PROOFS OF THE FUNDAMENTAL THEOREM OF ALGEBRA

MATTHEW STEED

Abstract. The fundamental theorem of algebra states that a polynomial of

degree n ≥ 1 with complex coefficients has n complex roots, with possible
multiplicity. Throughout this paper, we use f to refer to the polynomial
f : C −→ C defined by f (z) = z n + an−1 z n−1 + · · · + a0 , with n ≥ 1. We
provide several proofs of the fundamental theorem of algebra using topology
and complex analysis. We also suppose that a0 6= 0. Otherwise, 0 itself is
a root. The first proof is a topological proof. The next three use complex
analysis.

Contents
1. A Topological Proof 1
2. A Similar Proof Using the Language of Complex Analysis 3
3. A Proof Using the Maximum Modulus Principle 4
4. A Proof Using Liouville’s Theorem 4
Acknowledgments 5
References 5

1. A Topological Proof
Let f be the previously defined polynomial. We first show that there exists at
least one root of f in the complex numbers. With one root we can use an inductive
argument to show that there are n roots, possibly with multiplicity greater than
one. Let I be the interval [0, 1]. For positive real numbers r and R, let γ : I −→ C
and Γ : I −→ C be paths in the complex plane defined by γ(x) = re2πix and
Γ(x) = Re2πix . These paths are circles of radius r and R. They are homotopic in
C, so their images f ◦γ and f ◦Γ are homotopic because polynomials are continuous.
To show that there is at least one root, we show that there are values of r and R
such that f ◦ γ and f ◦ Γ are not homotopic in C \ {0}. This will be a contradiction
to what we have just shown, showing that 0 must lie in the image of f .
Let g1 : I −→ C and g2 : I −→ C be defined as g1 (x) = 1 and g2 (x) = e2πinx . We
show that there exist values of r and R such that both f ◦γ ' a0 g1 and f ◦Γ ' a0 g2 ,
and, moreover, that there exist homotopies between them that are nowhere zero.
We then show that a0 g1 and a0 g2 are not homotopic in C \ {0}, which will give our
desired contradiction.
The existence of the first homotopy follows from the continuity of f . Note that
f (0) = a0 . Then, there exists δ > 0 such that if |z| < δ, then |f (z) − a0 | < | a20 |.
Suppose that r < δ. If z is in the path γ, then |z| = r. Then |f (z) − a0 | < | a20 |

Date: January 10, 2015.

1
2 MATTHEW STEED

for z ∈ γ. By the reverse triangle inequality, ||f (z)| − |a0 || ≤ |f (z) − a0 | < | a20 |.
Then, −| a20 | < |f (z)| − |a0 | < | a20 |, and, adding |a0 | to both sides, | a20 | < |f (z)| <
| 3a20 |. Consider the homotopy between f ◦ γ and a0 g1 , h : I × I −→ C, defined as
h(x, t) = (1 − t)(f ◦ γ)(x) + a0 t. If r < δ, then |h(x, t)| > 0, as shown below:
|h(x, t)| = |(1 − t)(f ◦ γ)(x) + a0 t|
= |(f ◦ γ)(x) − t((f ◦ γ)(x) − a0 )|
≥ |(f ◦ γ)(x)| − t|(f ◦ γ)(x) − a0 |
a  a 
 0  0
>   − t 
2 2
≥0
Thus, with r chosen small enough, f ◦ γ ' a0 g1 with a nowhere zero homotopy.
To show that f ◦ Γ ' a0 g2 with a nowhere zero homotopy, we first show that
f ◦ Γ ' Rn g2 with a nowhere zero homotopy for large enough |R|. Then, we show
that Rn g2 ' a0 g2 with a nowhere zero homotopy.
For the first part, we make use of the fact that polynomials eventually behave
like their leading term. We can pick some real M such that if |z| > M , then
|z|n−j > 2n|aj | where aj is a coefficient in f and 0 ≤ j < n. Then, rearranging,
|z n | j
2n > |aj z |. By picking M large enough,n we make this inequality true for all
0 ≤ j < n. Then, adding the ineqalities, |z2 | > |an−1 z n−1 | + · · · + |a1 z| + |a0 | ≥
|an−1 z n−1 + · · · + a1 z + a0 |. Then, with z > M ,
|f (z)| = z n + an−1 z n−1 + · · · + a0 
 

