Fundamental Theorem of Algebra

Complex Analytic and Topological Proofs

Zhang Ruiyang
[email protected]

1
Contents

1 Introduction 3

2 Complex Analysis Proof 4

2.1 Complex Analysis Basics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4

2.2 Cauchy’s Theorem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8

2.3 Cauchy Integral Theorem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11

2.4 The Final Touch . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13

3 Topology Proof 14

3.1 Sketch Proof . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14

3.2 Algebraic Topology Basics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15

3.3 Fundamental Group of Circle . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17

3.4 The Final Touch . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18

References 19

2
1 Introduction

This notes is prepared as the supplementary material for a talk I gave at the Undergraduate
Mathematics Colloquium at UCL. It assumes little knowledge from the readers, and it will be
good to have some basic understanding of Analysis and Topology. Additionally, knowledge of
Complex Analysis and Algebraic Topology will be very helpful, and may make much of the
content of this notes trivial. Hopefully, it could serve as good expository material for novices of
Mathematics. Another aim of this notes is to explore how different branches of Mathematics
can come together to show the same result, illustrating the elegance of Mathematics.

The Fundamental Theorem of Algebra, or FTA, is a theorem that shows the existence of
root of a non-constant polynomial. To be more specific, it is about the existence of complex
root of a non-constant polynomial with complex coefficients. Moreover, the number of roots
is the same as the degree of the polynomial. The first proper proof was provided by Gauss,
and as time progresses, other Mathematicians provided other proofs of the same theorem. In
this notes, I will be presenting two (not original) proofs of FTA, the first one is using Complex
Analysis, and the second involves Algebraic Topology.

Recall that a complex polynomial

P (z) = an z n + · · · + a1 z + a0

where the coefficients a0 , · · · , an are complex numbers, and z ∈ C. The highest power of z with
non-zero coefficient is known as the degree. A root, or a zero, of the polynomial is a value
zo ∈ C such that P (z0 ) = 0. The existence of root can be extended to show that the number of
roots is the same as the degree. For a polynomial P (z) with degree n, if z0 is a root of it, we
can factorise it and get P (z) = (z − z0 )Q(z) for a polynomial Q(z) of degree n − 1. We can
continue this process of factorising and eventually get n roots.

Now, why do we have different kinds of proof for this theorem of Algebra? One reason is
that polynomial represents different things in different branches of Mathematics. In this notes,
we presented the Complex Analytic proof and the Algebraic Topological proof. In Complex
Analysis, a polynomial can be treated as a function that maps from C to C, and therefore we
can use properties of functions on it. In Algebraic Topology, we will plot a polynomial and treat
it as a shape. In Topology, a big area of interest is on the topological invariants, the properties
that will remain unchanged after homeomorphisms. Homeomorphism, roughly speaking, is the
action of continuous deforming. For example, the classic example is how in the topological
viewpoint, a mug and a donut are the same, since they can be continuously deformed into each
other.

For those who want to discover more about FTA, for example the alternative proofs, you may
refer to the book ‘The Fundamental Theorem of Algebra’ by Fine and Rosenberger [2], or this
post on MathOverflow.

3
2 Complex Analysis Proof

In this Chapter, we will be presenting the proof of FTA using Complex Analysis. As we assume
no prior knowledge beyond Mathematical Analysis and aim to make it as self-contained as
possible, we will start from the very beginning of Complex Analysis, and slowly move our way
to the proof of FTA. As the goal is to prove FTA, we will keep anything less relevant to the
proof to a minimum. Here, we will be looking at things such as Cauchy Integral Theorem,
Cauchy Integral Theorem for derivatives, Cauchy Inequalities, Liouville Theorem and so on. For
readers who are interested to have a more complete understanding of Undergraduate Complex
Analysis, I would recommend you to read a proper textbook, such as Serge Lang [4], or Elias
Stein [6] (both under the title ‘Complex Analysis’).

2.1 Complex Analysis Basics

From Mathematical Analysis, we learned about how to do calculus with real-valued function
f : R → R. Now, for Complex Analysis, we will be working with complex-valued functions
f : C → C. This is going to be different from the real-valued cases, but it is not that different.
Notice that we can write z ∈ C as x + iy for x, y ∈ R, we can somewhat think about C as R2 .
This may help us with some intuitions.

