The boundary correspondence of a conformal map #
A conformal map — a holomorphic injection f on an open set U ⊆ ℂ — is a proper map onto its
image: the preimage of a compact subset of f '' U is a compact subset of U. Properness is the
mechanism that forces the boundary to go to the boundary, and it is the first step of the
Carathéodory boundary correspondence, layer L5 of the conformal-mapping roadmap. This file
proves it, deduces that a point approached from inside U is carried out of every compact subset
of f '' U, and packages the consequences for a map that does extend continuously to closure U:
such an extension carries frontier U into frontier (f '' U), onto it once U is in addition
bounded, and — when the extension is injective on closure U — is a homeomorphism of the closures.
The existence of the continuous extension is the hard, hypothesis-laden half of Carathéodory's theorem (it needs the boundary of the image to be a Jordan curve, or at least locally connected), and it is not proved here; the L5 milestone is that existence statement. What is proved here is everything that holds unconditionally, plus the packaging that turns the extension, once it is available, into the boundary homeomorphism the milestone asks for.
The argument #
Properness is the inverse function theorem in disguise. An injective holomorphic map on an open set
is an open partial homeomorphism onto its image
(DifferentiableOn.toOpenPartialHomeomorph), and U ∩ f ⁻¹' K is exactly the image of K
under its inverse Function.invFunOn f U, so it is compact as the continuous image of a compact
set. Everything else follows from that one fact. If z i → w with z i ∈ U and w ∉ U, then
f (z i) cannot lie in a compact K ⊆ f '' U frequently: otherwise z i would frequently lie in
the compact — hence closed — set U ∩ f ⁻¹' K, which would then contain the limit w.
For a boundary point w ∈ frontier U, the filter 𝓝[U] w is the one to run this along: it is
NeBot precisely because w ∈ closure U. If a continuous extension F had F w inside the open
set f '' U, then a small closed ball around F w would be a compact subset of f '' U that
f = F enters eventually along 𝓝[U] w by continuity, and leaves eventually by properness —
impossible on a nontrivial filter. Hence F w ∉ f '' U, while F w ∈ closure (f '' U) because F
is continuous on closure U; that is membership in frontier (f '' U).
The argument beyond the two holomorphic inputs — the open mapping theorem, through
TauCeti.isOpen_image_of_differentiableOn_of_injOn, and its packaging as an open partial
homeomorphism — is purely topological, and no continuity hypothesis on the inverse of f appears
precisely because those two supply it. In accordance with the generality bar of
ConformalMapping/README.md, which fixes scalar ℂ for every theorem added in layers L0–L6, the
results are stated for conformal maps of ℂ rather than for an abstract proper map.
Main results #
TauCeti.isCompact_inter_preimage_of_differentiableOn_of_injOn— a conformal map is proper onto its image:U ∩ f ⁻¹' Kis compact for compactK ⊆ f '' U.TauCeti.eventually_notMem_of_tendsto_of_notMem— a family inUconverging to a point outsideU, in particular to a boundary point, is carried out of every compact subset of the image.TauCeti.notMem_image_of_mem_frontier— a continuous extension toclosure Usendsfrontier Uoutsidef '' U.TauCeti.image_closure_eq_closure_image— for boundedU, a continuous extension carriesclosure Uontoclosure (f '' U).TauCeti.image_frontier_subset_frontier_image— a continuous extension carriesfrontier Uintofrontier (f '' U); no boundedness is needed.TauCeti.image_frontier_eq_frontier_image— for boundedU, that inclusion is an equality: a continuous extension carriesfrontier Uontofrontier (f '' U).TauCeti.bijOn_closure_closure_imageandTauCeti.closureHomeomorph— for boundedU, an extension injective onclosure Uis a homeomorphismclosure U ≃ₜ closure (f '' U).
Coordination with upstream Mathlib #
Mathlib has no boundary correspondence for conformal maps, and — unlike the L0–L3 material — layer
L5 is absent from mathlib4#33505, the
in-progress human-curated Riemann-mapping-theorem effort, which stops at the mapping theorem itself.