≥ |z n | − an−1 z n−1 + · · · + a0 
 
 n   n  
z  z  
=   +   − an−1 z n−1 + · · · + a0 

2 2
 n
z 
>  
2
Let |R| > M , and consider the homotopy h : I × I −→ C defined by h(x, t) =
(1 − t)(f ◦ Γ)(x) + t(Rn e2πinx ). We show that h is nowhere zero:
|h(x, t)| = |(f ◦ Γ)(x) − t((f ◦ Γ)(x) − Rn e2πinx )|
= |(Rn e2πinx + · · · + a0 )
− t(an−1 Rn−1 e2πi(n−1)x + · · · + a1 Re2πix + a0 )|
 n 
R   
>   − an−1 Rn−1 e2πi(n−1)x + · · · + a1 Re2πix + a0 
2 
>0
We now show that Rn g2 ' a0 g2 with a nowhere zero homotopy. Let Rn = R∗ eiθ1 ,
and let a0 = a∗ eiθ2 , where R∗ and a∗ are positive real numbers. Consider the
homotopy h : I × I −→ C defined by h(x, t) = ((1 − t)R∗ + ta∗ )ei((1−t)θ1 +tθ2 ) e2πinx .
∗
The function (1 − t)R∗ + ta∗ has a zero at t = R∗R−a∗ . If R∗ = a∗ , then the function
is constant with no zero. If R∗ < a∗ , then the zero is negative. If R∗ > a∗ , then
the zero is greater than 1. In every case, t ∈/ I when the function is zero. Thus,
the homotopy h is never 0, and, by running each homotopy at twice speed, we have
that f ◦ Γ ' a0 g2 with a nowhere zero homotopy.
 PROOFS OF THE FUNDAMENTAL THEOREM OF ALGEBRA 3

We now consider the fundamental group π1 (C \ {0}, a0 ), and we show that a0 g1

and a0 g2 belong to different elements of the group. We use the fact that the complex
plane is a covering space of C \ {0} and that the exponential function is a covering
map. Also, the paths a0 g1 and a0 g2 have unique liftings to paths in C beginning
at a0 . We show that the endpoints of these paths are different. This is sufficient
to show that they belong to different equivalence classes in the fundamental group
[1].
Consider the covering map p : C −→ C \ {0} defined as p(z) = eiz . Let e0 be the
point in C such that a0 = eie0 . We consider liftings of a0 g1 and a0 g2 that begin
at a0 in C. They are ag 0 g1 : I −→ C, defined as ag 0 g1 (x) = e0 , and ag 0 g2 : I −→ C,
ie0
defined as ag 0 g2 (x) = 2πnx + e0 . As a check, (p ◦ ag 0 g1 )(x) = e = a0 = a0 g1 (x),
i(2πnx+e0 )
and (p ◦ ag 0 g2 )(x) = e = a0 e2πinx = a0 g2 (x), as required. Also, note
that ag 0 g1 (1) = e0 and that ag 0 g2 (1) = 2πn + e0 . Because n ≥ 1, the liftings have
different ending points, so the paths a0 g1 and a0 g2 are not homotopic in C \ {0}. It
follows from the transitivity of homotopy relations that f ◦ γ and f ◦ Γ are also not
homotopic in C \ {0}. This is a contradiction of the fact that they are homotopic.
Thus, the image of f must contain 0, so f has a root.
As mentioned previously, the existence of n roots follows from an inductive
argument once it is known that at least one root exists. Let P (z) = z n +an−1 z n−1 +
· · · + a0 be a polynomial of degree n. If z0 is a root of P , then P (z) = (z − z0 )Q(z),
where Q is a polynomial of degree n−1. Because Q also has a root, we can continue
expanding in this manner until P is written as the product of n linear factors. Thus,
P has a total of n roots. This completes our first proof of the fundamental theorem
of algebra.


