Alphabeta Math
Pipeline-generated
How statement and proof provenance work

The first chip identifies the source of the statement or construction; the second identifies the source of its local proof or verification.

  • Literature-sourced: the exact statement appears in a cited source; only wording and notation differ.
  • AI-adapted: a semantically identical restatement of literature-sourced material, modulo indexing, notation, and boundary cases adopted by the library.
  • AI-generated: a genuinely novel statement formulated by AI, with no source for the claim itself.

These labels describe origin, not correctness: citations and verification chips remain separate evidence.

✓ 13 results · all verified · 7 also independently AI-judged
Every result on this page is machine-checked by a proof checker and read in full and owner-audited; the judge is an additional, independent cross-model AI review of the proofs. The 6 not AI-judged were verified by owner audit (typically over a confirmed judge false positive), not failures.

Extremal Length and Planar Quasiconformality

1 · Prerequisites

2 · Summary

This page develops the extremal-length method in the plane and uses it to compare the geometric and analytic definitions of quasiconformality. Extremal length is the supremum over finite positive-area Borel densities of the squared family length divided by area, and the curve-family modulus is its extended reciprocal; Extremal length and the curve-family modulus of a path family fixes this convention before any computation, and The rho-length and the extremal length are well defined discharges the parameterization, ambient-domain and line-integral obligations of that definition.

Conformal invariance and monotonicity, the series law and the parallel law are proved in Conformal invariance, monotonicity, and the series and parallel laws for extremal length. The rectangle and round-annulus computations of Extremal length of the rectangle and of the round annulus then identify the joining-family value of a round annulus with its conformal parameter, and The conformal parameter of a round annulus is a complete invariant shows that this parameter is a complete invariant of finite round annuli while the punctured disc has infinite parameter and vanishing reciprocal modulus.

On the quasiconformal side, Orientation-preserving homeomorphisms and the geometric definition of quasiconformality defines orientation-preserving homeomorphisms whose quadrilateral moduli are distorted by at most K, and The ACL and Sobolev analytic definition of quasiconformality and The Beltrami coefficient and the maximal dilatation fix the analytic Wloc1,2 and Beltrami-coefficient formulations with the constant K=(1+k)/(1−k). The earlier Jordan boundary theorem Riemann maps of Jordan domains extend to homeomorphisms of the closures and the quadrilateral-only core Analytic quasiconformality gives both quadrilateral modulus bounds prove the two-sided geometric bounds through lower area and transverse reciprocity. The independent equivalence The geometric and analytic definitions of quasiconformality agree then supplies inverse regularity; The inverse of a quasiconformal map is quasiconformal with the same dilatation proves the signed-degree area formula and the inverse coefficient. The full wrapper An analytically quasiconformal homeomorphism distorts quadrilateral moduli by at most K retains arbitrary annular-end comparison and both null-set properties. Composition, conformality and normalized compactness follow without using the later general circular-dilatation branch, which uses the single explicitly authorized qualitative Gehring criterion alongside local sharp estimates.

The extremal-length items use Countable Choice only, through the length, measure and integration interfaces. The analytic quasiconformal items carry the Axiom of Choice inherited from the published ACL characterization of Wloc1,2, and each item states that assumption explicitly. The revised proofs are subject to the current owner review and stable certification; the later circular-dilatation branch identifies its single cited qualitative regularity interface and proves its sharp bounds and local quasisymmetry estimates.

3 · Logical flowchart

4 · Definitions, theorems and proofs

DefinitionDefinition: AI-adaptedProof: Not applicablejudge pass (gpt-6.1-sol)audited 2026-10-08Open item page →

The ACL and Sobolev analytic definition of quasiconformality

Sources

  • Mikhail Lyubich, Conformal Geometry and Dynamics of Quadratic Polynomials, vol. I, Ch. 2 §§11.3–11.5. Section 11.3 defines distributional partial derivatives and gives the ACL criterion for homeomorphisms of planar domains; §11.4 defines quasiconformality by local integrability of these derivatives and bounded dilatation; Proposition 2.11 proves local square-integrability for quasiconformal maps.
  • Christopher J. Bishop, Quasiconformal Mappings, Ch. 2 §1 and Ch. 3 §4, equation (4.1). The example there uses the Cantor singular function to show that the differential inequality alone does not imply ACL. Its printed derivative sentence has a typo: for f(x+iy)=x+g(x)+iy, one has ∂zf=1 and ∂zˉf=0 almost everywhere; these are the values used below.

Definition

Assume the Axiom of Choice (The Axiom of Choice). Let Ω,Ω′⊆C be complex domains (A complex domain is a nonempty connected open subset of C) and f:Ω→Ω′ a homeomorphism (Homeomorphism, open map, closed map, embedding, and what it means for a property to be topological). Read f as a map into R2; continuity makes its components locally square integrable. Recall that Wloc1,2(Ω) (Integer-order Sobolev spaces and their norms) means that the components of f admit weak first partial derivatives (Weak derivative of a locally integrable function) in Lloc2; equivalently, their almost-everywhere classes have ACL representatives in both coordinate directions whose measurable classical coordinate derivatives belong to Lloc2 (Absolute continuity on almost every coordinate line, The ACL characterisation of W1,p). ACL alone does not assert this derivative integrability. Countable Choice (The Axiom of Countable Choice (ACω)) is included in AC for the completed-product Fubini interfaces. On almost every coordinate line the continuous map f agrees almost everywhere with its ACL representative, hence everywhere by continuity of both restrictions; its classical line derivatives then represent its weak derivatives.

Let K≥1 and put k=(K−1)/(K+1)∈[0,1). The homeomorphism f is K-quasiconformal in the analytic sense when

(A1) f∈Wloc1,2(Ω); and

(A2) its weak Wirtinger derivatives ∂zf,∂zˉf (The Wirtinger derivatives ∂zf and ∂zˉf, and antiholomorphic functions) lie in Lloc2(Ω) and satisfy ∣∂zˉf(z)∣≤k∣∂zf(z)∣for almost every z∈Ω.

The inequality is a statement about Lloc2 classes: it is independent of the choice of representative of f and of the Borel representatives of its weak Wirtinger derivatives, since each such representative agrees almost everywhere with its class (Weak differentiation ignores null-set changes). The map is analytically quasiconformal if it is K-analytically quasiconformal for some finite K. Its minimal constant is recovered from the Beltrami coefficient defined later on this page.

Orientation and the regularity requirement. Inequality (A2) gives Jf=∣∂zf∣2−∣∂zˉf∣2≥(1−k2)∣∂zf∣2≥0 almost everywhere, with respect to planar Lebesgue measure (The nonnegative Lebesgue integral). A classical derivative inequality alone does not replace (A1): on (0,1)2, the map f(x+iy)=x+g(x)+iy, where g is the Cantor function, is a homeomorphism onto (0,2)×(0,1) (The map x↦x+c(x) is a homeomorphism from [0,1] onto [0,2]). The Cantor function is continuous, nonconstant, and locally constant off the null Cantor set, so g′=0 almost everywhere (The Cantor function is well defined, satisfies c(x)≤c(y) whenever x≤y, is surjective onto [0,1], and is constant on every interval removed from the Cantor set, The Cantor set is an uncountable subset of R of Lebesgue measure zero). Thus the classical Wirtinger derivatives are ∂zf=1 and ∂zˉf=0 almost everywhere. But g cannot be absolutely continuous on every compact subinterval of (0,1): the fundamental theorem would make it constant there, and then continuity would contradict g(0)=0, g(1)=1 (Fundamental theorem of calculus for absolutely continuous functions). Subtracting the identity shows that f fails ACL on every horizontal line and therefore fails (A1). Its classical inequality is not the weak-derivative assertion (A2). Orientation preservation follows from (A1) together with (A2), as proved by the equivalence theorem on this page.

DefinitionDefinition: Literature-sourcedProof: Not applicablejudge pass (gpt-6.1-sol)audited 2026-10-08Open item page →

Extremal length and the curve-family modulus of a path family

Sources

  • Christopher J. Bishop, Quasiconformal Mappings, Ch. 1 §1, printed pp. 1–3. Bishop defines admissibility by ℓρ(Γ)≥1, modulus by the infimum of ∫ρ2, and extremal length as the reciprocal modulus; he also states that the density may be taken Borel.
  • Mikhail Lyubich, Conformal Geometry and Dynamics of Quadratic Polynomials, vol. I, Ch. 1 §6.1, printed pp. 119–120. Lyubich defines Lρ(Γ)=ℓρ(Γ)2/mρ(U) and takes its supremum over finite-mass metrics; the reciprocal is extremal width and is also the infimum over metrics whose length on every curve is at least 1.
  • Lars Ahlfors and Arne Beurling, Conformal Invariants and Function-Theoretic Null-Sets, §4, printed pp. 114–115. Their supremum convention for extremal length agrees with λ below. Their Lemmas 4–5 give the rectangle and round-annulus constants after the curve family is specified.

Definition

Assume Countable Choice (The Axiom of Countable Choice (ACω)). Fix a complex domain Ω⊆C (A complex domain is a nonempty connected open subset of C), read as an open subset of the Euclidean plane, with its Borel σ-algebra (The Borel sigma-algebra of a topological space) and planar Lebesgue area measure (Lebesgue measure is a Radon measure on R^n).

A path in Ω is a continuous map γ:[a,b]→Ω. It is rectifiable when its two coordinate functions have bounded variation (A path in Rn is rectifiable exactly when every coordinate has bounded variation), and sγ denotes its arc-length function (The arc-length function sγ(t)=L(γ∣[a,t]) of a rectifiable path). A path family Γ is a set of paths. When curves joining specified boundary sets of Ω are used, allow paths γ:[a,b]→Ω‾ with γ((a,b))⊆Ω and endpoints in the named sets; read complex paths as planar rectifiable paths (Complex contours as planar rectifiable paths: the Euclidean, coordinate-BV, and piecewise-C1 dictionaries). Only the interior contributes to length, since each density below is extended by 0 outside Ω.

Let ρ:Ω→[0,+∞] be Borel measurable with respect to the Borel σ-algebra of Ω (Extended-real-valued measurable functions). Extending it by 0 on C∖Ω gives a Borel function on the plane by the trace identity for subspaces (The Borel sigma-algebra of a subspace is the trace of the ambient Borel sigma-algebra). For a rectifiable path γ:[a,b]→Ω‾, its arc-length function is continuous and nondecreasing (The arc-length function is continuous and nondecreasing, with increments equal to subpath lengths; it is strictly increasing exactly when no nondegenerate subpath is constant). Extend sγ to a continuous nondecreasing function Sγ:R→R by Sγ(t)=0 for t≤a, Sγ(t)=sγ(t) for a≤t≤b, and Sγ(t)=sγ(b) for t≥b. Countable Choice gives the associated Lebesgue-Stieltjes Borel measure dSγ (The Axiom of Countable Choice (ACω), Assuming countable choice, a nondecreasing right-continuous function defines a Borel measure on R). Define the ρ-length by ℓρ(γ):=∫[a,b]ρ(γ(t)) dSγ(t), using the nonnegative Lebesgue integral (The nonnegative Lebesgue integral). This value lies in [0,+∞]; set ℓρ(γ)=+∞ when γ is not rectifiable. For finite-valued continuous ρ and a rectifiable path whose full trace lies in Ω, this agrees with the published absolute line integral ∫γρ ∣dz∣ (The absolute line integral over a rectifiable path using its arc-length function, For a continuous integrand, the Riemann-Stieltjes and Lebesgue-Stieltjes integrals agree, Continuous integrands have complex and absolute line integrals along every rectifiable path). Continuity only on Ω does not assert that the zero extension is continuous at boundary endpoints. The continuity of Sγ makes dSγ atomless, so changing the integrand on finitely many parameter values does not change the length.

For a path family put ℓρ(Γ):=inf⁡γ∈Γℓρ(γ), with inf⁡∅:=+∞, and define its area by A(ρ):=∫Ωρ2 dA∈[0,+∞].

The extremal length of Γ is λ(Γ):=sup⁡ρℓρ(Γ)2A(ρ)∈[0,+∞], where the supremum is over Borel ρ:Ω→[0,+∞] with 0<A(ρ)<+∞. A supremum of an empty set is 0; since Ω is a nonempty open domain, it contains a nondegenerate rectangle Q, and its indicator has positive finite area (A box in Rn with parameters ai≤bi is Lebesgue measurable of measure ∏i<n(bi−ai), whichever of its faces are included). The value ℓρ(Γ)=+∞ is allowed. The curve-family modulus is the reciprocal μ(Γ):=1λ(Γ)∈[0,+∞], with 1/0:=+∞ and 1/+∞:=0. Thus λ(∅)=+∞ and μ(∅)=0, while any family containing a constant path has λ(Γ)=0 and μ(Γ)=+∞.

Conventions. (1) This library defines extremal length by the displayed supremum and calls its reciprocal the modulus. Sources that define modulus by inf⁡ρ∫ρ2 over metrics with ℓρ(Γ)≥1 call μ(Γ) the modulus and λ(Γ) the extremal length. (2) Extending ρ by zero outside Ω makes densities supported in Ω admissible and makes the value independent of an ambient enlargement; parameterization invariance, this ambient-domain independence, and agreement with the continuous line integral are the well-definedness obligations recorded for The rho-length and the extremal length are well defined ↗. (3) The page's rectangle and round-annulus constants are fixed by the computations stated there, and each cited source is translated into this library convention.

DefinitionDefinition: AI-adaptedProof: Not applicablejudge pass (gpt-6.1-sol)audited 2026-10-08Open item page →

The Beltrami coefficient and the maximal dilatation

Sources

  • Christopher J. Bishop, Quasiconformal Mappings, Ch. 2 §1, printed pp. 49–51. For an orientation-preserving nonsingular real-linear map, Bishop obtains the complex dilatation μ=fzˉ/fz, ∣μ∣<1, and the dilatation D=(1+∣μ∣)/(1−∣μ∣), equivalently ∣μ∣=(D−1)/(D+1).
  • Mikhail Lyubich, Conformal Geometry and Dynamics of Quadratic Polynomials, vol. I, Ch. 2 §§11.1.2–11.3, printed pp. 177–181. Section 11.1.2 defines the pointwise Beltrami coefficient; equation (11.3) on p. 178 gives Dil⁡=(1+∣μ∣)/(1−∣μ∣); §11.3 adds the ACL/distributional regularity and bounded-dilatation requirements. The last K-conversion on p. 181 prints denominator k−1; this is a sign typo, so the correct conversion from Bishop and the library definition is used here.

Definition

Assume the Axiom of Choice. Let f:Ω→Ω′ be a homeomorphism of complex domains whose components lie in Wloc1,2(Ω), with weak Wirtinger derivative classes as in The ACL and Sobolev analytic definition of quasiconformality, The Wirtinger derivatives ∂zf and ∂zˉf, and antiholomorphic functions, and Weak derivative of a locally integrable function.

Choose finite Borel representatives of the real and imaginary components of these derivative classes. Such representatives exist: extend each component by zero outside Ω, use that Lebesgue measure is the completion of its Borel restriction and that every completion-measurable function equals a Borel function almost everywhere (Borel measurable and Lebesgue measurable functions on Rn, L(Rn) is exactly the completion of the restriction of λn to the Borel sets, A function measurable for a completion is almost everywhere equal to one measurable for the original sigma-algebra), then restrict the resulting Borel representatives to Ω using the Borel trace identity (The Borel sigma-algebra of a topological space, The Borel sigma-algebra of a subspace is the trace of the ambient Borel sigma-algebra). Replacing infinite values on the resulting Borel null sets by zero gives finite representatives.

The Beltrami coefficient of f is the Borel function μf(z):={∂zˉf(z)/∂zf(z),∂zf(z)≠0,0,∂zf(z)=0. It is a finite-valued measurable function modulo equality almost everywhere. Changing the chosen Borel representatives changes μf only on a Lebesgue-null set, so this class is independent of those choices. For a general Wloc1,2 homeomorphism, μf need not be essentially bounded. If f is K-analytically quasiconformal and k=(K−1)/(K+1), then ∣μf∣≤k<1 almost everywhere and it defines a complex L∞(Ω) class (Complex Lp classes and Euclidean test-function conventions). The value zero on {∂zf=0} is a fixed convention; analytic quasiconformality gives ∂zˉf=0 almost everywhere on that set.

Let mf=∥μf∥∞ be the essential supremum of ∣μf∣ (The essential supremum of a measurable function with respect to a measure). Define the maximal dilatation by Kf:={1+mf1−mf,mf<1 and ∂zˉf=0 a.e. on {∂zf=0},+∞,otherwise. Then Kf∈[1,+∞], and for every finite K≥1, f is analytically K-quasiconformal exactly when Kf≤K. Indeed, off {∂zf=0} the analytic inequality is equivalent to ∣μf∣≤(K−1)/(K+1), and on that set the separate derivative condition in the definition of Kf is exactly what makes the inequality hold. For finite mf, intersecting the almost-everywhere bounds ∣μf∣≤mf+1/n gives ∣μf∣≤mf almost everywhere. For an analytically quasiconformal map, the separate derivative condition holds and mf<1, so Kf=(1+mf)/(1−mf) is the least admissible constant. In this class, Kf=1 exactly when μf=0 almost everywhere, equivalently when ∂zˉf=0 as an Lloc2 class. The separate one-quasiconformal theorem on this page supplies the holomorphic conclusion in that case; it is not an input to this definition. The nullity of {∂zf=0} will follow from the inverse theorem's area formula and null-set properties; it is not an assumption here.

LemmaStatement: AI-adaptedProof: AI-adaptedprecheck passjudge pass (gpt-6.1-sol)audited 2026-10-08Open item page →

The rho-length and the extremal length are well defined

Sources

  • Christopher J. Bishop, Quasiconformal Mappings, Ch. 1 §1, printed pp. 1–8. The notes define curve length by integrating a nonnegative Borel density against arclength and observe that densities may be set to zero outside the domain.
  • Mikhail Lyubich, Conformal Geometry and Dynamics of Quadratic Polynomials, vol. I, Ch. 1 §6.1 and Exercise 6.2, printed pp. 119–120. The exercise records ambient-surface independence; the arguments below supply the measure-theoretic details for the library's finite-positive-area convention, including its zero-area edge case.

Statement

Assume Countable Choice. Let Ω⊆Ω′ be complex domains, let Γ be a path family in Ω, and let ρ,σ:Ω→[0,+∞] be Borel with ρ≤σ. Let ℓρ, A, λ, and μ be as in Extremal length and the curve-family modulus of a path family. Then:

(i) Parameterization independence. If γ:[a,b]→Ω is rectifiable and h:[a′,b′]→[a,b] is continuous, nondecreasing, and onto, then ℓρ(γ∘h)=ℓρ(γ). With the arc-length parametrization γ~:[0,L]→Ω of Every rectifiable path factors through its arc-length function as a unit-speed path on [0,L], one has ℓρ(γ)=∫0Lρ(γ~(s)) ds.

(ii) Subpath additivity. Write νγ=dSγ for the Lebesgue-Stieltjes measure of the extended arc-length function in Extremal length and the curve-family modulus of a path family. For a≤u≤v≤b, ℓρ(γ∣[u,v])=∫[u,v]ρ(γ(t)) dνγ(t). Consequently, if [uj,vj]⊆[a,b] are pairwise disjoint parameter intervals, then ∑jℓρ(γ∣[uj,vj])≤ℓρ(γ).

