Round holes for every planar domain

← All digests

Family 071 · Real and complex analysis
Importance 5/5 · Provisional editorial rating
Sources consulted 2026-10-14

1. The question

Imagine a sheet of paper with holes of wildly different shapes, perhaps infinitely many holes accumulating everywhere along a complicated boundary. Can we change coordinates so that every hole becomes a round disk or a single point, while preserving angles inside the remaining region?

Koebe’s circle-domain conjecture asks precisely this question in complex analysis. Its answer is familiar when there is only one ordinary hole, but infinitely many interacting boundary components introduce a different difficulty: good approximations can develop bad boundaries in the limit.

Family 071 contains two manuscripts. The uniformization paper claims that every planar domain has a circle-domain model, with no restriction on the number or geometry of its complementary components. The rigidity paper claims that a circle domain with conformally removable boundary has only Möbius-equivalent circle-domain models. Existence and uniqueness are separate questions, and the second answer has an additional hypothesis.

2. Background and essential definitions

Domains and conformal coordinates

A domain is a nonempty connected open set. It is convenient to work on the Riemann sphere $\widehat{\mathbb C}=\mathbb C\cup\{\infty\}$, which adds one point at infinity to the complex plane. A conformal equivalence is a bijective holomorphic map with holomorphic inverse, interpreted in local coordinates near infinity. Its derivative multiplies all small directions at a point by the same nonzero complex number, preserving angles while allowing lengths to change.

A circle domain is a domain whose complementary connected components are closed round disks or individual points. “Round” is understood on the sphere, so a disk can contain infinity. The complement can also be empty. An annulus is a basic example, and a disk punctured at countably many isolated points gives a more complicated one.

The Riemann mapping theorem says that a simply connected proper plane domain other than the whole plane is conformally equivalent to the unit disk. It does not directly solve the many-hole problem. A conformal equivalence preserves the domain’s topology, so an annulus cannot become a disk; its inner and outer boundary components must both be represented.

For an annulus

$$A_r=\{z\in\mathbb C:r<|z|<1\},\qquad 0<r<1,$$

the ratio encoded by $r$ measures conformal information that cannot be removed by choosing coordinates. Roundness simplifies the description without erasing all geometry.

Removability and rigidity

A compact set $E$ on the sphere is conformally removable if every orientation-preserving homeomorphism of the sphere that is conformal outside $E$ must be Möbius. A Möbius transformation has the form

$$M(z)=\frac{az+b}{cz+d},\qquad ad-bc\ne0,$$

with the usual conventions at poles and infinity. These are exactly the orientation-preserving conformal automorphisms of the sphere.

A circle domain $\Omega$ is conformally rigid if every conformal equivalence from $\Omega$ to another circle domain is the restriction of such a transformation. The distinction matters: removability starts with a homeomorphism already defined on the whole sphere, whereas rigidity starts with a map defined only inside the domain. Extending that map across all boundary components is part of the problem.

3. Development history and previous partial results

Koebe posed the general circle-domain problem in 1908 and established finite-connectivity uniformization; his 1920 treatment develops the classical theory. Finite connectivity means that only finitely many complementary components have to be controlled.

He and Schramm’s 1993 theorem reached every countably connected domain. Their work also gave rigidity for countably connected circle domains. A later 1994 rigidity theorem allowed uncountably many components under a boundary-length hypothesis: the boundary had to be a countable union of sets of finite length.

Schramm’s transboundary extremal length extended a classical way of measuring families of curves so that complementary components themselves could contribute to the measurement. Subsequent results obtained circle-domain models under geometric restrictions on boundary components or on the internal geometry of the domain. Such hypotheses supplied control that mere compactness of conformal maps could not provide.

A specific warning comes from Rajala’s work on exhaustion: arbitrary approximations from inside need not produce a circle domain in the limit. Choosing finite approximations is therefore a substantive part of a proposed solution, rather than a routine last step.

The history of rigidity includes a change in the conjectural picture. He and Schramm proposed that rigidity was equivalent to removability of the boundary. Rajala constructed a rigid circle domain with non-removable boundary, disproving one direction. The new rigidity manuscript claims the remaining direction, from removability to rigidity. It does not restore the original equivalence.

4. What the new manuscripts claim

Unrestricted existence

The uniformization manuscript states that every nonempty connected open set $G\subset\widehat{\mathbb C}$ admits a conformal equivalence onto a circle domain. It imposes neither countability of complementary components nor separation, smoothness, or local regularity of the boundary.

Combined with an earlier exhaustion theorem, the paper obtains a carefully chosen increasing sequence of finitely connected subdomains whose normalized circle-domain maps converge to a map for $G$. Normalization fixes three distinct interior points, removing the freedom of postcomposing with Möbius transformations. The claim is existence of a suitable exhaustion, not successful convergence for every exhaustion.

The manuscript also derives a geometric consequence using Luo and Wu’s work: every complete orientable hyperbolic surface of genus zero can be realized intrinsically as the boundary of a suitable convex hull in hyperbolic three-space. “Hyperbolic” here means constant Gaussian curvature $-1$; the statement does not cover arbitrary metrics merely bounded below by $-1$.

