
◦ In the situation where b2 − 4ac < 0, the solutions are not real numbers but rather complex numbers. For
√
−2 ± −4 2
example, it indicates that the solutions to z + 2z + 2 = 0 are z = = −1 ± i.
2
◦ Indeed, we can check that if we evaluate the expression z 2 + 2z + 2 when z = −1 + i or −1 − i, we obtain
0.

◦ We can continue further by factoring the polynomial as

√ √ az 2 + bz + c = a(z − r1 )(z − r2 ) where r1 =
−b − b2 − 4ac −b + b2 − 4ac
and r2 = are the two roots.
2a 2a

6
 ◦ Using the formula for simplifying square roots given above, we may nd the roots of any quadratic
polynomial with complex coecients.

• Example: Solve the quadratic z 2 − (2 + i)z + (6 + 6i) = 0.

p √
(2 + i) ± (2 + i)2 − 4(6 + 6i) (2 + i) ± −21 − 20i
◦ The quadratic formula yields z = = so we must
2 2
compute the square roots of −21 − 20i.
r
−21 + 29 −20
◦ Per the formula with a = −21 and b = −20, we obtain c = = 2 and d = = −5, so the
2 2·2
square roots are ±(2 − 5i).

(2 + i) ± (2 − 5i)
◦ Thus we obtain the roots z = = 2 − 2i, 3i .
2
• More generally, the Fundamental Theorem of Algebra, which we will prove later, says that any polynomial
equation an z n + an−1 z n−1 + · · · + a0 can be completely factored as a product of linear terms over the complex
numbers. (This a foundational result in algebra and the rst complete and correct proof was given by Argand
and Gauss in the early 1800s.)

1.1.3 Basic Topology of C

• We now introduce some useful topological properties of subsets of C that are primarily for our convenience
later.

• Denition: If a ∈ C, the open disc with radius r>0 centered at z is the set Dr (a) = {z ∈ C : |z − a| < r}.
• If S is a subset of C, we may distinguish three dierent classes of points in C relative to S, based on their
behaviors when we draw balls around them.

1. If we can draw an open disc of positive radius around a that is entirely contained in S, then we call a an
interior point of S. Interior points are necessarily contained in S.
2. If we can draw an open disc of positive radius around a that is entirely contained in Sc, the complement
of S, then we call a an exterior point point of S. (Equivalently, it is an interior point of Sc, and thus is
c
necessarily contained in S .)

3. Otherwise, no matter what open disc we draw centered at a, it will always contain some points in S and
some points in Sc, in which case we call a a boundary point of S. Boundary points may be in S or in
Sc.
◦ Intuitively, the interior points of S are a positive distance away from S c , the exterior points of S
are a positive distance away from S , and the boundary points of S are an arbitrarily small distance
away from both S and Sc.

• Denition: If S is a subset of C, the interior of S, denoted int(S), is the set of its interior points. A set S is
open if all its points are interior points, which is equivalent to saying it contains none of its boundary points.
The boundary of S, denoted ∂(S), is the set of its boundary points. A set S is closed if it contains all its
boundary points, which is equivalent to saying its complement is open.

◦ There are various equivalent denitions and properties of open sets: for example, S is open if and only
if it is a union (not necessarily nite) of open discs.

◦ Example: For any a∈C and any r > 0, if S is the open disc |z − a| < r, then S actually an open set,
since all its points are interior points. Explicitly, for any z ∈ S , if r − |z − a| = s > 0, then by the triangle
inequality, the open disc Ds/2 (z) is contained in the set. Alternatively, the set of boundary points is the
circle |z − a| = r, and none of these points are in S.
◦ Example: For any a∈C and any r > 0, if S is the closed disc |z − a| ≤ r, then S actually is a closed
set, since its boundary is the circle |z − a| = r and all of these points are in S .
◦ Example: Any nite subset of C is closed, since all of the nitely many points are boundary points. As
a consequence, the complement of any nite subset of C is open.

7
 ◦ Example: The sets |z| > 1 and 0 < Re(z) < 1 are both open. Their boundaries are respectively the unit
circle |z| = 1 and the pair of lines Re(z) = 0, 1.
◦ Example: The sets |z| ≥ 1 and 0 ≤ Re(z) ≤ 1 are both closed. Their boundaries are respectively the
unit circle |z| = 1 and the pair of lines Re(z) = 0, 1.
◦ Example: The unit circle |z| = 1 is closed. This set is its own boundary.

◦ Example: The empty set is both open and closed, as is C itself. (In fact, these are the only two subsets
of C that are both closed and open, though this is harder to prove than it might seem .)
2

• Although there is much more to say about the topological properties of arbitrary subsets of C (which can have
quite pathological properties!), we are primarily interested in open regions and their associated boundaries,
which tend to behave much more nicely.

• Denition: A region in C consists of an open set along with a subset (possibly empty, possibly all) of its
boundary points.

◦ Examples: The sets |z| < 1, |z| ≤ 1, |z| > 1, |z| ≥ 1, and {z : z < 1} ∪ {−1, i} are all regions in C.
◦ Examples: The sets C and 0 < Re(z) ≤ 1 are regions in C.

• Another important notion is that of connectedness.

◦ Intuitively, a space is connected if it cannot be decomposed into separate pieces, in the sense that the
pieces do not overlap one another's closures.

◦ We also have another way to phrase this idea of connectedness: namely, if any two points in the set can
be joined by a continuous path inside the set.

• Denition: If S is a subset of C, S is connected if it cannot be written as the union of two nonempty subsets
with disjoint closures, and S is path-connected if for any two points z1 and z2 in S , there exists a continuous
function f : [0, 1] → C whose image is contained in S and such that f (0) = z1 and f (1) = z2 .

