Exponential and Trigonometric functions

Our toolkit of concrete holomorphic functions is woefully small. We will now remedy this by
introducing the classical exponential and trigonometric functions using power series.

1 The Exponential function

Definition 1. We define
∞ n
X z
exp(z) := .
n=0
n!

The fact that the series converges on the whole of C to a holomorphic function was shown in the
previous chapter. The famous product law for the exponential now follows from our discussion on
the Cauchy product.
Proposition 2. exp(z + w ) = exp(z) exp(w ).
Proof. Using the Cauchy product, we see that:
X X znw m X 1
exp(z) exp(w ) = = (z + w ) k = exp(z + w ).
n!m! k!
k n+m=k k
In the intermediate step, we used the binomial theorem. 

The following theorem captures all the familiar properties of the exponential function
Theorem 3. For all z, w ∈ C:
1. exp(z) , 0;
2. exp(−z) = exp(z1
);
3. exp |R is a positive and strictly increasing function;
4. exp(z) = exp(z);
5. if z ∈ R then |exp(iz)| = 1;
6. | exp(z)| = exp(Rez) ≤ exp(|z|).
Proof. The proof consists of a number of simple checks:
1. Observe that exp(0) = 1 from which it follows that exp(z) exp(−z) = 1.
2. Follows from product law.
3. If x > 0 then exp(x ) > 1 + x. If x < 0, then exp(x ) = 1/ exp(−x ) > 0 from the previous part.
If y > x then exp(y) = exp(x ) exp(y − x ) > exp(x ).
4. Obvious.
5. | exp(iz)| 2 = exp(iz) × exp(iz) = exp(0) = 1.
6. if z = x + iy then | exp(z)| = | exp(x ) exp(iy)| = | exp(x )| = exp(x ).


1
 2 Trigonometric functions

2 Trigonometric functions
We define the following two holomorphic functions:
Definition 4.
exp(iz) + exp(−iz)
cos(z) = ;
2
exp(iz) − exp(−iz)
sin(z) = .
2i
Remark 5. The definition above does not make any mention of angles. We will reverse-engineer this
definition and define angles in terms of trigonometric functions.
The following properties of the trigonometric follows directly from the basic properties of the
exponential function.
Theorem 6. For all z, w ∈ C:
1. (Euler’s formula) exp(iz) = cos(z) + i sin(z);
2. cos(z) is an even function and sin(z) is an odd function;
3. cos2 (z) + sin2 (z) = 1;
4. sin(z + w ) = sin(z) cos(w ) + cos(z) sin(w );
5. cos(z + w ) = cos(z) cos(w ) − sin(z) sin(w );
2n+1 z 2n+1
sin(z) = (−1)(2n+1)!
P
6. ;
P (−1) n z 2n
7. cos(z) = (2n)!
8. sin0 (z) = cos(z), cos0 (z) = sin(z).
Caution: The above theorem might give the impression that the trigonometric functions defined
above behave exactly the same as the classical trigonometric functions. However, this is untrue for
complex numbers. For instance, if y > 0 then it is easy to see that
exp(y) + exp(y) (1 + y)
cos(iy) = > ,
2 2
which shows that cos and hence sin is unbounded on C. In fact, we will show later in the course that
any bounded entire function is forced to be constant!

3 Periodicity and the definition of π

As we have deliberately avoided all mention of angles so as to have a self-contained development, the
properties of the number π are unavailable to us. In fact, we are yet to understand the behavior of
the trigonometric functions on the real-axis. We will now show that sin and cos are periodic function
with period 2π (we will define π shortly). We will also show that the circumference of the unit circle
is 2π .
Let us assume for the moment that we have found t ∈ C such that sin(t ) = 0. Then from the
trigonometric identities, we see that

cos(2t ) = 1 − 2 sin2 (t ) = 1, sin(2t ) = 0,

which means that sin(z + 2t ) = sin(z). Similarly, we can show that 2t is a period of cos as well. Thus,
we are turn our attention to determining the zeroes of sin.

2
 3 Periodicity and the definition of π

Let S denote the set of zeroes of sin. If z ∈ S then

exp(iz) = exp(−iz) = 0

and so exp(2iz) = 1 from which it follows that exp(−2y) = | exp(2iz)| = 1. This forces y to be real.
We have shown that S ⊂ R. Now, we need an useful inequality.
Lemma 7. If 0 < x < 2 then we have

sin(x ) > x (4 − x 2 )/4 > 0.

Proof. We have from the power series expansion that

x3 x5
| sin(x ) − x | ≤ + +...
3! 5!
" 2 # " 2 #2
x3 *1 + x + x
≤ +...+
3! , 20 20 -
x 3  1 1 
≤ 1+ + 2 +...
3! 5 5
< x /4,
3

Hence
x − sin(x ) < x 3 /4
from which the result follows. 