Rigidity with a boundary hypothesis

The companion states that if $\Omega$ and $\Omega'$ are circle domains, $\partial\Omega$ is conformally removable, and $f:\Omega\to\Omega'$ is a conformal equivalence, then

$$f=M|_\Omega$$

for a Möbius transformation $M$. No boundary extension of $f$ is assumed. There can be uncountably many complementary components.

The companion uses a quantitative theorem supplied by the uniformization paper. Thus the two claims form a connected family: the existence argument develops control of auxiliary functions that the rigidity argument then uses.

5. Why the advance matters

Importance: 5/5. If correct, the unrestricted existence theorem settles a central century-old uniformization problem. It supplies one geometric normal form for every planar domain, including boundaries too irregular for the earlier countability and geometric hypotheses.

The significance of a normal form is practical as well as conceptual. Round holes allow conformal geometry to be expressed through their positions and sizes, while point components preserve genuinely degenerate ends. Uniformization connects questions about analytic functions with concrete geometry and, through the stated application, with complete hyperbolic surfaces.

The rigidity result strengthens the usefulness of the model when its boundary is removable. In that situation the freedom in choosing a circle-domain representation is reduced to the explicit finite-dimensional family of Möbius transformations. The rating reflects the reach of the existence claim and the associated boundary methods, rather than an assumption that every resulting model is unique.

6. The main idea

Why finite round holes can lose roundness

One might try to uniformize increasingly large finite pieces and take a limit. A tracked round disk can converge to a disk $D_b$, yet the full complementary component $K_b$ of the limiting domain can be larger:

$$D_b\subseteq K_b.$$

A protrusion can belong to $K_b$ without appearing in the disk one followed. Other components may have no tracked disk at all. The paper’s strategy addresses these two problems separately.

First, collapse each source complementary component to a label. These labels form a compact totally disconnected set on a quotient sphere. This construction organizes the ends without assuming their original shapes are well behaved.

Next, use auxiliary scalar functions as probes. Their Dirichlet energy,

$$\mathcal E(u)=\int_G |\nabla u|^2\,dA,$$

measures how much the function changes. In two dimensions this energy is unchanged by conformal coordinates. A function taking one value near an interior anchor and another near an end acts as a barrier separating them.

The manuscript describes a selection argument using barriers and probability distributions on paths. Small-energy barriers force most limiting components to collapse to points. A countable set of marked components remains. Additional probes with prescribed change around a marked component detect any protrusion beyond its tracked disk; an integral estimate is designed to rule out that protrusion.

The key claimed transfer theorem preserves previously chosen probes on larger and larger compact cores while adding very little energy elsewhere. This permits the requirements to accumulate without losing the properties already arranged. The target conclusions are that every unmarked component is a point and every marked component equals its round disk.

How the companion uses this control

The rigidity argument uses retained scalar probes to establish a Sobolev extension statement across a removable totally disconnected set. A Sobolev function has square-integrable weak first derivatives; this is an energy-based notion of differentiability that allows nonsmooth functions.

Boundary components create a further complication because a point component on one side can correspond to a disk on the other. The paper separates the countably many pairs involving disks, controls traces on the remaining point components, and derives a flux inequality. Comparing that inequality with area bounds is the described mechanism forcing the conformal map to be Möbius. These are the manuscripts’ organizing ideas, rather than a proof of the full limiting argument.

7. Scope, limitations, and open questions

The existence claim is unrestricted within planar domains, but it is not a higher-dimensional uniformization theorem. Complex analyticity and the conformal invariance of two-dimensional Dirichlet energy are essential to the setting.

Existence does not imply that an arbitrary circle-domain model is rigid. Removability is an explicit hypothesis of the companion theorem, and the known failure of the reverse implication remains part of the picture. Nor does a conformal equivalence inside the domain automatically provide a homeomorphic correspondence between all original boundary points.

The approximation consequence selects an exhaustion; it supplies no universal numerical algorithm or rate of convergence. The hyperbolic realization consequence concerns intrinsic distances and existence, not a prescribed boundary identification or a uniqueness theorem. Questions about broader curvature bounds remain separate from the constant-curvature application stated here.

8. Sources and further reading

The source manuscripts below contain the existence theorem, the retained-test construction, and the rigidity theorem. For the classical transition from finite to countable connectivity, see He and Schramm. For the distinction between rigidity and removability, Rajala’s counterexample is essential context. The hyperbolic application uses Luo and Wu.

Optional introductions to the Riemann mapping theorem and Möbius transformations provide familiar examples of conformal coordinates.

Source manuscripts

• M1. OpenAI, Removable Boundaries and Rigidity of Circle Domains (September 23, 2026).
• M2. OpenAI, Koebe’s Circle-Domain Conjecture (September 23, 2026).

Sources consulted on 14 October 2026 at repository revision 1581d4c. Download the manuscript citations.