◦ There are various equivalent formulations of these conditions: for example, a connected set is equivalently
one that cannot be written as the union of two nonempty disjoint proper open subsets.

◦ In general any path-connected subset is necessarily connected (the proof follows by the same argument
used to show that ∅ and C are the only subsets of C that are both closed and open), but the converse is
not true in general .
3

◦ However, connectedness and path-connectedness are equivalent for open sets, and thus also for regions .
4

◦ Since path-connectedness is much more concrete and easier to visualize, we will generally phrase things
in terms of path-connectedness.

◦ Example: The regions C, |z| < 1, |z| ≤ 1, |z| > 1, |z| ≥ 1, {z : z < 1} ∪ {−1, i}, and 0 < Re(z) ≤ 1 are
all path-connected, since in fact any two points in these regions are joined by a line inside the region.

◦ Example: The region with |z| < 1 or |z| > 2 is not path-connected, since there is no way to draw
a continuous path from a point in the rst component to a point in the second one without passing
through points with 1 ≤ |z| ≤ 2, which are not in the region. (Here, we can see that this region consists
of two connected components, namely |z| < 1 and |z| > 2.)
2 For completeness: if S or S c is empty the result is trivial so suppose
S is both open and closed in C and let z ∈ S and w ∈ S c .
Consider the real-valued function f : [0, 1] → C given by f (t) = tz + (1 − t)w , which is simply the graph of the line segment from z to w ,
and dene T to be the subset of elements t ∈ [0, 1] such that f (t) ∈ S . Then since f (1) = w ∈ S c , T is a nonempty proper subset of S .
If α is the least upper bound of T , then by denition there exist elements of T and of T c that are arbitrarily close to α (otherwise, we
could decrease or increase α respectively). Applying f shows that there are elements of S and of S c that are arbitrarily close to f (α),
and so f (α) is a boundary point of S . But if f (α) ∈ S then S is not open, while if f (α) ∈ S c then S is not closed, contradiction.
3 A standard counterexample example is the infamous topologist's sine curve given by the graph of y = sin(1/x) for x 6= 0 along
with the origin. This set is not path-connected (there is no way to draw a path from (0, 0) to any other point on the graph) but it is
connected (there is no way to separate the graph into two pieces without causing the two pieces' closures to intersect).
4 For S is a nonempty connected open set and let z ∈ S . Dene U to be the set of points in S that can be joined
this, suppose that
by a path to z V to be the subset of points in S that cannot be joined by a path to z . For any w ∈ U , since S is open there
and dene
exists an open disc of radius r > 0 such that Dr (w) ⊆ S . But for every point in this disc, there is a path joining it to w (namely, along
an appropriate radius) and then following this path from w to z shows that this point is also connected to z . Thus in fact Dr (w) ⊆ U
so U is open. Similarly, V is open: for w 0 ∈ V we obtain a similar disc Dr (w 0 ) inside S , and then any point in the disc Dr (w 0 ) cannot
be connected to z , since otherwise w 0 would be. But then S is the union of two disjoint open subsets, contradicting connectedness.

8
 • There is one other useful notion regarding regions: that of boundedness.

• Denition: If S is a subset of C, S is bounded if there exists some positive radius R such that S is contained
in the open disc of radius R centered at 0, and otherwise S is unbounded.

◦ Equivalently, S is bounded if there exists some R > 0 with |z| < R for all z ∈ S . Inversely, S is unbounded
when there exists a sequence of elements z1 , z2 , . . . , zn , . . . in S such that |zn | → ∞ as n → ∞.
◦ Example: The regions |z| < 1, |z| ≤ 1, and {z : z < 1} ∪ {−1, i} are all bounded, since they all lie inside
the disc of radius 2 centered at 0.

◦ Example: The regions C, |z| > 1, |z| ≥ 1, and 0 < Re(z) ≤ 1 are all unbounded, since they contain
elements of arbitrarily large absolute value.

1.2 The Complex Derivative and Holomorphic Functions

• Now that we have discussed the arithmetic of the complex numbers, we begin our study of complex-valued
functions f : C → C.

• Our rst main goal is to generalize the notion of the derivative of a function to the complex case. There is a
very natural way to try to do this; namely, by dening the complex derivative of a function as a limit in the
same way as with a real-valued function.

◦ Explicitly, iff : C → C, we would like to try dening the complex derivative at a point z0 ∈ C as

0 f (z) − f (z0 )
f (z0 ) = lim .
z→z0 z − z0
◦ This is in fact the denition we will eventually use, but we must rst review some properties regarding
limits of complex-valued functions.

1.2.1 Limits

• We begin by dening limits of complex-valued functions. We can use essentially the same denition as in the
case of real-valued functions:

• Denition: A function f :C→C has the limit L as z → a, written as lim f (z) = L if for any >0 (no
z→a
matter how small) there exists a δ>0 (depending on ) such that for all z∈C with 0 < |z − a| < δ , we have
|f (z) − L| < .

◦ This is simply the usual -δ denition of limit but with the variables having domain C rather than R.
◦ The usual intuition is as follows: suppose you claim that the function f (x) has a limit L, as x gets close
to a. In order to prove to me that the function really does have that limit, I challenge you by handing
you some value  > 0, and I want you to give me some punctured disc 0 < |z − a| < δ (a disc centered
at z=a with radius δ , excluding the center) with the property that f (z) is always within  for z in that
disc, except possibly at a.

◦ If f (z) really does stay close to the limit value L as z gets close to a, then, no matter what value of  I
picked, you should always be able to answer my challenge with some punctured disc, because the values
of f (z) should stay near L when z is near a.

