Complex Analysis
The Schwarz Lemma and Automorphisms of the Disk
Here is a fact that ought to feel like cheating: if a holomorphic map sends the unit disk into itself and fixes the center, then simply knowing it fixes the origin forces |f(z)| ≤ |z| everywhere and |f′(0)| ≤ 1 — a single boundary and a single normalization pin down the map's size across the whole disk. That is the Schwarz Lemma (Hermann Amandus Schwarz, 1869), one of the most disproportionately powerful one-line theorems in mathematics.
Precisely: let 𝔻 = {z ∈ ℂ : |z| < 1} and let f: 𝔻 → 𝔻 be holomorphic with f(0) = 0. Then |f(z)| ≤ |z| for all z ∈ 𝔻 and |f′(0)| ≤ 1. Moreover, if equality holds at even one interior point — |f(z₀)| = |z₀| for some z₀ ≠ 0, or |f′(0)| = 1 — then f is a rotation, f(z) = eiθz.
- FieldComplex analysis (geometric function theory)
- First provedHermann Amandus Schwarz, ~1869; named by Carathéodory
- Key hypothesesf: 𝔻 → 𝔻 holomorphic, f(0) = 0
- Statement|f(z)| ≤ |z| and |f′(0)| ≤ 1; equality ⟹ f is a rotation
- Proof techniqueDivide by z, apply the Maximum Modulus Principle
- Generalizes toSchwarz–Pick lemma (invariant under the hyperbolic metric)
Interactive visualization
Press play, or step through manually. The visualization is yours to drive — try it before reading on.
Watch the 60-second explainer
A condensed visual walkthrough — narrated, captioned, under a minute.
What the lemma actually claims
Fix the open unit disk 𝔻 = {z ∈ ℂ : |z| < 1}. Suppose f : 𝔻 → 𝔻 is holomorphic (complex-differentiable, hence given locally by a convergent power series) and satisfies the single normalization f(0) = 0. The Schwarz Lemma delivers three conclusions:
- Pointwise size bound: |f(z)| ≤ |z| for every z ∈ 𝔻.
- Derivative bound: |f′(0)| ≤ 1.
- Rigidity (equality case): if |f(z₀)| = |z₀| for some z₀ ≠ 0, or if |f′(0)| = 1, then f is a pure rotation: f(z) = eiθz for a fixed real θ.
The astonishing part is how little is assumed. We do not assume f is a polynomial, injective, or continuous up to the boundary — only that it maps the disk into itself and fixes the center. Holomorphy plus these two facts is enough to control |f| globally, and equality anywhere inside forces total rigidity.
The picture: rigidity from a single boundary
Real-variable intuition badly underestimates this. A smooth real function [−1,1] → [−1,1] with f(0)=0 can wiggle wildly; there is no reason its slope at 0 should be ≤ 1. Holomorphic functions are far more rigid: their values on any small set determine them everywhere, so a global constraint ("never leave the disk") propagates inward as a hard bound.
Geometrically, think of g(z) = f(z)/z. Because f maps into the disk, on a circle of radius r just under 1 we have |g| = |f|/r ≤ 1/r. The Maximum Modulus Principle says a holomorphic function attains its largest modulus on the boundary, so |g| ≤ 1/r holds on the entire disk of radius r — not just its edge. Now let r → 1. The bound tightens to |g| ≤ 1 everywhere, i.e. |f(z)| ≤ |z|. The center-fixing hypothesis is what makes g holomorphic in the first place: f(0)=0 removes the potential pole at the origin.
The key idea of the proof
The whole proof is one clever construction plus one classical theorem. Because f(0) = 0, the power series of f has no constant term: f(z) = a₁z + a₂z² + ⋯. Define
- g(z) = f(z)/z for z ≠ 0, and g(0) = f′(0) = a₁.
