1.

Introduction and Basic Properties

If C denotes the set of complex numbers, then
(1.1.1)
When we restrict our attention to polynomials with real coefficients we will
use the notation
(1.1.2)
where IR is the set of real numbers. Rational functions of type (m, n) with
complex coefficients are then defined by

(1.1.3) R~,n := {~ : p E p~, q E p~} ,

while their real cousins are denoted by
(1.1.4)
The distinction between the real and complex cases is particularly important
for rational functions (see E.4).
The set of trigonometric polynomials T/: is defined by
(1.1.5)
A real trigonometric polynomial of degree at most n is an element of T/:
taking only real values on the real line. We denote by Tn the set of all real
trigonometric polynomials of degree at most n. Other characterizations of
Tn are given in E.g. Note that if z := eiO , then an arbitrary element of T/:
is of the form
(1.1.6)
and so many properties of trigonometric polynomials reduce to the study
of algebraic polynomials of twice the degree on the unit circle in iC.
The most basic theorem of this book, and arguably the most basic
nonelementary theorem of mathematics, is the FUndamental Theorem of
Algebra. It says that a polynomial of exact degree n (that is, an element

The Fundamental Theorem of Algebra appears to have been given its name
by Gauss, although the result was familiar long before; it resisted rigorous
proof by d'Alembert (1740), Euler (1749), and Lagrange (1772). It was more
commonly formulated as a real theorem, namely: every real polynomial factors
completely into real linear or quadratic factors. (This is an essential
result for the integration of rational functions.) Girard has a claim to priority
of formulation. In his "Invention Nouvelle en L'Algebra" of 1629 he
wrote "every equation of degree n has as many solutions as the exponent
of the highest term." Gauss gave the first satisfactory proof in 1799 in his
doctoral dissertation, and he gave three more proofs during his lifetime. His
first proof, while titled "A new proof that every rational integral function of
one variable can be resolved into real factors of the first or second degree,"
was in fact the first more-or-less satisfactory proof. Gauss' first proof is a
geometric argument that the real and imaginary parts of a polynomial, u
and v, have the property that the curves u = 0 and v = 0 intersect, and by
modern standards has some topological problems. His third proof of 1816
amounts to showing that

There is a sharpening of Lucas' theorem for real polynomials formulated

by Jensen. We need to introduce the notion of Jensen circles for a
polynomial p E Pn . For p E Pn the nonreal roots of p come in conjugate
pairs. For each such pair, a +i(3, a - i(3, form the circle centered at a with
radius 1(31. So this circle centered on the x-axis at a has a+i(3 and a-i(3 on
the opposite ends of its perpendicular diameter. The collection of all such
circles are called the Jensen circles for p.
Theorem 1.3.3 (Jensen's Theorem). Let p E Pn- Each nonreal zero of p'
lies in or on some Jensen circle for p.
The proof, which is similar to the proof of Lucas' theorem, is left for
the reader as E.3.
We state the following pretty generalization of Lucas' theorem due to
Walsh [21]. The proof is left as EA. Proofs can also be found in Marden
[66J and P61ya and Szego [76].

If all the zeros of p E Pn are real, then all the zeros of p~ are also real.
b] Suppose all the zeros of both p E Pn and q E Pm are real, and all the
zeros of Pn are smaller than any of the zeros of qn' Show that all the zeros
of (p/q)' are real.
Hint: Consider the graph of
(p/q), p' q'
(p/q) p q
D
Define W(p), the Wronskian of p, by
W(p)(z) = p(z)p"(z) - (p'(Z))2
_I p(z) p'(z) I
- p'(z) p"(z)
= p2(z) (p'(Z))'
p(z)
c] Prove that if p E Pn has only distinct real zeros, then W(p) has no real
zeros.
In Craven, Csordas, and Smith [87] it is conjectured that, for p E Pn ,
the number of real zeros of W (p) / p2 does not exceed the number of nonreal
zeros of p (a question they attribute to Gauss).
d] Let p E Pn . Show that any real zero of W(p) lies in or on a Jensen
circle of p.
Proof. See Dilcher [91]. D
e] Show that Lucas' theorem does not hold for rational functions.
Hint: Consider r(x) = x/(a2
- x2
). D
The next exercise is a weak form of Descartes' rule of signs.
E.6 Positive Zeros of Muntz Polynomials. Suppose 60 < 61 < ... < 6n
and
Show that either f = 0 or f has at most n zeros in (0, 00).

Suppose n ?: 3. Prove that L1(p) < L1(p').

