Complex Analysis

The Maximum Modulus Principle: Peaks Live on the Boundary

Here is a fact that has no analogue in the real world: if a holomorphic function on a disk ever attains its largest magnitude at an interior point, then that function is constant — it has no variation whatsoever. There is no such thing as a genuine interior "peak" of |f|. Every honest maximum of a non-constant analytic function is forced out onto the boundary of the region.

Precisely: if f is holomorphic on a domain (open connected set) Ω ⊂ ℂ and |f| attains a local maximum at some point z₀ ∈ Ω, then f is constant on Ω. Equivalently, on a bounded domain with f continuous up to the closure, max over the closure of |f| equals max over the boundary ∂Ω. Rigidity, not smoothness, is doing the work.

  • FieldComplex analysis
  • Key hypothesisf holomorphic on a domain (open + connected)
  • ConclusionInterior max of |f| ⟹ f constant
  • Proof techniqueMean value property / open mapping theorem
  • Sharpened bySchwarz lemma, Hadamard three-lines theorem
  • Boundary versionmax_{Ω̄} |f| = max_{∂Ω} |f| on bounded Ω

Interactive visualization

Press play, or step through manually. The visualization is yours to drive — try it before reading on.

Open visualization fullscreen ↗

Watch the 60-second explainer

A condensed visual walkthrough — narrated, captioned, under a minute.

The precise statement

Let Ω ⊆ ℂ be a domain — a non-empty open connected set — and let f: Ω → ℂ be holomorphic (complex-differentiable, equivalently analytic). The principle comes in two standard forms.

  • Interior (rigidity) form: If |f| attains a local maximum at some point z₀ ∈ Ω — i.e. |f(z)| ≤ |f(z₀)| for all z in some disk around z₀ — then f is constant on all of Ω.
  • Boundary form: If Ω is bounded and f is continuous on the closure Ω̄ and holomorphic on Ω, then max over Ω̄ of |f| is attained on the boundary ∂Ω, so max_{Ω̄} |f| = max_{∂Ω} |f|.

Connectedness is essential to the interior form: it is what upgrades "constant near z₀" to "constant on Ω." Note the conclusion is about constancy of f itself, not merely of |f| — although for holomorphic functions |f| constant already forces f constant.

The picture: analytic functions can't bulge

Think of the graph of |f| over the plane as a taut rubber surface. For a general smooth function you can push up a bump anywhere — a smooth interior peak is normal. Holomorphic functions forbid this. The real part u and imaginary part v of f are harmonic: they satisfy Laplace's equation Δu = 0, meaning the value at any point equals the average over any surrounding circle. A harmonic function is perfectly "balanced" — it cannot sit strictly above all its neighbors, because then it would exceed its own average.

|f| inherits a subharmonic version of this averaging: |f(z₀)| ≤ average of |f| on any small circle around z₀. If z₀ were a strict interior max, that average would be strictly less than |f(z₀)| — a contradiction. So the surface |f| behaves like a soap film stretched on a wire frame ∂Ω: its highest point is always on the wire. The only way to have a flat plateau at the top is for the whole film to be flat, i.e. f constant.

The key idea of the proof

The cleanest proof runs through the mean value property. By Cauchy's integral formula, for any disk of radius r inside Ω centered at z₀,

f(z₀) = (1/2π) ∫₀²π f(z₀ + r e) dθ.

Taking absolute values and using the triangle inequality for integrals,

|f(z₀)| ≤ (1/2π) ∫₀²π |f(z₀ + r e)| dθ ≤ max on the circle of |f|.

Now suppose |f(z₀)| is the maximum, M. Then the average of |f| on the circle is ≥ M, but every value on the circle is ≤ M, so the average is ≤ M. Both together force |f| ≡ M on the entire circle, for every small r — hence on a whole disk. A holomorphic function with constant modulus is constant (write f = M e; the Cauchy–Riemann equations kill φ). Finally, the identity theorem propagates constancy from that disk across the connected Ω. An alternative one-line proof: the open mapping theorem says f(Ω) is open, so f(z₀) can't be a maximum-modulus point unless the image is a single point.

A worked example and the canonical special case

Concrete example. Take f(z) = z² on the closed unit disk Ω̄ = {|z| ≤ 1}. Then |f(z)| = |z|², which is ≤ 1 everywhere and equals 1 exactly on the boundary circle |z| = 1. The maximum, 1, lives entirely on ∂Ω, precisely as promised — never at an interior point. The interior value at z = 0 is 0, the minimum, sitting at a zero of f.

Canonical special case — Schwarz lemma. The maximum modulus principle is the engine behind the Schwarz lemma: if f: 𝔻 → 𝔻 is holomorphic with f(0) = 0, then |f(z)| ≤ |z| and |f′(0)| ≤ 1. The trick: g(z) = f(z)/z is holomorphic (removable singularity at 0), and on |z| = r we have |g| ≤ 1/r. Letting r → 1 and applying the maximum modulus principle to g gives |g| ≤ 1 throughout 𝔻, i.e. |f(z)| ≤ |z|. Equality anywhere forces g constant, so f(z) = ez, a rotation. This lemma classifies the automorphisms of the disk.

Why the hypotheses matter

Holomorphy is indispensable. The result is false for merely smooth, or even real-analytic, functions of a real variable. Consider g(x, y) = 1 − (x² + y²) on ℝ²: it is smooth and has a strict interior maximum of 1 at the origin, with no drama. What breaks is the averaging property — an ordinary smooth function need not equal its circle-average.