From the above inequality, it follows that sin(1) > 3/4 and therefore cos(1) < 3/4. We apply the
intermediate value theorem on the interval [0, 1] to the function sin − cos to conclude that sin(t ) =
cos(t ) for some t ∈ [0, 1]. Using the trigonometric identities cos(2t ) = sin(4t ) = 0. This shows
that sin has a zero in [0, 4]. Clearly, S is a subgroup. Any additive subgroup of R is either dense or
cyclic. Clearly, S is not dense. Let π be the generator of S. From the inequality above, π > 2 and the
discussion above shows π < 4. This means that 0 < π /2 < 2 and therefore sin(π /2) > 0. But

0 = sin(π ) = 2 sin(π /2) cos(π /2)

which along with the identity sin2 (π /2) + cos2 (π /2) = 1 forces sin(π /2) = 1 and cos(π /2) = 0. This
implies that
sin(π /2 + x ) = sin(π /2 − x )
and as sin is positive on [0, π /2], it follows that sin is positive on (0, π ). It now follows from the
identity
sin(y + x ) sin(y − x ) = sin2 (y) − sin2 (x )
that if 0 < x < y < π /2 then sin2 (y) > sin2 (x ) and hence sin is an increasing function on [0, π /2]. If
p is a period of sin(z) then

sin(z) = sin(z + p) = sin(z) cos(p) + cos(z) sin(p).

If z = 0 then p = nπ , n ∈ Z. However as sin is an odd function, the period of sin must be 2nπ . In
a similar way, we can prove that the periods of cos is also 2nπ . We can summarize this discussion
succinctly in the following:

3
 4 Argument of a complex number

Theorem 8. The following are equivalent:

1. p is a period of exp;
2. exp(p) = 1;
3. p ∈ {2nπi : n ∈ Z}.

Henceforth, we shall freely use the basic properties of trigonometric functions without fuss.

4 Argument of a complex number

Given any non-zero complex number z, we can define the unit vector z/|z| = u + iv. Now |u | ≤ 1, we
can find a unique θ ∈ [0, π ] such that cos(θ ) = 1. This is because cos is a strictly-decreasing function
from 1 to −1 in this interval. Now, either sin(θ ) = v or sin(θ ) = −v. If sin(θ ) = −v, we replace θ by
−θ . In either case, we have found a θ such that z/|z| = cos(θ ) + i sin(θ ). Thus, using Euler’s theorem,
we can write

z = |z| exp(iθ ),
which is called the polar representation of the number z. The “angle” θ is called the argument of
the number z. This is the angle the vector z makes with the x-axis. The choice of θ —as one would
anticipate—is unique only up to an addition by 2nπ by the periodicity of sin and cos.
Definition 9. Given z , 0, we define

Arg(z) := {θ ∈ R : z = |z| exp(iθ )}.

Note that the capital letter in Arg is intentional and indicates that the object is a set and not a
number. When we use the notation arg(z), we mean an arbitrary fixed choice made from Arg(z).
By considering a unit vector z, we can form the triangle using the x-axis and the vector z and a
line parallel to the y-axis. It is now clear that the new definitions of sin and cos are the same as the
classical ones defined as the ratios of sides of right-triangles.
The argument satisfies the following properties:
Proposition 10. Let z, w ∈ C \ {0}. Then
1. Arg(zw ) = Arg(z) + Arg(w );
2. Arg(z −1 ) = − Arg(z);
3. Arg(z/w ) = Arg(z) − Arg(w ).
Remark 11. The algebraic signs appearing in the above proposition have to be interpreted as element-
wise operations. In particular, the ‘−’ sign has nothing to do with set-theoretic complements.
We can now define the general notion of an angle. Given a two complex numbers z, α with α , 0,
we define the ray
L(z, α ) := {z + tα : t ≥ 0}.
It is easy to see that if L(z 1 , α 1 ) = L(z 2 , α 2 ) iff z 1 = z 2 and Arg(α 1 ) = Arg(α 2 ). We define the angle

(L(z, α 1 ), L(z, α 2 )) = Arg(α 2 /α 1 ).

Remark 12. Note that the notion of angle we have defined is an oriented angle. Specifically, (L(z, α 1 ), L(z, α 2 )) =
−(L(z, α 2 ), L(z, α 1 )).

4
 5 The complex logarithm

5 The complex logarithm

Definition 13. For z ∈ C \ {0}, we define

Log(z) := {w ∈ C : exp(w ) = z}.

Remark 14. The logarithm is not defined for z = 0 for obvious reasons: the above set will be empty.
Also keep in mind that, as per our notational convention, the complex logarithm is not a function but
a set.

Proposition 15. Let z = r exp(iθ ). Then

Log z := {x + iy : log x = r, y ∈ Arg(z)}.

Proof. If w = x + iy then exp(w ) = exp(x ) exp(iy) from which the proposition follows immediately.


6 Roots of unity
Let us determine the n-th roots of unity for 0 < n ∈ N. If w n = 1, then w = exp(iθ ) where
( )
2kπ
θ ∈ :k ∈Z .
n

The above set is an infinite set. However, the set of distinct values of w that arise from the set corre-
spond to k = 0, 1, . . . , n − 1. The quantity ω := exp(2π /n) generates all such values: the n-th roots
of unity are precisely 1, ω, ω 2 , . . . , ω n−1 . All the basic properties about the n-th roots can be easily
proved. The n-th roots of an arbitrary complex number can also be easily computed.
We have already defined π using the sin function. We relate this to the circumference of the circle.
Observe that the n-th roots of unity form a regular polygon. The circumference of this polygon is
clearly
Ln = n|1 − ω | = n|1 − (cos(2π /n) + i sin(2π /n))|.
A strightforward computation now gives

|1 − ω | 2 = 2(1 − cos(2π /n)) = 4 sin2 (π /n).

This shows that Ln = 2n sin(π /n). Recall that if 0 < x < 2, we have

| sin(x ) − x | < x 3 /4.

Setting x = π /n, we see that

|Ln − 2π | < π 3 /2n3 .
This shows that as n → ∞, Ln → 2π .