• We can mostly avoid using the formal denition as a practical matter by instead establishing a number of
basic limit evaluations and limit rules, as follows:

• Proposition (Basic Limits): Let a∈C and suppose f and g have lim f (z) = Lf and lim g(z) = Lg .
z→a z→a

1. We have limz→a z = a and for any c∈C we have limz→a c = c.

2. For any c∈C we have limz→a cf (z) = cLf , limz→a |f (z)| = |Lf |, and limz→a f (z) = Lf .
3. We have limz→a [f (z) + g(z)] = Lf + Lg , limz→a [f (z) − g(z)] = Lf − Lg , and limz→a f (z)g(z) = Lf Lg .

9
 f (z) Lf
4. If Lg 6= 0, we have lim = .
z→a g(z) Lg
5. If L = x + yi, then limz→a f (z) = L if and only if limz→a Re[f (z)] = x and limz→a Im[f (z)] = y .
◦ These results are all standard applications of the limit denition, and the proofs are essentially the
same as the corresponding results for real-valued functions.

◦ For example, for the rst part of (2), let  > 0. If c = 0 the result is trivial since cf (z) = 0 = cLf
so we may choose any positive δ . Otherwise with c 6= 0, since limz→a f (z) = Lf there exists δ > 0
such that |f (z) − Lf | < / |c| for 0 < |z − a| < δ . Then we have |cf (z) − cLf | = |c| · |f (z) − Lf | < 
as desired.

◦ We remark that (5), in particular, allows us to reduce any question about limits of complex functions
to the corresponding real-valued limits of its real and imaginary parts.

5−z
• Example: Find limz→2−i .
2|z|

5−z limz→2−i (5 − z) limz→2−i 5 − limz→2−i z 5 − (2 − i)

◦ Using the limit rules we have limz→2−i = = = =
2|z| limz→2−i 2|z| 2| limz→2−i z| 2|2 − i|
3+i
√ .
2 5

• We also have the natural notion of continuity:

• Denition: If f :C→C is a complex-valued function, we say f is continuous at a∈C if limz→a f (z) = f (a).
If f is continuous on its entire domain, we say f is continuous everywhere (or often, just continuous).

◦ In other words, a continuous function is one whose limit as z→a is simply the value of the function at
a.
◦ Per limit properties (1)-(3), we see that sums, dierences, products, and quotients (with nonzero denom-
inator) of continuous functions are continuous.

◦ Example: Since polynomials are constructed from constants and the variable z using addition, subtrac-
tion, and multiplication, by the above we see that any polynomial p(z) is continuous everywhere. More
p(z)
generally, by property (4) we see that any rational function is continuous whenever q(z) 6= 0.
q(z)

• Even more generally, by (5), for any continuous real-valued functions p(x, y) and q(x, y) we see that the
function f (x + iy) = p(x, y) + iq(x, y) is a continuous complex-valued function.

◦ Example: The complex-conjugation function f (z) = z is continuous, since it is given by f (x+iy) = x−iy
and the real and imaginary components are both continuous.
p
◦ Example: The modulus function f (z) = |z| is continuous, since it is given by f (x + iy) = x2 + y 2 and
the real and imaginary components are both continuous.

◦ Example: The complex exponential function f (z) = ez is continuous, since it is given by f (x + iy) =
x x
e cos y + ie sin y and the real and imaginary components are both continuous.

• We also have various substitution results for continuous functions:

• Proposition (Substitution and Continuity): Let f : C → C.

1. For any g : C → C, if limz→a g(z) = L exists and f is continuous at L, then limz→a f (g(z)) = f (L).
◦ Proof: Let  > 0. Since f L, there exists δ1 > 0 such that |w − L| < δ1 implies
is continuous at
|f (w) − f (L)| < . w = L since f is continuous.)
(Note that we can include

◦ Additionally, since limz→a g(z) = L, there exists δ > 0 such that 0 < |z − a| < δ implies |g(z) − L| <
δ1 .
◦ Thus, taking w = g(z), we see that if 0 < |z − a| < δ then |g(z) − L| < δ1 hence |f (g(z)) − f (L)| < ,
so limz→a f (g(z)) = f (L) as desired.

10
 2. If g is continuous at z = a and f is continuous at g(a), then the composition f ◦g is continuous at z = a.
Thus, the composition of continuous functions is continuous.

◦ Proof: Immediate from (1).

3. (Limits Along Curves) If limz→a f (z) exists and c : R → C is continuous with c(0) = a, then limz→a f (z) =
limt→0 f (c(t)).
◦ In other words, if c is a continuous curve passing through z = a, then if f (x, y) has a limit at z = a,
we may compute the limit by approaching a along the curve c.
◦ Proof: The same as (1) with g:C→C changed to c : R → C.
4. (Two-Paths Test) If c : R → C and d : R → C are continuous with c(0) = d(0) = a, and limt→0 f (c(t)) 6=
limt→0 f (d(t)), then limz→a f (z) does not exist.
◦ In other words, if f has two dierent limits along the paths c, d as z → a, then f has no limit at
z = a.
◦ Proof: By (3), if limz→a f (z) exists, then limt→0 f (c(t)) and limt→0 f (d(t)) would both equal
limz→a f (z). Since they are not equal, the limit cannot exist.

• We can use (4) above to give a convenient way of establishing that a limit limz→a f (z) does not exist: namely,
by nding two dierent paths approaching a along which f (z) has dierent limits.

z2 z
• Example: Show that limz→0 2 = limz→0 does not exist.
|z| z

◦ We try various paths approaching zero. Along the real axis, with c(t) = t as the real parameter t → 0,
t2
we obtain the limit limt→0 2 = limt→0 1 = 1.
|t|
(it)2
◦ Along the imaginary axis, with c(t) = it as t → 0, we obtain the limit limt→0 2 = limt→0 −1 = −1.
|it|
◦ Since the limits along these paths are dierent, the original limit does not exist.