We will start with some of the basics of complex-valued functions. Recall that f : R → R will
be differentiable at a point x if

f (x + h) − f (x)
lim
h→0 h
exists. Similarly, we will say f : C → C is differentiable, or more specifically complex
differentiable, at a point z if both

f (z + h) − f (z) f (z + ih) − f (z)

lim and lim
h→0 h h→0 ih
exist and are equal for h ∈ R. This does not seem to be too accessible. Let us see if we can
simplify it a bit.

As we know that we can write z ∈ C as x + iy for x, y ∈ R, we can tweak things a bit, and
rewrite f (z) ∈ C as u(x, y) + iv(x, y) where u(x, y), v(x, y) ∈ R2 and x, y ∈ R. If we assume
u(x, y) and v(x, y) are real differentiable at that point z, we will have

f (z + h) − f (z) u(x + h, y) + iv(x + h, y) − u(x, y) − iv(x, y)

lim = lim
h→0 h h→0 h
u(x + h, y) − u(x, y) iv(x + h, y) − iv(x, y)
= lim + lim
h→0 h h→0 h
∂u ∂v
= +i
∂x ∂x

4
and

f (z + ih) − f (z) u(x, y + h) + iv(x, y + h) − u(x, y) − iv(x, y)

lim = lim
h→0 ih h→0 ih
u(x, y + h) − u(x, y) iv(x, y + h) − iv(x, y)
= lim + lim
h→0 ih h→0 ih
∂u ∂v
= −i + .
∂y ∂y

∂u ∂v
For these two to be equal, we need to have ∂x + i ∂x = −i ∂u
∂y +
∂v
∂y , or simply

(
∂u ∂v
∂x = ∂y
∂v
∂x = − ∂u
∂y .

This is known as the Cauchy-Riemann Equations. The Cauchy-Riemann equations provide

us with an alternative way to check for complex-differentiability. If we can write f (z) as
u(x, y) + iv(x, y) for u, v differentiable functions on R2 at z = x0 + iy0 , as long as u, v satisfy
the Cauchy-Riemann equations for (u0 , v0 ), it is complex differentiable. If this is true for every
point of the domain, the function is holomorphic. If a holomorphic function has domain of
the entire C, it is entire

We have quite a bit of differentiation. Now let us move on to integration. Just like what we did
for differentiation, let us recall how we did integration for real-valued function f : R → R.

For (Riemann) integrable functions f : [a, b] → R, since the domain is a subset of R, we can
visualise the domain as a portion from the real line. For complex-valued function f : C → C,
the domain will be a subset of C, and instead of for example a segment of the real line, we will
have a more complicated domain like a path or a loop. We have not properly defined what a
path or a loop may be, but by common sense we could have a rough picture of a smooth curve
connecting two points for a path, and a circle-ish curve starting and ending at the same point
for a loop, all happening in the complex plane. Now, let us try to formally define them.
Definition 2.1. A smooth path in the complex plane is a map γ : [A, B] → C that is
continuously differentiable on [A, B].

Sometimes a smooth path is too much to ask for, and we will have a slightly weaker form that
will allow us to do integration just fine.
Definition 2.2. A continuous map γ : [A, B] → C, where [A, B] ⊂ R is a closed interval, is
called a piecewise smooth path in C if there exists a finite partition A = A0 < A1 < · · · <
An = B such that the restriction of γ to each segment [Aj , Aj+1 ] is a smooth path.

5
In the following, we will use ‘path’ when we really mean ‘piecewise smooth path’. Now, we are
ready to define integration over path.
Definition 2.3. Let γ : [A, B] → C be a piecewise smooth path and f be a continuous
complex-valued function defined on the set γ([A, B]) ⊂ C. Then, the integral of f over γ is
the number Z Z B
f (z)dz = f (γ(t))γ 0 (t)dt.
γ A

Remark. The equality is simply a change of variable using the fact that z = γ(t).

Notice that we are using a parametrisation of the path to define the integral. The choice of
parametrisation is mostly independent to the integral.
Proposition 2.4. Assume that the conditions of Definition 2.3 are satisfied and ϕ : [A1 , B1 ] →
[A, B] is a bijective differentiable function whose derivative is everywhere positive. Then,
Z Z
f (z)dz = f (z)dz.
γ◦ϕ γ

If the derivative is everywhere negative, then

Z Z
f (z)dz = − f (z)dz.
γ◦ϕ γ

Proof. If ϕ is increasing with ϕ(A1 ) = A and ϕ(B1 ) = B, then we will have

Z Z B
f (z)dz = f (γ(t))γ 0 (t)dt
γ A
Z B1
= f (γ(ϕ(u)))γ 0 (ϕ(u))φ0 (u)dt
A1
Z B1
= f (γ ◦ ϕ(u))(γ ◦ ϕ)0 (u)du
ZA1
= f (z)dz.
γ◦ϕ

The decreasing case can be shown in a similar manner.

