The Fundamental Theorem of Algebra and Linear Algebra

Author(s): Harm Derksen

Source: The American Mathematical Monthly, Vol. 110, No. 7 (Aug. - Sep., 2003), pp. 620-623
Published by: Mathematical Association of America
Stable URL: http://www.jstor.org/stable/3647746
Accessed: 03/03/2009 10:47

Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at
http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unless
you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you
may use content in the JSTOR archive only for your personal, non-commercial use.

Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at
http://www.jstor.org/action/showPublisher?publisherCode=maa.

Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed
page of such transmission.

JSTOR is a not-for-profit organization founded in 1995 to build trusted digital archives for scholarship. We work with the
scholarly community to preserve their work and the materials they rely upon, and to build a common research platform that
promotes the discovery and use of these resources. For more information about JSTOR, please contact [email protected].

Mathematical Association of America is collaborating with JSTOR to digitize, preserve and extend access to
The American Mathematical Monthly.

http://www.jstor.org
 NOTES
Edited by William Adkins

The Fundamental Theorem of Algebra

and Linear Algebra
Harm Derksen

1. INTRODUCTION. The first widely accepted proof of the fundamental theorem

of algebra was published by Gauss in 1799 in his Ph.D. thesis, although by today's
standards this proof has gaps. In 1814 Argand gave a proof (with only small gaps)
that was based on a flawed 1746 proof of d'Alembert. Many other proofs followed,
including three more by Gauss. For more about the history of the fundamentaltheorem
of algebra, see [5] or [6].
Proofs of the fundamentaltheorem of algebra can be divided roughly into three cat-
egories (see [3] for a collection of proofs). First there are the topological proofs (see [1]
or [8]). These proofs are based on topological considerations such as the winding num-
ber of a curve in 12 around 0. Gauss's original proof might fit under this heading as
well. Then there are analytical proofs (see [9]), which are related to Liouville's theo-
rem: a nonconstant entire function on C is unbounded. Finally, there are the algebraic
proofs (see [4] or [10]). These proofs use only the fact that every polynomial of odd de-
gree with real coefficients has a real root and that every complex number has a square
root. The deeper reasons why these arguments work can be understood in terms of
Galois theory.
Since the fundamental theorem of algebra is needed in linear algebra courses, it
would be desirable to have a proof of it in terms of linear algebra. In this paper we
prove that every squarematrix with complex coefficients has an eigenvector. This state-
ment is equivalent to the fundamental theorem of algebra. In fact, we will prove the
slightly stronger result that any number of commuting square matrices with complex
entries have a common eigenvector. The proof lies entirely within the framework of
linear algebra, and unlike most other algebraic proofs of the fundamental theorem of
algebra, it does not require Galois theory or splitting fields. Another (but longer) proof
using linear algebra can be found in [7].

2. PRELIMINARIES. For the proof we use only the following elementary proper-
ties of real and complex numbers.

Lemma 1. Every polynomial of odd degree with real coefficients has a zero.

Proof. It is enough to prove that a monic polynomial

P(x) = xn + alxn- + - .- + an

with a..., . an in R and n odd has a zero. If a = la,l * + lan

+ a I + 1, then it is easy
to see that P(a) > 0 and P(-a) < 0. By the intermediate value theorem there exists
X in the interval [-a, a] such that P (X) = 0. X

620 t? THE MATHEMATICAL ASSOCIATION OF AMERICA [Monthly 110

Lemma 2. Every complex number has a square root.

Proof. Consider a + fBiwith a and 15real. If y = Va2 + /2, then

( +a + =a i.
2 + 2 i) +

3. THE PROOF. For a field K and for positive integers d and r, consider the follow-
ing statement:
P(K, d, r): Any r commuting endomorphisms A1, A2, ... , Ar of a K-vector space V
of dimension n such that d does not divide n have a common eigenvector.

