A global bound controls a local derivative
A bounded differentiable real function need not be constant: is an immediate example. For an entire function, however, differentiability in every complex direction and a bound on the whole plane together impose much stronger control. Cauchy's estimate lets a circle of arbitrarily large radius constrain the derivative at its center.
Theorem 1.10.1
(Liouville's theorem)
If is entire and there is a finite constant such that for every , then is constant.
Proof
(Let the circle grow)
Fix any . For every , the entire function is holomorphic on a neighborhood of the closed disk . The Cauchy estimate on its boundary gives
The same fixed works for every radius. Sending shows . Since was arbitrary, everywhere. For any , integrate the derivative along the line segment from to ; the chain rule gives . Thus is constant.
Proof X-ray
What the growing circle uses
The numerator stays bounded independently of , while the denominator grows. A different bound for each circle is not enough unless . Each closed disk must have a neighborhood on which is holomorphic, so this proof also uses that is entire.
The hypothesis is a single bound on the whole plane, and a bound on a smaller set does not suffice.
Counterexample mode
A bound on the real axis is not a plane bound
The entire function satisfies for real , but , whose modulus is unbounded as . Thus real-axis boundedness cannot replace boundedness on .
Common mistake
(Bounded on every disk is not bounded)
Every entire function is continuous, hence bounded on each closed disk by some . That is not the hypothesis of Liouville's theorem: for one has , and the proof's bound does not tend to zero. The theorem needs one constant that serves every radius.
Why every nonconstant complex polynomial has a zero
The fundamental theorem of algebra states that every nonconstant polynomial with complex coefficients has at least one complex zero. The idea is to suppose that there is no zero. Its reciprocal would then be entire; polynomial growth at infinity would make the reciprocal small there, while compactness controls it on the remaining disk.
Theorem 1.10.2
(Fundamental theorem of algebra)
Let with and . Then for some .
Proof
(A zero-free polynomial would have a bounded reciprocal)
Suppose has no zero. Then is entire. Since
there is such that whenever . Consequently as and is bounded outside that disk. On the closed disk , the continuous function is bounded by compactness. Hence is bounded on all of . Liouville's theorem makes constant; its limit at infinity forces that constant to be zero, contradicting .
Repeated division turns the existence of one root into a complete factorization, once multiplicity is made precise.
Definition 1.10.3
(Multiplicity of a root)
Let be a nonzero polynomial and . The multiplicity of as a root of is the largest integer such that divides . Equivalently, .
Corollary 1.10.4
(Complete factorization over )
If with and , there are , unique up to order, with
Thus has exactly roots counted with multiplicity.
Proof
(Induction on the degree)
For , . For , the theorem gives a root , and polynomial division writes with constant; evaluating at gives . The quotient has degree and leading coefficient , so the induction hypothesis factors it. Uniqueness holds because has no zero divisors: if two such products agree, any in the first is a root of the second, so it equals some factor there; cancel and repeat.
A nonzero constant polynomial has no roots; the zero polynomial vanishes everywhere. Neither is a counterexample, because the theorem assumes a nonconstant polynomial. The statement concerns complex roots: has no real root but has the complex roots and .
Common mistake
(Exactly roots means counted with multiplicity)
A polynomial of degree need not have distinct roots. For , of degree , the root has multiplicity , so there are only two distinct roots. The count is the sum of the multiplicities.
Worked examples
Worked example 1.10.5
(Factoring a cubic after finding one root)
Consider . A direct calculation gives . Dividing by yields
The theorem guarantees a complex root before we find one; it does not identify that root or say it is real. Here the factorization tells us all three roots, each of multiplicity one.
Worked example 1.10.6
(Linear growth forces a linear function)
Let be entire with for all . The Cauchy estimate for the second derivative on gives
Hence and . The bound is satisfied by , so the conclusion cannot be improved to “constant”: Liouville's theorem is the edge case in which the growth is bounded.
Worked example 1.10.7
(A one-sided bound on the real part)
Let be entire with for all . No bound on is given, but is entire with . Liouville makes constant, so , and since never vanishes, . Thus is constant. The same argument with , handles a lower bound on or a one-sided bound on .
Practice with solutions
Quick check
(An entire function satisfies for all . If , find .)
Which theorem applies?
Solution · Answer
Liouville's theorem gives , so .
Quick check
(Let be entire with for constants . Show that is constant. When , why does this bound alone not directly justify Liouville's theorem?)
Repeat the proof of Liouville's theorem with this bound.
Solution · Answer
At any , Cauchy's estimate on gives . Hence and is constant. The displayed upper bound itself grows without limit when , so it does not by itself supply the uniform bound that Liouville's theorem requires; the proof needs only .
Quick check
(Explain why cannot be used as a bounded entire function in the polynomial proof. Factor over .)
Where is the reciprocal undefined?
Solution · Answer
Since , the reciprocal has poles at and and is not entire. The reciprocal argument applies only under the temporary assumption that the polynomial has no zero, which this polynomial violates.
Quick check
(Count the complex roots of , with and without multiplicity.)
Read the exponents in the factorization.
Solution · Answer
There are three roots counted with multiplicity: twice and once. There are two distinct roots.