We will now introduce another definition.

Definition 2.5. A piecewise smooth curve is a subset in C of the form γ([A, B]) where
γ : [A, B] → C is an injective piecewise smooth path and γ 0 (t) 6= 0 ∀ t ∈ [A, B] except for
possibly finitely many points.

A curve will be called closed if we have γ(A) = γ(B). The direction of a closed curve will
matter here. We will call the counterclockwise direction as positive, and clockwise as negative.
Readers can refer to the way sign of the angles is determined in the polar coordinates.

We will now try to do an example of path integral.

6
Example 2.6. If γ is a positively oriented circle of radius r > 0 centred at a ∈ C, then
Z
dz
= 2πi.
γ z −a

A way to parametrise the circle is γ(t) = a + reit where t ∈ [0, 2π]. Then, we have z = a + reit
and dz = ireit dt. So,
Z 2π
ireit dt
Z
dz
=
γ z−a 0 reit
Z 2π
= idt
0
= 2πi.

This example will appear again in the following sections.

Next, we will study some of the properties of integration over curves. We first make Ra definition
B
of length. For path γ : [A, B] → C, the length of it is defined to be length(γ) = A |γ 0 (t)|dt
for a particular parametrisation of the curve. Then, we have the following inequality.

Proposition 2.7. For a continuous function defined on a curve γ([A, B]) with γ : [A, B] → C,
we have Z 
 f (z)dz  ≤ sup |f (z)| · length(γ).
 
γ z∈γ([A,B])

Proof. Let M = supz∈γ([A,B]) |f (z)|. We have

Z  Z B 
 f (z)dz  =  f (γ(t))γ 0 (t)dt
   
γ A
Z B 
≤ f (γ(t))γ 0 (t)dt
 
A
ZB 
≤ M γ 0 (t)dt
 
A
Z B 
 0 
≤M· γ (t)dt = M · length(γ)
A

Recall that we have a Fundamental Theorem of Calculus for real-valued functions. We will
have that for complex-valued ones.

7
Definition 2.8. let U ⊂ C be an open set and f : U → C be a continuous function. We say
that a function F : U → C is an antiderivative (or primitive) of f if F is holomorphic on U
and F 0 (z) = f (z) for every z ∈ U .
Theorem 2.9. (Fundamental Theorem of Calculus for Complex Functions) Let U ⊂
C be an open set and f : U → C be a continuous function that has an antiderivative F in U . If
γ is a path in U joining points p ∈ U and q ∈ U , then
Z
f (z)dz = F (q) − F (p).
γ

Proof. Let γ : [A, B] → U be a path such that γ(A) = p and γ(B) = q. Then,
Z Z
f (z)dz = F 0 (z)dz
γ γ
Z B
= F 0 (γ(t))γ 0 (t)dt
A
ZB
d
= F (γ(t))dt
A dt
= F (γ(B)) − F (γ(A))
= F (q) − F (p).

Corollary 2.10. If a continuous function f defined on an open set U ⊂ C has an antiderivative

on this set, then the integral of f over every closed path in U vanishes.

This is obvious to see, as F (q) = F (p) for a closed path. Also, if you ever wonder why the
integral over the circle in the example above does not vanish, check the conditions of this
Corollary and see if they are all satisfied, especially when z = a.

2.2 Cauchy’s Theorem

From the end of the previous section, we saw that a function defined on an open set with
an antiderivative will have vanishing integral over any closed path. Having an antiderivative
everywhere on an open set is quite a strong condition, and we do not always have that. The
generalisation that remove the condition on existence of antiderivative is the Cauchy’s Theorem.
However, due to the expository nature of this text, we will not be showing and proving the full
form of this theorem, and instead we will only be looking at the minimum form to proceed with
our proof of the Fundamental Theorem of Algebra. For those readers who would like to know
the complete version, do learn some Topology (especially homotopy) and refer to a Complex
Analysis textbook. Let us start with the following theorem.
Theorem 2.11. (Goursat’s Theorem) If U is an open set in C, and T ⊂ U is a triangle
whose interior is also contained in U , then
Z
f (z)dz = 0
∂T

8
wherever f is holomorphic in U .

Proof. First assume that T ⊂ C is an arbitrary triangle and f : T → C is an arbitrary

continuous function. The midsegments divide T into four congruent triangles Ta , Tb , Tc , and
Td as shown below.

