TheoremComplete

Montel's Theorem and Normal Families

Normal families provide the compactness framework for spaces of holomorphic functions, analogous to the Arzela-Ascoli theorem for continuous functions. Montel's theorem is the key tool for extracting convergent subsequences.


Normal Families

Definition7.5Normal family

A family F\mathcal{F} of holomorphic functions on a domain DD is normal if every sequence in F\mathcal{F} has a subsequence that converges uniformly on every compact subset of DD. The limit function is holomorphic by the theorem on uniform limits.

Theorem7.6Montel's theorem

A family F\mathcal{F} of holomorphic functions on a domain DD is normal if and only if it is locally bounded: for every compact KDK \subset D, there exists MKM_K such that f(z)MK|f(z)| \leq M_K for all fFf \in \mathcal{F} and zKz \in K.


Proof of Montel's Theorem

Proof

Step 1: Equicontinuity from local boundedness. Fix a compact KDK \subset D and let δ=dist(K,D)/2\delta = \text{dist}(K, \partial D)/2. For z1,z2Kz_1, z_2 \in K with z1z2<δ|z_1 - z_2| < \delta, Cauchy's integral formula on wz1=δ|w-z_1| = \delta gives:

f(z1)f(z2)=12πiwz1=δf(w)(1wz11wz2)dwMz1z2δ22πδ=2πMz1z2δ.|f(z_1) - f(z_2)| = \left|\frac{1}{2\pi i}\oint_{|w-z_1|=\delta} f(w)\left(\frac{1}{w-z_1} - \frac{1}{w-z_2}\right)dw\right| \leq \frac{M |z_1-z_2|}{\delta^2} \cdot 2\pi\delta = \frac{2\pi M|z_1-z_2|}{\delta}.

This gives uniform equicontinuity of F\mathcal{F} on KK.

Step 2: Arzela-Ascoli. Choose a countable dense subset {qn}D\{q_n\} \subset D. For any sequence {fk}F\{f_k\} \subset \mathcal{F}, use a diagonal argument: extract a subsequence converging at q1q_1, then a sub-subsequence converging at q2q_2, etc. The diagonal subsequence converges at all qnq_n.

Step 3: Uniform convergence. Equicontinuity plus pointwise convergence on a dense set implies uniform convergence on compact subsets (standard argument via finite ε\varepsilon-nets). \blacksquare


Montel's Great Theorem

Theorem7.7Montel's theorem (strong form)

A family F\mathcal{F} of holomorphic functions on a domain DD that all omit two fixed values a,bCa, b \in \mathbb{C} (aba \neq b) is normal.

RemarkConnection to Picard

Montel's great theorem implies Picard's little theorem: if ff is entire and omits two values, consider F={fn}\mathcal{F} = \{f_n\} where fn(z)=f(nz)f_n(z) = f(nz). Each fnf_n omits the same two values, so by Montel, {fn}\{f_n\} is normal. Analysis of the limit gives ff constant. The proof uses the theory of the modular function λ\lambda.


Applications

ExampleFixed-point theorem for holomorphic self-maps

Let f:DDf: \mathbb{D} \to \mathbb{D} be holomorphic. The iterates fn=fff^n = f \circ \cdots \circ f form a normal family (bounded by 11). If ff has no fixed point in D\mathbb{D}, the Denjoy-Wolff theorem says fncf^n \to c uniformly on compacts for some cDc \in \partial\mathbb{D}. If ff has a unique fixed point z0Dz_0 \in \mathbb{D} and ff is not a rotation, then fnz0f^n \to z_0 uniformly on compacts.

ExampleVitali's convergence theorem

If {fn}\{f_n\} is a normal family on DD and fnff_n \to f pointwise on a set with an accumulation point in DD, then fnff_n \to f uniformly on compact subsets. (Any convergent subsequence must have the same limit by the identity theorem.)

RemarkCompactness in function spaces

Montel's theorem is the complex-analytic analogue of the Arzela-Ascoli theorem. It provides the compactness needed for existence proofs throughout complex analysis: the Riemann mapping theorem, normal forms for conformal maps, and value distribution theory all rely on it fundamentally.