(iii) Agreement with the absolute line integral. If ρ is finite-valued and continuous on Ω, then for every rectifiable path γ in Ω the value ℓρ(γ) equals the published absolute line integral ∫γρ ∣dz∣ (The absolute line integral over a rectifiable path using its arc-length function, Continuous integrands have complex and absolute line integrals along every rectifiable path). In particular, ℓ1(γ) is the arc length of γ.

(iv) Monotonicity and area additivity. For every path, ℓρ(γ)≤ℓσ(γ), hence ℓρ(Γ)≤ℓσ(Γ). If pairwise disjoint Borel sets E1,…,Ek⊆Ω satisfy ρ=0 off their union, then A(ρ)=∑j=1k∫Ejρ2 dA.

(v) Independence of the ambient domain. Extending ρ by zero to Ω′ does not change the ρ-length of any path in Ω or its area. Consequently, λ(Γ) and μ(Γ) computed using metrics on Ω equal those computed using metrics on Ω′.

(vi) Nondegeneracy and scaling. A(ρ)=0 if and only if ρ=0 almost everywhere. If 0<A(ρ)<+∞, then for every real scalar c>0 the quotient ℓcρ(Γ)2/A(cρ) equals ℓρ(Γ)2/A(ρ). This includes ℓρ(Γ)=0 and ℓρ(Γ)=+∞; the denominator is always finite and positive.

Facts & Assumptions

Given: Countable Choice, the domains and path family in the Statement, Borel densities ρ≤σ, and the definitions of the path length, area, extremal length, and modulus.

[F1]

For a rectifiable path γ:[a,b]→Ω, its arc-length function sγ is continuous nondecreasing with sγ(a)=0, sγ(b)=L(γ), and sγ(v)−sγ(u)=L[u,v](γ) (The arc-length function sγ(t)=L(γ∣[a,t]) of a rectifiable path, The arc-length function is continuous and nondecreasing, with increments equal to subpath lengths; it is strictly increasing exactly when no nondegenerate subpath is constant).

[F2]

Extending sγ constantly to the left and right of [a,b] gives a nondecreasing right-continuous real function Sγ; Countable Choice supplies its finite-on-compact Borel Lebesgue-Stieltjes measure νγ with νγ((u,v])=Sγ(v)−Sγ(u) (The Axiom of Countable Choice (ACω), Assuming countable choice, a nondecreasing right-continuous function defines a Borel measure on R).

[F3]

The interval formulas give νγ([u,v])=Sγ(v)−Sγ(u−) and νγ({u})=Sγ(u)−Sγ(u−). Hence continuity of Sγ makes νγ atomless (Interval formulas and atoms for a Lebesgue-Stieltjes measure).

[F4]

Every rectifiable path has a unique arc-length factorization γ=γ~∘sγ with γ~:[0,L]→Ω (Every rectifiable path factors through its arc-length function as a unit-speed path on [0,L]).

[F5]

Arc length is unchanged by a continuous surjective nondecreasing reparameterization, including pauses; applying this result to every restriction of γ also gives sγ∘h=sγ∘h (Arc length is invariant under every continuous surjective monotone reparametrization, including pauses and reversal).

[F6]

Two Borel measures on R that are finite on compact sets and agree on all half-open intervals (u,v] agree on every Borel set (The interval data on (a,b] determines the Borel measure uniquely). Lebesgue measure assigns (u,v] the length v−u (A box in Rn with parameters ai≤bi is Lebesgue measurable of measure ∏i<n(bi−ai), whichever of its faces are included).

[F7]

Every nonnegative measurable function has an increasing simple approximation, and increasing limits pass through the nonnegative integral (Every nonnegative measurable function admits an explicit increasing sequence of simple approximations, Monotone convergence for the integral).

[F8]

For nonnegative measurable functions, the integral is monotone and positively homogeneous; it is additive on finite sums (Monotonicity and nonnegative homogeneity of the nonnegative integral, Additivity of the nonnegative Lebesgue integral). For measurable E, ∫Ef=∫f1E (Integral over a measurable subset).

[F9]

A nonnegative measurable function has integral zero exactly when it vanishes almost everywhere (A nonnegative measurable function has integral 0 exactly when it vanishes almost everywhere).

[F10]

For a finite continuous real integrand g and nondecreasing right-continuous integrator F, the Riemann-Stieltjes and Lebesgue-Stieltjes integrals agree (For a continuous integrand, the Riemann-Stieltjes and Lebesgue-Stieltjes integrals agree). The absolute complex line integral is defined using the Riemann-Stieltjes integral against sγ (The absolute line integral over a rectifiable path using its arc-length function).

Proof

technique · push forward the arc-length measure, then use monotonicity and finite additivity
1.1F1F2F3F6constructalgebra

Fix a rectifiable γ:[a,b]→Ω, let s=sγ and L=L(γ), and let νγ be the measure of [F2]. Define a finite Borel measure η on R by η(B):=νγ({t∈[a,b]:s(t)∈B}); it is a measure because inverse images preserve Borel sets, disjointness, and countable unions. For t∈[0,L], let ut be the largest point in the nonempty compact set {u∈[a,b]:s(u)≤t}. Continuity and surjectivity of s give s(ut)=t and {s≤t}=[a,ut]. Therefore, for 0≤r<q≤L, s−1((r,q])∩[a,b]=(ur,uq] and [F2] gives η((r,q])=S(uq)−S(ur)=q−r. The measure η is supported on [0,L]. Each level set s−1({t}) is a closed interval or a singleton; on a nondegenerate level interval [vt,ut], [F3] gives νγ([vt,ut])=S(ut)−S(vt−)=t−t=0, and on a singleton [F3] gives zero mass as well. Thus η has no endpoint atoms. Clipping each half-open interval to [0,L] shows that its mass agrees with Lebesgue measure restricted to [0,L]; [F6] gives η=λ∣[0,L].

1.2F1F2F3F6F7F8given

Fix a≤u≤v≤b. By [F1], the arc-length function of γ∣[u,v] is sγ(t)−sγ(u) on [u,v], extended constantly outside that interval. Its associated Lebesgue-Stieltjes measure agrees with νγ restricted to [u,v]: both are finite Borel measures supported there and have the same values on every half-open subinterval by [F2] and the interval formulas [F3], so [F6] identifies them. The definition of ℓρ then gives the displayed restriction formula in (ii). For finitely many subintervals with disjoint interiors, their common endpoints have zero νγ-measure by [F3], and additivity of the nonnegative integral [F8] bounds the sum of their path lengths by the integral over [a,b]. Taking increasing finite partial sums gives the same bound for a countable family by [F7], proving (ii).

1.3F1F3F10algebra

If ρ is finite-valued and continuous, then g=ρ∘γ is a continuous real function. By [F3], νγ has no atoms, so its mass at the endpoints is zero. The continuous-integrand comparison in [F10] identifies ℓρ(γ)=∫(a,b]g dνγ with the Riemann-Stieltjes absolute line integral, proving the first assertion of (iii). When ρ=1, every Riemann-Stieltjes sum against sγ telescopes to sγ(b)−sγ(a)=L(γ), proving the final assertion.

1.4F8givenalgebra

If ρ≤σ, [F8] gives ℓρ(γ)≤ℓσ(γ) for each path, and taking infima over Γ preserves the inequality. For pairwise disjoint Borel E1,…,Ek with ρ=0 off their union, ρ2=∑j=1kρ21Ej pointwise, so finite additivity and the definition of the restricted integral in [F8] give the area sum in (iv).

1.5F8F9givenalgebra

Apply [F9] to ρ2 to obtain A(ρ)=0 iff ρ2=0 almost everywhere, which is equivalent to ρ=0 almost everywhere. For c>0, [F8] gives ℓcρ(γ)=cℓρ(γ) for each path and hence ℓcρ(Γ)=cℓρ(Γ); it also gives A(cρ)=c2A(ρ). For 0<A(ρ)<+∞ and real c>0, the area A(cρ)=c2A(ρ) remains finite and positive. If the path-family length is finite, cancellation of c2 proves quotient invariance, including length zero. If the length is +∞, both quotients are +∞ because their denominators are finite and positive. This proves (vi) on the full admissible-density domain without an undefined +∞/+∞ expression.

2.1F4F7step 1.1givenalgebra

For an indicator g=1B, the definition of η gives ∫[a,b]g(s(t)) dνγ(t)=η(B)=∫[0,L]g ds. Finite sums give the same identity for nonnegative simple g, and [F7] extends it to every nonnegative Borel g by increasing simple approximation and monotone convergence. Taking g=ρ∘γ~ and using γ=γ~∘s from [F4] yields ℓρ(γ)=∫0Lρ(γ~(s)) ds.

2.2F8F9F11step 1.4constructalgebra

Extending ρ by zero from Ω to Ω′ preserves each path integral for paths in Ω and preserves area, because the new integrand is zero off Ω. Extension therefore gives λΩ′(Γ)≥λΩ(Γ). Conversely, take any ρ′:Ω′→[0,+∞] with 0<AΩ′(ρ′)<∞, and put ρ=ρ′∣Ω. Its path-length infimum on Γ is unchanged and AΩ(ρ)≤AΩ′(ρ′). If AΩ(ρ)>0, this restriction is admissible and its quotient is at least the quotient from Ω′. If AΩ(ρ)=0, [F9] gives ρ=0 almost everywhere. When ℓρ(Γ)=0, the quotient from Ω′ is zero. When ℓρ(Γ)>0, choose the rectangle Q of [F11] and, for ε>0, set ρε=ρ+ε1Q on Ω. Then AΩ(ρε)=ε2∣Q∣ and ℓρε(Γ)≥ℓρ(Γ) by [F8], so the quotients on Ω are unbounded as ε↓0. Thus every admissible quotient on Ω′ is at most λΩ(Γ), proving equality of extremal lengths; their reciprocals agree as well.

3.1F4F5step 2.1given

Let h:[a′,b′]→[a,b] be continuous, nondecreasing and onto. By [F5], the total lengths of γ∘h and γ agree, and applying [F5] to each restriction gives sγ∘h=sγ∘h. Thus γ∘h=γ~∘sγ∘h with the same unique arc-length parametrization γ~ as in [F4]. The formula of step 2.1 applied to both paths proves ℓρ(γ∘h)=ℓρ(γ), establishing (i).

4.1step 1.1step 2.1step 3.1step 1.2step 1.3step 1.4step 2.2step 1.5∎

Steps 1.1, 2.1, and 3.1 prove parameterization independence and the arc-length formula in (i); step 1.2 proves (ii), step 1.3 proves (iii), step 1.4 proves (iv), step 2.2 proves (v), and step 1.5 proves (vi).

DefinitionDefinition: AI-adaptedProof: Not applicablejudge pass (gpt-6.1-sol)audited 2026-10-08Open item page →

Orientation-preserving homeomorphisms and the geometric definition of quasiconformality

Sources

  • Christopher J. Bishop, Quasiconformal Mappings, Ch. 2 §2, printed pp. 51–53. Bishop defines a quadrilateral as a Jordan domain with two disjoint closed boundary arcs marked, assigns its modulus by a conformal rectangle, and defines geometric quasiconformality by the two-sided modulus bound for every quadrilateral.
  • Mikhail Lyubich, Conformal Geometry and Dynamics of Quadratic Polynomials, vol. I, Ch. 2 §12.5, printed p. 188. QC2 states that moduli of quadrilaterals and annuli are K-quasi-invariant for an orientation-preserving homeomorphism.
  • Lars Ahlfors and Arne Beurling, Conformal Invariants and Function-Theoretic Null-Sets, §§4–5, printed pp. 114–120, for the extremal-length convention and its conformal invariance.

Definition

Let Ω,Ω′⊆C be complex domains and f:Ω→Ω′ a homeomorphism (A complex domain is a nonempty connected open subset of C, Homeomorphism, open map, closed map, embedding, and what it means for a property to be topological).

Orientation. For a∈Ω, f induces an isomorphism f∗:H2(Ω,Ω∖{a};Z)⟶H2(Ω′,Ω′∖{f(a)};Z). Excision and the local-homology calculation identify both groups with Z (Excision for singular homology, Local homology detects manifold dimension, interior, and boundary). The generators are those determined by the standard orientation of C≅R2, with positively oriented basis (1,i) (Orientation of a finite-dimensional real vector space, R-orientation of a topological manifold); the restriction isomorphisms from coordinate balls to points are supplied by Coordinate-ball classes identify local homology stalks. Write εf(a)∈{+1,−1} for the multiplier of f∗. Naturality of relative homology (Functoriality of relative homology) makes these maps compatible with coordinate-ball restrictions. In the local-system charts of R-orientation of a topological manifold, their multiplier is locally constant. Since Ω is connected, εf is constant. The map f is orientation-preserving when this sign is +1, and orientation-reversing when it is −1.

A quadrilateral. A quadrilateral in Ω is a set Q=φ(Π‾), where Π=(0,w)×(0,h) with w,h>0 and φ:Π‾→C is continuous and injective. Its two marked sides are either φ({0}×[0,h]) and φ({w}×[0,h]), or the other pair of opposite sides. For one such choice, let Γ(Q) be the family of continuous paths in Q whose endpoints lie on different marked sides and whose interior lies in int⁡Q. This is a curve family in Ω; its modulus μ(Γ(Q)) is defined in Extremal length and the curve-family modulus of a path family. The image f(Q) is again a quadrilateral with marked sides carried by f, and fΓ(Q)=Γ(f(Q)).

Geometric definition. Let K≥1. The homeomorphism f is K-geometrically quasiconformal when it is orientation-preserving and, for every quadrilateral Q with Q‾⊆Ω and for each choice of its marked sides, 1K μ(Γ(Q))≤μ(Γ(f(Q)))≤K μ(Γ(Q)), using the conventions of Extremal length and the curve-family modulus of a path family for zero and infinite values. It is geometrically quasiconformal if it is K-geometrically quasiconformal for some finite K; the least such K is Kf≥1, its maximal dilatation.

The modulus condition uses the extremal-length construction and its Countable-Choice hypothesis; the orientation sign itself uses no choice (The Axiom of Countable Choice (ACω)). The inverse of a K-geometrically quasiconformal map is also K-geometrically quasiconformal: its local-homology map is the inverse isomorphism, and the two modulus inequalities rearrange to the same bounds for f−1. Thus Kf−1=Kf. The local orientation clause and the modulus inequality are distinct parts of this definition.

TheoremStatement: Literature-sourcedProof: AI-adaptedprecheck passjudge pass (gpt-6.1-sol)audited 2026-10-08Open item page →

Conformal invariance, monotonicity, and the series and parallel laws for extremal length

Sources

  • Christopher J. Bishop, Quasiconformal Mappings, Ch. 1 §1, printed pp. 2–4. Lemma 1.1 proves conformal invariance by transforming lengths and areas and then applying the inverse map; Lemma 1.2 proves overflow monotonicity; Lemma 1.3 and Corollary 1.4 give the disjoint-support modulus addition rule; Lemma 1.5 gives the series inequality.
  • Mikhail Lyubich, Conformal Geometry and Dynamics of Quadratic Polynomials, vol. I, Ch. 1 §§6.1–6.2, printed pp. 119–121. The definition identifies extremal width with the infimum of area over metrics whose length on every curve is at least one. The text then proves conformal invariance, the series law by normalizing two metrics to have equal length, area and extremal length, and the parallel law by restriction to disjoint supporting sets.
  • Lars Ahlfors and Arne Beurling, Conformal Invariants and Function-Theoretic Null-Sets, §4, printed p. 115. Lemmas 1–3 state overflow monotonicity, the series inequality, and the parallel harmonic-sum law. The arguments below retain the extended-value and finite-positive-area conventions of this library.

Statement

Assume Countable Choice, and let λ and μ be the extremal length and curve-family modulus of Extremal length and the curve-family modulus of a path family. Let Ω,Ω′⊆C be complex domains. A path family in a domain consists of paths whose full traces lie in that domain.

(a) Conformal invariance. If f:Ω→Ω′ is a biholomorphism (Biholomorphic maps between complex domains) and Γ is a path family in Ω, then fΓ:={f∘γ:γ∈Γ} satisfies λ(fΓ)=λ(Γ) and μ(fΓ)=μ(Γ). Thus these parameters are invariants of conformal equivalence (Conformal equivalence and the automorphism group of a domain).

(b) Monotonicity and overflow. If Γ⊆Γ′, then λ(Γ′)≤λ(Γ). If every γ∈Γ0 contains a subpath belonging to Γ1, then λ(Γ0)≥λ(Γ1) and μ(Γ0)≤μ(Γ1).

(c) Series law. Let Γ0,Γ1,Γ be path families in Ω. Suppose there are disjoint Borel sets E0,E1⊆Ω such that every path of Γj has trace in Ej, and every path in Γ has restrictions to two disjoint closed parameter intervals that belong respectively to Γ0 and Γ1. Then λ(Γ)≥λ(Γ0)+λ(Γ1).

(d) Parallel law (Grötzsch). If Γ0,Γ1 are path families in Ω and their traces lie respectively in disjoint Borel sets E0,E1⊆Ω, then μ(Γ0∪Γ1)=μ(Γ0)+μ(Γ1). Equivalently, λ(Γ0∪Γ1) is the harmonic sum of λ(Γ0) and λ(Γ1), where the harmonic sum of x,y∈[0,+∞] means 1/(1/x+1/y) with the reciprocal conventions of the definition.

All assertions use the extended-real conventions of Extremal length and the curve-family modulus of a path family, in particular λ(∅)=+∞, μ(∅)=0, and addition of +∞ to a nonnegative value gives +∞.

Facts & Assumptions

Given: Countable Choice, complex domains Ω,Ω′, the path families and the biholomorphism in the Statement.

[F1]

The path metric length is parameterization independent, is additive on disjoint subpath intervals, is monotone in the density, and has area additivity on disjoint Borel supports. The area-zero and density-scaling cases are also part of the well-definedness result (The rho-length and the extremal length are well defined).

[F3]

A C1 diffeomorphism satisfies the change-of-variables identity for every nonnegative Borel density, with equality of extended integrals and under Countable Choice (Borel change of variables from the compact-support formula and Radon uniqueness).

[F5]

Arc length is additive over adjacent parameter intervals; a Lipschitz map multiplies length by at most its Lipschitz constant, while a similarity multiplies length by its absolute scale (Arc length is additive across every subdivision point and decreases under restriction, A C-Lipschitz map multiplies path length by at most C; isometries preserve length and scalar dilation multiplies it by the absolute scale).

[F6]

The interval data (u,v]↦v−u determines Lebesgue measure uniquely among Borel measures finite on compact sets (The interval data on (a,b] determines the Borel measure uniquely).

[F8]

The supremum and reciprocal definitions of λ and μ include empty families, constant paths, zero and infinite values, and use Borel densities with finite positive area (Extremal length and the curve-family modulus of a path family).

Proof

technique · transport path metrics, then establish the two circuit laws from the metric definitions
1.1F2F5givenconstruct

Fix a rectifiable γ:[a,b]→Ω with positive length L, let α:[0,L]→Ω be its unit-speed arc-length parametrization, and put w(s)=∣f′(α(s))∣. By [F2], w is continuous and strictly positive. Define T(s) to be the arc length of f∘α on [0,s]. It is finite: locally on a convex disk around a point of the compact trace of α, boundedness of f′ makes f Lipschitz by the real one-variable mean-value theorem; finitely many such disks and a subdivision of [0,L] then give finite length. Arc-length additivity makes T(v)−T(u) the length of f∘α on [u,v].