Lemma 3. If'P(K, d, 1) holds, then P(K, d, r) holds for all r > 1.

Proof We prove the lemma by induction on r.

Assume that P(K, d, r - 1) holds. Suppose that A1, A2, ..., Ar are commuting
endomorphisms of a K-vector space V of dimension n such that d does not divide n.
By induction on n we prove that AI, A2,..., Ar have a common eigenvector. The case
n = 1 is trivial.
Because P(K, d, 1) holds, Ar has an eigenvalue X in K. Let W be the kernel and Z
the image of Ar - 3I. Note that W and Z are stable under AI, A2, ..., Ar-1.
Suppose that W = V. Because dim W + dim Z = dim V, either d does not divide
dim W or d does not divide dim Z. Since dim W < n and dim Z < n, we may assume
by induction on n that A1, ..., Ar already have a common eigenvector in W or in Z.
In the remaining case, W = V. Because P(K, d, r - 1) holds, we may assume that
A, ..., Ar-1 have a common eigenvector in V, say v. Since Arv = Xv, v is a common
eigenvector of A, ..., Ar. U

Lemma 4. P(1R, 2, r) holds for all r, i.e., if A1, , Ar are commuting endomor-
phisms on an odd dimensional R-vector space, then they have a common eigenvector.

Proof: By Lemma 3 it is enough to show that P(RI, 2, 1) is true. If A is an endomor-

phism of an odd dimensional IE-vectorspace, then det(xI - A) is a polynomial of odd
degree, which has a zero X by Lemma 1. Then X is a real eigenvalue of A. X

Lemma 5. P(C, 2, 1) holds, i.e., every endomorphism of a C-vector space of odd

dimension has an eigenvector

Proof Suppose that A : Cn -> Cn is a C-linear map with n odd. Let V be the IR-
vector space Hermn(C), the set of n x n Hermitianmatrices. We can define commuting
endomorphisms L1 and L2 of V by
AB + BA*
L1(B)=
2
and
AB - BA*
L2(B) =
A*=
Here
A is the transpose of the complex conjugate of the matrix A.
Here A* = A is the transpose of the complex conjugate of the matrix A.

August-September 2003] NOTES 621

 Observe that dimR V = n2 is odd. Now P(IR, 2, 2) (see Lemma 4) implies that L1
and L2 have a common eigenvector B, say L1(B) = XB and L2(B) = I B with X and
,/ real. But then

(L1 + iL2)(B) = AB = (X + q,i)B,

and any nonzero column vector of B gives an eigenvector for the matrix A.1

Lemma 6. P(C, 2k, r) holds for all k and r.

Proof We prove the lemma by induction on k. The case k = 1 follows from Lemmas 5
and 3. Assume that P(C, 21, r) holds for I < k. We will establish P(C, 2k, r). In view
of Lemma 3, it suffices to prove P(C, 2k, 1). Suppose that A : Cn - Cn is linear,
where n is divisible by 2k-1 but not by 2k. Let V be the C-vector space Skewn(C), the
set of n x n skew-symmetric matrices with complex entries. Define two commuting
endomorphisms L1 and L2 of V by

L1(B) =AB - BAt

and

L2(B) = ABAt.

Note that dim V = n(n - 1)/2, which ensures that 2k-1 does not divide dim V.
By P(C, 2k-1, 2), L1 and L2 have a common eigenvector B, say Li(B) = XB and
L2(B) = ftB, where X and /t are now complex numbers. It follows that

IgB = ABAt = A(AB - XB),

so

(A2 - XA - gII)B = 0.

Let v be a nonzero column of B. Then

(A2 - XA - IlI)v = 0.