Then, we have
Z Z Z Z Z
f (z)dz = f (z)dz + f (z)dz + f (z)dz + f (z)dz
∂T ∂Ta ∂Tb ∂Tc ∂Td

where the ∂ represents the boundary of the shape. This is true by looking at the direction of
the arrows. The boundaries of Tc will go in the opposite direction as the sides of the other
three, so they will cancel out eventually.

We will try to prove by contradiction.

R  Assume that the integral of f over the boundary of T
does not vanish, and set  ∂T f (z)dz  = C > 0 for some C. Next, we know that at least one of
the four smaller triangles (denote by T1 ) will have
Z  C
f (z)dz  ≥
 
4

∂T1

or else the quality above will not hold. We will then repeat the process of dividing a triangle
into four to T1 , and get a T2 that satisfies
Z  C
f (z)dz  ≥ 2.
 
4

∂T2

By iterating this process, we will get a sequence of triangles with T = T0 ⊃ T1 ⊃ · · · ⊃ Tn ⊃ · · ·

where Z  C
f (z)dz  ≥ n .
 
4

∂Tn

It is also easy to spot that there exists a point a ∈ U in every Tn . Also, if we let the perimeter
of T to be p, the perimeter of Tn will then be p/2n since we are bisecting each of the sides to
form the smaller triangles. Now, since f is holomorphic in all of U , it must has a derivative
at a, and we can then have a function ϕ by f (z) = f (a) + f 0 (a)(z − a) + ϕ(z)(z − a) where
ϕ(z) → 0 as z → a. Notice that both f (a) and f 0 (a)(z − a) have antiderivative in U , so by
Corollary 2.10, we have
Z Z Z
0
f (z)dz = (f (a) + f (a)(z − a) + ϕ(z)(z − a))dz = ϕ(z)(z − a)dz.
∂Tn ∂Tn ∂Tn

9
Some more, as a ∈ Tn and z is on the boundary of Tn , we have |z − a| ≤ p/2n . Then, by
Proposition 2.7, we can have the estimate
Z  p p
ϕ(z)(z − a)dz  ≤ εn n · n
 
2 2

∂Tn

where εn = supz∈Tn |ϕ(z)| → 0 as n → ∞ by the way ϕ is defined.

Combining what we have gotten so far, we will have

Z  Z 
n
f (z)dz  ≤ 4  f (z)dz  ≤ εn p · p.
  

∂T ∂Tn

By having n → ∞, the absolute of the integral will be 0, since ϕn → 0. Thus we complete the
proof.

Now, we will try to generalise the theorem to a broader range of shapes. But first, we need
another result.
Proposition 2.12. Let U ⊂ C be a convex open set and f : U → C be a holomorphic function.
Then f has an antiderivative.

Proof. First, we will use the notation [a, b] to denote the line segment from a to b here.
R
Choose a point z0 ∈ U and define a function F : U → C by the formula F (z) = [z0 ,z] f (t)dt.
We will then show that this F is an antiderivative of f .

Let a ∈ U , and a + h ∈ U . Let T be a triangle with vertices z0 , a and a + h. By convexity, it is

contained in U . Then, using Goursat’s Theorem, we have
Z Z Z Z
0= f (t)dt = f (t)dt + f (t)dt + f (t)dt
∂T [z0 ,a] [a,a+h] [a+h,z0 ]

when we also have

Z Z Z
F (a + h) − F (a) = − f (t)dt − f (t)dt = f (t)dt.
[z0 ,a+h] [z0 ,a] [a,a+h]

Note that
Z Z Z Z
f (t)dt = f (a)dt + (f (t) − f (a))dt = f (a) · h + (f (t) − f (a))dt.
[a,a+h] [a,a+h] [a,a+h] [a,a+h]

Using a similar estimation as the proof of Goursat’s Theorem, we get

R 
 F (a + h) − F (a)   [a,a+h] (f (t) − f (a))dt
− f (a) =
 
h |h|


|h| · supt∈[a,a+h] |f (t) − t(a)|

≤
|h|
= sup |f (t) − t(a)|.
t∈[a,a+h]

10
Since the function f is holomorphic and then continuous, the right-hand side above will tend
to zero as h → 0, which means the derivative of F at a is f (a), so the proof is completed.

With this result, we can improve Goursat’s Theorem and get a form of Cauchy’s Theorem for
convex open set.