1.2F2F5givenalgebra

Fix s0∈(0,L) and ε>0. Choose a convex disk D about α(s0) on which ∣f′(z)−f′(α(s0))∣<ε. On D, the map g(z)=f(z)−f′(α(s0))z is ε-Lipschitz by the real mean-value estimate applied along line segments. For u<v sufficiently close to s0, α([u,v])⊂D; polygonal sums and [F5] then give ∣T(v)−T(u)−w(s0)(v−u)∣≤ε(v−u), since α has length v−u on [u,v]. Hence T′(s0)=w(s0); as s0 varies this derivative is continuous.

1.3F2F5givenalgebra

On every interval [u,v]⊆[0,L], apply the real mean-value theorem to T on each interval of a partition. The resulting sums for T(v)−T(u) are Riemann sums for the continuous function w, so refinement gives T(v)−T(u)=∫uvw(s) ds. The positive continuous function w has a positive minimum on [0,L], so T is strictly increasing. Its range is [0,T(L)], and the arc-length parametrization of f∘γ is β=f∘α∘T−1.

1.4F1F8given

If Γ⊆Γ′, then for each Borel density ρ, ℓρ(Γ′)≤ℓρ(Γ), so each quotient for Γ′ is at most the corresponding quotient for Γ; taking suprema gives λ(Γ′)≤λ(Γ). If every path of Γ0 contains a subpath in Γ1, [F1] gives ℓρ(Γ0)≥ℓρ(Γ1) for every ρ, hence λ(Γ0)≥λ(Γ1). Reciprocal order gives μ(Γ0)≤μ(Γ1), including zero and infinite values.

1.5F1F7F8givenconstructalgebra

For a path family Γ, call a Borel density σ width-admissible when ℓσ(Γ)≥1, and let W(Γ) be the infimum of A(σ) over such densities. Then W(Γ)=μ(Γ). Indeed, if A(ρ)∈(0,∞) and 0<ℓρ(Γ)<∞, scaling by 1/ℓρ(Γ) gives a width-admissible density with area the reciprocal of its extremal-length quotient; if ℓρ(Γ)=+∞, arbitrary positive rescalings have areas tending to zero. Taking the supremum over ρ proves W(Γ)≤μ(Γ). Conversely, any width-admissible σ with finite positive area gives λ(Γ)≥1/A(σ), so μ(Γ)≤A(σ). If A(σ)=0, [F1] makes σ=0 almost everywhere; adding ε1Q for the rectangle in [F7] preserves width-admissibility and has area ε2∣Q∣, forcing λ(Γ)=+∞ and again μ(Γ)≤W(Γ). If λ(Γ)=0, no width-admissible density can have finite area by these same implications; thus W(Γ)=+∞=μ(Γ). These cases establish the claimed equality with all extended values.

2.1F1F4F6step 1.3given

For a nonnegative Borel function q on [0,T(L)], define ν(B)=∫T−1(B)w(s) ds. This is a finite Borel measure by [F4]. For 0≤u<v≤L, step 1.3 gives ν((T(u),T(v)])=T(v)−T(u); endpoint singletons have zero measure because w is bounded. The measure is supported on [0,T(L)]; clipping any half-open interval to this range and using the interval identity and the zero endpoint atoms shows that ν agrees with Lebesgue measure restricted to [0,T(L)] on every half-open interval in R. Hence [F6] gives equality of these two restricted measures on Borel sets. Indicators, simple functions and increasing simple approximations now yield ∫0T(L)q(t) dt=∫0Lq(T(s))w(s) ds. Taking q=ρ′∘β and using [F1] proves for every Borel ρ′:Ω′→[0,+∞] that ℓρ′(f∘γ)=ℓ(ρ′∘f)∣f′∣(γ).

2.2F1F7F8step 1.4step 1.5given

For the series law, if either λ(Γj)=0, overflow in step 1.4 gives the desired lower bound by the other term. If either value is +∞, the same overflow gives λ(Γ)=+∞. It remains to consider 0<λ(Γ0),λ(Γ1)<∞. Choose for each j a Borel density with finite positive area whose quotient qj is arbitrarily close from below to λ(Γj). Restrict it to Ej; its path infimum on Γj is unchanged, and its area can only decrease. The restricted area is positive, since zero area together with a positive path infimum would, by [F1] and the rectangle perturbation of step 1.5, force λ(Γj)=+∞. Thus the restricted quotient remains positive and finite.

2.3F1step 1.5given

For the parallel law, take any width-admissible density σ for Γ0∪Γ1. Its restrictions σj=σ1Ej remain width-admissible for Γj, since every path of Γj lies in Ej. By nonnegative-integral monotonicity and additivity on the disjoint sets [F1, F4], A(σ)≥A(σ0)+A(σ1)≥μ(Γ0)+μ(Γ1), using step 1.5. Taking the infimum over σ gives μ(Γ0∪Γ1)≥μ(Γ0)+μ(Γ1); if there is no width-admissible density the left side is +∞ and the inequality still holds.

3.1F1F2F5step 2.1

The same local Lipschitz estimate applies to f−1 on compact subsets of Ω′. Consequently f∘γ is rectifiable exactly when γ is rectifiable: each direction follows by covering the compact path trace with finitely many convex disks and subdividing its parameter interval. Nonrectifiable paths have infinite length by definition on both sides of the last identity, and a zero-length path is constant, for which both integrals vanish because the arc-length measure is zero. Thus the length identity holds for every path.

3.2F1step 2.2givenalgebra

Write Lj=ℓρj(Γj) and Aj=A(ρj) for the restricted densities in step 2.2, and replace ρj by (Lj/Aj)ρj. The scaling law [F1] makes both its path infimum and its area equal to qj=Lj2/Aj. Put ρ=ρ0+ρ1; its supports are disjoint, so [F1] gives A(ρ)=q0+q1. Each γ∈Γ contains subpaths from Γ0,Γ1 on disjoint parameter intervals. Subpath additivity and nonnegativity show ℓρ(γ)≥q0+q1, hence ℓρ(Γ)2/A(ρ)≥q0+q1. Letting the two quotients approach their suprema proves λ(Γ)≥λ(Γ0)+λ(Γ1).

3.3F1step 1.5step 2.3given

If both μ(Γj) are finite, choose width-admissible densities σj with areas arbitrarily close above their infima, restrict them to Ej, and put σ=σ0+σ1. Every path of either family has σ-length at least one, while disjoint area additivity gives A(σ)=A(σ0)+A(σ1). Therefore μ(Γ0∪Γ1)≤μ(Γ0)+μ(Γ1) by step 1.5 and passage to arbitrarily small errors. If either summand is infinite this upper bound is automatic in the extended order. Combined with step 2.3 this proves equality.

4.1F2F3F4F8step 3.1

Given a Borel ρ′ of finite positive area on Ω′, put ρ=(ρ′∘f)∣f′∣ on Ω. The function is Borel by [F2, F4], and [F3] with det⁡Df=∣f′∣2 (The Jacobian determinant of a holomorphic map is ∣f′∣2 and is positive exactly where f′≠0) gives AΩ(ρ)=AΩ′(ρ′). The two areas are therefore finite and positive, while step 3.1 gives ℓρ(Γ)=ℓρ′(fΓ). Hence the corresponding extremal-length quotients agree. Applying the same construction to f−1 shows λ(fΓ)=λ(Γ); taking reciprocals gives μ(fΓ)=μ(Γ). The definition of conformal equivalence then gives the stated invariance.

5.1F8step 4.1step 1.4step 3.2step 2.3step 3.3∎

By definition λ=1/μ with 1/0=+∞ and 1/(+∞)=0. Thus the reciprocal of μ(Γ0)+μ(Γ1) is the harmonic sum of λ(Γ0),λ(Γ1), including when either modulus is zero or infinite. Steps 4.1, 1.4, 3.2, 2.3 and 3.3 establish (a), (b), (c) and (d), respectively.

TheoremStatement: Literature-sourcedProof: AI-adaptedprecheck passjudge pass (gpt-6.1-sol)audited 2026-10-08Open item page →

Extremal length of the rectangle and of the round annulus

Sources

  • Christopher J. Bishop, Quasiconformal Mappings, Ch. 1 §1, printed pp. 4–5 (PDF pp. 9–10). Lemma 1.6 proves the rectangle modulus with a constant test density and horizontal Cauchy–Schwarz slices. Lemma 1.7 proves the annulus connecting-family modulus by radial slices and the density (slog⁡(R/r))−1.
  • Mikhail Lyubich, Conformal Geometry and Dynamics of Quadratic Polynomials, vol. I, Ch. 1 §6.3.1, printed pp. 121–122. Proposition 6.6 proves the vertical-family extremal length in a flat cylinder by integrating over almost every vertical leaf; in a round annulus these leaves are radial. Exercise 6.8 gives the dual horizontal-family width, whose leaves are concentric circles.
  • Lars Ahlfors and Arne Beurling, Conformal Invariants and Function-Theoretic Null-Sets, §4, printed p. 115. Lemma 4 gives the rectangle joining-family constant, and Lemma 5 gives the separating-family constant in a round annulus. The present proof fixes the library's orientation-specific winding-one family and its reciprocal-modulus notation directly.

Statement

Assume Countable Choice and use the conventions of Extremal length and the curve-family modulus of a path family.

(i) Rectangle. Let 0<w,h<∞ and Π=(0,w)×(0,h). Let ΓΠv be the family of paths with interior in Π and one endpoint on each vertical side {0}×(0,h) and {w}×(0,h). Let ΓΠh be the analogous family joining the two horizontal sides. Then λ(ΓΠv)=wh,μ(ΓΠv)=hw, and λ(ΓΠh)=hw,μ(ΓΠh)=wh.

(ii) Round annulus. Let 0<r<R<∞ and A(r,R)={z:r<∣z∣<R}. Let Γr,R be the family of paths with interior in A(r,R) and one endpoint on each boundary circle. Let Θr,R be the family of rectifiable closed paths in A(r,R) whose winding number about 0 is 1 (The winding number of a closed contour about a point off its trace). Then λ(Γr,R)=12πlog⁡Rr,μ(Γr,R)=2πlog⁡(R/r), and λ(Θr,R)=2πlog⁡(R/r),μ(Θr,R)=12πlog⁡Rr. For a loop based at 1, winding number 1 is equivalently the positive generator under the standard identification of π1(C×,1) with Z (Winding number identifies the fundamental group of C times with the integers); for a loop based elsewhere, first change basepoint in C×.

All four families are nonempty, and the displayed values are finite and strictly positive.

Facts & Assumptions

Given: Countable Choice, the dimensions and radii in the Statement, and the curve-family length, area, extremal length, and modulus conventions.

[F1]

For a Borel density ρ, ℓρ is monotone in ρ; continuous densities agree with the absolute line integral; the arc-length parametrization formula and Lebesgue–Stieltjes interval uniqueness give parameterized length integrals; and A(ρ) is the nonnegative area integral (The rho-length and the extremal length are well defined, Extremal length and the curve-family modulus of a path family).

[F2]

Cauchy–Schwarz applies to L2 functions, and Tonelli interchanges the nonnegative area integrals over sigma-finite product spaces (Cauchy-Schwarz inequality for L2, Tonelli's theorem for nonnegative measurable functions on a sigma-finite product). If a slice has infinite square integral, the corresponding Cauchy–Schwarz upper bound is automatic.

[F3]

The polar map P(s,θ)=s(cos⁡θ+isin⁡θ) is C1 with determinant s>0. The Euclidean inverse-function theorem gives a local C1 inverse at every point; the polar-form theorem, on the branch 0<θ<2π, makes P one-to-one and onto the annulus minus the positive ray, so these local inverses combine to a global C1 inverse (The Euclidean inverse function theorem, Every nonzero complex number has a unique polar form r(cos⁡θ+isin⁡θ) with r>0 and −π<θ≤π, If γ:[a,b]→Rn is continuous, differentiable on (a,b), and γ′ extends continuously to [a,b], then L(γ)=∫ab∥γ′(t)∥2 dt). The cut is a planar null set under Countable Choice (A box with a degenerate side is Lebesgue null, and so is every coordinate hyperplane in Rn).

[F4]

For every nonnegative Borel g on A(r,R), the change of variables through P gives ∫A(r,R)g(z) dA(z)=∫02π∫rRg(seiθ) s ds dθ. This is the nonnegative Borel change-of-variables theorem on the cut annulus, followed by ignoring the null cut (A C^1 diffeomorphism satisfies the change-of-variables formula for nonnegative Lebesgue measurable functions, A continuous map has Borel preimages of Borel sets, Arithmetic and lattice operations preserve measurability whenever they are defined, [F3]).

[F6]

A nowhere-zero holomorphic function on a disc has a holomorphic logarithm (A nonvanishing holomorphic function on a disc has a holomorphic logarithm); holomorphic composition obeys the chain rule and (exp⁡L)′=(exp⁡L)L′ (The chain rule for complex derivatives, The complex exponential is entire and its complex derivative is itself). Also ∣exp⁡(x+iy)∣=ex (exp⁡(x+iy)=ex(cos⁡y+isin⁡y), ∣exp⁡(x+iy)∣=ex, and eiπ+1=0).

[F8]

For a positively oriented once-traversed circle, (2πi)−1∫dz/z=1 (The normalized integral around a positively oriented circle centred at a is 1). A closed contour's winding number is defined by that normalized integral (The winding number of a closed contour about a point off its trace); for based loops at 1, this integer classifies the positive generator of π1(C×,1) (Winding number identifies the fundamental group of C times with the integers).

[F9]

The supremum for λ is over Borel densities of finite positive area, permits extended path-length infima, and μ=1/λ with 1/0=+∞ and 1/(+∞)=0 (Extremal length and the curve-family modulus of a path family).

[F10]

The rectangle Π is nonempty open and convex; the round annulus is nonempty and open because ∣∣z∣−∣w∣∣≤∣z−w∣, and it is path-connected by radial paths to an intermediate circle and arcs on that circle. Thus both are complex domains (A complex domain is a nonempty connected open subset of C, Annuli in the complex plane, Every convex subset of Rn, in particular every ball and Rn itself, is path-connected and hence connected, Every path-connected space is connected, and every path component lies inside a component, Conjugation is an involutive real-field automorphism, zz‾=∣z∣2, and modulus is definite, multiplicative, and subadditive).

Proof

technique · compute area in polar coordinates and use Cauchy–Schwarz on the path foliations
1.1F3F4givenalgebra

Put P(s,θ)=s(cos⁡θ+isin⁡θ) on (r,R)×(0,2π). Its Jacobian determinant is s>0, so the inverse-function theorem in [F3] gives local C1 inverses. The principal polar-form theorem gives a unique angle in (0,2π) for every point off the positive real ray; thus P is bijective onto the cut annulus and its local inverses combine to a global C1 inverse. The omitted ray is null by [F3]. The nonnegative Borel change-of-variables theorem [F4] therefore gives the displayed polar area identity for every nonnegative Borel integrand, including extended-valued ones.

1.2F1F2F9given

Let ρ be any Borel density on Π with 0<A(ρ)<∞, and put L=ℓρ(ΓΠv). For each y∈(0,h) the horizontal segment from (0,y) to (w,y) belongs to ΓΠv, so L≤∫0wρ(x,y) dx. For almost every y, Tonelli and finite area make ρ(⋅,y) square-integrable; Cauchy–Schwarz then gives L2≤w∫0wρ(x,y)2 dx. In particular L<∞. Integrating over y yields hL2≤wA(ρ), so every quotient is at most w/h.

1.3F1F2F4F7F9given

Let ρ be any Borel density on A(r,R) with 0<A(ρ)<∞, and put L=ℓρ(Γr,R). For each θ∈(0,2π) the radial segment s↦seiθ belongs to Γr,R, so L≤∫rRρ(seiθ) ds. For almost every θ, the square integral is finite; weighted Cauchy–Schwarz gives L2≤(∫rRdss)∫rRρ(seiθ)2s ds=log⁡(R/r)∫rRρ(seiθ)2s ds. Integrating in θ and applying [F2, F4] yields 2πL2≤log⁡(R/r)A(ρ); in particular L<∞. Hence every quotient for Γr,R is at most (2π)−1log⁡(R/r).

1.4F1F5F6F7given

For a rectifiable path γ joining the two annulus boundary circles, its compact trace lies in {∣z∣≥r}⊂C×. Cover the trace by discs avoiding 0 and subdivide its parameter interval so each subpath lies in one such disc, using compactness and the Lebesgue number lemma (Heine-Borel in Rn: with the Euclidean metric a subset of Rn is compact if and only if it is closed and bounded, and the proof by bisection uses no choice principle; the same holds on the real line, The image of a compact metric space under a continuous map is compact, and so is the image of any compact subset, Every open cover of a compact metric space has a Lebesgue number: a δ>0 such that every nonempty subset of diameter less than δ lies inside a single member of the cover). On each disc take a holomorphic logarithm Gj of z by [F6]; differentiating eGj(z)=z gives Gj′(z)=1/z. Since ∣eGj(z)∣=eRe⁡Gj(z)=∣z∣, [F6, F7] identify Re⁡Gj(z) with log⁡∣z∣. The fundamental theorem for complex line integrals [F5] evaluates each subpath integral, and additivity [F5] gives Re⁡∫γdzz=log⁡∣γ(b)∣−log⁡∣γ(a)∣=±log⁡(R/r). The sign depends on the initial endpoint. The fundamental inequality in [F5] therefore yields ℓ1/∣z∣(γ)=∫γ∣dz∣∣z∣≥∣∫γdzz∣≥log⁡(R/r). For nonrectifiable paths the left side is +∞ by definition.

1.5F5F8F10givenalgebra

Put S=R/r and D(z)=z/r. Both D:A(r,R)→A(1,S) and its inverse w↦rw have constant complex difference quotients, so they are holomorphic and D is biholomorphic (Biholomorphic maps between complex domains, Complex differentiability at a point, the complex derivative, holomorphic functions, and entire functions). In the componentwise Riemann–Stieltjes definition, replacing γ by D∘γ multiplies its coordinate integrators by 1/r, while the pulled-back integrand 1/(D∘γ) is multiplied by r; therefore ∫D∘γdw/w=∫γdz/z. Hence D maps Θr,R bijectively onto Θ1,S. Conformal invariance Conformal invariance, monotonicity, and the series and parallel laws for extremal length gives equality of their extremal lengths and moduli. It remains to calculate the normalized family in A(1,S).

1.6F1F4F5F7F8givenalgebra

For the normalized winding-one family Θ1,S, every γ satisfies ∫γdz/z=2πi by [F8]. The continuous density ρ(z)=1/∣z∣ therefore has ℓρ(γ)≥2π by [F1, F5]. Its area is A(ρ)=∫02π∫1S1s2s ds dθ=2πlog⁡S by [F4, F7]. It is finite and positive, so λ(Θ1,S)≥2π/log⁡S.

2.1F1F9step 1.2givenalgebra

The constant density ρ=1/w on Π gives every path in ΓΠv length at least 1, because its endpoint displacement is w and Euclidean path length is at least that distance. Its area is h/w, so its quotient is at least w/h. Together with step 1.2 this gives λ(ΓΠv)=w/h and μ(ΓΠv)=h/w. Interchanging the two coordinates gives the horizontal formulas.