2. A Similar Proof Using the Language of Complex Analysis

We now present a proof of the fundamental theorem of algebra that is similar
to the above but written in the language of complex analysis. We show that for a
large enough circle centered at the origin, the image of the circle will wrap around
the origin n times, but, under the assumption that the image of the polynomial lies
in C \ {0}, it does not wrap around the origin at all.
Consider a circle Γ = Re2πix , with x ∈ I and R chosen large enough that
|Rn e2πixn | > |an−1 R(n−1) e2πix(n−1) + · · · + a0 |. Then, using Rouche’s Theorem,

f 0 (z)
Z Z
dz
= dz
f (Γ) z Γ f (z)
(z n )0
Z
= dz
zn
ZΓ
n
= dz
Γ z
Z 1
n
= 2πix
2πiRe2πix dx
0 Re
Z 1
= 2πindx
0
= 2πin
4 MATTHEW STEED

We now show that under the assumption that f is never 0, the integral must be 0.
Polynomials are holomorphic functions, and the inverses of holomorphic functions
are holomorphic wherever the function is nonzero. If the image of f is contained
in C \ {0}, then f1 is holomorphic everywhere. Also, f 0 is a polynomial of degree
f0
n − 1, so it is holomorphic everywhere. Then, f is holomorphic everywhere. By
f0
Cauchy’s integral theorem, the integral of over a closed path is 0. Then,
f
Z 0 Z
f (z) dz
dz =
Γ f (z) f (Γ) z
=0
This is a contradiction, so the image of f must contain 0. Then, there is at
least one zero of f . The existence of n zeros, with possible multiplicity, follows by
induction as in the previous proof. 

3. A Proof Using the Maximum Modulus Principle

We now provide a proof of the fundamental theorem of algebra that makes use
of the maximum modulus principle, i.e., the modulus of a holomorphic function on
a connected, open set has no local maximum unless the function itself is constant.
Consider the disk of radius R centered at the origin such that |f (Reiθ )| > 21 |Reiθ |n
and 12 |Reiθ |n > |a0 |. There exists an R that satisfies these conditions, as shown in
the first proof. Then, for z ≥ |R|, |f (z)| > |a0 |. The function |f | is continuous and
the disk is compact, so, by the extreme value theorem, |f | has a minimum on the
disk. Call it a, and let α be a point such that f (α) = a. For any z on the boundary
of the disk, |f (z)| > |a0 | ≥ a. Thus, the minimum is not achieved on the boundary.
It is achieved in the open disk of radius R centered at the origin.
Suppose that a 6= 0. Then, f is never 0, so the function f1 is holomorphic. This
allows us to apply the maximum modulus principle to it. The open disk of radius
R is an open connected subset of C, and a1 is a local maximum of the function | f1 |.
The maximum modulus principle then implies that | f1 | is constant. However, |f |
is then constant, but this is a contradiction because polynomials are not constant.
Thus, a = 0, and α is a root of f . The existence of n roots follows as it does in the
first proof.

4. A Proof Using Liouville’s Theorem

Liouville’s Theorem, i.e. that a bounded, entire function is constant, provides
a proof of the fundamental theorem of algebra that is very similar to the proof
using the maximum modulus principle. Consider the same disk of radius R used
in the previous proof. As above, there exists some α on the disk such that |f (α)|
is a minimum on the disk. We suppose again that f (α) 6= 0. For any z such
1 1 1
that |z| ≥ |R|, |f (z)| > |f (α)|, so | f (α) | > | f (z) |. Then, | f (α) | is a maximum
1 1
of | f | over the whole complex plane. As above, | f | is holomorphic on all of C.
Then, by Liousville’s Theorem, | f1 | is constant, so |f | is constant. This is again a
contradiction, so f (α) = 0. There are again n roots, following from an induction
argument. [2]

 PROOFS OF THE FUNDAMENTAL THEOREM OF ALGEBRA 5

Acknowledgments. It is a pleasure to thank my mentor, Yiwen Zhou, for sug-

gesting as a project proofs of the fundamental theorem of algebra and for helping
me learn some of the basics of topology and complex analysis. Our meetings were
always a lot of fun, and he showed me how interesting topology is. Before the
REU I did not know anything about it, but it is now one of my favorite subfields of
mathematics. I am glad to have had Yiwen as my mentor. It was a great experience.

References
[1] James Munkres, Topology, Prentice-Hall, 2000.
[2] Lars Ahlfors, Complex Analysis, McGraw-Hill, 1979.