The removable-singularity theorem makes g holomorphic on all of 𝔻 (dividing by z is legal precisely because the constant term vanished). Fix any r with |z| < r < 1. On the circle |ζ| = r, |g(ζ)| = |f(ζ)|/r < 1/r, since f lands in 𝔻. By the Maximum Modulus Principle, |g(z)| ≤ 1/r throughout |z| ≤ r. Holding z fixed and letting r ↑ 1 gives |g(z)| ≤ 1, i.e. |f(z)| ≤ |z| and (at z=0) |f′(0)| = |g(0)| ≤ 1.
The rigidity case is the same principle run backwards: equality means |g| attains its maximum 1 at an interior point, which by the Maximum Modulus Principle forces g to be a constant of modulus 1. Hence g(z) ≡ eiθ and f(z) = eiθz.
Worked example: classifying the automorphisms of 𝔻
The lemma's headline payoff is a complete list of the biholomorphic self-maps (automorphisms) of the disk. Consider the Blaschke / Möbius factors
φa(z) = (z − a) / (1 − ā z), |a| < 1.
A direct check shows |φa(z)| = 1 exactly when |z| = 1, so φa maps 𝔻 bijectively to 𝔻, with inverse φ−a. Multiplying by a unimodular constant eiθ gives more automorphisms.
Claim: every automorphism has the form f(z) = eiθ φa(z). Proof via Schwarz: let f ∈ Aut(𝔻) and set a = f−1(0). Then F = f ∘ φ−a fixes 0, so |F′(0)| ≤ 1; but F−1 also fixes 0, so |(F−1)′(0)| ≤ 1 too. Since these derivatives are reciprocal, both equal 1 — the equality case fires, forcing F(z) = eiθz. Unwinding, f = eiθ φa. So Aut(𝔻) is exactly this 3-real-parameter group.
Why the hypotheses matter — and what breaks
Every hypothesis is load-bearing.
- Drop f(0) = 0: the conclusion |f(z)| ≤ |z| fails. Take f(z) = (z+½)/(1+½z), an automorphism. It maps 𝔻 → 𝔻 but f(0) = ½ > 0, violating the bound at the origin. The correct replacement is the Schwarz–Pick Lemma, which drops the fixed-point requirement by measuring distance in the hyperbolic metric ds = 2|dz|/(1 − |z|²): every holomorphic self-map is a non-expansion, and the automorphisms are exactly its isometries.
- Drop "into 𝔻": if f only maps into ℂ, nothing bounds it; f(z)=2z fixes 0 yet |f′(0)|=2.
- Drop holomorphy: a merely smooth self-map fixing 0 obeys no such bound — the argument is powered entirely by the Maximum Modulus Principle, which is false for non-holomorphic functions.
Schwarz sits alongside the Cauchy integral formula, the maximum modulus principle, and Liouville's theorem as an expression of holomorphic rigidity, and it is the seed of Pick's, Carathéodory's, and Ahlfors' later refinements.
Applications and significance
The lemma's reach far exceeds its two-line proof:
- Riemann Mapping Theorem: Schwarz supplies the uniqueness half. Any two biholomorphisms of a simply connected domain onto 𝔻 differ by an automorphism, and Schwarz pins that ambiguity to the 3-parameter Aut(𝔻) — normalizing f(z₀)=0 and f′(z₀)>0 makes the map unique.
- Hyperbolic geometry: Schwarz–Pick shows holomorphic maps contract the Poincaré metric, making 𝔻 the model of the hyperbolic plane and Aut(𝔻) ≅ PSL(2,ℝ) its orientation-preserving isometry group.
- Fixed points and dynamics: the Denjoy–Wolff theorem — iterating a holomorphic self-map converges to a single boundary or interior point — is built on Schwarz–Pick contraction.
- Estimates and bounds: it seeds Carathéodory's inequality, the Koebe distortion theorem, subordination theory, and countless coefficient and derivative estimates.