2.2F4F7F9step 1.1step 1.3step 1.4givenalgebra

The Borel density ρ(z)=1/(2π∣z∣) on A(r,R) has ℓρ(Γr,R)≥(2π)−1log⁡(R/r) by step 1.4. Its area, by step 1.1 and [F7], is A(ρ)=∫02π∫rR14π2s2s ds dθ=12πlog⁡(R/r), which is finite and positive. Its quotient is therefore at least (2π)−1log⁡(R/r). With step 1.3 this proves λ(Γr,R)=(2π)−1log⁡(R/r) and μ(Γr,R)=2π/log⁡(R/r).

2.3F1F2F4F8F9step 1.5step 1.6given

For any finite-positive-area Borel density ρ on A(1,S) put L=ℓρ(Θ1,S). For each s∈(1,S) the circle γs(θ)=seiθ, 0≤θ≤2π, has winding number 1 by [F8], so L≤ℓρ(γs). Its speed is s, hence its arc-length function is sθ; the Stieltjes interval formula, finite-measure uniqueness, and increasing simple approximation give ℓρ(γs)=s∫02πρ(seiθ) dθ. For almost every s, the finite area and [F4] make the circle density square-integrable, and Cauchy–Schwarz gives L2≤2πs2∫02πρ(seiθ)2 dθ. Rearranging this as L22πs≤s∫02πρ(seiθ)2 dθ and integrating over s∈(1,S) gives log⁡S2πL2≤∫1S∫02πρ(seiθ)2s dθ ds=A(ρ) by [F4]. Therefore every quotient is at most 2π/log⁡S, which with step 1.6 proves λ(Θ1,S)=2π/log⁡S and μ(Θ1,S)=(2π)−1log⁡S. Step 1.5 and log⁡S=log⁡(R/r) transfer these two values to Θr,R.

3.1F7F9step 2.1step 2.2step 2.3∎

The positive rectangle segments, radial annulus segments, and once-traversed circles used above witness that all assigned families are nonempty. The formulas are finite and strictly positive because w,h>0 and log⁡(R/r)>0 by [F7]. Taking reciprocals under the conventions of [F9] gives the four displayed modulus values, while steps 2.1, 2.2 and 2.3 give all four extremal lengths.

TheoremStatement: Literature-sourcedProof: AI-adaptedprecheck passjudge pass (gpt-6.1-sol)audited 2026-10-08Open item page →

The conformal parameter of a round annulus is a complete invariant

Sources

  • Mikhail Lyubich, Conformal Geometry and Dynamics of Quadratic Polynomials, vol. I, Ch. 1 §6.3.1, printed pp. 121–122. Proposition 6.6 gives the vertical-family value for a conformal annulus, and Exercise 6.8 gives the dual circular-family width. Section 6.3.6, printed p. 124, Corollary 6.20 records the related shrinking-nest consequence when the sum of annular moduli diverges.
  • Christopher J. Bishop, Quasiconformal Mappings, Ch. 1 §1, printed pp. 2–5. Lemma 1.2 gives overflow monotonicity, and Lemma 1.7 computes the annulus connecting-family modulus.
  • Lars Ahlfors and Arne Beurling, Conformal Invariants and Function-Theoretic Null-Sets, §4, printed p. 115. Lemma 5 computes the extremal length of curves separating the two boundary circles. The arguments below use the library's winding-one family and reciprocal convention directly.

Statement

Assume Countable Choice. For 0<r<R<∞ let A(r,R)={z:r<∣z∣<R} and let Γr,R be the family of paths with interior in A(r,R) joining the two boundary circles. Define the conformal parameter M(A(r,R)):=λ(Γr,R)=12πlog⁡Rr, the extremal length of the joining family (Extremal length of the rectangle and of the round annulus); it is the classical conformal modulus of a round annulus in the normalization for which the connecting-family modulus is 1/M=2π/log⁡(R/r) in the reciprocal library convention of Extremal length and the curve-family modulus of a path family. Then:

(i) For 0<r<R<∞ and 0<r′<R′<∞, A(r,R) and A(r′,R′) are conformally equivalent (Conformal equivalence and the automorphism group of a domain) if and only if M(A(r,R))=M(A(r′,R′)), equivalently R/r=R′/r′. When the parameters agree, z↦(r′/r)z is a conformal equivalence.

(ii) Let D∗={0<∣z∣<1} and let ΓD∗ be the family of paths γ:[0,1]→C with γ(0)=0, γ(1)∈∂D, and γ((0,1))⊆D∗. Then the limiting conformal parameter is infinite: M(D∗):=λ(ΓD∗)=+∞.

(iii) The punctured disc D∗ is not conformally equivalent to any round annulus A(r,R) with 0<r<R<∞, nor to the unit disc D, nor to the plane C. The punctured plane C∗ is likewise not conformally equivalent to any round annulus.

Facts & Assumptions

Given: Countable Choice, the annuli and domains in the Statement, and the extremal-length conventions.

[F1]

Extremal length is the supremum of ℓρ(Γ)2/A(ρ) over Borel densities of finite positive area; it is monotone under family inclusion in the reverse direction, and its value is independent of an ambient enlargement when the family lies in the smaller domain (Extremal length and the curve-family modulus of a path family, The rho-length and the extremal length are well defined, Conformal invariance, monotonicity, and the series and parallel laws for extremal length).

[F2]

For a finite round annulus, λ(Γr,R)=12πlog⁡Rr,λ(Θr,R+)=2πlog⁡(R/r), where Θr,R+ is the family of rectifiable closed paths in A(r,R) with winding number 1 about 0 (Extremal length of the rectangle and of the round annulus).

[F3]

A conformal equivalence preserves extremal lengths of path families whose full traces lie in its domains (Conformal invariance, monotonicity, and the series and parallel laws for extremal length).

[F4]

Based loops modulo endpoint-fixed homotopy form π1; continuous pointed maps induce homomorphisms that respect composition and based homotopy, and a homeomorphism induces an isomorphism with inverse induced by its inverse map (Based loops and the fundamental group, The homomorphism on fundamental groups induced by a pointed continuous map, Induced fundamental-group maps are well defined, functorial and invariant under based homotopy). For a path c from a to b, conjugation [γ]↦[c−1∗γ∗c] changes the basepoint from a to b; conjugation by the reverse path is its inverse, since a path followed by its reverse is homotopic relative endpoints to a constant path.

[F5]

The unit circle has fundamental group Z, and the winding number is the corresponding integer for based loops in C× at 1 (The trigonometric loops give π1({(x,y):x2+y2=1},(1,0))≅Z, Winding number identifies the fundamental group of C times with the integers, The winding number of a closed contour about a point off its trace). Scaling a circle and its contour by a positive factor leaves ∫dz/z unchanged by the componentwise Riemann–Stieltjes definition. Every automorphism of (Z,+) is multiplication by +1 or −1 (The integers form a commutative ring).

[F6]

For Ω=A(r,R) choose s0=(r+R)/2; for D∗ choose s0=1/2; for C∗ choose s0=1. Each is nonempty and open because its radial interval is open and z↦∣z∣ is continuous; radial segments to the circle of radius s0, followed by circle arcs, show path-connectedness and hence the complex-domain property (A complex domain is a nonempty connected open subset of C, Annuli in the complex plane, Paths, path-connected spaces and path components, Every path-connected space is connected, and every path component lies inside a component, t↦(cos⁡t,sin⁡t) is a bijection from [0,2π) onto the real unit circle, Radial normalisation x↦x/∥x∥2 is continuous on Rn∖{0}, Conjugation is an involutive real-field automorphism, zz‾=∣z∣2, and modulus is definite, multiplicative, and subadditive). The homotopy H(z,t)=((1−t)∣z∣+ts0)z/∣z∣ stays in the radial interval, fixes that circle, and deformation retracts Ω onto it.

[F8]

For every nonnegative Borel density, length on a path is the integral along an arc-length parametrization and is additive over subpath intervals. A zero extension from a Borel subdomain is Borel and preserves area; endpoint values do not affect path length because the arc-length Stieltjes measure is atomless (The rho-length and the extremal length are well defined, The Borel sigma-algebra of a subspace is the trace of the ambient Borel sigma-algebra).

[F9]

The integral logarithm agrees with the natural logarithm, is strictly increasing, and satisfies log⁡(2n)=nlog⁡2 with log⁡2>0 (The integral logarithm L is the published natural logarithm, The integral logarithm is continuous and strictly increasing on (0,∞), L(1/x)=−L(x), L(xn)=nL(x), and in particular L(2n)=nL(2)). The natural numbers are unbounded in R (Every complete ordered field is Archimedean).

[F10]

A nonvanishing holomorphic function on a homologically simply connected domain has a holomorphic logarithm, and a holomorphic logarithm of w has derivative 1/w (Star-shaped plane domains are homologically simply connected, A nonvanishing holomorphic function on a homologically simply connected domain has a holomorphic logarithm, A holomorphic logarithm is a primitive of the logarithmic derivative).

[F11]

A bounded entire function is constant (Liouville's theorem: every bounded entire function is constant). The integral of the derivative of a holomorphic function over a closed rectifiable contour is zero (The integral of a continuous complex derivative over every closed rectifiable contour is zero), while (2πi)−1∫∣w∣=1/2dw/w=1 (The normalized integral around a positively oriented circle centred at a is 1).

Proof

technique · winding-number classes, monotonicity, and a holomorphic-logarithm obstruction
1.1F4F5F6F7givenalgebra

The radial deformation retraction in [F6], based at s0, induces inverse homomorphisms on the domain and circle fundamental groups by [F4]. Scaling that circle to the unit circle identifies its group with Z by [F5]; the scaling leaves ∫dz/z unchanged because the integrand and coordinate integrators acquire reciprocal factors. Changing basepoint along a radial/circular path also preserves the winding integer, since the path and its reversal contribute opposite contour integrals. Thus winding identifies the fundamental group of each radial domain with Z. A biholomorphism between two such domains and its inverse induce inverse group isomorphisms, so it acts on winding numbers by an automorphism of Z, necessarily multiplication by a sign ε∈{+1,−1}. Since Z is abelian, the conclusion is independent of the basepoint paths; it sends the winding-one closed-loop family onto the winding-ε family. Rectifiability is preserved in both directions by the conformal path transport in Conformal invariance, monotonicity, and the series and parallel laws for extremal length.

1.2F1F7givenalgebra

For any rectifiable closed γ, complex conjugation c(z)=zˉ satisfies ∫c∘γdw/w=∫γdz/z‾, by expanding the componentwise Riemann–Stieltjes definition, so it interchanges winding +1 and −1. If γ~ is an arc-length parametrization of γ, then c∘γ~ is one for c∘γ because c is a Euclidean isometry. The arc-length integral formula in [F1] gives ℓρ(c∘γ)=ℓρ∘c(γ). Also A(ρ∘c)=A(ρ) by the Borel change-of-variables formula in [F7]. Since ρ↦ρ∘c is a bijection on finite-positive-area Borel densities, the two sign families have equal extremal length.

1.3givenalgebra

If R/r=R′/r′, the map f(z)=(r′/r)z is a bijective holomorphic map A(r,R)→A(r′,R′) with holomorphic inverse w↦(r/r′)w, so the annuli are conformally equivalent.

1.4F8F13givenconstruct

Fix n≥1 and put εn=2−n. Given γ∈ΓD∗, continuity and ∣γ(0)∣=0<εn<1=∣γ(1)∣ give a nonempty compact level set {t:∣γ(t)∣=εn}; compactness of [0,1] gives its largest member tn. For tn<t<1, one has εn<∣γ(t)∣<1, since another value at or below εn would force a later hit of that level, so γ∣[tn,1] belongs to Γεn,1. If γ is nonrectifiable, its assigned length is already +∞. If it is rectifiable, subpath additivity in [F8] gives the length comparison below.

1.5F11given

If a biholomorphism h:D∗→C existed, its inverse h−1:C→D∗ would be entire and bounded by 1. Liouville's theorem [F11] would make it constant, contradicting bijectivity.

1.6F10F11F12givenconstructalgebra

If a biholomorphism h:D→D∗ existed, then h would be nowhere zero. By [F12], the disc is homologically simply connected, so [F10] supplies a holomorphic g:D→C with eg(z)=h(z). Set L(w)=g(h−1(w)) on D∗. Then eL(w)=w, and [F10] gives L′(w)=1/w. Integrating around the positively oriented circle ∣w∣=1/2⊂D∗, [F11] gives 0=∫L′(w) dw=∫dw/w=2πi, a contradiction.

2.1F2F3F9step 1.1step 1.2

Conversely, let h:A(r,R)→A(r′,R′) be a biholomorphism. By step 1.1, it maps Θr,R+ onto one of the two target sign families. Conformal invariance [F3] and the sign equality in step 1.2 give λ(Θr,R+)=λ(Θr′,R′+). Using [F2], this is 2πlog⁡(R/r)=2πlog⁡(R′/r′). Both logarithms are positive by [F9]; cancellation and strict monotonicity in [F9] give R/r=R′/r′. This proves (i).

2.2F1F2F8F9step 1.4given

Let ρ be any Borel density on A(εn,1) with 0<A(ρ)<∞, and extend it by zero to a Borel density ρ~ on D∗. Its area is unchanged. Step 1.4 and [F8] show ℓρ~(ΓD∗)≥ℓρ(Γεn,1), because every path contains the annular subpath and the endpoint values carry no length mass. Thus the extremal-length quotient of ρ~ on ΓD∗ is at least the quotient of ρ on Γεn,1. Taking suprema and applying [F2] yields λ(ΓD∗)≥λ(Γεn,1)=log⁡(2n)2π=nlog⁡22π. As n is unbounded and log⁡2>0, this proves M(D∗)=+∞.

2.3F1F2F3F9step 1.2

Define ΘD∗+ and ΘC∗+ as the families of rectifiable closed paths of winding +1 in the indicated domains. For every n≥1, ΘA(2−n,1)+⊆ΘD∗+ and ΘA(2−n,2n)+⊆ΘC∗+. These subannular families have full traces in the larger domains, so [F1] and monotonicity [F3] give 0≤λ(ΘD∗+)≤2πnlog⁡2,0≤λ(ΘC∗+)≤2π2nlog⁡2. Letting n grow proves both extremal lengths are zero. By step 1.2 the corresponding winding-−1 families also have extremal length zero.

3.1F2F3step 1.1step 1.2step 2.3

If either Ω=D∗ or Ω=C∗ were conformally equivalent to a finite round annulus A(r,R), step 1.1 would map its winding-one family onto one of the two target sign families. Conformal invariance [F3] and step 1.2 would then equate its zero extremal length from step 2.3 with the strictly positive value 2π/log⁡(R/r) in [F2], a contradiction. This proves both finite-annulus exclusions in (iii).

4.1step 1.3step 2.1step 2.2step 3.1step 1.5step 1.6∎

Steps 1.3 and 2.1 prove (i), step 2.2 proves (ii), and steps 1.5, 1.6, and 3.1 prove every exclusion in (iii).

LemmaStatement: Literature-sourcedProof: AI-adaptedprecheck passaudited 2026-10-08Open item page →

Riemann maps of Jordan domains extend to homeomorphisms of the closures

Statement

Assume the Axiom of Choice. Let Γ⊆C^ be a Jordan curve, and let Ω1,Ω2 be the two complementary components. Each component admits a conformal equivalence from D, and every conformal equivalence ψ:D→Ωj extends uniquely to a homeomorphism ψ‾:D‾→Ωj‾.

After a Möbius change of coordinates making Γ⊂C, write Ω for the bounded component and Ω∗ for the component containing ∞. Then there is a conformal equivalence ψ∗:C^∖D‾→Ω∗ with ψ∗(∞)=∞, and every such exterior equivalence extends uniquely to a homeomorphism of the closures. If the original curve contains ∞, this normalization is understood after the stated Möbius change; an exterior map fixing ∞ in the original coordinate cannot exist in that case.

Facts & Assumptions

Given: The Axiom of Choice; a Jordan curve Γ; its complementary components; and the conformal maps in the statement.

[F1]

Jordan–Brouwer separation gives two complementary components with common boundary Γ. The plane Jordan–Schönflies theorem extends curve homeomorphisms to the plane and makes each closed bounded Jordan region a closed topological disk. (Jordan–Brouwer separation, Jordan–Schönflies extension for plane curves)

[F2]

A plane domain whose complement is connected and unbounded is homologically simply connected: for any complex cycle in the domain, its index is locally constant off its trace and zero sufficiently far away; the unbounded complement contains a point sufficiently far from the trace, and connectedness makes the locally constant index zero throughout the complement. (Homologically simply connected complex domains, The index of a cycle is locally constant off its trace and vanishes far from it)

[F3]

Under AC, every proper homologically simply connected plane domain is biholomorphic to D, with any chosen base point normalized to 0. (Every proper homologically simply connected plane domain is conformally equivalent to the unit disc)

[F4]

A Möbius transformation is a biholomorphism of the sphere; in particular J(z)=1/z exchanges D and C^∖D‾ and swaps 0 and ∞. (Möbius transformations of the Riemann sphere, Every Möbius transformation is a biholomorphism of the Riemann sphere, The standard holomorphic charts on the Riemann sphere, with holomorphy and poles at infinity)

[F5]

A holomorphic bijection is a real C1 diffeomorphism with Jacobian ∣Φ′∣2 (The Jacobian determinant of a holomorphic map is ∣f′∣2 and is positive exactly where f′≠0, An injective holomorphic map has no critical point and is biholomorphic onto its image). Nonnegative C1 change of variables on compact interior exhaustions therefore gives ∫G∣Φ′∣2dA=∣Φ(G)∣ (A C^1 diffeomorphism satisfies the change-of-variables formula for nonnegative Lebesgue measurable functions).

[F6]

Cauchy–Schwarz and nonnegative polar Fubini apply (Cauchy-Schwarz inequality for L2, Tonelli and Fubini for the completed product, with only almost-everywhere section measurability). The polar map on a half-annulus has determinant rho and a smooth inverse on its angle branch; applying [F5]'s change of variables gives the polar area integral. On each compact semicircle subarc, the C1 path length equals the integral of its speed (If γ:[a,b]→Rn is continuous, differentiable on (a,b), and γ′ extends continuously to [a,b], then L(γ)=∫ab∥γ′(t)∥2 dt); increasing those subarcs gives the improper length.

[F9]

Write z=x+iy and w=u+iv. Substitution into the stereographic coordinates in Stereographic projection identifies the Riemann sphere with the unit two-sphere and expansion of the squared Euclidean distance give χ(z,w)=2∣z−w∣(1+∣z∣2)(1+∣w∣2)≤2∣z−w∣, where the inequality uses (1+∣z∣2)(1+∣w∣2)≥1. Indeed, the dot product of the two unit-sphere images is 4Re⁡(zw‾)+(∣z∣2−1)(∣w∣2−1)(1+∣z∣2)(1+∣w∣2), and ∥Σ(z)−Σ(w)∥2=2−2Σ(z)⋅Σ(w) simplifies to the square of the displayed formula. The chordal metric is defined by that Euclidean distance (The chordal metric on the Riemann sphere).

[F10]

A holomorphic function on an upper half-disc, continuous on its closure and real on its diameter, extends holomorphically by Schwarz reflection. A holomorphic function vanishing on an interval inside a domain is identically zero by the identity theorem (Harmonic and holomorphic Schwarz reflection across the real axis, Identity theorem for holomorphic functions). A Möbius coordinate flattens any small arc of the unit circle to an interval, so this applies to a continuous boundary function that is zero on that arc.