Outline. It is required to show that {3j - (3j-1 > L1(p) for each j ?: 2. Let
2 :::; j :::; n be fixed. Since
we have
n1

"~" =0 ({3j-1 - ai)({3j - ai) .

Now let Uj := aj - {3j-l, Vj:= {3j - aj. Also for each i, let di := aj - aj-i,
ei := aj+i - aj. Then the above can be rewritten as
j-l 1 1 n-j 1
"L"- (d - u)(d + v) + (-u·v·) +L""- (e· + u·)(e· - v·) =0 .
i=l ' J 'J J J i=l ' J ' J
Define
j-l n-j
"" uv "" uv
F(u, v) := ~ (di _ u)(di + v) + ~ (ei + u)(ei - v) .
Note that F is increasing in each variable (0 :::; u < dl , 0:::; V < el) and
observe that
To prove the result, it suffices to show that if u and v are nonnegative
numbers satisfying u + v = L1(p), then F(u,v) < 1.
The Chebyshev polynomials Tn are named after the versatile Russian mathematician,
P. L. Chebyshev (1821-1894). The T comes from the spelling
Tchebychef (or some such variant; there are many in the literature). A
wealth of information on these polynomials may be found in Rivlin [90].
Throughout later sections of this book the Chebyshev polynomials will
keep recurring. The initial exercises explore elementary properties of the
Chebyshev polynomials.
Erdos [39] proved that for t E Tn with IItlllR. :::; 1, the length of the graph
of t on [0, 21f] is the longest if and only if t is of the form t(B) = cos(nB +a)
with some a E JR (see E.6). He conjectured that for any p E Pn with
Ilpll[-I,I] :::; 1, the maximum arc length is attained by the nth Chebyshev
polynomial Tn. This is proved in Bojanov [82b]. Kristiansen [79] also claims
a proof. In E.9 the reducibility of Tn is considered, and in E.ll the basic
properties of the transfinite diameter are established.
E.1 Basic Properties.
a] Establish the equivalence of the three representations of Tn given in
equation (2.1.1).
Hint: cosnB = ~[(cosB + i sin B

Let E be a compact subset of C. Let

The points Zi at which the above maximum are obtained are called nth
Fekete points for E. If the points Zi are the nth Fekete points for E, then
the polynomial
n

qn(z) := II (z - Zi)
i=1
is called an nth (monic) Fekete polynomial for E. The transfinite diameter
or logarithmic capacity of E is defined by
1

cap(E):= lim (.~1n(E)) n(n-l) ,

n--->oo
where the limit exists by part c] (below).
a] Let Z1, Z2," . ,Zn be nth Fekete points for E. Then
1 Z1 Z1n-1
1 n-1
(.<1n(E))1/2 = abs
Z2 Z2
1 Zn z~-1
Hint: See E.2 b] (Vandermonde determinant) of Section 3.2. 0
b] Let qn(z) := TI~=1 (z - Zi) be an nth Fekete polynomial for E. Let
ffi n:= . min Iq~(Zi)1 and Mn := IlqnllE'
~=ll'" ,n

Then
Outline. We have
n

Iqn(z)1 2.<1n(E) = II Iz - Zi!2 II IZi - zjl :=; .<1n+l(E)

i=1 1:C;i,j:C;n
i#j
and

II
1:C;i,j:C;n+1
i#,j#k,i#j

:=;.<1n(E) II IZk
Hint: Without loss of generality we may assume that the arc An is symmetric
with respect to the x-axis and 1 E An. Now use part h] and the
transformation x = ~ (z + Z-1 ). D
2.2 Orthogonal Functions
The most basic properties of orthogonal functions are explored in this section.
The following section specializes the discussion to polynomials.
In this section the functions are complex-valued and the vector spaces
are over the complex numbers. All the results have obvious real analogs
and in many later applications we will restrict to these corresponding real
cases.
An inner product on a vector space V is a function (".) from V x V
to C that satisfies, for all i, g, h E V and a, (3 E C,
(2.2.1) (j,1) > 0 unless i = 0 (positivity)
(2.2.2) (j, g) = (g,1) (conjugate symmetry)
(2.2.3) (ai + (3g, h) = a(j, h) + (3(g, h) (linearity).
A vector space V equipped with an inner product is called an inner
product space. It is a normed linear space with the norm 11.11 := (', Y/2.
The canonical example for us will be the space C[a, b] of all complexvalued
continuous functions on [a, b] with the inner product

(2.2.4) (j, g) :=
i(x)g(x)w(x) dx,
l
b

where w(x) is a nonnegative integrable function on [a, b] that is positive

except possibly on a set of measure zero. It is a normed linear space with