In short, Schwarz is the workhorse behind the statement that the disk is rigid: its holomorphic self-maps can only shrink hyperbolic distance, and the rare ones that preserve it form a beautifully small, explicitly described group.
| Result | Hypotheses | Conclusion | When is it sharp? |
|---|---|---|---|
| Schwarz Lemma | f: 𝔻 → 𝔻 holomorphic, f(0) = 0 | |f(z)| ≤ |z|, |f′(0)| ≤ 1 | Equality ⟺ f(z) = e^{iθ} z (a rotation) |
| Rigidity / equality case | Above, plus |f(z₀)| = |z₀| for some z₀ ≠ 0 | f is forced to be a rotation everywhere | Always — any interior equality collapses f |
| Schwarz–Pick Lemma | f: 𝔻 → 𝔻 holomorphic (no fixed point needed) | d_hyp(f(z),f(w)) ≤ d_hyp(z,w); |f′(z)|/(1−|f(z)|²) ≤ 1/(1−|z|²) | Equality ⟺ f ∈ Aut(𝔻) (a hyperbolic isometry) |
| Disk automorphisms Aut(𝔻) | f: 𝔻 → 𝔻 bijective and holomorphic | f(z) = e^{iθ} (z − a)/(1 − ā z), a ∈ 𝔻, θ ∈ ℝ | These are exactly the equality maps above |
Frequently asked questions
Why is the hypothesis f(0) = 0 essential?
It lets you divide: writing g(z) = f(z)/z is only legal because f(0)=0 kills the constant term of the power series, so g has a removable singularity at 0 rather than a pole. Without it the bound |f(z)| ≤ |z| is simply false — for instance f(z) = (z+½)/(1+½z) is a disk self-map with f(0)=½. The fixed-point-free generalization is the Schwarz–Pick Lemma.
What exactly happens in the equality case?
If |f(z₀)| = |z₀| for any single interior point z₀ ≠ 0, or if |f′(0)| = 1, then f must be a rotation f(z)=e^{iθ}z. The reason is that equality means |g(z)|=|f(z)/z| attains its maximum value 1 at an interior point, and the Maximum Modulus Principle says a holomorphic function whose modulus peaks inside its domain is constant. This is called the rigidity phenomenon.
How does Schwarz give all automorphisms of the disk?
Every automorphism is f(z) = e^{iθ}(z−a)/(1−āz) for some a ∈ 𝔻 and θ ∈ ℝ. Given any automorphism f, precompose with a Blaschke factor to fix 0; then both that map and its inverse fix 0 and have derivative of modulus ≤ 1 by Schwarz. Since the derivatives are reciprocal, both equal 1, so the equality case forces a rotation. Unwinding gives the Blaschke-times-rotation form.
What is the Schwarz–Pick Lemma and how does it improve Schwarz?
Schwarz–Pick removes the fixed-point requirement. For any holomorphic f: 𝔻 → 𝔻, |f′(z)|/(1−|f(z)|²) ≤ 1/(1−|z|²), meaning f never increases the hyperbolic (Poincaré) distance. Equality at one point holds exactly when f is an automorphism, i.e. a hyperbolic isometry. Classical Schwarz is the special case z=0, f(0)=0, where the hyperbolic metric reduces to the ordinary derivative bound.
Does the Schwarz Lemma work for domains other than the disk?
Not verbatim, but morally yes. By the Riemann Mapping Theorem any simply connected proper subdomain of ℂ is biholomorphic to 𝔻, so you can transport the lemma. More intrinsically, every hyperbolic domain carries a Poincaré metric, and the Ahlfors–Schwarz Lemma states holomorphic maps contract it. Rigidity of self-maps is thus a general feature of hyperbolic geometry, not special to the round disk.
Is the derivative bound |f′(0)| ≤ 1 sharp, and by which maps?
Yes, it is attained exactly by the rotations f(z)=e^{iθ}z, which have |f′(0)|=1. These are the only self-maps fixing 0 that achieve equality — the rigidity clause guarantees no non-rotation can. This sharpness is what makes Schwarz a genuine extremal principle rather than a loose estimate, and it drives extremal problems throughout geometric function theory.