[F11]

The Riemann sphere is compact Hausdorff, and C is an open dense subspace. (The Riemann sphere is the published one-point compactification of the complex plane)

Proof

technique · Riemann mapping, short crosscuts, Jordan separation and Schwarz reflection
1.1F1F4choose

If ∞∈Γ, choose q∉Γ and a Möbius map M with pole at q; then M(Γ)⊂C. Prove the assertion in this normalized coordinate and transport maps and closures back by M−1. Hence assume Γ⊂C.

2.1F1F4step 1.1choose

By [F1], the finite plane complement has bounded component Omega and unbounded component Ustar. Set Ω∗=Ustar∪{∞}, the corresponding sphere domain; its closure is a topological closed disk. Choose a∈Ω and τa(z)=1/(z−a); then V=τa(Ω∗) is the bounded Jordan component containing0. Omega and V are open connected plane domains; Omega-star is an open connected sphere domain.

3.1F2step 2.1

The complements C∖Ω and C∖V are connected and unbounded because each is the closure of the unbounded Jordan component, while Ω and V are bounded. For every cycle in either domain its index is locally constant off its trace and vanishes far away by [F2]; unboundedness supplies an omitted point with index zero, and connectedness then makes the index zero at every omitted point. Thus Ω and V are proper homologically simply connected domains.

4.1F3F4step 2.1step 3.1

Apply [F3] to Ω with base point a and invert the resulting map to obtain ψ:D→Ω. Apply [F3] to V with base point 0 to obtain g:V→D with g(0)=0. With I(z)=1/z, the map ψ∗(z):=τa−1(g−1(I(z))) is a conformal equivalence from C^∖D‾ onto Ω∗ and satisfies ψ∗(∞)=∞.

5.1F4givenstep 2.1step 4.1algebra

Fix any conformal equivalence h:D→Ω and any ξ∈∂D. The Möbius map αξ(z)=ξ(i−z)/(i+z) sends H to D and 0∈∂H to ξ; put Φ=h∘αξ. For r>0, let Πr={r/2<∣z∣<r, Im⁡z>0} and let Gr be its half-circle family.

6.1F5F6step 5.1algebra

For each r>0 and r/2<ρ<r, define the open semicircle image length L(ρ)=∫0π∣Φ′(ρeiθ)∣ρ dθ, possibly infinite. On compact angle subintervals this is exactly its C1 length by [F6], and exhaustion gives the improper length. Cauchy–Schwarz gives L(ρ)2≤πρ∫Sρ∣Φ′∣2ds. Integrating dρ/ρ and applying polar Fubini yields ∫r/2rL(ρ)2dρ/ρ≤π∫Πr∣Φ′∣2dA=π∣Φ(Πr)∣ by [F5]. No conformal invariance theorem for a not-yet-extended boundary family is used.

7.1F5F6F7F9step 5.1step 6.1choose

For dyadic rn=2−n the half-annuli Πrn are disjoint. Their images are disjoint measurable subsets of bounded Omega, so [F7] gives An=∣Φ(Πrn)∣→0. Step 6.1 and ∫rn/2rndρ/ρ=log⁡2 allow a radius ρn with L(ρn)2≤πAn/log⁡2+4−n. Countable Choice selects one at each scale. Hence these image arcs have finite length tending to zero and their source caps shrink to xi. Their Euclidean diameters are at most their lengths, and their spherical diameters also tend to zero by [F9]. Finite image length supplies endpoint limits in the next step, before any boundary family is invoked.

8.1F1step 2.1step 5.1step 7.1construct

Each finite-length image arc in step 7.1 has endpoint limits, since the remaining variation of a rectifiable arc tends to zero. Those limits lie on Γ: an interior limit, transported by the continuous inverse map, would make a sequence approaching the source boundary converge to an interior point. The arc interior is embedded; if the endpoints coincide it is a Jordan loop based at that boundary point. If they differ, uniform continuity of the inverse of a Jordan parameterization shows that they are joined by one Γ subarc of diameter tending to zero. Join this small subarc to the image arc to form a small Jordan loop Ln; in the coincident case take the image arc itself. Its bounded interior has diameter tending to zero, since it lies in the convex hull of its boundary (every affine coordinate attains its extreme on that boundary). The connected exterior Ω∗ contains infinity and avoids Ln, hence lies on its unbounded side. Every point of the other Γ subarc, or of Γ minus the coincident endpoint, is approached from Ω∗ and misses Ln, so also lies on that unbounded side. Thus the bounded interior of Ln is contained in Ω. For large n it excludes the fixed point h(0). Crosscut separation in the topological Jordan disk therefore identifies it with the image of the source cap adjacent to ξ. These image caps have diameter tending to zero. Nested source caps then show that h has a unique limit at ξ, and that its continuous extension is continuous there. Apply this at every boundary point.

9.1F1F10F11step 8.1construct

Suppose two distinct boundary points have the same image a. The image of their straight source chord is a Jordan loop L meeting Γ only at a. Its bounded interior is contained in Ω: the connected exterior of Ω contains infinity and avoids L, and exterior approximation puts every point of Γ∖{a} on the unbounded side of L. One source chord cap maps to this bounded interior. Continuity forces its source boundary-circle arc to map into Γ∩int⁡L‾={a}. Flatten a subarc by a Möbius coordinate; Schwarz reflection [F10] extends h−a across the interval where it is zero, and the identity theorem forces h constant, a contradiction. Hence the boundary extension is injective. It is onto Γ, because inverse images of a sequence in Ω approaching any given boundary point have a subsequence in the compact closed disk, and its limit cannot be interior. The extension is a continuous bijection of compact Hausdorff closures, hence a homeomorphism.

10.1F4step 2.1step 3.1step 5.1step 9.1

For any conformal equivalence h∗:C^∖D‾→Ω∗, the map τa∘h∗∘I is a conformal equivalence D→V; repeating the crosscut and reflection argument of steps 5.1–9.1 for V=τa(Ω∗) gives a homeomorphic extension, and conjugating back by τa−1 and I gives the required extension of h∗.

11.1F11givenstep 1.1step 9.1step 10.1∎

Any two continuous extensions agree on the dense open domain of their conformal equivalence, so agree on its closure because the sphere is Hausdorff; the extension is unique. Transporting the normalized conclusions from step 1.1 proves the theorem for the original sphere curve; the exterior normalization at ∞ is asserted only in a coordinate where ∞∉Γ.

LemmaStatement: Literature-sourcedProof: AI-adaptedprecheck pendingjudge pass (gpt-6.1-sol)audited 2026-10-08Open item page →

Analytic quasiconformality gives both quadrilateral modulus bounds

Statement

Assume the Axiom of Choice. Let f:Ω→Ω′ be an orientation-preserving analytically K-quasiconformal homeomorphism, K≥1, in the sense of The ACL and Sobolev analytic definition of quasiconformality. For every Jordan quadrilateral Q⋐Ω with either pair of opposite marked sides, write ΓQ for its joining family and Γf(Q) for the corresponding image family (Orientation-preserving homeomorphisms and the geometric definition of quasiconformality). Then K−1μ(ΓQ)≤μ(Γf(Q))≤Kμ(ΓQ), equivalently K−1λ(ΓQ)≤λ(Γf(Q))≤Kλ(ΓQ). This result supplies quadrilateral bounds; no inverse regularity, inverse null-set property or arbitrary-ring comparison is assumed.

Facts & Assumptions

Given: AC, the orientation-preserving analytic homeomorphism, its constant K, and a relatively compact marked Jordan quadrilateral.

[F1]

The analytic definition supplies Wloc1,2 and ACL coordinate representatives. The weak partials agree with their classical line derivatives. Put k=(K−1)/(K+1); the Wirtinger identities give Jf=∣fz∣2−∣fzˉ∣2≥0 and ∥Df∥op2≤KJf (The ACL and Sobolev analytic definition of quasiconformality, The Wirtinger derivatives ∂zf and ∂zˉf, and antiholomorphic functions, The ACL characterisation of W1,p). The smooth local-homology multiplier is the determinant sign (A real linear isomorphism preserves or reverses orientation according to the sign of its determinant, Smooth orientation sign is the local integral homology multiplier); step 2.1 transfers this to a homeomorphism at a nonsingular differentiability point by a nonvanishing boundary homotopy.

[F2]

Egorov and Lusin give uniform convergence and continuous restrictions on compact sets outside sets of arbitrarily small measure. One-dimensional differentiation and completed-product Fubini give density one on almost every horizontal and vertical slice of such sets (Egorov's theorem, Assuming countable choice, Lusin's theorem on finite-measure subsets of R^n, Lebesgue differentiation theorem on Rn, Tonelli and Fubini for the completed product, with only almost-everywhere section measurability).

[F3]

Differentiation of locally finite positive Borel measures identifies their absolutely continuous densities through small-ball ratios. Nonnegative change of variables holds for a C1 diffeomorphism (Differentiation of sigma-finite Borel measures finite on compact sets, A C^1 diffeomorphism satisfies the change-of-variables formula for nonnegative Lebesgue measurable functions).

[F5]

Rectangle joining-family extremal lengths are its aspect ratio and reciprocal; the straight foliation has the same value by the identical slice estimate. Modulus is the infimum of density area under length admissibility, obtained by scaling the defining quotient. Family inclusion reverses extremal length; arc-length integration is invariant under parameterization (Extremal length of the rectangle and of the round annulus, Extremal length and the curve-family modulus of a path family, The rho-length and the extremal length are well defined, Conformal invariance, monotonicity, and the series and parallel laws for extremal length).

[F6]

Complex Hölder and Minkowski control products and sums in L2 (Complex Holder, Minkowski, and the quotient norm). Countable Choice and the assumed AC permit the countable compact covers and subsequences used below (The Axiom of Countable Choice (ACω), The Axiom of Choice). Interior mollification and Lp approximate identities give derivative convergence; for continuous functions they converge uniformly on compacta (Interior mollification commutes with weak derivatives, Every L1 approximate identity converges to the identity in Lp for 1≤p<∞, L1 approximate identities converge uniformly on compacta for bounded continuous functions). The AC fundamental theorem is Fundamental theorem of calculus for absolutely continuous functions.

Proof

technique · prove the lower area and exceptional-curve interfaces without an inverse, rectify quadrilaterals, and obtain the second inequality by transverse reciprocity
1.1F1F2givenconstruct

A continuous planar homeomorphism with finite partial derivatives almost everywhere is totally differentiable almost everywhere. Here are the needed details. On an interior rectangle, Egorov and Lusin give a compact set E outside arbitrarily small area on which the two partials are continuous and their directional difference quotients converge uniformly. The measurable error is the supremum over rational 0<∣t∣<1/n; continuity in nonzero t makes this the full supremum. By [F2], almost every p∈E is a density-one point of both coordinate slices. Translate p to 0, and put A(x+iy)=f(0)+xfx(0)+yfy(0). Uniform quotients and continuous partials give ∣f(w)−A(w)∣≤3ε∣w∣ whenever the horizontal or vertical projection of w belongs to the corresponding slice of E, for sufficiently small w. For any fixed z with r=∣z∣, slice density supplies rectangle sides at coordinates within εr on either side of each coordinate of z. Every boundary point w of that rectangle has one of the good projections. Openness of f makes the maximum of ∣f(w)−A(z)∣ occur on the boundary: A(z) is a fixed constant, and an open image cannot have an interior maximum of distance from it. Thus ∣f(z)−A(z)∣≤3ε(1+2ε)r+2ε(∣fx(0)∣+∣fy(0)∣)r. Let ε↓0. Taking compact sets with excluded area tending to zero proves the assertion; [F1] makes it applicable to f. No maximum principle for a variable affine difference is used.

1.2F2F3F5F6construct

We record the precise exceptional-curve argument. On each compact interior neighborhood mollify a continuous W1,2 map u to smooth un with uniform convergence and derivative convergence in L2. Choose a subsequence with ∑n∥Dun−Du∥2<∞ and put G=∑n∣Dun−Du∣+∣Du∣. This is in L2. Rectifiable curves with ∫γG=∞ have modulus zero, because G/m is admissible on them with area tending to zero. On every remaining curve, the derivatives of un∘γ converge in L1 to Du(γ)γ′ along arc-length parameterization; uniform convergence and the fundamental theorem of calculus identify the limit as the derivative of the absolutely continuous path u∘γ. Borel null representatives are handled by an infinite density on their null set. A countable compact exhaustion and positive summable multiples of the local barriers give one L2 barrier for all bad compact restrictions. If u has finite total energy on a bounded domain and extends continuously to its boundary, include ∣Du∣ in the barrier. Good boundary-joining paths then have finite derivative integral on the whole interval; local absolute continuity and endpoint continuity imply absolute continuity including endpoints. For an AC path, its variation on intervals is the integral of its speed, first by partitions and the fundamental theorem and then by differentiation. Uniqueness of measures and simple approximation therefore give the weighted arc-length identity for every nonnegative Borel density. This proves the chain-rule length inequality outside a modulus-zero family, not merely on coordinate lines.

1.3F4givenconstruct

Every marked Jordan quadrilateral rectifies to a rectangle. Map its interior to the upper half-plane with three marked boundary points normalized to 0,1,∞; the fourth is a<0, by [F4]. On the upper half-plane take a primitive P of 1/z(z−1)(z−a), choosing the analytic square root by a logarithm. Its derivative is nonzero. The inverse-square-root singularities at a,0,1 are integrable; at infinity the derivative is O(∣z∣−3/2), so the primitive has a common finite limit there. Each of the four real boundary intervals maps monotonically to a straight side; the directions alternate by π/2. Equality of the two limits at infinity closes the polygon, and its four positive side lengths give a rectangle with equal opposite sides. On upper half-disks indented about the three branch points, image boundaries converge to that rectangle boundary. The argument principle [F4] gives exactly one preimage for every interior point and none for an exterior point. Thus P is a conformal bijection, extending homeomorphically to the four sides.

2.1F1F3step 1.1givenalgebra

Define the positive locally finite Borel measure ν(E)=∣f(E)∣. At a differentiability point with invertible derivative A, the expansion f(p+v)=f(p)+Av+o(∣v∣) traps the image of a radius-r disk between the ellipses A(Dr) enlarged and contracted by o(r). The inner containment follows by Jordan separation, since the image boundary is an o(r) perturbation of the ellipse and has the same winding on its contracted interior. Hence its area ratio tends to ∣det⁡A∣. At a singular derivative the image is contained in an o(r) neighborhood of a bounded segment or point, so that ratio tends to zero. Since f preserves orientation, a nonsingular derivative has positive determinant: interpolation to Av on a small boundary circle preserves the local orientation sign. Measure differentiation [F3] therefore identifies ν's absolutely continuous density with Jf. Its singular part is nonnegative, giving ∫EJf≤∣f(E)∣ for every relatively compact Borel E. Simple approximation and exhaustion give ∫(g∘f)Jf≤∫g for every nonnegative Borel target function. This is only the lower area inequality.

2.2F3F4F5step 1.2step 1.3algebra

The rectangle map and its inverse have finite conformal energy: nonnegative change of variables on compact interior exhaustions gives ∫∣ϕ′∣2=∣Q∣ and the corresponding finite rectangle area for its inverse. Their boundary extensions are continuous. Step 1.2 therefore transports boundary-joining families outside modulus-zero exceptional families; the weighted length identity and the conformal area identity give invariance in both directions. Consequently [F5] transfers the rectangle joining-family values, equality with its straight foliation, and transverse reciprocity to every Jordan quadrilateral. This explicitly extends the compact-interior conformal-invariance supplier to the boundary families used here.

3.1F1F5F6step 2.1step 1.2algebra

Removing a zero-modulus family does not change modulus: add to any good-family admissible density a bad-family admissible density with arbitrarily small L2 norm and use Minkowski [F6]. A density of infinite length on all bad curves and arbitrarily small norm is obtained by summing a sequence of bad-family admissible densities with summable norms. Apply step 1.2 to f and a target admissible density τ. Its pullback ρ=(τ∘f)∥Df∥op satisfies A(ρ)≤KA(τ) by step 2.1 and [F1]. Weighted length on every good source curve is at least the target image length. Adding a vanishing-cost bad-family density gives μ(Γ)≤Kμ(fΓ), equivalently λ(fΓ)≤Kλ(Γ), for the compact-trace families used here. This establishes one direction only.

4.1F1F2F3F5step 2.1step 3.1step 2.2algebra

Rectify the source Q and take its straight joining foliation F. In conformal coordinates, f∘ϕ has finite energy on the whole rectangle: step 2.1 and ∥Df∥2≤KJf bound it by a constant times ∣f(Q)∣ after conformal change of variables. Sobolev coordinate change follows by mollification on compact subsets, the classical chain rule and L2 convergence there. Fubini gives a Borel null set N of bad leaf parameters; good leaves are AC on the entire closed interval with the weighted chain rule. The bad leaves have modulus zero: if the leaf width is w, the density w−11(0,w)×O for an open O⊃N has length one on each bad leaf and area ∣O∣/w→0. Removing them leaves the source foliation's value unchanged. The full target joining family contains the image good foliation. Step 3.1, or its identical good-leaf pullback, thus gives λ(Γf(Q))≤λ(fFgood)≤Kλ(Fgood)=Kλ(ΓQ). No claim that images of bad leaves have zero modulus is needed.

5.1F5step 2.2step 4.1algebra∎

Repeat step 4.1 for the transverse source foliation. By step 2.2 the two joining-family extremal lengths multiply to one in both quadrilaterals. Inverting the transverse upper bound therefore gives λ(Γf(Q))≥K−1λ(ΓQ). Their values are finite and positive because the rectifying rectangles are nondegenerate. Taking reciprocals gives both displayed modulus bounds. The second bound came from transverse reciprocity, not an assumed inverse-null property.

Remark

The local arguments used in this proof establish the following auxiliary interfaces under the stated AC assumption. A continuous planar homeomorphism with finite coordinate partial derivatives almost everywhere is totally differentiable almost everywhere. For an orientation-preserving such map, the image-area measure has absolutely continuous density Jf, so ∫EJf≤∣f(E)∣ on relatively compact Borel sets; the corresponding nonnegative weighted inequality follows by simple approximation and exhaustion. This is a lower inequality, not area equality or inverse-null.

For a continuous Wloc1,2 map, a summable-gradient mollification barrier gives the AC chain rule and the nonnegative Borel weighted speed identity on every rectifiable compact restriction outside a modulus-zero curve family. If its total energy is finite and it extends continuously to the boundary, including the total derivative in the barrier gives the same assertion on good compact boundary-joining paths. Jordan quadrilaterals rectify to nondegenerate rectangles, their straight joining foliation has the full joining-family extremal length, and their two transverse joining-family values multiply to one. These are the proved inputs used above; none grants a second arbitrary-family distortion bound or inverse regularity.

TheoremStatement: Literature-sourcedProof: AI-adaptedprecheck passaudited 2026-10-08Open item page →

The geometric and analytic definitions of quasiconformality agree

Sources

  • Christopher J. Bishop, Quasiconformal Mappings, Ch. 2 §2 and Ch. 3 §4, printed pp. 51–52 and 91–96. The geometric definition is quasi-invariance of every quadrilateral's modulus; Theorem 4.1 proves ACL from that condition, and Lemmas 4.4–4.6 give the Jacobian and area estimates used in the analytic direction.
  • Mikhail Lyubich, Conformal Geometry and Dynamics of Quadratic Polynomials, vol. I, Ch. 2 §12.5, printed p. 188, QC1–QC2 and Proposition 12.15; §§11.3–11.4 and §§12.1–12.4 contain the analytic and geometric regularity arguments.
  • Lars Ahlfors and Arne Beurling, Conformal Invariants and Function-Theoretic Null-Sets, §§4–5, printed pp. 114–120, for the extremal-length convention and model quadrilateral/annulus values.