So this file is new Lean formalization rather than a temporary shim. It does consume the L0–L3 shims
TauCeti.isOpen_image_of_differentiableOn_of_injOn and
DifferentiableOn.toOpenPartialHomeomorph, which are to be refactored onto Mathlib once the
upstream work lands.
References #
- C. Carathéodory, Über die gegenseitige Beziehung der Ränder bei der konformen Abbildung, Math. Ann. 73 (1913).
- J. B. Conway, Functions of One Complex Variable I (GTM 11), Ch. IX.
- Ch. Pommerenke, Boundary Behaviour of Conformal Maps, Ch. 2.
Properness #
A conformal map is proper onto its image. If f is holomorphic and injective on an open
set U and K ⊆ f '' U is compact, then the part of the preimage of K lying in U is compact.
Properness is what distinguishes f '' U as the whole image: the preimage cannot escape to the
boundary of U while its image stays in a compact part of f '' U.
A conformal map carries a family converging out of U out of every compact subset of the
image. If z i ∈ U eventually, z i → w and w ∉ U, then eventually f (z i) avoids any
compact K ⊆ f '' U.
Continuous extensions to the closure #
A continuous extension of a conformal map sends the boundary off the image. If F is
continuous on closure U and agrees with f on U, then F maps no boundary point of U into
the open set f '' U.
For open U the hypothesis w ∈ frontier U is w ∈ closure U together with w ∉ U; both halves
are used, the first to make the filter 𝓝[U] w nontrivial and the second to invoke
TauCeti.eventually_notMem_of_tendsto_of_notMem.
A continuous extension of a conformal map carries the closure onto the closure of the
image, for a bounded U. Only boundedness and continuity are used; no holomorphy hypothesis
appears.
A continuous extension of a conformal map carries the boundary into the boundary. No boundedness is needed.
A continuous extension of a conformal map carries the boundary onto the boundary, for a
bounded U. No injectivity of the extension is needed, only the injectivity of f on U that
makes it conformal: the inclusion of TauCeti.image_frontier_subset_frontier_image is an equality
because a point of frontier (f '' U) lies in closure (f '' U) = F '' closure U, so it is F z
for some z ∈ closure U, and z ∈ U is impossible — it would put F z = f z back inside the open
set f '' U, which frontier (f '' U) avoids.
The boundary homeomorphism #
An injective continuous extension is a bijection of the closures. If F is continuous on
closure U for a bounded U, agrees with f on U, and is injective on closure U, then it maps
closure U bijectively onto closure (f '' U). The set-level form of
TauCeti.closureHomeomorph: injectivity is the hypothesis, and surjectivity onto
closure (f '' U) is TauCeti.image_closure_eq_closure_image. Neither holomorphy of f nor
openness of U is needed; a conformal f is the intended application.
The homeomorphism of closures induced by an injective continuous extension. If F is
continuous on closure U for a bounded U, agrees with f on U, and is injective on
closure U, then F is a homeomorphism of closure U onto closure (f '' U). Neither holomorphy
of f nor openness of U is assumed; the intended instance is a conformal map f on an open U
together with a continuous extension F.
For a Riemann map of a Jordan domain this is the conclusion the Carathéodory correspondence
(layer L5) asks for; what that milestone adds is the existence of such an F, which is not
proved here. Continuity of the inverse is free: closure U is compact and ℂ is Hausdorff.
The definition is not exposed; TauCeti.coe_closureHomeomorph_apply and
TauCeti.coe_closureHomeomorph_symm_apply are its characterizations.
Equations
- TauCeti.closureHomeomorph hUb hFc hFf hFi = ⋯.homeoOfEquivCompactToT2
Instances For
The boundary homeomorphism is the extension F itself.
The inverse of the boundary homeomorphism is the set-level inverse
Function.invFunOn F (closure U).