  • Drop analyticity: f(z) = z̄ (complex conjugate) has |f| = |z|, still peaking on the boundary of a disk — but f(z) = 1 − |z|² peaks strictly at the interior origin. Non-holomorphic functions bulge freely.
  • Drop connectedness: on a disjoint union of two disks, f could be one constant on one disk and larger elsewhere; the local max on the first disk needn't force global constancy.
  • Drop boundedness (boundary form): f(z) = ez on the right half-plane is bounded by 1 on the imaginary axis (the boundary) yet blows up inside — the boundary form fails on unbounded domains without a growth condition, which is exactly what the Phragmén–Lindelöf theorems repair.

Applications and significance

The maximum modulus principle is one of the most-used tools in analysis because it converts an interior estimate into a boundary estimate — usually a huge simplification.

  • Schwarz lemma & disk automorphisms: as above, it pins down all holomorphic self-maps of the disk and underpins the Schwarz–Pick lemma and hyperbolic geometry.
  • Hadamard three-lines / three-circles theorems: refinements bounding |f| on intermediate lines/circles by a log-convex interpolation of boundary bounds — foundational in interpolation theory and in bounding the Riemann zeta function.
  • Uniqueness in PDE: because Re f is harmonic, the principle gives uniqueness for the Dirichlet problem — two harmonic functions with the same boundary data must coincide.
  • Operator theory & functional analysis: the von Neumann inequality and spectral estimates lean on maximum-modulus reasoning; it also drives H control theory.
  • Numerical estimates: to bound a holomorphic function on a region, one only needs to check the boundary — the workhorse behind countless a priori bounds.

Historically it crystallized in the work of Bernhard Riemann and was made precise by mathematicians including Carathéodory in the early 20th century, tightly linked to Schwarz (Schwarz lemma, 1869–1890s) and Hadamard (three-circles, 1896).

The maximum modulus principle and its close relatives — hypotheses and what each concludes.
ResultHypothesesConclusion
Maximum modulus principlef holomorphic on domain Ω; |f| has interior local maxf is constant on Ω
Boundary formΩ bounded, f holomorphic on Ω and continuous on Ω̄max_{Ω̄} |f| = max_{∂Ω} |f|
Minimum modulus principlef holomorphic, non-vanishing on Ω; |f| has interior local minf is constant (apply max principle to 1/f)
Open mapping theoremf holomorphic and non-constant on domain Ωf(Ω) is open — so f(z₀) is never a boundary point of the image
Schwarz lemmaf: 𝔻→𝔻 holomorphic, f(0)=0|f(z)| ≤ |z| and |f′(0)| ≤ 1, equality ⟹ rotation
Real-variable analogueu harmonic on Ω (real part of holomorphic)u has no interior max or min unless constant

Frequently asked questions

Does the maximum modulus principle say |f| has no interior maximum, ever?

It says a non-constant holomorphic function has no interior maximum of |f|. A constant function trivially attains its (equal) maximum everywhere, including the interior. So the honest statement is: an interior local maximum of |f| forces f to be constant. For any genuinely varying analytic function, the sup of |f| is only approached on (or as you head toward) the boundary.

Is there a corresponding minimum modulus principle?

Yes, but with an extra hypothesis: if f is holomorphic and non-vanishing on a domain Ω and |f| attains an interior local minimum, then f is constant. You prove it by applying the maximum modulus principle to 1/f, which is holomorphic precisely because f has no zeros. The non-vanishing hypothesis is essential — f(z) = z has a strict interior minimum of |f| at z = 0 (a zero), and is obviously non-constant.

Why does the proof need connectedness of Ω?

The mean value argument first shows f is constant on a disk around the maximum point z₀. To conclude f is constant on the whole domain you invoke the identity theorem, which says two holomorphic functions agreeing on a set with a limit point agree everywhere — but only on a connected set. On a disconnected open set, f could be constant on the component containing z₀ and do something completely different on another component.

What is the relationship to the open mapping theorem?

They are essentially two faces of the same rigidity. The open mapping theorem says a non-constant holomorphic f sends open sets to open sets. If |f(z₀)| were a maximum, then f(z₀) would be a point of the open image f(Ω) of largest modulus — but an open set in ℂ contains a small disk around each of its points, including points of strictly larger modulus. Contradiction. So maximum modulus is an immediate corollary of open mapping, and vice versa the max principle helps prove open mapping.

Does the boundary form require f to be continuous up to the boundary?

Yes — that continuity is what lets you actually evaluate |f| on ∂Ω and take a maximum there. Without it, sup over the interior can exceed anything achieved on the boundary. On unbounded domains you also need a growth condition: e^z on the right half-plane is bounded by 1 on the boundary imaginary axis yet unbounded inside, so the naive boundary form fails. Phragmén–Lindelöf theorems restore it under controlled growth.

How is this used to bound the Riemann zeta function?

Through the Hadamard three-lines theorem, a quantitative descendant of the maximum modulus principle. It states that for f holomorphic and bounded on a vertical strip a ≤ Re s ≤ b, the quantity log(sup of |f| on the line Re s = x) is a convex function of x. Applied to ζ(s), this interpolates growth bounds between the two edges of the critical strip (the convexity bound), a starting point for subthreshold and subconvexity estimates.