Function-Space Formulations of Weak Cosmic Censorship

Weak cosmic censorship is not the sentence “singularities are hidden.” A mathematical formulation must specify an Einstein–matter system, a complete constraint-satisfying asymptotically flat data space, a topology and generic subset, a maximal globally hyperbolic development, and a precise regularity or completeness condition at future null infinity. Changing any one of those choices can change the truth value. This page supplies the function-space formulation; the physical causal interpretation remains with Global Structure.

Required background. Analysis for Mathematical Relativity supplies weighted function spaces and genericity notions, while the Einstein equation supplies the constraints and maximal development.

Helpful background. Causality and Global Structure supplies conformal infinity and visibility; PDEs and Green Functions for Gravitation supplies well-posedness topologies; and Black-Hole Geometry and Mechanics supplies the exterior/horizon interpretation.

A parameterized conjecture

Fix the tuple

$$\mathfrak C_{\mathrm W} =(\mathcal E,\mathcal D_{\mathrm{AF}},\mathcal T, \mathcal G,\mathcal A,\mathcal Q).$$

Here $\mathcal E$ is an Einstein–matter system; $\mathcal D_{\mathrm{AF}}$ is a class of regular, complete, constraint-satisfying asymptotically flat initial data; $\mathcal T$ is a topology on that constraint set modulo the selected gauge equivalence; $\mathcal G$ is a declared generic subset; $\mathcal A$ specifies the asymptotic end and conformal regularity; and $\mathcal Q$ is the exterior conclusion.

For $d\in\mathcal D_{\mathrm{AF}}$, write $\operatorname{MGHD}(d)$ for the maximal globally hyperbolic development. A null-infinity completeness formulation is

For every $d\in\mathcal G$, the designated component of future null infinity $\mathscr I^+$ exists in the prescribed asymptotic class and is complete.

“Complete” concerns the future extent of null infinity, not just the radial distance traveled by one escaping light ray. In a regular Bondi description, observers receiving radiation at infinity must have an unlimited future range of the asymptotically normalized retarded time. A general definition must specify the allowed conformal gauge or an equivalent physical normalization. In spherical symmetry, Christodoulou’s formulation uses the divergence of affine lengths of a family of ingoing null segments as those segments approach null infinity, with their normalization transported from one fixed outgoing ray Dafermos 2005, §5.4.

This is different from asking whether each outgoing physical null ray that reaches a given cut has infinite affine length. In Minkowski space an outgoing ray at fixed $u=t-r$ has $r\to\infty$ and infinite affine length. Keeping only the portion of the conformal boundary $u\lt u_*$ leaves that fact unchanged, but removes its future cuts. This deliberately truncated example illustrates the distinction; it is not an MGHD counterexample to censorship. Future asymptotic predictability is another, causal formulation requiring a regular conformal completion and domain-of-dependence control. An equivalence between such formulations requires additional hypotheses.

The clause that $\mathscr I^+$ exists is indispensable. A formal Bondi coordinate expansion or a candidate conformal factor does not construct the global boundary. The asymptotic theorem must control the completion and the remainder in the same data class used by the conjecture.

The data space is part of the claim

For vacuum data on a three-manifold $\Sigma$, a typical weighted Sobolev model has

$$(h-\delta,K) \in H^s_{\delta_0}(\Sigma) \times H^{s-1}_{\delta_0-1}(\Sigma),$$

subject to the Hamiltonian and momentum constraints, parity or end conditions as needed, and quotient by asymptotically trivial diffeomorphisms. The indices $s$ and $\delta_0$, the number and topology of the ends, and the permitted ADM charges cannot be omitted. A scalar-field or kinetic model enlarges the phase space and changes both the constraints and topology.

The subscript convention also matters. One common choice, using dimensionless asymptotic coordinates, is

$$\|f\|_{H^s_\delta}^2 =\sum_{|\alpha|\le s}\int \langle x\rangle^{-2\delta-3+2|\alpha|} |\partial^\alpha f|^2\,d^3x, \qquad \langle x\rangle=(1+|x|^2)^{1/2},$$

for integer $s$; fractional orders require the corresponding interpolation or Fourier definition. Here one derivative lowers the asymptotic weight by one, which explains the displayed $K$ space. A function behaving as $r^p$ has a finite zeroth-order weighted norm at infinity when $p\lt\delta$; the borderline $p=\delta$ gives a logarithmic divergence. Authors also use the opposite weight sign, so indices cannot be compared without their norm definitions.

The genericity quantifier must then be one of, for example:

• open dense in $\mathcal T$;
• residual, meaning a countable intersection of open dense sets;
• complement of a finite-codimension submanifold in a chart;
• full measure for an explicitly supplied probability measure.

These are inequivalent in infinite dimensions. A residual set can have empty interior; “codimension two” in an affine slice gives no canonical probability; and there is no natural measure on the full constraint manifold. The topology must also be compatible with local evolution so that a perturbation of data has a controlled meaning in spacetime.

A proved spherical result and its boundary

Christodoulou’s spherically symmetric massless Einstein–scalar work uses a characteristic profile in $BV\cap L^1$. Within that model he isolates an exceptional set $E$ of naked-singularity-producing data. Through every $\vartheta\in E$ he constructs a two-dimensional affine plane $\Pi_\vartheta$ such that

$$(\Pi_\vartheta\setminus\{\vartheta\})\cap E=\varnothing,$$

with distinct exceptional planes disjoint. Perturbations in the two selected directions form trapped surfaces. This is a rigorous positive-codimension instability statement in that characteristic BV framework Christodoulou 1999, Theorem 4.1, pp. 214–216.

It is not an open-dense theorem for unrestricted vacuum Cauchy data. Spherical symmetry removes gravitational-wave degrees of freedom, and a two-dimensional plane inside the reduced phase space does not define codimension in the full constraint manifold.