(x2 − y 2 ) + 2xyi
◦ Remark: In terms of real and imaginary parts, this limit is equivalent to lim(x,y)→(0,0)
x2 + y 2
where x = x + iy . To establish the nonexistence of the given limit, we could equivalently have shown
x2 − y 2
that the real limit lim(x,y)→(0,0) does not exist.
x2 + y 2

z2 − z2
• Example: Show that limz→0 does not exist.
zz
◦ We try various paths approaching zero. Along the real axis, with c(t) = t as the real parameter t → 0,
t2 − t2
we obtain the limit limt→0 = limt→0 0 = 0.
t·t
(it)2 − (−it)2
◦ Along the imaginary axis, with c(t) = it as t → 0, we obtain the limit limt→0 = limt→0 0 =
it · (−it)
0.
◦ Although the limits along these paths are equal, this does not mean the original limit exists.

(t + it)2 − (t − it)2
◦ Let us try along the line c(t) = t + it as t → 0. We obtain the limit limt→0 =
(t + it)(t − it)
4it2
limt→0 = 2i.
2t2
◦ Since the limit along this path is dierent from the limit along the other paths, the original limit does
not exist.

11
1.2.2 The Complex Derivative

• We now have enough background on limits to dene the derivative. We can use the same denition as the
usual one for real-valued functions:

• Denition: If f : C → C is a complex-valued function, the complex derivative f 0 (a) at a point z0 ∈ C is the

f (z) − f (z0 )
limit lim , assuming it exists.
z→z0 z − z0

◦ As with the derivative of a real-valued function, we can equivalently state this denition as f 0 (z) =
f (z + h) − f (z)
lim .
h→0 h
◦ In some cases the rst denition is easier to use, while in others the second denition is easier to use.

• Example: Verify that the complex derivative of f (z) = z 2 exists everywhere and compute it.

f (z) − f (z0 ) (z − z0 )(z + z0 )

◦ We compute lim = lim = lim (z + z0 ) = 2z0 .
z→z0 z − z0 z→z0 z − z0 z→z0

◦ Since the limit always exists, we can say that f 0 (z0 ) = 2z0 everywhere.

◦ Note that this agrees with, and in fact extends, the (ordinary) real-valued derivative of the function
f (x) = x2 on the real line.

• Example: Verify that the complex derivative of f (z) = 2z 3 + 5 exists everywhere and compute it.

f (z) − f (z0 ) (2z 3 + 5) − (2z03 + 5)

◦ We compute lim = lim = lim (2z 2 + 2zz0 + 2z03 ) = 3z02 .
z→z0 z − z0 z→z0 z − z0 z→z0

◦ Since the limit always exists, we can say that f 0 (z0 ) = 3z02 everywhere.

◦ Notice again that this agrees with the real-valued derivative of the function f (x) = 2x3 + 5.

• Example: Determine at which points the complex derivative of f (z) = z exists.

f (z) − f (z0 ) z − z0
◦ The required limit is lim = lim . There does not seem to be a natural way to
z→z0 z − z0 z→z0 z − z0
simplify this expression further.

◦ Let us try to compute the limit along dierent paths approaching z0 .

(z0 + t) − z0 t
◦ Along the horizontal line z = z0 +t, for a real parameter t → 0, the limit becomes lim = lim = 1.
t→0 (z0 + t) − z0 t→0 t

(z0 + it) − z0 −it

◦ Along the vertical line z = z0 +it for a real parameter t → 0, the limit becomes lim = lim = −1.
t→0 (z0 + it) − z0 t→0 it

◦ Along these two paths the limit has dierent values, so the overall limit does not exist at any point z0 .
This means the derivative does not exist at any point.

• Example: Determine at which points the complex derivative of f (z) = zz exists.

 
(z + h)(z + h) − zz hz + hz + hh h h
◦ The required limit is lim = lim = lim z + h + z = z + lim z .
h→0 h h→0 h h→0 h h→0 h

h
◦ We can see that if z=0 then the limit lim z is simply zero (since the expression itself is zero), so for
h→0 h
z=0 the derivative exists and is zero.

h
◦ Otherwise, if z 6= 0, then the limit lim z does not exist since the limit along the real axis is z while the
h→0 h
value on the imaginary axis is −z .
◦ We conclude that f (z) is only dierentiable at z=0 , and its derivative there is 0.

12
• Since our denition of the complex derivative is exactly the same as that of the derivative of a real-valued
function, it is quite reasonable to expect that all of the usual dierentiation rules will apply. This is indeed
the case.

• Proposition (Dierentiation Rules): Suppose f and g are both complex-dierentiable at z = a. Then the
following hold:

1. If f is dierentiable at a then f is continuous at a.

2. (Sums/Dierences) We have (f + g) (a) = f (a) + g 0 (a)
0 0
and (f − g)0 (a) = f 0 (a) − g 0 (a).
3. (Product Rule) We have (f g)0 (a) = f 0 (a)g(a) + f (a)g 0 (a).
f 0 (a)g(a)−f (a)g 0 (a)
4. (Quotient Rule) If g(a) 6= 0, we have (f /g)0 (a) = g(a)2 .

5. (Chain Rule) If g is dierentiable at f (a) then (f ◦ g) (a) = f (g(a))g 0 (a).

0 0

◦ The proofs of all of these are exactly the same as the usual proofs in single-variable calculus.
f (z) − f (a)
◦ For example, for (1) we have limz→a [f (z) − f (a)] = limz→a · (z − a) = f 0 (a) · 0 = 0 hence
z−a
limz→a f (z) = f (a) as claimed.