Statement

Assume the Axiom of Choice. Let Ω,Ω′⊆C be complex domains, let K≥1, and put k=(K−1)/(K+1). For a homeomorphism f:Ω→Ω′, the following are equivalent.

(a) f is K-geometrically quasiconformal in the sense of Orientation-preserving homeomorphisms and the geometric definition of quasiconformality: it preserves orientation and, for every quadrilateral Q with Q‾⊆Ω and either choice of opposite marked sides, K−1μ(Γ(Q))≤μ(Γ(f(Q)))≤Kμ(Γ(Q)).

(b) f is K-analytically quasiconformal in the sense of The ACL and Sobolev analytic definition of quasiconformality: f∈Wloc1,2(Ω) and ∣fzˉ∣≤k∣fz∣almost everywhere.

Consequently the least geometric constant equals the analytic maximal dilatation Kf of The Beltrami coefficient and the maximal dilatation, analytic quasiconformal homeomorphisms preserve orientation, and the quasiconformal class is closed under inverses with the same maximal dilatation. Also Kf=1 exactly when μf=0 almost everywhere.

Facts & Assumptions

Given: The Axiom of Choice, complex domains, a homeomorphism, and either the geometric or analytic K-quasiconformality condition.

[F1]

The geometric definition imposes both modulus bounds for every relatively compact Jordan quadrilateral and each pair of opposite sides. A homeomorphism carries the corresponding path family onto the family in its image quadrilateral (Orientation-preserving homeomorphisms and the geometric definition of quasiconformality).

[F2]

If f is analytic K-quasiconformal, then the modulus-distortion lemma gives the two-sided bounds of the geometric definition on each quadrilateral (Analytic quasiconformality gives both quadrilateral modulus bounds).

[F3]

The geometric bounds apply to every thin rectangle compactly inside the domain (Orientation-preserving homeomorphisms and the geometric definition of quasiconformality). The explicit area-function argument in step 1.2 proves ACL; it assumes no inverse-null or circular-dilatation result. The fundamental theorem reconstructs each AC line from its integrable derivative (Fundamental theorem of calculus for absolutely continuous functions); the general ACL characterization identifies locally L2 line derivatives as weak derivatives (The ACL characterisation of W1,p).

[F4]

The auxiliary differentiability and lower-area interfaces in the Remark of Analytic quasiconformality gives both quadrilateral modulus bounds apply to continuous planar homeomorphisms with finite partials almost everywhere: the maximum is taken after subtracting a fixed constant, and the image-area density is the absolute Jacobian. For an orientation-preserving map it is Jf.

[F5]

If an orientation-preserving homeomorphism is differentiable almost everywhere, the pushforward measure E↦area⁡(f(E)) has absolutely continuous density Jf by differentiation of measures; its singular part is nonnegative, so ∫EJf dA≤area⁡(f(E)) for relatively compact Borel E (Differentiation of sigma-finite Borel measures finite on compact sets). Bishop, Lemma 4.4, printed pp. 95–96, gives the square estimate by a Vitali covering. Together with ∣Df∣op2≤KJf, this controls local L2 energy. Its differentiability input is the Gehring–Lehto theorem in [F4].

[F6]

At a differentiability point the singular-value ratio of the real derivative is ∣fz∣+∣fzˉ∣∣fz∣−∣fzˉ∣. The inequality that this ratio is at most K is equivalent to ∣fzˉ∣≤((K−1)/(K+1))∣fz∣ (The Wirtinger derivatives ∂zf and ∂zˉf, and antiholomorphic functions, algebra).

[F7]

If a homeomorphism is differentiable at z with invertible derivative A, then its local-homology orientation multiplier is sgn⁡det⁡A: on a sufficiently small sphere, the straight homotopy from f(z+v)−f(z) to Av avoids zero because the differentiability remainder is smaller than 12min⁡∣u∣=1∣Au∣ ∣v∣. Homotopy invariance, excision and functoriality identify this sphere degree with the local homology map (The singular chain homotopy formula, Functoriality of relative homology, Excision for singular homology, Local homology detects manifold dimension, interior, and boundary, A real linear isomorphism preserves or reverses orientation according to the sign of its determinant, Smooth orientation sign is the local integral homology multiplier).

[F8]

The Beltrami coefficient and least analytic dilatation are defined in The Beltrami coefficient and the maximal dilatation. Geometric inversion preserves both bounds and orientation by rearranging the definition; apply the equivalence proved here to obtain its analytic inverse, without citing a later inverse-coefficient theorem.

Proof

technique · use annular and quadrilateral modulus distortion in the analytic direction; use ACL, the local Jacobian-area estimate and infinitesimal quadrilaterals in the geometric direction
1.1F1F2F3F4F7given

Assume (b). By [F3, F4], f is differentiable almost everywhere. At any point z where Df(z) exists and is nonsingular, write f(z+v)−f(z)=Av+o(∣v∣) with A=Df(z). If c=min⁡∣u∣=1∣Au∣>0, then for all sufficiently small r, the straight homotopy from v↦f(z+v)−f(z) to v↦Av stays nonzero on ∣v∣=r, since the remainder is less than cr/2. Hence the induced map on local homology has the same sign as A, namely sgn⁡det⁡A; this is the determinant orientation rule in [F7]. At singular differentiability points Jf=0. If f were orientation-reversing, it would follow that Jf≤0 almost everywhere. The analytic inequality gives Jf=∣fz∣2−∣fzˉ∣2≥(1−k2)∣fz∣2≥0, so Jf=0 and Df=0 almost everywhere. By ACL and the one-dimensional fundamental theorem, on almost every horizontal segment in any small rectangle the restriction of f would be constant, contradicting injectivity. Thus f preserves orientation. Finally [F2] gives the modulus bounds in (a).

1.2F1F3F4F5givenconstruct

Assume (a). Fix a rectangle R⋐Ω and define the finite Borel measure ν(B)=∣f({(x,y)∈R:y∈B})∣ on R. Let A(y)=ν((−∞,y)). Applying the measure-differentiation theorem in [F5] to the forward and backward half-intervals, which shrink nicely to y, shows that A′(y) exists and is finite for almost every y. At such a height choose finitely many disjoint intervals (uj,vj), of total length l, and put dj=∣f(vj,y)−f(uj,y)∣. For strips of height t above these intervals, uniform continuity makes every path joining the image vertical sides have length at least (dj−ε)+ when t is sufficiently small. Constant density one and the geometric lower modulus bound give (dj−ε)+2≤K(vj−uj)∣f(Rj)∣/t. Their open images are disjoint and lie in the full strip, so Cauchy–Schwarz gives (∑j(dj−ε)+)2≤Kl(A(y+t)−A(y))/t. Let t↓0 and then ε↓0. The bound ∑jdj≤KlA′(y) is exactly absolute continuity on the horizontal line. Repeat vertically and cover by countably many interior rectangles. Partial derivatives exist almost everywhere; [F4]'s fixed-constant rectangle argument gives total differentiability almost everywhere.

1.3F1F4F6givenalgebra

At a differentiability point with singular values s1≥s2>0, use a small square aligned with the right singular vectors and rescale by its side length. The image lies in an o(1) neighborhood of the linear rectangle of dimensions s1,s2, so its area is at most s1s2+o(1). Every path joining the image sides perpendicular to the first singular vector has length at least s1−o(1), by endpoint separation. Constant density one therefore gives image modulus at most (s1s2+o(1))/(s1−o(1))2. The source square modulus is one, so its geometric lower bound gives 1/K≤s2/s1 in the limit. Hence s1/s2≤K, equivalent to the Beltrami inequality in [F6]. If the derivative has rank one, the same rescaled image has area o(1) while the joining length remains bounded below, contradicting that lower modulus bound. Rank zero satisfies the inequality directly. This proves the sharp differential bound without presuming continuity at a degenerate quadrilateral or a later inverse result.

1.4F3F4F5givenalgebra

By the derivative bound, ∥Df∥op2≤KJf almost everywhere. The independently proved lower area inequality [F4]–[F5] gives ∫Q∥Df∥op2≤K∣f(Q)∣ on every interior square. The Hilbert–Schmidt square is at most twice this, so both partials are locally square integrable. The ACL characterization now identifies them as weak derivatives; the bounded continuous map is also locally square integrable. Thus f∈Wloc1,2, proving (b).

2.1F1F2step 1.1step 1.4F8given∎

The two implications hold for each K≥1, so the least geometric constant and the least analytic constant coincide. The geometric definition gives the same bounds and orientation for the inverse. Applying the implication just proved to that inverse gives its analytic K-quasiconformality with the same least constant. Finally, the Beltrami definition gives Kf=1 iff ∥μf∥∞=0, which is equivalent to μf=0 almost everywhere.

LemmaStatement: Literature-sourcedProof: AI-adaptedprecheck passaudited 2026-10-08Open item page →

The inverse of a quasiconformal map is quasiconformal with the same dilatation

Sources

  • Mikhail Lyubich, Conformal Geometry and Dynamics of Quadratic Polynomials, vol. I, Ch. 2 §12.5, printed p. 188, Proposition 12.15, for inverse and composition quasiconformality; These earlier source sections are contextual; the present proof obtains inverse regularity from the independent quadrilateral equivalence, then proves its area and chain-rule interfaces locally.
  • Christopher J. Bishop, Quasiconformal Mappings, Ch. 2 §1, printed pp. 49–51, for the real-linear inverse dilatation and the Beltrami chain identity.

Statement

Assume the Axiom of Choice. Let f:Ω→Ω′ be an analytically K-quasiconformal homeomorphism (The ACL and Sobolev analytic definition of quasiconformality) with Beltrami coefficient μf (The Beltrami coefficient and the maximal dilatation). Then g=f−1:Ω′→Ω is analytically K-quasiconformal, Kg=Kf, and μg(f(z))=−μf(z) fz(z)fz(z)‾for almost every z∈Ω. Consequently ∣μg∣∘f=∣μf∣ almost everywhere, and f is analytically quasiconformal exactly when f−1 is, with the same maximal dilatation.

Facts & Assumptions

Given: The Axiom of Choice, complex domains Ω,Ω′, and an analytic K-quasiconformal homeomorphism f:Ω→Ω′.

[F1]

Analytic and geometric quasiconformality agree with the same constant. The inverse geometric bounds follow by rearrangement and its orientation sign is the inverse positive local-homology map; hence the inverse is independently analytically K-quasiconformal (The geometric and analytic definitions of quasiconformality agree, Orientation-preserving homeomorphisms and the geometric definition of quasiconformality).

[F2]

The earlier core gives total differentiability a.e. and the lower area inequality, without assuming inverse regularity (Analytic quasiconformality gives both quadrilateral modulus bounds, Remark). The mollifications are smooth (Convolution with a mollifier is smooth, and derivatives pass under the integral sign). Smooth Sard and ordinary nonnegative change of variables apply to its smooth approximants (Morse-Sard for smooth manifolds, A C^1 diffeomorphism satisfies the change-of-variables formula for nonnegative Lebesgue measurable functions). The complete reverse area argument is supplied below, separately for each already-regular inverse. Interior mollification and Lp approximate identities give derivative convergence; for continuous functions they converge uniformly on compacta (Interior mollification commutes with weak derivatives, Every L1 approximate identity converges to the identity in Lp for 1≤p<∞, L1 approximate identities converge uniformly on compacta for bounded continuous functions). The AC fundamental theorem is Fundamental theorem of calculus for absolutely continuous functions.

[F3]

For a real-differentiable homeomorphism f at z with invertible derivative, differentiability of g=f−1 at f(z) gives Dg(f(z))=(Df(z))−1. In Wirtinger form, the real chain rule is (g∘f)z=(gw∘f)fz+(gwˉ∘f)fzˉ‾,(g∘f)zˉ=(gw∘f)fzˉ+(gwˉ∘f)fz‾, and the Wirtinger coefficients uniquely determine a real-linear map (The chain rule for total derivatives: D(g∘f)(a)=Dg(f(a))∘Df(a), The Wirtinger chain rule for compositions of real-differentiable complex-valued maps, The Wirtinger derivatives ∂zf and ∂zˉf, and antiholomorphic functions). The Beltrami coefficient and least-dilatation conventions are those of The Beltrami coefficient and the maximal dilatation.

[F4]

The local-homology multiplier of an invertible smooth derivative is its determinant sign (Smooth orientation sign is the local integral homology multiplier); the connecting-map identification in step 2.1 translates that multiplier into local winding. Smooth Euclidean inverse branches are supplied by Choice-free smooth inverse function theorem in Euclidean space. Stokes holds for C1 complex forms on bounded C1 Euclidean domains under AC (Stokes for complex forms on a bounded C1 Euclidean domain); rounding finitely many rectangle corners and then letting their radii shrink gives the rectangle-with-disks formula below. Measure uniqueness applies on a generating pi-system with a finite-measure exhaustion (Measures agreeing on a generating pi-system are equal under an increasing finite-measure exhaustion from that pi-system).

Proof

technique · obtain inverse regularity from the quadrilateral equivalence, establish the area formula for both maps, and then differentiate on a common full-measure set
1.1F1F2given

By [F1], g=f−1 is analytically K-quasiconformal independently of any area equality or inverse-null assumption. Both maps are orientation-preserving, belong locally to W1,2, and are differentiable almost everywhere by [F2].

2.1F1F2step 1.1construct

We prove area equality for either map u. On an interior rectangle, choose countable dense sets of good ACL horizontal and vertical levels. Their subrectangles form a basis with rectifiable Jordan image boundaries. These image boundaries have area zero: divide each finite-length arc into pieces of length at most δ, cover by squares of side 2δ, and let the total cost Cδ(length⁡+δ) tend to zero. Let D be such a closed rectangle and T⋐u(D∘). Smooth mollifications un converge uniformly on a neighborhood of D, with derivatives converging in L2; determinants Jn converge to Ju in L1, and since Ju≥0, the integrals of (Jn)− tend to zero. Uniform convergence makes the straight homotopy of the boundary images avoid every y∈T for large n. The positively oriented rectifiable Jordan contour u(∂D) has winding one at y: triangulate D into a singular2-chain, transport it by u, and apply naturality of the pair boundary map to its positive local orientation class. Excision identifies that class with the positive generator of H2(R2,R2∖{y}); the connector is an isomorphism because R2 is contractible, giving the positive punctured-plane H1 generator. Degree-one Hurewicz and the winding/degree identification give analytic winding one. This uses Long exact sequence of a pair, Naturality of the pair long exact sequence, Excision for singular homology, The first Hurewicz map is abelianization, For loops in C times, the winding number about 0 equals the circle degree, Winding number identifies the fundamental group of C times with the integers; translating by −y and a nonzero complex scaling normalizes the contour basepoint to1 without changing its integral. Thus un(∂D) also has winding one about T.

3.1F2F4step 2.1constructalgebra

For a regular value y of un away from its boundary, its fibre in D is finite. Delete disjoint small disks around these preimages. Round the four rectangle corners in neighborhoods avoiding its finite fibre, and apply C1 Stokes [F4] to the pulled-back closed angular form (−v du+u dv)/(2π(u2+v2)), after translating by y. This gives boundary winding as the sum of the small-circle windings. The corner-arc integrals tend to zero, since the form is bounded there and their lengths tend to zero; passage to the limit recovers the rectangle formula. Interpolation to the invertible derivative on each small circle makes each term sgn⁡Jn there. Hence a regular y∈T has signed fibre count1 and at least one positive-Jacobian preimage. Sard [F2] makes the exceptional target values null. Use the smooth local inverse theorem [F4] and a countable rational-basis cover to partition the open set {Jn>0}∩D∘ into disjoint Borel inverse-branch pieces. Nonnegative change of variables on each branch and countable additivity give ∫D(Jn)+≥∣T∣. Therefore ∫DJn≥∣T∣−∫D(Jn)−; passage to the limit and compact exhaustion of u(D∘) give ∫D∘Ju≥∣u(D∘)∣. The opposite lower area inequality is [F2]. On each chosen basis rectangle, its subrectangles from the same dense good levels, together with the whole rectangle, form a generating pi-system. Both restricted measures are finite and agree there, so [F4] applies with the constant whole-rectangle exhaustion. Countably many such rectangles cover the domain; disjointizing that cover extends equality to all relatively compact Borel sets. Exhaustion gives the area formula wherever needed. In particular u sends null Borel sets to null sets. Apply this argument to both f and the already-regular g. This supplies N and inverse-N without assuming either.

4.1F1F2F3step 3.1algebra

If fz vanished on a positive-area Borel set E inside a compact exhaustion, the Beltrami inequality would make Jf=0 there. Step 3.1 gives ∣f(E)∣=0, while the null-set property of g would force ∣E∣=0, a contradiction. Thus fz≠0 a.e. and Df is nonsingular a.e. The non-differentiability set of g has null preimage under g by the same null-set property, so almost every source point is a common differentiability point. At such a point the chain rule [F3] for g∘f=id⁡ gives 0=(gw∘f)fzˉ+(gwˉ∘f)fz‾ and the corresponding z equation equals1. Division yields μg(f(z))=−fzˉ(z)/fz(z)‾=−μf(z)fz(z)/fz(z)‾.

5.1F1F3step 3.1step 4.1algebra∎

The modulus of the unimodular factor in step 4.1 is one. Null-set preservation in both directions transports the coefficient equality and its essential bounds, so ∥μg∥∞=∥μf∥∞ and Kg=Kf. Step 1.1 already supplies inverse Sobolev regularity, so the coefficient calculation does not circularly assume it. Repeating the result for g proves both directions of the final equivalence.

Remark

The area argument above proves, for every relatively compact Borel set E and either of the independently regular maps u=f or u=f−1, ∣u(E)∣=∫EJu dA. Consequently each map sends area-null Borel sets to null sets, and Jf>0 almost everywhere. The proof first obtains inverse regularity from quadrilateral equivalence, then proves this formula for both maps by ACL-selected rectangles and signed local winding. Neither inverse-null nor a later MRMT result is an input.

LemmaStatement: Literature-sourcedProof: AI-adaptedprecheck passaudited 2026-10-08Open item page →

An analytically quasiconformal homeomorphism distorts quadrilateral moduli by at most K

Sources

  • Christopher J. Bishop, Quasiconformal Mappings, Ch. 2 §§1–2, printed pp. 50–53, for the length–area method and quadrilateral distortion; Ch. 3 §4, printed pp. 94–96, for the differentiability and Jacobian-area estimates.
  • Mikhail Lyubich, Conformal Geometry and Dynamics of Quadratic Polynomials, vol. I, Ch. 2 §11.4, printed pp. 181–182, Proposition 11.14, for the area formula; §11.5, printed p. 183, Proposition 11.18, for total differentiability; §12.1, printed pp. 183–184, Lemma 12.1 and Proposition 12.3; Ch. 1 §6.3.1, printed pp. 121–122, Proposition 6.6 and Exercise 6.8. The full cited passages were reread. Proposition 11.18 is stated but its proof is deferred to Project 11.19, so it does not close the differentiability obligation.
  • F. W. Gehring and O. Lehto, On the total differentiability of functions of a complex variable, Ann. Acad. Sci. Fenn. Ser. A I Math. 272 (1959), pp. 1–9. The original Gehring–Lehto paper was not read. The earlier core supplies a complete fixed-constant maximum proof; the inverse item supplies the signed-degree area correction.