Theorem 2.13. (Cauchy’s Theorem for convex open set) Let U ⊂ C be a convex open
set and f : U → C be a holomorphic function. Then, (a) the integral of f over every closed
path in U vanishes; (b) if p, q ∈ U and γ1 , γ2 are two paths in U joining p and q, then the
integrals of f over γ1 and γ2 will be identical.

To generalise this version of the Cauchy’s Theorem even further, we will need to know more
Topology, mostly homotopy. We do not need much of that to continue this proof to the
Fundamental Theorem of Algebra, so we will omit it here. We just need one more corollary to
finish this section.

Corollary 2.14.
R Suppose f is holomorphic in an open set containing the circle C and its
interior, then C f (z)dz = 0.

Proof. Let D be a disc (open circle) with boundary circle C. Then, there exists a slightly larger
disc D; which contains D and so that f is holomorphic on D0 . D0 is an open convex set, so we
can apply Cauchy’s Theorem for convex open set to complete the proof.

2.3 Cauchy Integral Theorem

Theorem 2.15. (Cauchy Integral Theorem) Suppose f is holomorphic in an open set that
ocntians the closure of a disc D. If C denotes the boundary circle of this discs with the positive
orientation, then Z
1 f (z)
f (w) = dz
2πi C z − w
for any w ∈ D.

Proof. Fix w ∈ D and consider the keyhole Γδ,ε which omits the point w as below.

Here, we let δ be the width of the corridor and ε be the radius of the small circle centred at w.
f (z)
Since the function F (z) = z−w is holomorphic everywhere except w, we will have
Z
F (z)dz = 0
Γδ,ε

by Cauchy’s theorem. Now, we make the corridor narrower by letting δ → 0, and eventually
these two sides will cancel out over the integrals. The remaining part consists of two curves,

11
one is the large circle C with positive orientation, and the other is a small circle, called Cε ,
centred at w with radius ε and negative orientation. The integral then becomes

f (z) − f (w) f (w)

F (z) = + .
z−w z−w
Notice that the integral of the first term on the right will go to 0 as ε → 0. Thus, we have
Z Z
f (w) dz
dz = f (w)
Cε z − w Cε z −w
= −f (w)2πi.

So, we have Z Z
f (z)
0= F (z)dz = dz − f (w)2πi,
Γδ,ε C z−w
1
R f (z)
or simply f (w) = 2πi C z−w dz.

If we think about the this theorem for a while, we will realise how remarkable it is! The theorem
is saying that if we know the values of f on the boundary curve C, then we can know everything
about f inside C. Next, we will have a corollary of it.

Corollary 2.16. (Cauchy Integral Theorem for derivatives) If f is holomorphic in an

open set U , then f has infinitely many complex derivatives in U . Moreover, if C ⊂ U is a circle
whose interior is also contained in U , then
Z
(n) n! f (z)
f (w) = dz
2πi C (z − w)n+1

for all w in the interior of C.

This is quite easy to show using induction. This will be left as an exercise for the readers. ω 

After knowing the formula for derivatives, we can find a bound for it.

12
Corollary 2.17. (Cauchy Inequalities) If f is holomorphic in an open set that contains the
closure of a disc D centred at z0 and of radius R, then
n!MR
|f (n) (z0 )| ≤
Rn
where MR = supz∈C |f (z)| denotes the supremum of |f | on the boundary circle C.

Proof. Applying the Cauchy Integral Theorem for derivatives, we have

|f (w)|
Z
(n) n!
|f (z0 )| ≤ |dw|
2π C |w − z0 |n+1
Z
n! MR
≤ |dw|
2π Rn+1 C
n! MR
= · 2πR
2π Rn+1
n!MR
= .
Rn

2.4 The Final Touch

Theorem 2.18. (Liouville’s Theorem) Assume f (z) is entire and suppoes it is bounded in
the complex plane, namely |f (z)| < M for all z ∈ C, then f (z) is constant.

Proof. For any circle of radius R around z0 , the Cauchy inequality tells us that |f 0 (z0 )| ≤ M
R
when we set n = 1. But, R can be any number, so we can let it be extremely large and get
|f 0 (z0 )| = 0 for every z0 ∈ C. If the derivative is 0, the function is a constant.

Finally, we have all we need to prove the Fundamental Theorem of Algebra.

Theorem 2.19. (Fundamental Theorem of Algebra) Every non-constant polynomial
P (z) = an z n + · · · + a0 with complex coefficients has a root in C.