6. (Polynomials) Any complex polynomial p(z) = an z n + an−1 z n−1 + · · · + a1 z + a0 is dierentiable with

0 n−1
complex derivative p (z) = nan z + (n − 1)an−1 z n−2 + · · · + a1 .
◦ Example: The complex derivative of p(z) = iz 4 − (3 + i)z + 5 is p0 (z) = 4iz 3 + (3 + i).
◦ Observe in particular that this is exactly the same formula for dierentiating a real polynomial.

◦ Proof: Apply (1)-(4) and the fact that the complex derivative of z is 1.

• As a consequence of (6) above we also see that rational functions of z are complex-dierentiable on large
regions (namely, their entire domains, which consist of the entire complex plane except the nite number of
points where their denominators are zero).

◦ Our primary objects of study will be functions that are complex-dierentiable on such large regions,
rather than functions like f (z) = zz that are only dierentiable on a small set, so we give them a simpler
name.

• Denition: A function whose complex derivative exists on a region U is said to be holomorphic on U.

◦ Although being holomorphic seems to be a relatively mild condition, it actually turns out to be quite
restrictive. As we will see in later chapters, holomorphic functions have a large number of unexpectedly
pleasant properties.

◦ For example, even though by denition a holomorphic function only possesses a rst derivative, in fact
a holomorphic function necessarily has derivatives of all orders (compare with the situation with real-
dierentiable functions, which may not even have a second derivative).

◦ Furthermore, as essentially an immediate consequence of having derivatives of all orders, holomorphic

functions may be represented locally by power series which (as we will show) necessarily have a positive
radius of convergence on which they converge to the value of the original function.

• We would like to be able to establish the holomorphicity of most nice functions, such as ez .

◦ However, dierentiating more complicated functions directly from the denition is quite a bit more
painful. (Try dierentiating ez using the denition, for example!)

• We can also attempt to compute derivatives for functions written in terms of real and imaginary parts.
However, when we try to compute the limit using the denition, we will end up with a mess, since it requires
writing everything in terms of the original limit variable z.

◦ We can always convert any function of the form p(x, y) into one in terms of z = x + iy and z = x − iy
by substituting x = (z + z)/2 and y = (z − z)/(2i).
◦ Example: The function f (x+iy) = 4x2 −6iy is equivalent to f (z) = 4[ z+z 2 z−z 2
2 ] −6i[ 2i ] = (z +z) −3(z −z).
◦ In some cases we can still use the two-paths approach directly to see that the derivative does not exist.

13
 • Example: Show that the complex derivative of f (x + iy) = (x2 + y) + (2x + y 2 )i does not exist at any point
z = x + iy .

◦ Let us try to compute the limit along dierent paths approaching z0 = x0 + y0 i.

◦ Along the horizontal line z = z0 + t (i.e., x = x0 + t, y = y0 ) for a real parameter t → 0, the limit
(x0 + t)2 + 2i(x0 + t) − (x20 + 2ix0 ) 2x0 t + t2 + 2it
becomes lim = lim = 2x0 + 2i.
t→0 t t→0 t
◦ Along the vertical line z = z0 + it (i.e., x = x0 , y = y0 + it) for a real parameter t → 0, the limit becomes
(y0 + t) + i(y0 + t)2 − (y0 + iy02 ) t + i(2y0 t + t2 )
lim = lim = −i + 2y0 .
t→0 it t→0 it
◦ Along these two paths the limits are always dierent (since x0 and y0 are real), so the complex derivative
does not exist .

• It is much harder to show that a complex derivative does exist when the function is described in terms of real
and imaginary parts. One way to do this is to convert back into an expression involving z and z.
x − iy
• Example: Show that f (x + iy) = has a complex derivative for x + iy 6= 0, and compute it.
x2 + y 2
z 1
◦ We rewrite f in terms of z and z: noting that x − iy = z and x2 + y 2 = zz , we see that f (z) = = .
zz z
1 1
◦ Thus, f (z) = , so by the usual dierentiation rules, we have f 0 (z) = − for z 6= 0.
z z2

1.2.3 Partial Derivatives and the Cauchy-Riemann Equations

• There is another possible approach to complex dierentiation that is motivated by multivariable calculus.

◦ As we have already been discussing, we can equivalently consider any complex-valued function f (z) =
f (x + iy) as a function of the two variables x and y representing the real and imaginary parts of z.
◦ This separation of the real and imaginary parts allows us to calculate partial derivatives of f with respect
to x and y.
∂f ∂f
• Denition: If f (z) = f (x + iy) is a complex-valued function, we dene the partial derivatives and at a
∂x ∂y
∂f f (a + h + bi) − f (a + bi) ∂f f (a + hi + bi) − f (a + bi)
point z = z+bi to be (a+bi) = lim and (a+bi) = lim ,
∂x h→0 h ∂y h→0 h
where both limits are single-variable real limits.

∂f
◦ These complex partial derivatives have the same interpretation as in multivariable calculus: represents
∂x
the rate of change of f as the variable x changes but y remains constant (i.e., the rate of change in the
∂f
real direction), while represents the rate of change of f as the variable y changes but x remains
∂x
constant (i.e., the rate of change in the imaginary direction).

◦ When computing partial derivatives, we may simply dierentiate the real and imaginary parts separately
with respect to the appropriate variable, and all of the other familiar properties of partial derivatives
also apply here (e.g., the product rule, the quotient rule, and the chain rule).

