What Counts as Singularity Resolution?

A proposal has not resolved a black-hole singularity until it says which failure has disappeared. Geodesic incompleteness, divergent curvature, unbounded tidal response, nonunique quantum evolution, loss of asymptotic information, and absence of an operational continuation are distinct criteria. They can disagree in one model. The scientifically useful question is therefore not “is the singularity gone?” but “which data and observables are continued, by what dynamics, in what regularity class, and with what error?”

Helpful background. Semiclassical Gravity supplies effective field equations and renormalized stress tensors; Quantum Information supplies unitary channels and information-loss tests; Quantum Fields in Curved Spacetime supplies state-dependent field observables; Black-Hole Information supplies the asymptotic recovery question; Quantum Gravity supplies microscopic completion programmes; and Causality and Global Structure supplies geodesic completeness, extensions, and Cauchy data.

A hierarchy of resolution criteria

Geometric continuation. A specified family of causal geodesics is complete, or the metric admits an extension in a declared differentiability class. A $C^0$ extension can exist without classically defined curvature. Continuity alone does not even guarantee a well-defined distributional curvature tensor; additional weak differentiability is needed to control products of connection coefficients.

Curvature bounds and tidal observables. A curvature bound may constrain scalar invariants, parallelly propagated tidal components, or integrated detector responses. No single scalar detects every singularity, and a finite invariant does not ensure finite tidal forces in every frame.

Quantum determinism. A declared quantum evolution is unitary or otherwise uniquely fixed on its physical Hilbert space. In a fixed-background probe test, this may reduce to essential self-adjointness of a spatial operator. In full gravity it requires constraints, inner product, observables, and boundary data.

Information and operational continuation. On the declared complete asymptotic system, evolution is reversible—unitary when input and output spaces agree, or isometric with the full output included—and a finite observer can in principle identify the relevant output observables. A merely normalized completely positive channel need not preserve information. A unitary S-matrix can coexist with large interior curvature; an operationally smooth infall can coexist with inaccessible global information.

These criteria are not equivalent. A page must state the implication it has proved rather than use one as shorthand for all the others.

Two discriminating model checks

Flat spacetime written in a chart that ends at a coordinate boundary supplies the removable case: curvature vanishes and a larger chart extends every geodesic. The “singularity” was a coordinate artifact. By contrast, Schwarzschild’s $r=0$ region has incomplete timelike geodesics and divergent tidal curvature; changing coordinates does not repair it.

A quantum probe supplies a different test. If the reduced wave operator on the incomplete space has more than one self-adjoint extension, initial data do not select a unique unitary evolution without extra boundary conditions. Finite expectation values for some states do not cure that ambiguity. Horowitz and Marolf formulate this essential-self-adjointness criterion and examples in Horowitz and Marolf 1995, §§ 2–3. The criterion is valuable but limited: a test field that evolves uniquely on a fixed background does not prove that quantum gravity resolves the geometry.

Results and open questions through 2026-08-19

This discussion reflects results available through 2026-08-19. A permissive interpretation accepts satisfaction of one physically motivated criterion as a meaningful partial resolution. A conservative interpretation reserves the unqualified word resolved for a stable microscopic evolution that also reproduces the appropriate semiclassical exterior. These competing positions become compatible once the criterion is included in every claim.

Criterion and evidence	Result	Scope or limitation
Geodesic and boundary analysis in Geroch 1968, pp. 526–540	Singular behavior is characterized without pretending that a missing spacetime point is an ordinary tensor location.	Different completions and causal curves can diagnose different failures; no universal local scalar criterion follows.
Essential-self-adjointness probe in Horowitz and Marolf 1995, §§ 2–3	Some classically incomplete static backgrounds admit unique quantum probe evolution without added boundary data.	This is fixed-background one-particle or field-mode evidence, not full quantum geometry or backreaction.
Low-regularity charged interior in Dafermos 2005, main theorems, pp. 445–504	Cauchy-horizon formation, blow-up, and extendibility can coexist in different regularity senses.	The hypotheses are classical spherical Einstein–Maxwell–scalar data and do not settle quantum continuation.
Black-to-white transition model in Haggard and Rovelli 2015, §§ 2–4	A bounded quantum region can replace the classical endpoint in a specified matching construction.	Stability, lifetime, state propagation, and derivation of matching data require additional arguments.

The evidence shows that partial criteria can be met in controlled settings. It remains unresolved whether a generic evaporating black hole satisfies all geometric, quantum-evolution, information, and operational criteria in one single-theory completion.

A proposal-reporting protocol

First state the classical failure: incomplete curves, curvature blow-up, Cauchy-horizon nonuniqueness, or another invariant. Next identify the modified equations or amplitudes and the regime in which corrections dominate. Specify initial and boundary data on both sides, including any self-adjoint extension, contour, final-state condition, or stochastic rule.

Then propagate at least one complete set of observables and show normalization, linearity, constraint preservation, and a controlled classical limit. Test generic perturbations rather than a symmetric solution alone. Finally state what an asymptotic or infalling observer could measure and what resource is required. A bounded effective curvature with an arbitrary matching surface is a regularization prescription, not derived quantum dynamics.

The strongest adverse check changes the criterion. A model with bounded scalar curvature but two inequivalent outgoing states passes that scalar test and fails quantum determinism. A unitary exterior channel that requires divergent local stress passes information conservation and fails semiclassical smoothness. Reporting both results is more informative than assigning one verdict.

Generic quantum-gravity construction, classical causal theory, stable thermodynamics, and string microphysics remain outside this page. The bounded outcome is a common vocabulary for comparing them without erasing their different hypotheses.

The conclusions could change after a theorem establishes implications between two previously distinct criteria, or a controlled backreacted model propagates generic data through the endpoint uniquely, or an explicit counterexample shows that a widely used curvature, self-adjointness, or S-matrix test fails to capture a consequential observable.

Exercises

A model keeps every curvature scalar finite and lets wave packets cross the core, but its wave operator has two self-adjoint extensions producing different outgoing phases. Which criteria pass and fail?

Solution

The model passes the stated scalar-curvature criterion and may pass a weak wave-packet continuation test for each chosen extension. It fails unique quantum evolution because initial data alone do not select one extension; the outgoing phase depends on extra boundary data. Geodesic completeness and tidal boundedness still require their own checks, and no asymptotic information claim is justified until the extension is derived and the resulting channel is normalized and linear.

References

• Dafermos, Mihalis. 2005. “The Interior of Charged Black Holes and the Problem of Uniqueness in General Relativity.” Communications on Pure and Applied Mathematics 58, 445–504. arXiv:gr-qc/0307013.
• Geroch, Robert. 1968. “What Is a Singularity in General Relativity?” Annals of Physics 48, 526–540. DOI.
• Haggard, Hal M., and Carlo Rovelli. 2015. “Black Hole Fireworks: Quantum-Gravity Effects outside the Horizon Spark Black to White Hole Tunneling.” Physical Review D 92, 104020. arXiv:1407.0989.
• Horowitz, Gary T., and Donald Marolf. 1995. “Quantum Probes of Spacetime Singularities.” Physical Review D 52, 5670–5675. arXiv:gr-qc/9504028.