A 2024 v1 preprint claims a charged analogue in the spherically symmetric Einstein–Maxwell–charged-complex-scalar system. Its precise theorem uses characteristic profiles $\vartheta\in BV([0,\infty),dv)\cap L^1((0,\infty),dv/v)$: data outside an exceptional set have complete $\mathscr I^+$, while through each exceptional datum passes a complex two-dimensional transverse family containing no other exceptional datum An–Tan 2024, Theorem 1.1 and Remark 1, pp. 2–3, and Theorem 9.1, p. 136. This is an unrefereed theorem claim in another spherical characteristic data space, not a measure statement or a result for unrestricted matter models.

The topology dependence is now explicit. A 2026 preprint claims that a family of continuously self-similar spherical Einstein–scalar naked solutions is nonlinearly stable in a localized Hölder topology adapted to their $C^{1,\alpha}$ regularity: a small open ball has asymptotically flat MGHDs with no trapped surface and incomplete $\mathscr I^+$ Zheng 2026, Definitions 2.7 and 2.11, pp. 18–20, and Theorem 3.1, pp. 23–25. This recent claim has an evidence cutoff of 2026-08-12. It does not contradict the BV instability theorem because the admissible perturbation spaces differ at the distinguished singular sphere.

Existence constructions require a separate data-surface label. The vacuum exterior of Rodnianski and Shlapentokh-Rothman can be glued to a companion interior to give a maximal development from a complete asymptotically flat characteristic cone with incomplete $\mathscr I^+$ Shlapentokh-Rothman 2022, Theorem 1.3, pp. 5–6. Its initial data are not smooth: they have arbitrarily high angular regularity but only $C^{1,s}$ regularity in the null direction Shlapentokh-Rothman 2022, Remark 1.2, pp. 5–6. An and Wu instead construct a nonspherical Einstein–scalar spacetime from structured $\kappa$-self-similar incoming data and small outgoing null perturbations, again with incomplete $\mathscr I^+$ An–Wu 2026, Theorem 1.3, pp. 3–4. Neither existence theorem supplies an open stable neighborhood in a smooth complete-Cauchy-data topology.

Visibility, completeness, and exterior predictability

Let a conformal completion identify the chosen $\mathscr I^+$. The black-hole region is

$$\mathcal B=M\setminus J^-(\mathscr I^+).$$

This definition organizes visibility but does not define a singular subset of $M$: a singular endpoint is not normally a manifold point. Three possible failure statements must be distinguished:

• the designated future null infinity ends after a finite normalized future duration, in the precise completeness sense chosen above;
• a singular ideal endpoint lies in $J^-(\mathscr I^+)$ in a chosen boundary construction;
• the exterior fails future asymptotic predictability from the initial surface.

These address related censorship mechanisms, but none is an automatic substitute for another. An exact overextremal metric or a spacetime created by deleting a point is also not a counterexample unless it arises as an MGHD of admissible complete regular data.

Boundaries of the formulation

The asymptotically flat definition does not port unchanged to $\Lambda\ne0$. For de Sitter asymptotics, future conformal infinity is spacelike and observer horizons matter. For anti-de Sitter asymptotics, conformal infinity is timelike and its boundary condition is part of the evolution. A WCC statement in either setting requires a new $\mathcal A$ and $\mathcal Q$.

Weak censorship also does not imply deterministic interiors. Complete $\mathscr I^+$ can coexist with a continuously extendible Cauchy horizon inside a black hole. Conversely, an inextendible MGHD need not have a complete visible exterior. Weak and strong censorship answer different questions Isenberg 2015, §3, pp. 6–7.

Check the formulation

A paper proves that a small open ball in a localized Hölder topology on spherically symmetric characteristic Einstein–scalar data produces incomplete $\mathscr I^+$. Which WCC statement has been refuted, and which has not?

Solution

Assuming the data are regular, complete, and asymptotically flat in the paper’s sense, the result refutes a generic-completeness formulation whose phase space is that spherical characteristic Hölder class: an open set of counterexamples cannot be removed as nongeneric there. It does not refute WCC for smooth weighted-Sobolev Cauchy data, nonsymmetric data, vacuum gravity, another matter model, another asymptotic class, or a different definition of genericity. Those are different tuples $\mathfrak C_{\mathrm W}$.

References

• An, Xinliang, and Hong Kiat Tan. “A Proof of Weak Cosmic Censorship Conjecture for the Spherically Symmetric Einstein–Maxwell–Charged Scalar Field System.” arXiv:2402.16250v1 (2024). arXiv v1.
• An, Xinliang, and Shengrong Wu. “Naked Singularities beyond Spherical Symmetry: Singular Inner Cauchy Horizons for the Einstein–Scalar Field System.” arXiv:2607.07134v1 (2026). arXiv v1.
• Christodoulou, Demetrios. “The Instability of Naked Singularities in the Gravitational Collapse of a Scalar Field.” Annals of Mathematics 149 (1999): 183–217. arXiv.
• Dafermos, Mihalis. “Spherically Symmetric Spacetimes with a Trapped Surface.” Classical and Quantum Gravity 22 (2005): 2221–2232. DOI.
• Isenberg, James. “On Strong Cosmic Censorship.” Surveys in Differential Geometry 20 (2015): 17–36. arXiv.
• Shlapentokh-Rothman, Yakov. “Naked Singularities for the Einstein Vacuum Equations: The Interior Solution.” arXiv:2204.09891v1 (2022). arXiv v1.
• Zheng, Weihao. “Nonlinear Stability of Continuously Self-Similar Naked Singularities for the Einstein–Scalar Field Equations I: Main Results.” arXiv:2605.16235 (2026). arXiv.