∂f ∂f
• Example: For f (x + iy) = (x2 + y) + (2x + y 2 )i we have = 2x + 2i and = 1 + 2yi.
∂x ∂y
∂f ∂f
• Example: For f (x + iy) = (ye2x + x2 y) + (2 sin(x + y) − 4y 2 )i, nd the partial derivatives and .
∂x ∂y
∂f ∂f
◦ Dierentiating yields = (2ye2x + 2xy) + (2 cos(x + y))i and = (e2x + x2 ) + (2 cos(x + y) − 8y)i .
∂x ∂y

14
 ∂f ∂f
• Example: For f (z) = z 2 , nd the partial derivatives and .
∂x ∂y

◦ We rst write f write out explicitly in terms of real and imaginary parts as f (x + iy) = (x2 − y 2 ) + 2xyi.
∂f ∂f
◦ Now evaluating the partial derivatives yields = 2x + 2yi and = −2y + 2xi .
∂x ∂y
• More fruitfully, instead of working with the variables x and y, we could apply the change of variables x=
(z + z)/2 and y = (z − z)/(2i) to write the original function in terms of z and z.

◦ By doing this, we can now view our complex-valued function f as a function of the two variables z and
z, and take partial derivatives with respect to z and z.
◦ When f is written in terms of z
z already, these partial derivatives are easy to evaluate
and (we just
view f as a two-variable function of z and z and evaluate partial derivatives, just as above):
∂f ∂f
◦ Example: For f (z) = z 2 , we have = 2z and = 0.
∂z ∂z
∂f ∂f
◦ Example: For f (z) = z , we have = 0 and = 1.
∂z ∂z
∂f ∂f
◦ Example: For f (z) = zz , we have = z and = z.
∂z ∂z
• When f is written in terms of x and y instead, we can evaluate the partial derivatives by direct substitution,
but we can avoid some of the work by using the multivariable chain rule.

∂f
◦ x = (z + z)/2 and y = (z − z)/(2i), by the multivariable chain rule we
Explicitly, since have =
    ∂z
∂f ∂x ∂f ∂y 1 ∂f ∂f ∂f ∂f ∂x ∂f ∂y 1 ∂f ∂f
· + · = −i and = · + · = +i .
∂x ∂z ∂y ∂z 2 ∂x ∂y ∂z ∂x ∂z ∂y ∂z 2 ∂x ∂y
∂f ∂f
◦ Remark: Formally, these are actually the denitions of the partial derivatives and .
∂z ∂z
∂f ∂f
• Example: For f (x + iy) = (x2 + y) + i(2x + y 2 ), nd and .
∂z ∂z
∂f ∂f
◦ First, we compute = 2x + 2i and = 1 + 2yi.
∂x ∂y
 
∂f 1 ∂f ∂f 1 1 1 1 1 1
◦ Thus, = −i = [(2x + 2i) − i(1 + 2yi)] = [(2x − 2y) + 3i] = ( z + z) + (− z + z +
∂z 2 ∂x ∂y 2 2 2 2 2 2
1
)i.
2
 
∂f 1 ∂f ∂f 1 1 1 1 1
◦ Likewise, = +i = [(2x + 2i) + i(1 + 2yi)] = [(2x − 2y) + 3i] = ( z + z) + ( z −
∂z 2 ∂x ∂y 2 2 2 2 2
1 3
z + )i.
2 2
◦ We may check this by substituting in x = z+z z−z
2 and y = 2i and expanding to obtain f (z) = ( 14 z 2 +
1 1 2
2 zz + 4z ) + ( 12 z − 1 2
4z + 3
2z + 1
2 zz − 1 2
4 z )i, and then dierentiate explicitly.
∂f 1 1 1 1 1 ∂f 1 1 3 1 1
◦ This yields = ( z + z) + ( − z + z)i and = ( z + z) + ( + z − z)i, which (of course)
∂z 2 2 2 2 2 ∂z 2 2 2 2 2
agrees with the calculations above.

∂f ∂f
• Example: For f (x + iy) = ex (cos y + i sin y), nd and .
∂z ∂z
◦ Notice that this function is simply the complex exponential f (z) = ez , and it is very tempting simply
to dierentiate formally to obtain the requested derivatives. However, we have not established the usual
dierentiation rule for the complex exponential yet, so we must instead revert to using real and imaginary
parts.

15
 ∂f ∂f
◦ We have = ex (cos y + i sin y) = ez and = ex (− sin y + i cos y) = iez .
∂x ∂y
   
∂f 1 ∂f ∂f 1 z z z ∂f 1 ∂f ∂f 1
◦ Thus = −i = [e − i(ie )] = e and = +i = [ez + i(iez )] = 0.
∂z 2 ∂x ∂y 2 ∂z 2 ∂x ∂y 2
◦ Of course, these do agree with the intuitively-sensible partial derivatives for the complex exponential!

∂f
• Let us now investigate the relationship between the complex derivative f 0 (z) and the partial derivatives
∂z
∂f
and .
∂z
◦ First, observe that possessing a complex derivative is a much stronger property than simply possessing
partial derivatives.
∂f ∂f
◦ For example, the partial derivatives of f (z) = z are certainly dened, since =0 and = 1, but
∂z ∂z
this function is not complex-dierentiable anywhere.
∂f ∂f
◦ Likewise, for f (z) = zz , the partial derivatives = z and = z are dened, but the complex
∂z ∂z
derivative is only dened when z = 0.
◦ f (x + iy) = (x2 + y) + i(2x + y 2 ), the complex derivative does not exist anywhere while the
Similarly, for
∂f 1 1 1 1 1 ∂f 1 1 3 1 1
partial derivatives are = ( z + z) + ( − z + z)i and = ( z + z) + ( + z − z)i.
∂z 2 2 2 2 2 ∂z 2 2 2 2 2
◦ On the other hand, if f (z) is a polynomial in z , say f (z) = an z n + an−1 z n−1 + · · · + a1 z + a0 , then it is
0 n−1
complex-dierentiable everywhere with derivative f (z) = nan z + (n − 1)an−1 z n−2 + · · · + a1 , while
∂f ∂f
its partial derivatives are = nan z n−1 + (n − 1)an−1 z n−2 + · · · + a1 and = 0.
∂z ∂z
∂f ∂f
◦ Additionally, for the complex exponential f (z) = ez , we computed = ez and = 0. We did not
∂z ∂z
compute the complex derivative using the denition, but it seems reasonable to expect that it should
exist and equal ez (which in fact it does).