1
Proof. If P has no roots, then P (z) is a bounded holomorphic function, since we can have
P (z) an−1 a0
= an + ( + ··· + n)
zn n z
whenever z 6= 0. Each term in the bracket will go to 0 as |z| → ∞, and we can say that
there exists R > 0 such that if c = |a2n | |, then |P (z)| ≥ c|z|n whenever |z| > R and P (z)
1
is
1
therefore bounded. Also, since P is continuous and has no roots in the disc |z| ≤ R, P (z) is
1
bounded too. Thus, the claim earlier of P (z) is bounded is shown. Then, according to Liouville’s
1
Theorem, P is a constant. This contradicts with the condition that P is non-constant. Thus,
by contradiction, the theorem is proved.

13
3 Topology Proof

In this Chapter, we will be presenting another proof of FTA using Topology, and specifically
winding number. We will start with a sketch proof that lacks some rigour and proper definitions
of terms. We will then fill up the gaps and complete the proof in the following sections.

3.1 Sketch Proof

To start, we first need the concept of ‘winding number’. It is quite easy to see what this term
means via a graphical example.

Winding number of a loop γ is the number of times that γ winds anticlockwise around a
point, for example 0. We will denote this number by #γ, and we can see that in the above
figure, #α = 3, #β = 0, #γ = −1, and #δ = −1. More over, if two loops can be continuously
deformed into each other without crossing zero, these two loops will have the same winding
number. This relationship between the two loops is known as homotopty, and we will go into
more detail later on this Chapter. In this example, γ and δ are homotopic, so they have the
same winding number -1.

Next, take a complex polynomial f (z) = a0 + a1 z + · · · + an z n of degree n, and suppose that

f has no complex root. Then, we take an R ∈ [0, ∞). As the input z goes around in circle
{z||z| = R} in the anticlockwise direction, f (z) will form a loop γR in C and it does not cross
0, since f (z) has no complex root and no z0 ∈ C will let f (z0 ) = 0.

(1) We know that as R increases, γR will only change continuously, since f is a continuous
function. So, all possible γR can be deformed into each other and are thus homotopic, which
implies that #γR is independent of the value of R.

(2) When |z| is large, f (z) will behave like an z n , then the remaining terms will be negligible in
this case. For z travelling on the circle {z||z| = R} in the anticlockwise direction, it will wind
around 0 n times. So, for big enough R, #γR = n.

(3) When R = 0, γ0 will be a constant at f (0) 6= 0, so #γ0 = 0.

If we compile all these information, we will get #γ0 = #γR = 0 ∀R ∈ [0, ∞), so n = 0 and f is
a constant. This contradicts with the condition of FTA, so there exists z0 ∈ C as the root of

14
any non-constant complex polynomial.

3.2 Algebraic Topology Basics

Now, let us begin to define things properly and rigorously. Topology deals with different shapes,
and one of the simplest shape is that of path and loop. It is essentially the same concept as
the path and loop we defined earlier in the previous Chapter, but there is some very minor
difference in their definitions. A path in a space X is a continuous map f : I → X where
I is the unit interval [0, 1]. If the starting and ending points of the path are identical, i.e.
f (0) = f (1), then it becomes a loop.

Next, we will start to define what we mean by a continuous deformation. A continuous

deformation, or a homotopy, of paths in X is a family fs : I → X, 0leS ≤ 1 such that (1)
the endpoints fs (0) = x0 and fs (1) = x1 are independent of t, meaning that they are all fixed
regardless of the value of t. (2) the associated map F : I × I → X defined by F (s, t) = fs (t) is
continuous. When two paths f0 and f1 are connected in this way by a homotopy fs , they are
said to be homotopic, and we denote this relationship by f0 ' f1 .

Example 3.1. (Linear Homotopies) Any two paths f0 and f1 in Rn having the same
endpoints x0 and x1 are homotopic via the homotopy fs (t) = (1 − s)f0 (t) + sf1 (t). The
endpoints will not be changed during this homotopy, and each point will travel along the line
segment at a constant speed. This fs (t) is continuous since the vector addition and scalar
multiplication of f0 and f1 are continuous. This construction can be generalised to any convex
subspace X ⊂ Rn , since this homotopy will work given the convexity of space.

We can find some algebraic structure of homotopies.

Proposition 3.2. The relation of homotopy on paths with fixed endpoints in any space is an
equivalence relation.

Proof. Recall that an equivalence relation needs to satisfy three conditions (1) reflexivity, (2)
symmetry, and (3) transitivity.