By Lemma 3 there is a a in C such that 82 = A-2 + 4/. We can write (x2 - Xx - ,I) =
(x - a)(x - ,), where a = (X + 3)/2 and = ((X - S)/2. We then have

(A -aI)w = 0,

where w = (A - flI)v. If w = 0, then v is an eigenvector of A with eigenvalue P; if

w : 0, then w is an eigenvector of A with eigenvalue a. a

Theorem 7. If A1, A2, . .., Ar are commuting endomorphisms of a finite dimensional

nonzero C-vector space V, then they have a common eigenvector.

Proof: Let n be the dimension of V. There exists a positive integer k such that 2k does
not divide n. Since P(C, 2k, r) holds by Lemma 6, the theorem follows. 1

Corollary 8 (Fundamental Theorem of Algebra). If P (x) is a nonconstant polyno-

mial with complex coefficients, then there exists a X in C such that P(X) = 0.

622 ? THE MATHEMATICAL ASSOCIATION OF AMERICA [Monthly 110

Proof. It suffices to prove this for monic polynomials. Suppose that

P(x) = n + alxn-1 + a2xn-2 + * ** + an.

Then P(x) = det(xI - A), where A is the companion matrix of P:

0 ... 0 -an
1 0 0 -an-_l
A= 0 1 0 -an-2

0 0 ... 1 -al

Theorem 7 implies that A has a complex eigenvalue X in C, from which it follows that
P(X) = 0.

Remark. As for all algebraic proofs of the fundamentaltheorem of algebra, the state-
ment can be generalized to more general fields. An orderedfield R is a field with the
following properties: (i) for every a in R \ {0}, either a or -a is a square and (ii) the
sum of any two squares in R is also a square. On such an ordered field there is a total
ordering defined by a < f3 if and only if B - a is a square. If a in R is a square in R,
then we define Xa to be the unique f3in R such that 82 = a and ,8 is itself a square
in R. The element -1 is not a square in an ordered field. We can construct a field
C by adjoining an element i with i2 = -1 to R in a fashion similar to the way C is
constructed from R. It can be shown (just as for C) that any element of C has a square
root. If we assume R is an ordered field such that every polynomial of odd degree has
a zero, then the foregoing proof goes through with IRreplaced by R and C replaced
by C. In particular,C is algebraically closed.

ACKNOWLEDGMENT. The authorwas partially supportedby NSF AwardDMS-0102193.

REFERENCES

1. B. H. Arnold, A topological proof of the fundamentaltheorem of algebra, this MONTHLY56 (1949)

465-466.
2. J. L. Brennerand R. C. Lyndon, Proof of the fundamentaltheorem of algebra,this MONTHLY88 (1981)
253-256.
3. B. Fine and G. Rosenberger,The FundamentalTheoremof Algebra, Springer-Verlag,New York, 1997.
4. L. Horowitz, A proof of the "fundamentaltheorem of algebra"by means of Galois theory and 2-Sylow
groups, Nieuw Arch. Wisk.(3) 14 (1966) 95-96.
5. H.-D. Ebbinghaus, H. Hermes, F. Hirzebruch,M. Koecher, K. Mainzer, A. Prestel, and R. Remmert,
Zahlen, K. Lamotke, ed., Springer-Verlag,Berlin, 1983.
6. C. Gilain, Sur l'histoire du thdoremefondamentalde l'algebre: theorie des dquationset calcul intdgral,
Arch. Hist. Exact Sci. 42 (1991) 91-136.
7. H.-J. Kowalsky,Zum Fundamentalsatzder Algebra,Abh. Braunschweig. Wiss. Ges. 35 (1983) 111-120.
8. S. Stein, The fundamentaltheorem of algebra,this MONTHLY 61 (1954) 109.
9. F. Terkelsen,The fundamentaltheorem of algebra,this MONTHLY83 (1976) 647.
10. H. Zassenhaus, On the fundamentaltheorem of algebra,this MONTHLY 74 (1967) 485-497.

Department of Mathematics, Universityof Michigan, East Hall, 525 East University,Ann Arbor, MI 48109-
1109
[email protected]

August-September 2003] NOTES 623