◦ Notice that in all of the examples we have listed where the function is complex-dierentiable (or where
we expect it to be), the partial derivative with respect to z is zero and the partial derivative with respect
to z is equal to the actual complex derivative.

◦ Inversely, in all of the situations where the function is not complex-dierentiable, the partial derivative
with respect to z is not zero.

∂f
• In fact, we can show that if the complex derivative exists, then in fact it equals the partial derivative ,
∂z
∂f
and also the partial derivative must equal zero.
∂z
◦ Informally, what this means is that in order for f to be complex-dierentiable, it must be a function of
z alone, not involving any z terms.

∂f ∂f
• Theorem (Cauchy): Suppose that f is complex-dierentiable at z = z0 . Then (z0 ) = f 0 (z0 ) and (z0 ) = 0.
∂z ∂z
◦ Proof: Let us explicitly write f (x + iy) = u(x, y) + iv(x, y) where u and v are real-valued, and suppose
that f is complex-dierentiable at z0 = x0 + y0 i.
0 f (z) − f (z0 )
◦ Consider the values of the limit f (z0 ) = L = lim along a horizontal and vertical line
z→z0 z − z0
through z0 .
◦ Along the horizontal line the limit is

u(x0 + t, y0 ) + iv(x0 + t, y0 ) − u(x0 , y0 ) − iv(x0 , y0 )

L = lim
t→0 t
u(x0 + t, y0 ) − u(x0 , y0 ) v(x0 + t, y0 ) − v(x0 , y0 )
= lim + i lim
t→0 t t→0 t
∂u ∂v
= +i
∂x ∂x

16
 while along the vertical line the limit is

u(x0 , y0 + t) + iv(x0 , y0 + t) − u(x0 , y0 ) − iv(x0 , y0 )

L = lim
t→0 it
u(x0 , y0 + t) − u(x0 , y0 ) v(x0 , y0 + t) − v(x0 , y0 )
= −i lim + lim
t→0 t t→0 t
∂u ∂v
= −i + .
∂y ∂y

∂u ∂v ∂u ∂v
◦ Since these expressions must be equal and u, v are real-valued, we require = =− .
and
∂x ∂y ∂y ∂x
∂f ∂u ∂v ∂f ∂u ∂v ∂f
◦ But now since = +i and = + i , we see that the partial derivative =
  ∂x ∂x ∂x ∂y ∂y ∂y   ∂z
1 ∂f ∂f ∂u ∂v 0 ∂f 1 ∂f ∂f
−i = +i = L = f (z0 ), while the partial derivative = −i = 0, as
2 ∂x ∂y ∂x ∂x ∂z 2 ∂x ∂y
claimed.

• These relations on the real and imaginary parts of f that we identied above are called the Cauchy-Riemann equations.
It turns out that having a complex derivative on a region is actually equivalent to satisfying the Cauchy-
Riemann equations:

• Theorem (Looman-Mencho ): Suppose that f (x+iy) = u(x, y)+iv(x, y). Then f has a complex derivative at
every pointz0 in an open region U if and only if f is continuous on U and u and v satisfy the Cauchy-Riemann
∂u ∂v ∂u ∂v ∂f
equations = and = − , or equivalently when = 0, at all points in U , and in such a case the
∂x ∂y ∂y ∂x ∂z
0 ∂u ∂v ∂v ∂u ∂f
complex derivative is given by f (z) = +i = −i = .
∂x ∂x ∂y ∂y ∂z
◦ This theorem is a fundamental result in complex analysis, and allows us to rapidly determine whether a
function is holomorphic.

◦ We showed the forward direction already, but we will defer the proof of the reverse direction for now,
since the most natural approach requires integration in the complex plane. (Our proof later will also
assume that the partial derivatives of f are continuous, rather than just f itself, which substantially
simplies the argument.)

◦ From the result above, we can immediately see that all of the usual properties of the derivative of a
real-valued function (such as the product rule and the chain rule) extend to the complex derivative, since
the complex derivative can be written in terms of real-valued partial derivatives.

• Example: Show that f (x + iy) = ex (cos y + i sin y) is holomorphic and that it equals its own derivative.

◦ We simply check the Cauchy-Riemann equations: we have u = ex cos y and v = ex sin y .

∂u ∂v ∂u ∂v
◦ Then = ex cos y = and = −ex sin y = − .
∂x ∂y ∂y ∂x
∂u ∂v
◦ The equations are satised, so f is holomorphic and its derivative is +i = ex cos y + iex sin y ,
∂x ∂x
which is indeed equal to the original function.

◦ Thus, we see that the complex exponential f (z) = ez is indeed holomorphic and its derivative is itself.

y + ix
• Example: Determine whether f (x + iy) = is holomorphic, and if so nd its derivative.
x2 + y 2
y x
◦ We simply check the Cauchy-Riemann equations: we have u= and v= .
x2 + y2 x2 + y2
∂u 2xy ∂v ∂u x2 − y 2 ∂v
◦ Then = 2 2 2
= and = 2 2 2
=− .
∂x (x + y ) ∂y ∂y (x + y ) ∂x
∂u ∂v 2xy + i(x2 − y 2 )
◦ The equations are satised, so f is holomorphic and its derivative is +i = .
∂x ∂x (x2 + y 2 )2

17
 i −i
◦ Remark: Note in fact that f (z) = from which we can easily compute f 0 (z) = .
z z2
• Example: Determine whether f (x + iy) = ey (cos x + i sin x) is holomorphic, and if so nd its derivative.