Reflexivity is easy to spot, since we can have the constant homotopy fs = f to construct the
relation. Symmetry is also obvious, as we can get f1 ' f0 from f0 ' f1 with homotopy fs by
having the inverse homotopy f1−s . Transitivity is the one that needs some work.

If we have f0 ' f1 via fs , and g0 = f1 ' g1 via gs , we need to show that f0 ' g1 with homotopy
hs . This homotopy can be found by connecting fs and gs , by having f2s for 0 ≤ s < 1/2, and
g2s−1 for 1/2 ≤ s ≤ 1. These two components agree on s = 1/2, since g0 = f1 . The continuity is
obvious since it is the union of two continuous functions with the same value at the intersection
point.

Knowing the equivalence relation of the path homotopy, we will denote the equivalence class of
a path f by homotopy by [f ], and it is called the homotopy class of f .

15
From the above proof, we can give a name to what we did to show the transitivity of homotopy.
Given two paths f, g : [0, 1] → X such that f (1) = g(0), there is a composition product
path f ◦ g that traverses first f and then g, defined by
(
f (2t), 0 ≤ t ≤ 12
f ◦ g(t) = 1
g(2t − 1), 2 ≤ t ≤ 1.

Remark. Do take note that we are traversing first with f then with g, it is not how we
normally think about compositions of functions. It is not the most intuitive way to think about
it, but this has been the case for the textbooks I am referring to, so I guess that is the norm.

This product operation respects homotopy classes, since if f0 ' f1 and g0 ' g1 via the homotopy
fs and gs respectively, and if we have f0 (1) = g0 (0) so f0 ◦ g0 is defined, fs ◦ gs is defined and
provides a homotopy f0 ◦ g0 ' f1 ◦ g1 . If instead of paths, we have a homotopy of loops, then
the common starting and ending points x0 is known as the basepoint. The set of all homotopy
classes [f ] of loops f : I → X at the basepoint x0 is denoted by π1 (X, x0 ). This set has some
interesting algebraic structure too, and in particular:
Proposition 3.3. π1 (X, x0 ) is a group with respect to the product operation [f ] ◦ [g] = [f ◦ g].

Proof. Recall the axioms of the group structure, we need (1) associativity, (2) existence of
identity, and (3) existence of inverse.

We first check if the operation is associative, meaning that [α ◦ β] ◦ [γ] = α ◦ [β ◦ γ] for any three
loops α, β, γ based at the same point p. We will do a reparametrisation of a path f to make
it as f ϕ where ϕ : [0, 1] → [0, 1] is a continuous map such that ϕ(0) = 0 and ϕ(1) = 1. The
reparametrisation will preserve the homotopy class, as f ϕ ' f via the homotopy f ϕt where
ϕs (t) = (1 − s)ϕ(t) + st such that ϕ0 = ϕ and ϕ1 (t) = t. This method should not be brand
new any more, as we have just used this for the example on linear homotopies. With this in
mind, we can do a reparametrisation to (α ◦ β) ◦ γ using

2t
 0 ≤ t ≤ 14
ϕ(t) = t + 14 1
4 ≤t≤ 2
1

 t+1 1
2 2 ≤t≤1

, which will show us the associativity.

The identity element of the group will be the constant path defined to be c(t) = f (1) for all
t ∈ [0, 1]. With this path, we can get f ◦ c ' f for a path f : [0, 1] → X. Also, we can get
another constant path c(t) = f (0) for all t ∈ [0, 1]. With this path, we can get c ◦ f ' f for a
path f : [0, 1] → X. Since we are dealing with loops, f (0) = f (1) and the two constant paths
are identical, making it the identity element.

For path f (t), f (1 − t) will be the inverse path of it. The remaining details can be filled by the
readers without much difficulties.

This group π1 (X, x0 ) is called the fundamental group of X at the basepoint x0 . The subscript
1 of π is used to denote this group as the first group in a sequence of groups πn (X, xn ), called

16
the homotopy groups. We will not be discussing them in this notes. Interested readers can
refer to any textbook on Algebraic Topology, for example the classic by Allen Hatcher[3].

Let us think about what this group is trying to do for a bit. In Topology, we are frequently, in fact
always, dealing with topological invariants - properties that will be preserved by homeomorphism.
If you do not know already, a homeomorphism is a bijective map f : X → Y for topological
spaces X and Y when both f and its inverse are continuous. One goal of Topology is to classify
different shapes and spaces up to homeomorphism. If two shapes have different values for a
particular topological invariant, they will not be the same in the topological sense. Among
the various invariants, one common one, constructed by Poincaré, is the fundamental group.
For example, do you think the fundamental group of a disc will be the same as that of an
annulus? You can try to draw them out and play around with it, hopefully you will get the
answer of “No”. In the same example, do you think the basepoint you choose will affect the
fundamental group? For points in a disc, do you think the fundamental group will be different
for two distinct points? This brings us to the question on change of basepoint.