Statement

Assume the Axiom of Choice. Let f:Ω→Ω′ be a homeomorphism of complex domains which is K-quasiconformal in the analytic sense, with K≥1, and let λ,μ denote the extremal length and reciprocal curve-family modulus of Extremal length and the curve-family modulus of a path family.

(i) Quadrilaterals. Let Q be a quadrilateral with Q‾⊂Ω and marked opposite sides E0,E1, as in Orientation-preserving homeomorphisms and the geometric definition of quasiconformality, and let Γ(Q;E0,E1) consist of paths in Q joining those sides. Then fΓ(Q;E0,E1)=Γ(f(Q);f(E0),f(E1)) and K−1μ(Γ(Q;E0,E1))≤μ(Γ(f(Q);f(E0),f(E1)))≤Kμ(Γ(Q;E0,E1)), equivalently K−1λ(Γ(Q;E0,E1))≤λ(Γ(f(Q);f(E0),f(E1)))≤Kλ(Γ(Q;E0,E1)). The same inequalities hold with the other pair of opposite marked sides.

(ii) Annuli. Let A~⊂Ω be a doubly connected domain, meaning that C^∖A~ has exactly two connected components (The Riemann sphere is the published one-point compactification of the complex plane), and put A=f(A~). Let Γ(A~) and Γ(A) be the path families joining the two annular ends. Then K−1μ(Γ(A~))≤μ(Γ(A))≤Kμ(Γ(A~)),K−1λ(Γ(A~))≤λ(Γ(A))≤Kλ(Γ(A~)). For a round annulus A(r,R) the connecting-family extremal length is the conformal parameter (2π)−1log⁡(R/r) and the reciprocal modulus is 2π/log⁡(R/r), by Extremal length of the rectangle and of the round annulus and The conformal parameter of a round annulus is a complete invariant.

(iii) Area and null sets. For every relatively compact Borel set E⊂Ω, area⁡(f(E))=∫EJf dA. In particular, f−1 maps Lebesgue-null Borel subsets of Ω′ to null subsets of Ω.

Annular end-path convention. In (ii), paths have parameter interval (0,1), are proper, and their tails escape into different ends, defined by the two-collar exhaustion proved below. Local rectifiability means rectifiability on each compact parameter interval. Their nonnegative Borel weighted length is the increasing supremum of those compact-restriction arc-length integrals; a nonrectifiable restriction has infinite length. The displayed extremal-length supremum over finite positive-area densities and reciprocal modulus convention are unchanged. The proof verifies compatibility with ordinary finite-boundary families and transport of the two ends, including punctured and infinite rings.

Facts & Assumptions

Given: AC, the analytic homeomorphism and K, quadrilateral or arbitrary doubly connected subdomain, and the stated path families.

[F1]

Both quadrilateral bounds are proved without inverse-null in Analytic quasiconformality gives both quadrilateral modulus bounds. The rectangle and round-annulus numerical constants are Extremal length of the rectangle and of the round annulus; The conformal parameter of a round annulus is a complete invariant fixes the finite-round parameter convention.

[F2]

The independent quadrilateral equivalence gives orientation and same-K inverse regularity (The geometric and analytic definitions of quasiconformality agree, The inverse of a quasiconformal map is quasiconformal with the same dilatation). The auxiliary Remark of the inverse lemma records the proved area equality and both null-set properties for the two already-regular maps; it does not assume inverse-null to obtain regularity.

[F3]

The core's auxiliary Remark supplies the summable-gradient barrier, weighted AC speed identity and one-direction density pullback. At good differentiability points ∥Df∥op2≤KJf by the analytic inequality (The ACL and Sobolev analytic definition of quasiconformality, The Wirtinger derivatives ∂zf and ∂zˉf, and antiholomorphic functions, Analytic quasiconformality gives both quadrilateral modulus bounds).

[F4]

Jordan separation and Schönflies give closed Jordan disks and two-sided boundary collars; open connected planar sets are polygonally connected (Jordan–Brouwer separation, Jordan–Schönflies extension for plane curves, Every connected component of an open subset of Rn is open and polygonally connected). We supply the two-end exhaustion below instead of assuming finite-round uniformization.

Proof

technique · use the earlier quadrilateral core and independent inverse, then compare improper end-path integrals outside a zero-modulus exceptional family
1.1F1F3givenconstruct

For this ring clause, an end path is a proper continuous γ:(0,1)→A~ whose two tails approach different exhaustion ends. It is locally rectifiable if all compact parameter restrictions are rectifiable; define ℓρ(γ)=sup⁡0<a<b<1ℓρ(γ∣[a,b]), setting this to infinity if any such restriction is nonrectifiable. Define λ by the same supremum of ℓρ(Γ)2/A(ρ) over finite positive-area Borel densities and μ=1/λ, including zero/infinite reciprocal conventions. This local end-path convention allows nonlanding tails, punctures and infinity; it requires no finite global Euclidean length. For a globally rectifiable compact physical-boundary path, atomless arc-length measure makes this improper integral equal its ordinary integral. Compact-trace paths of infinite Euclidean length form a zero-modulus family: for traces in a fixed ball, a constant density on that ball has infinite improper length and arbitrarily small area after scaling; take the countable union over balls. Removing this family preserves modulus by the vanishing-cost barrier argument. Thus the finite-boundary family values agree with the compact-path convention even though end paths allow merely local rectifiability. Scaling densities gives the admissible-area characterization exactly as in the compact convention Extremal length and the curve-family modulus of a path family.

2.1F4step 1.1construct

Write the spherical complement of A~ as disjoint compact connected sets K0,K1. Each is nonseparating: the connected domain lies in one component of the open complement of K0; any other component U would lie in the closed K1, while its nonempty boundary lies in K0, contradicting disjointness. For a finite-chart nonseparating continuum K, cover it by finitely many small disks centered on K. Their union is connected because each component meets the covered connected K. Slightly perturb radii within the positive cover margin to avoid the finitely many tangencies/triple intersections. Exposed boundary arcs then form disjoint finite degree-two cycles, hence Jordan curves. Connectedness gives one outer cycle; the others bound holes. Fill those holes to obtain a closed Jordan neighborhood. These neighborhoods can be nested: choose the next disks in the preceding interior, and filling cannot cross the preceding outer boundary. They shrink to K, since each outside point has a path to infinity avoiding K at positive distance, which small enough disk unions miss, placing that point in their exterior. Normalize each Ki in a chart avoiding the other and choose disjoint nested neighborhoods. The closed cores between their Jordan boundaries are compact, nested and exhaust A~. Their complement in the domain has exactly two connected collars: any point in a neighborhood minus K can follow a path avoiding K to its first outer-boundary hit, and all hits connect through the thin inner boundary collar, disjoint from K. Thus each proper tail eventually lies in one consistent collar, and there are exactly two ends. A homeomorphism is proper as a map of these locally compact domains and carries this exhaustion and its two collar components to an exhaustion and two collar components of A. Therefore f induces an end bijection and maps their end-path families bijectively. No extension to individual wild boundary points is asserted.

3.1F2F3step 1.1step 2.1constructalgebra

On a countable compact-neighborhood exhaustion of the source use the local barriers of [F3]. Extend each by zero and multiply by positive constants so their L2 norms have finite sum. Their sum G belongs to L2. Every path with a bad compact restriction has infinite G integral; the family Z of such end paths has modulus zero, since G/m is admissible there with area tending to zero. Outside Z, each image compact restriction is AC and has the weighted speed identity. For a nonnegative Borel target density τ, the local chain rule gives ℓτ(f∘γ∣[a,b])≤ℓ(τ∘f)∥Df∥op(γ∣[a,b]). Increasing compact parameter intervals to (0,1) proves the same inequality for the improper lengths, including infinite values. The lower-area inequality and ∥Df∥op2≤KJf, exhausted over compact source sets, give pullback area at most KA(τ). If τ is target admissible, the pullback is admissible outside Z; add G/m to handle Z and let its L2 norm tend to zero. Taking infima gives μ(Γ(A~))≤Kμ(Γ(A)). This is only one direction.

4.1F1F2step 2.1step 3.1algebra

Apply step 3.1 separately to the independently K-QC inverse in [F2]. Its end correspondence is the inverse of that in step 2.1, so it gives μ(Γ(A))≤Kμ(Γ(A~)). Together the two inequalities give the claimed annular bounds. Reciprocals give the extremal-length bounds even when one value is zero or infinite. The finite-round numerical constants follow from [F1]. This argument uses exhaustion for individual lengths and density areas, not an unproved continuity of modulus under arbitrary ring-core exhaustion.

5.1F1F2step 4.1given∎

The quadrilateral assertion is [F1], and f carries its joining family onto the full image family because it is a homeomorphism on a neighborhood of its compact closure. Area equality on all relatively compact Borel sets is [F2]; applying it to both f and its independently regular inverse proves their null-set preservation. This establishes the entire area and inverse-null clause and completes all three assertions without using any circular-dilatation or MRMT supplier.

TheoremStatement: Literature-sourcedProof: AI-adaptedprecheck passjudge pass (gpt-6.1-sol)audited 2026-10-08Open item page →

Composition and inversion of quasiconformal maps and their Beltrami coefficients

Sources

  • Mikhail Lyubich, Conformal Geometry and Dynamics of Quadratic Polynomials, vol. I, Ch. 2 §12.5, printed p. 188, Proposition 12.15; §11.1 for the composition and inverse coefficient identities.
  • Christopher J. Bishop, Quasiconformal Mappings, Ch. 2 §1, printed pp. 49–51, for the linear dilatation product bound and the general Beltrami chain identity.

Statement

Assume the Axiom of Choice. Let Ω→ g Ω′′→ f Ω′ be analytically quasiconformal homeomorphisms, with g K1-quasiconformal and f K2-quasiconformal (The ACL and Sobolev analytic definition of quasiconformality).

(i) Composition. The composite f∘g is analytically K1K2-quasiconformal, with Kf∘g≤K1K2, and almost everywhere μf∘g(z)=gzˉ(z)+μf(g(z)) gz(z)‾gz(z)+μf(g(z)) gzˉ(z)‾.

(ii) Inverse. The inverse f−1 is K2-quasiconformal, Kf−1=Kf, and μf−1(f(z))=−μf(z)fz(z)fz(z)‾for almost every z.

Consequently quasiconformal homeomorphisms are closed under inverses and composition, and the 1-quasiconformal self-maps of a domain form a group.

Facts & Assumptions

Given: The Axiom of Choice, two analytic quasiconformal homeomorphisms as in the Statement, and their Beltrami representatives.

[F1]

Analytic and geometric quasiconformality agree with the same least constant; geometric quasiconformality is closed under composition because the two modulus inequalities multiply, and orientation signs multiply (The geometric and analytic definitions of quasiconformality agree, Orientation-preserving homeomorphisms and the geometric definition of quasiconformality).

[F2]

The inverse of an analytic K-quasiconformal map is analytic K-quasiconformal with the same maximal dilatation; the proof gives the a.e. inverse Beltrami formula (The inverse of a quasiconformal map is quasiconformal with the same dilatation).

[F3]

The real chain rule holds at common differentiability points. A real-linear map z↦az+bzˉ has Wirtinger coefficients a,b; composition and inversion are calculated by the two Wirtinger equations (The chain rule for total derivatives: D(g∘f)(a)=Dg(f(a))∘Df(a), The Wirtinger chain rule for compositions of real-differentiable complex-valued maps, The Wirtinger derivatives ∂zf and ∂zˉf, and antiholomorphic functions). The Beltrami coefficient and least-dilatation conventions are those of The Beltrami coefficient and the maximal dilatation.

[F4]

Analytic quasiconformal homeomorphisms and their inverses map area-null Borel sets to null sets by the area clause of An analytically quasiconformal homeomorphism distorts quadrilateral moduli by at most K. This allows the exceptional differentiability sets of the factors to be pulled back when applying the a.e. chain rule. The supplier proves the area formula through independent inverse regularity and signed local winding, so this is an earlier proved interface.

[F5]

For 0≤a,b<1 and ∣ω∣=1, ∣a0+ωb01+a0‾ωb0∣≤∣a0∣+∣b0∣1+∣a0∣∣b0∣(a0,b0∈C, ∣a0∣=a, ∣b0∣=b). Indeed, the squared ratio is (a2+b2+2abcos⁡θ)/(1+a2b2+2abcos⁡θ), increasing in cos⁡θ; its maximum is at cos⁡θ=1. Moreover, if t=(a+b)/(1+ab), then (1+t)/(1−t)=((1+a)/(1−a))((1+b)/(1−b)).

Proof

technique · compose the geometric modulus inequalities, then compute the almost-everywhere Beltrami chain rule
1.1F1givenalgebra

By [F1], f and g are geometrically quasiconformal with constants K2 and K1. Applying their two-sided modulus bounds successively to any quadrilateral gives the two-sided bound with constant K1K2 for f∘g; the orientation signs multiply, so the composite is geometrically K1K2-quasiconformal. The equivalence in [F1] makes it analytically K1K2-quasiconformal.

2.1F2F3F4step 1.1given

By [F2], g=f−1 in the inverse case has the same analytic maximal dilatation and the stated inverse coefficient identity. For the composition formula, take the full-measure set where g is differentiable, f is differentiable at g(z), and the weak derivatives agree with the classical derivatives. The exceptional set for f pulls back to a null set by [F4]. The composite is analytic by step 1.1, so its weak and classical derivatives also agree almost everywhere.

3.1F2F3F4F5step 1.1step 2.1given∎

At each point of the common set, the real chain rule gives (f∘g)z=fw(g)gz+fwˉ(g)gzˉ‾,(f∘g)zˉ=fw(g)gzˉ+fwˉ(g)gz‾. Writing μf(g)=fwˉ(g)/fw(g) and dividing the second equation by the first yields the displayed formula in (i); the denominator is nonzero almost everywhere because each analytic quasiconformal homeomorphism has positive Jacobian almost everywhere. Put ν=μg(z), μ=μf(g(z)), and ω=gz‾/gz, so ∣ω∣=1. The composition formula becomes μf∘g=ν+ωμ1+ων‾μ. By [F5], ∣μf∘g∣≤(k1+k2)/(1+k1k2), where kj=(Kj−1)/(Kj+1). The identity in [F5] converts this to Kf∘g≤K1K2, consistent with step 1.1. Part (ii) and the group assertion follow from [F2] and the identity map's coefficient 0.

TheoremStatement: Literature-sourcedProof: AI-adaptedprecheck passjudge pass (gpt-6.1-sol)audited 2026-10-08Open item page →

Every 1-quasiconformal homeomorphism is conformal

Statement

Assume the Axiom of Choice. Let Ω,Ω′⊆C be complex domains (A complex domain is a nonempty connected open subset of C) and f:Ω→Ω′ a homeomorphism. The following are equivalent.

(a) f is 1-quasiconformal, in either the geometric or the analytic sense (Orientation-preserving homeomorphisms and the geometric definition of quasiconformality, The ACL and Sobolev analytic definition of quasiconformality, The geometric and analytic definitions of quasiconformality agree).

(b) f∈Wloc1,2(Ω) and its weak Wirtinger derivative satisfies ∂zˉf=0 almost everywhere; under this Sobolev hypothesis this is equivalent to μf=0 almost everywhere together with ∂zˉf=0 almost everywhere on {∂zf=0} (The ACL and Sobolev analytic definition of quasiconformality, The Beltrami coefficient and the maximal dilatation).

(c) f is holomorphic; equivalently, it is a biholomorphism of Ω onto Ω′ (Biholomorphic maps between complex domains).

Thus the conformal maps in this library's orientation-preserving sense are exactly the 1-quasiconformal homeomorphisms; in particular, they preserve angles.

Facts & Assumptions

Given: Choice, complex domains Ω,Ω′, and a homeomorphism f:Ω→Ω′.

[F1]

The geometric and analytic definitions have the same least dilatation. In the analytic class, Kf=1 iff μf=0 almost everywhere; the defining inequality then gives fzˉ=0 almost everywhere. Conversely, if f is analytic quasiconformal and fzˉ=0 almost everywhere, then its Beltrami coefficient is zero (including the set where fz=0), so Kf=1 (The ACL and Sobolev analytic definition of quasiconformality, The Beltrami coefficient and the maximal dilatation, The geometric and analytic definitions of quasiconformality agree).

[F2]

Distributional derivatives commute, and Δ=4∂z∂zˉ on distributions (Linearity, locality, and commutation of weak derivatives, Distributional harmonicity and Poisson's equation on an open subset of Rn).

[F3]

A locally integrable distribution with zero Laplacian has a smooth harmonic representative; if the original function is continuous, it equals that representative everywhere (Weyl's lemma for the Laplacian, Locally integrable weakly harmonic functions are smooth).

[F4]

For a smooth function, the Cauchy–Riemann equation ∂zˉf=0 is equivalent to holomorphy. An injective holomorphic map on a complex domain has nowhere-vanishing derivative and a holomorphic inverse onto its open image (Complex differentiability is equivalent to real total differentiability together with a complex-linear derivative, with ∂zˉf=0, or with the Cauchy–Riemann equations, An injective holomorphic map has no critical point and is biholomorphic onto its image).

[F5]

At a differentiability point, a real-linear derivative given by multiplication by a nonzero complex number has determinant ∣f′(z)∣2>0 and hence preserves the local orientation (The Wirtinger derivatives ∂zf and ∂zˉf, and antiholomorphic functions, A real linear isomorphism preserves or reverses orientation according to the sign of its determinant, Smooth orientation sign is the local integral homology multiplier).

[F6]

A one-to-one holomorphic map preserves extremal length of every path family by conformal invariance, and thus satisfies the geometric modulus inequalities with constant 1 (Conformal invariance, monotonicity, and the series and parallel laws for extremal length, Orientation-preserving homeomorphisms and the geometric definition of quasiconformality).

Proof

technique · reduce the unit-dilatation condition to the weak Cauchy–Riemann equation, apply Weyl's lemma, and use conformal invariance for the converse
1.1F1given

By [F1], an analytically 1-quasiconformal map satisfies (b) and has μf=0 almost everywhere. If the hypothesis in (a) is geometric, the equivalence theorem first supplies the analytic condition with the same constant. Conversely, (b) is exactly the analytic 1-quasiconformal condition, so its least constant is 1. Under the Sobolev hypothesis, μf=0 forces fzˉ=0 off {fz=0}; the additional condition on that set gives fzˉ=0 almost everywhere, proving the coefficient reformulation in (b).

1.2F2given

Assume (b). The map f is continuous, hence locally integrable, and its weak Wirtinger derivative vanishes as a distribution. By [F2], Δf=4∂z(∂zˉf)=0 distributionally, componentwise.

2.1F3F4step 1.2given

By [F3], f agrees almost everywhere with a smooth harmonic function. The representative is actually f everywhere: the difference of two continuous functions that vanishes almost everywhere must vanish everywhere, since any point where it were nonzero would have a neighborhood of positive area where it remained nonzero. Thus f is smooth and harmonic. Its classical ∂zˉf is continuous and represents the zero distribution, so it vanishes pointwise; [F4] gives that f is holomorphic. Since f is injective, [F4] also shows its inverse is holomorphic onto its open image; surjectivity identifies that image with Ω′. Therefore f is a biholomorphism, proving (b)⇒(c).