◦ We simply check the Cauchy-Riemann equations: we have u = ey cos x and v = ey sin x.

∂u ∂v
◦ Then = −ey sin x whereas = ey sin x. Since these are not equal, we see that f is not holomorphic .
∂x ∂y

z3
• Example: Determine whether f (z) = is holomorphic, and if so nd its derivative.
z2
◦ f is given in terms of z and z we simply compute its partial derivative with respect to z : this yields
Since
∂f z3
= −2 3 . Since this quantity is nonzero, we see that f is not holomorphic .
∂z z
◦ Observe that if we dene f (0) = 0, then we have limz→0 |f (z)| = limz→0 |z| = 0 and hence limz→0 f (z) =
0 = f (0), so f is continuous at 0.
∂f f (h)−f (0) f (ih)−f (0)
◦ Also,
∂x (0)h = limh→0 = limh→0 hh = 1 and ∂f
∂y (0) = limh→0 = limh→0 ih
h = i, so
ih h
∂f 1 ∂f ∂f ∂f
∂z (0) = 2 ∂x (0) − i ∂y (0) = 0. Thus ∂z is also zero at z = 0, meaning that f satises the Cauchy-
Riemann equations at 0.
z2
◦ However, the complex derivative of f does not exist at z =0 since the limit limz→0 f (z)
z = limz→0 z2
does not exist: its value along the real axis is 1 while its value along the line θ = π/4 is −1.
◦ Thus, this example illustrates that we cannot simply check the Cauchy-Riemann equations at a single
point to determine complex dierentiability at that point.

1.2.4 Holomorphic Functions and Angles

• Now that we have studied holomorphic functions in general, we will now discuss one very interesting geometric
property of holomorphic functions: namely, that they preserve angles.

◦ To make this more precise, suppose that we have a continuous curve c:R→C with c(t) = x(t) + iy(t).
◦ If we further assume that c is a dierentiable function (i.e., that both x(t) and y(t) are dierentiable
real-valued function), then we have a well-dened notion of the tangent vector to the curve c given by
c0 (t) = x0 (t) + iy 0 (t).
◦ Example: For c(t) = eit = cos t + i sin t, we have c0 (t) = ieit = − sin t + i cos t, which indeed yields
complex vectors tangent to the curve as is readily apparent from a graph:

◦ If we have two dierentiable curves c1 and c2 passing through the same point z0 ∈ C, we dene the angle
between c1 and c2 to be the angle between their tangent vectors at z0 .

18
 ◦ Our claim is that if f :R→C is holomorphic on a region R and z0 ∈ R, then the angle between two
curves c1 and c2 is preserved upon applying f, which is to say, the angle between c1 and c2 is the same
as the angle between f ◦ c1 and f ◦ c2 .

• Theorem (Holomorphic Functions Preserve Angles): Suppose that f : R → C is holomorphic on a region R

and that two dierentiable curves c1 and c2z0 ∈ R. Then if f 0 (z0 ) 6= 0, the angle
pass through the point
between c1 and c2 at z0 is the same as the angle between f ◦ c1 and f ◦ c2 at f (z0 ): in other words, f preserves
angles.

◦ The main idea of the proof is simply to use the chain rule, which in this situation is as follows: for a
d
dierentiable curve c and a holomorphic function f :R →C on a region R, we have [(f ◦ c)(t)] =
dt
f 0 (c(t)) · c0 (t).
◦ Proof: By the chain rule, the tangent vector to f ◦ c at f (z0 ), as a complex number, is simply the tangent
vector to c at z0 multiplied by the nonzero value f 0 (z0 ).
◦ By expressing this multiplication in polar (or exponential) form we see that it simply dilates by a xed
factor and rotates by a xed angle, so it preserves the relative angle between any two vectors.

◦ In particular, scaling by f 0 (z0 ) preserves the angle between the tangent vectors to c1 and c2 , as claimed.

◦ Remark: The argument here (that complex scalings preserve angles) is informal. We can formalize it
using the language of inner products.

◦ Explicitly, forz, w ∈ C dene the pairing hz, wi = Re(zw), which for z = a + bi and w = c + di is simply
hz, wi = ac + bd, the usual dot product of two vectors in R2 .
◦ Then by an application of the law of cosines, one may show that hz, wi = |z| |w| cos θ, where θ is the
hz, wi
directed angle between z and w, and so for nonzero z, w we have cos θ = .
|z| |w|
hαz, αwi
◦ For any nonzero α ∈ C, if ϕ is the directed angle between αz and αw we see cos ϕ = =
|αz| |αw|
2
Re(|α| zw) Re(zw)
2 = = cos θ, and thus we have ϕ=θ as claimed.
|α| |z| |w| |z| |w|

• It is very useful to work with maps preserving angles, since they tend to preserve other convenient physical
properties. We give such maps a name:

• Denition: A conformal map is a function f : R → S that preserves angles, in the sense for any dierentiable
curves c1 and c2 in R, the angle between c1 and c2 at z0 ∈ R is the same as the angle between f ◦ c1 and
f ◦ c2 at f (z0 ) ∈ S .

◦ Our result above shows that holomorphic functions with nonzero derivatives are always conformal.

◦ Conformal maps (particularly, conformal maps whose inverses exist and are also conformal) are useful
because they can allow us to transport questions back and forth between two regions, one example of
which is studying solutions boundary-value problems arising in physics.

◦ Another common application of conformal maps is in constructing (literal!) maps in cartography: al-
though a conformal function may distort distances, it will still faithfully preserve angles.

◦ We will return to discuss various methods for constructing conformal maps on particular regions later.

Well, you're at the end of my handout. Hope it was helpful.

Copyright notice: This material is copyright Evan Dummit, 2015-2022. You may not reproduce or distribute this
material without my express permission.

19