We define a change of basepoint map βh : π1 (X, x0 ) → π1 (X, x1 ) by βh [f ] = [h ◦ f ◦ h−1 ],

where h : [0, 1] → X is an invertible path from x0 to x1 . We let the inverse of h be h−1 = h(1−t)
that goes from x1 to x0 . We can then associate each loop f based at x1 the loop h ◦ f ◦ h−1
based at x0 .

Proposition 3.4. The map βh : π1 (X, x0 ) → π1 (X, x1 ) is an isomorphism.

Proof. First, we see that βh is a homomorphism since βh [f ◦ g] = [h ◦ f ◦ g ◦ h−1 ] = [h ◦ f ◦

h−1 ◦ h ◦ g ◦ h−1 ] = βh [f ] ◦ βh [g]. It is an isomorphism with inverse βh−1 since βh βh−1 [f ] =
βh [h−1 ◦ f ◦ h] = [h ◦ h−1 ◦ f ◦ h ◦ h−1 ] = [f ]. Similarly, βh−1 βh [f ] = [f ]. Thus, the map is indeed
an isomorphism.

If X is path-connected, meaning that we can connect any pair of points in the space using a
path, we can have the above map for every pair points. Since we are working algebraic objects,
two fundamental groups being isomorphic means that they are the ‘same’. So, the basepoint is
not too important for a path-connected space X, and we can sometimes drop x in the notation
and just write the fundamental group as π1 (X).

Now, if a space is both path-connected and has trivial fundamental group of 0, then it is called
simply-connected.

3.3 Fundamental Group of Circle

Theorem 3.5. π1 (S 1 ) is an infinite cyclic group generated by the homotopy class of the loop
ω(t) = (cos 2πt, sin 2πt) based at (1, 0).

Remark. The proof of this theorem will be omitted here. Interested readers should refer to
[1] or [3]. I will add this part later if there is such a demand. Do let me know via email if you
would like to see the proof of it in this notes.

17
3.4 The Final Touch

Theorem 3.6. (Fundamental Theorem of Algebra) Every non-constant polynomial with

coefficients in C has a root in C.

Proof. We may assume the polynomial is of the form p(z) = z n + a1 z n−1 + · · · + an . If p(z)
has no roots in C, then for each real number r ≥ 0, the formula

p(re2πit )/p(r)
fr (t) =
|p(re2πit )/p(r)|

defines a loop in the unit circle S 1 ⊂ C bases at 1. As r varies, fr is a homotopy of loops based
at 1. Since f0 is the trivial loop, we deduce that the class [fr ] ∈ π1 (S 1 ) is zero for all r. Now,
fix a large value of r, bigger than |a1 | + · · · + |an | and bigger than 1. Then, for |z| = r, we have

|z n | ≥ (|a1 | + · · · + |an |)|z n−1 | > |a1 z n−1 | + · · · + |an | ≥ |a1 z n−1 + · · · + an |.

This inequality is saying that the polynomial pt (z) = z n + t(a1 z n−1 + · · · + an ) has no roots
on the circle |z| = r when 0 ≤ t ≤ 1. Replacing p by pt in the formula for fr to the loop
ωn (t) = esπint . Using the fundamental group of circle, ωn represents n times a generator of the
infinite cyclic group π1 (S 1 ). Since we have shown that [ωn ] = [fr ] = 0, we conclude that n = 0.
Thus, the only polynomials without roots in C are constants.

18
References

[1] M. A. Armstrong. Basic Topology, Undergraduate Texts in Mathematics. Springer, 1979.

[2] Benjamin Fine & Gerhard Rosenberger. The Fundamental Theorem of Algebra, Undergrad-
uate Texts in Mathematics. Springer, 1997.

[3] Allen Hatcher. Algebraic Topology. Cambridge University Press, 2001.

[4] Serge Lang. Complex Analysis, Fourth Edition, Graduate Texts in Mathematics 103. Springer,
1999.

[5] Serge Lvovski. Principles of Complex Analysis, Moscow Lectures 6. Springer, 2020.

[6] Elias M. Stein & Rami Shakarchi. Complex Analysis, Princeton Lectures in Analysis III.
Princeton University Press, 2003.

19