3.1F1F4F5F6step 1.1step 1.2step 2.1given∎

Assume (c). Then f is injective and holomorphic, so [F4] gives f′(z)≠0 everywhere and a holomorphic inverse. By [F5], f preserves orientation. By [F6], it preserves the modulus of every quadrilateral family exactly, hence is geometrically 1-quasiconformal; the equivalence theorem in [F1] makes it analytically 1-quasiconformal as well. This proves (c)⇒(a); steps 1.1 and 1.2 prove (a)⇒(b), and step 2.1 proves (b)⇒(c).

TheoremStatement: Literature-sourcedProof: AI-adaptedprecheck passaudited 2026-10-08Open item page →

Compactness of the normalized K-quasiconformal self-maps of the sphere

Statement

Assume the Axiom of Choice. Fix K≥1 and let C^ carry its chordal metric χ (The chordal metric on the Riemann sphere, The chordal metric induces the standard topology of the Riemann sphere, The Riemann sphere is the published one-point compactification of the complex plane). Let FK be the set of orientation-preserving, K-quasiconformal homeomorphisms f:C^→C^ satisfying f(0)=0, f(1)=1, and f(∞)=∞, where quasiconformality is understood in the geometric sense of Orientation-preserving homeomorphisms and the geometric definition of quasiconformality and equivalently in the analytic sense by The geometric and analytic definitions of quasiconformality agree. Then:

(i) FK is equicontinuous in χ: for every ε>0 there is δ>0 such that χ(f(z),f(w))<ε whenever χ(z,w)<δ, for every f∈FK.

(ii) Every sequence in FK has a subsequence converging uniformly on C^ to an element of FK. Thus FK is compact in the uniform topology.

(iii) If fn are orientation-preserving quasiconformal sphere homeomorphisms, fn→f uniformly in χ, and f is a homeomorphism, then its maximal dilatation, assigned Kf=+∞ when f is not quasiconformal, is lower semicontinuous: Kf≤lim inf⁡n→∞Kfn.

The normalization is essential: it removes the noncompact Möbius freedom. The compactness and lower-semicontinuity claims apply equally to the equivalent analytic class.

Facts & Assumptions

Given: AC, normalized orientation-preserving sphere homeomorphisms, and a common K when proving compactness.

[F1]

Geometric and analytic constants agree; inverses have the same constant. The full area formula gives ∫EJf=∣f(E)∣ and its weighted version by simple approximation (The geometric and analytic definitions of quasiconformality agree, The inverse of a quasiconformal map is quasiconformal with the same dilatation, An analytically quasiconformal homeomorphism distorts quadrilateral moduli by at most K).

[F2]

Stereographic coordinates have conformal scale 2/(1+∣w∣2); this follows by differentiating the explicit map in Stereographic projection identifies the Riemann sphere with the unit two-sphere. Its area weight is 4/(1+∣w∣2)2, whose total planar integral is 4π by polar coordinates. Chordal distance is the Euclidean distance of the sphere images (The chordal metric on the Riemann sphere).

[F3]

The core's exceptional-curve argument supplies AC and weighted speed on almost every circular leaf after a smooth polar coordinate change. Jordan separation identifies the small side of a loop lying in a sufficiently small spherical cap (Analytic quasiconformality gives both quadrilateral modulus bounds, Jordan–Brouwer separation).

[F5]

Real L2 is a Hilbert space (L2 with the integral pairing is a Hilbert space). The four matrix entries form a finite Hilbert direct sum: its sum-of-squares inner product is complete because each component is complete. Hilbert spaces are reflexive and bounded sequences have weakly convergent subsequences under the stated AC consequences. Convex norm-continuous functionals are weakly lower semicontinuous (Hilbert spaces are reflexive by Riesz representation, Reflexivity is equivalent to weak subsequential compactness of bounded sequences, A convex norm-lower-semicontinuous functional is weakly lower semicontinuous). We apply this to the weighted integral of the squared matrix operator norm on real L2 matrix fields; it is convex and norm-continuous by the matrix norm triangle inequality and Cauchy–Schwarz. Interior mollification and Lp approximate identities give derivative convergence; for continuous functions they converge uniformly on compacta (Interior mollification commutes with weak derivatives, Every L1 approximate identity converges to the identity in Lp for 1≤p<∞, L1 approximate identities converge uniformly on compacta for bounded continuous functions). The AC fundamental theorem is Fundamental theorem of calculus for absolutely continuous functions. The general Sobolev/ACL equivalence is The ACL characterisation of W1,p, applied before asserting quasiconformality.

Proof

technique · obtain equicontinuity by weighted spherical energy and short circles, then preserve the exact analytic bound by weak derivatives and distributional Jacobians
1.1F1F2F3givenalgebra

A normalized map fixes infinity and is a plane homeomorphism on the finite chart. In either of the two bounded stereographic source charts, [F1]–[F2] and ∥Df∥op2≤KJf give total weighted energy at most 4πK: integrate the target weight 4/(1+∣f∣2)2 against Jf. For a circle of source radius r centered at x, let L(r) be its spherical image length. Almost every circle is AC by [F3], and Cauchy–Schwarz gives L(r)2≤2πr∫∣z−x∣=r4∥Df∥op2/(1+∣f∣2)2 ds. Integrating with dr/r between δ and δ yields ∫L(r)2dr/r≤8π2K. Hence one such circle has L(r)≤4πK/log⁡(1/δ), uniformly in f and the center x.

2.1F1F2F3step 1.1construct

The three fixed points have positive minimum pairwise chordal distance c. At each x at least two of them stay a fixed positive distance from x; a uniformly small source disk avoids those two. Use the short circle from step 1.1 surrounding the smaller disk of radius delta. Its image lies in a spherical cap of diameter at most twice its length. For sufficiently small delta this cap cannot contain both avoided fixed points. The complement of the cap is connected, so Jordan separation makes one image complementary component lie inside the cap and the other contain its exterior. The image of the source disk cannot be the latter component, because it would contain at least one of the two fixed points which the source disk avoids. Thus its diameter tends uniformly to zero. Euclidean and chordal distances are uniformly comparable on the two bounded source charts, proving common chordal equicontinuity. The inverse maps are normalized and have the same K by [F1], so the identical argument gives their equicontinuity.

3.1F4step 2.1construct

Apply [F4] to a sequence and then its inverses on the obtained subsequence. We obtain uniform limits f and g. Uniform convergence and continuity show g(f(z))=z and f(g(z))=z, so f is a homeomorphism with inverse g and fixes the three points. It preserves orientation: uniformly close sphere maps are homotopic by normalized straight-line interpolation of their unit-sphere values; homotopy invariance of the sphere degree preserves degree one, and for a homeomorphism this is the positive local orientation sign (The singular chain homotopy formula, Global sphere degree is the sum of local degrees).

4.1F1F5step 3.1constructalgebra

We prove closure with the exact K, independently of circular-dilatation or quadrilateral-modulus continuity. On any compact source patch choose a target chart avoiding its compact f-image complement point. Uniform convergence gives bounded finite coordinate values for f_n there for large n. The area formula and distortion bound give a uniform local L2 derivative bound. By [F5], pass to weak L2 derivative limits on a smaller patch; uniform convergence and integration against test functions identify them as Df. Write fn=un+ivn. The distributional identity Jfn=∂x(un∂yvn)−∂y(un∂xvn) follows by smooth approximation and commutation of mixed weak derivatives. Uniform convergence of u_n and weak L2 convergence of the derivatives of v_n show that these Jacobians converge distributionally to Jf. For every nonnegative smooth compactly supported phi, weak lower semicontinuity and the bound on f_n give ∫ϕ∥Df∥op2≤lim inf⁡∫ϕ∥Dfn∥op2≤Klim⁡∫ϕJfn=K∫ϕJf. Hence ∥Df∥op2≤KJf almost everywhere. If J_f is zero this forces Df zero; otherwise the singular-value ratio is at most K, equivalently the analytic Beltrami bound. The general ACL characterization applies to the resulting W1,2 class, and continuity identifies f pointwise with its ACL representative on almost every line. Thus f is analytically K-QC by its definition and geometrically K-QC by [F1], proving closure.

5.1F1F4F5step 3.1step 4.1algebra∎

The subsequential limit is therefore in the normalized family. Sequential compactness and the supremum chordal metric give compactness by [F4]. For general uniformly convergent quasiconformal homeomorphisms with a homeomorphic limit, if L=lim inf⁡Kfn<∞, take a subsequence whose constants tend to L. The local derivative argument in step 4.1 uses the uniformly bounded constants and passes their limit to give ∥Df∥op2≤LJf. It follows that Kf≤L; if L is infinite the inequality is automatic. Orientation is preserved by the same homotopy argument. This proves the full lower-semicontinuity claim without a normalization assumption on that sequence.

LemmaStatement: Literature-sourcedProof: AI-adaptedprecheck passaudited 2026-10-08Open item page →

Circular dilatation, quasisymmetry and the analytic definition

Sources

  • Mikhail Lyubich, Conformal Geometry and Dynamics of Quadratic Polynomials, vol. I, Ch. 2 §§12.2–12.5, printed pp. 185–188. Lemma 12.6 bounds the macroscopic circular dilatation of an analytic QC map by the annular modulus inequality; Proposition 12.7 records that bound. Lemma 12.11 and Proposition 12.13 give compact-set and normalized quasisymmetry. Proposition 12.14 gives an ACL* argument from bounded upper circular dilatation. Its p. 188 statement omits orientation preservation, although §12.5 defines quasiconformality for orientation-preserving homeomorphisms; the reflection h(z)=zˉ has circular dilatation 1 but fails the library's analytic Beltrami inequality, so part (iii) includes the orientation hypothesis explicitly.
  • The same volume, Ch. 2 §11.4, printed pp. 181–182, Proposition 11.14, gives the Jacobian-area estimates; §11.5, printed p. 183, states Proposition 11.18 for total differentiability but labels its proof as Project 11.19, “Fill in details.” The cited project is not used as a proof: the earlier quadrilateral core independently supplies total differentiability.
  • Christopher J. Bishop, Quasiconformal Mappings, Ch. 3 §4, printed pp. 91–96, for ACL, differentiability, and the Jacobian area estimates. The proof of Theorem 4.2 is not used as certified evidence here; its maximum argument is an open source obligation.

Statement

Assume the Axiom of Choice. Let h:U→V be a homeomorphism between plane domains. For z∈U and r>0 with D(z,r)‾⊂U, set Mh(z,r)=max⁡∣w−z∣=r∣h(w)−h(z)∣,mh(z,r)=min⁡∣w−z∣=r∣h(w)−h(z)∣, and define the macroscopic circular dilatation Dil⁡(h,z,r)=Mh(z,r)mh(z,r),Dil⁡(h,z)=lim sup⁡r↓0Dil⁡(h,z,r),Dil⁡(h)=sup⁡z∈UDil⁡(h,z).

(i) If h is analytically K-quasiconformal, then Dil⁡(h)≤eCK for an absolute constant C. More precisely, for every z∈U there is r0(z)>0 such that whenever 0<r<r0(z) and D(z,r)‾⊂U, the inner and outer radii s,R of h(D(z,r)) about h(z) satisfy log⁡(R/s)≤CK.

(ii) If Dil⁡(h)<∞, then h is quasisymmetric on compact subsets, with control depending on Dil⁡(h) and the relative distances of the compact set and its image from the domain boundaries. In particular, a K-quasiconformal homeomorphism is locally quasisymmetric with control depending only on K and these distances, and its inverse has the corresponding inverse control function.

(iii) If h is orientation-preserving and Dil⁡(h)≤L<∞, then h is analytically L-quasiconformal: it is ACL on almost every horizontal and vertical line, belongs to Wloc1,2, is differentiable almost everywhere, and satisfies ∣hzˉ∣≤L−1L+1∣hz∣a.e.

The metric assertions (i) and (ii) do not need an added orientation premise; the analytic conclusion (iii) requires orientation preservation, since complex conjugation has circular dilatation one. The constants in (i) and (ii) are not sharp; the Beltrami constant in (iii) follows from the pointwise differential eccentricity bound.

Facts & Assumptions

Given: The Axiom of Choice, plane domains U,V, the homeomorphism h, and the circular dilatations in the Statement. Part (iii) also assumes the orientation-preserving condition of Orientation-preserving homeomorphisms and the geometric definition of quasiconformality.

[F1]

At a differentiability point with nonsingular derivative, the limsup circular ratio equals the derivative singular-value ratio. A rank-one derivative forces that ratio to infinity by testing kernel and transverse directions, while a zero derivative already satisfies the Beltrami inequality. Thus once the necessary ACL/weak regularity is established, the sharp analytic L bound follows from the circular bound and the Wirtinger identities (The Wirtinger derivatives ∂zf and ∂zˉf, and antiholomorphic functions).

[F2]

The full earlier ring theorem gives both annular end-family distortion bounds (An analytically quasiconformal homeomorphism distorts quadrilateral moduli by at most K). The round-annulus value is λ(A(s,R))=(2π)−1log⁡(R/s) (Extremal length of the rectangle and of the round annulus). The continuum-circle argument below supplies the required absolute source-ring bound locally; no cited annulus-geometry lemma is used. The circle estimate uses completed-product Fubini and Jordan separation (Tonelli and Fubini for the completed product, with only almost-everywhere section measurability, Jordan–Schönflies extension for plane curves, Jordan–Brouwer separation).

[F3]

A map is η-quasisymmetric when ∣x−a∣≤t∣x−b∣ implies ∣h(x)−h(a)∣≤η(t)∣h(x)−h(b)∣ for distinct triples, where η:[0,∞)→[0,∞) is an increasing homeomorphism with η(0)=0. Reversing this ratio inequality gives the inverse control η∗(t)=1/η−1(1/t) for t>0, with η∗(0)=0. Step 2.1 proves the local control needed here. Lyubich, §12.3, supplies the convention; Lemma 12.11 concerns embeddings of the whole Euclidean space with an all-scale circular bound, rather than this compact-set statement.

[F4]

The sole delegated citation input is qualitative: an orientation-preserving homeomorphism between finite planar domains with bounded upper infinitesimal circular dilatation everywhere is analytically quasiconformal with some finite constant. This is Gehring, Definitions for a Class of Plane Quasiconformal Mappings (1967), §9 Definition8′, printed p.179, with §3 Definitions2/2′, p.176 and the orientation/domain convention on p.175. The complete ten-page paper was read; §14 points elsewhere for the equivalence proof, so this is a statement citation under the exact root-recorded last-resort authorization, not a claim that this paper supplies that proof. The sharp L bound, area arguments, modulus comparison and quasisymmetry estimates below are local. Total differentiability after this qualitative regularity is the earlier independently proved core Remark (Analytic quasiconformality gives both quadrilateral modulus bounds).

Proof

technique · use only the authorized qualitative metric criterion, derive its sharp bound locally, and prove analytic circular/quasisymmetry control by continuum-circle modulus and dyadic shrink
1.1F1F4givenalgebra

Assume h preserves orientation and its upper circular dilatation is at most L everywhere. The exact qualitative citation [F4] gives ACL, local W1,2 and finite analytic distortion. The general differentiability argument in the core Remark applies, without assuming a sharp analytic constant. At every nonsingular differentiability point, [F1] identifies the circular limsup with the singular-value ratio, bounded by L. A rank-one derivative would force an infinite circular ratio and is excluded; rank zero satisfies the Beltrami inequality. Thus ∣hzˉ∣≤(L−1)(L+1)−1∣hz∣ a.e., and the already-obtained W1,2 regularity makes h analytically L-quasiconformal. This proves (iii); no sharp bound is obtained merely by citing [F4].

1.2F2givenconstructalgebra

Fix a source point x and a sufficiently small radius r so that the disk of outer radius R=Mh(x,r) about h(x) lies inside V; put s=mh(x,r). The preimage of the round ring centered at h(x) with radii s,R has inner Jordan continuum E containing x and a point at distance r from x, and outer closed Jordan exterior F containing a point at distance r and infinity. Thus diam(E) is at least r and d=dist⁡(E,F)≤2r. Choose a closest pair e0,f0 and its midpoint c. There is e in E with ∣e−e0∣≥r/2; a compact path in the Jordan exterior F from f0 toward infinity gives an analogous far point. Since the pair is closest, ∣e−f0∣≥d, whence ∣e−c∣2≥d2/4+∣e−e0∣2/2≥d2/4+r2/8, and likewise on F. Connectedness makes every circle centered at c with radius between d/2 and d2/4+r2/8 meet both continua. One complementary circle arc has one endpoint on each and lies inside the ring, so an admissible density has integral at least one on that whole circle. Cauchy–Schwarz and polar Fubini give source modulus at least 14πlog⁡(1+r2/(2d2))≥c0:=14πlog⁡(9/8)>0. If R=s the ratio is one and no ring is needed. Otherwise [F2] gives (2π)−1log⁡(R/s)≤K/c0. Thus R/s≤eCK with the absolute constant C=2π/c0. The closest pair exists because E is compact and F closed; the exterior path and circle-arc separation use Jordan–Schönflies. For a fixed compact source set, choose the small radii uniformly by continuity on a larger compact neighborhood and its positive image distance from the target boundary. This proves (i).

2.1F1F3step 1.1step 1.2givenconstructalgebra∎

For an analytic map, step 1.2 gives a uniform small-scale circular bound H on each compact neighborhood. Openness makes the maximum image distance in a closed source ball occur on its boundary. Put d=Lx(r/2) and choose y on that inner circle realizing d, with midpoint yprime of x,y. The equal-radius comparison at yprime gives d≤(H+1)∣h(y)−h(yprime)∣, and the comparison at y gives ∣h(y)−h(yprime)∣≤Hly(r/4). The image of B(y,r/4) contains the disk of its inner radius, is inside h(B(x,r)), and is centered a distance d from h(x). Hence Lx(r)≥(1+c)d with c=1/[H(H+1)]. Iteration gives Lx(tr)≤CHtαlx(r) for 0<t≤1, where α=log⁡2(1+c)>0. The reverse doubling bound is Lx(2r)≤(H+1)Lx(r): take a maximizing point at radius2r, its midpoint, and compare the two equal-radius increments about that midpoint. Iteration supplies an increasing power bound for t at least1. These give the local quasisymmetry control in the analytic case; finite compact collars extend the control over the compact set. The inverse control formula is η∗(t)=1/η−1(1/t) by reversing the ratio inequality. For a general bounded-circular map, apply step 1.1; when its orientation is reversed first postcompose with complex conjugation, which preserves every distance ratio and changes the orientation. The resulting analytic constant is L, so step 1.2 and the same shrink argument apply. On each compact collar the small-scale estimate handles triples of sufficiently small diameter; for the remaining triples the fixed positive minimum image separation completes a continuous control function. The input/output collar data enter this compact control. If the domains are the whole plane, step 1.2 has no small-radius restriction, and the shrink and doubling inequalities hold at all scales with control depending only on K. This proves the claimed local metric control and its inverse formula; the global-plane conclusion uses no domain-boundary data.

Remark

For an analytically K-quasiconformal homeomorphism of the whole plane, the continuum-circle proof has no boundary restriction. Put H=eCK, c=1/[H(H+1)], α=log⁡2(1+c) and β=log⁡2(H+1). The proved all-scale shrink and doubling inequalities give a global Euclidean quasisymmetry control η(t)=CHtα for 0≤t≤1 and η(t)=CHtβ for t≥1, with CH=Hmax⁡(1+c,H+1). This conclusion uses the local analytic modulus argument; the delegated citation is only the initial qualitative criterion for a general metric map.

5 · Examples, counterexamples and false statements

None yet.

Sources