How much mass does a black hole require? Penrose inequalities beyond time symmetry

← All digests

Family 260 · Mathematical physics
Importance 5/5 · Provisional editorial rating
Sources consulted 2026-10-25

1. The question

How much mass must a universe contain if a black-hole boundary of a given size is already present? In three spatial dimensions, the simplest proposed answer is

$$m\geq\sqrt{\frac{A}{16\pi}}.$$

Here units are chosen so that the gravitational constant and speed of light are one. Mass therefore has units of length. The Schwarzschild black hole, the ideal nonrotating and uncharged black hole, attains equality: its horizon has area $16\pi m^2$.

The difficult words are “mass,” “boundary,” and “size.” General relativity allows moving spatial slices, electric and magnetic fields, rotating configurations, and different geometries far away. A surface indicating trapped light need not itself be the smallest surface enclosing that region. Changing one of these choices can change a true inequality into a false one.

Family 260 contains thirteen manuscripts addressing these distinctions. They claim spacetime Penrose inequalities, classifications of equality, charged and rotating extensions, results for negative cosmological constant, and a Bondi-mass version. One manuscript supplies counterexamples to an insufficiently qualified rotating formulation. The common aim is to relate quantities on a single spatial slice without first solving the entire future evolution.

2. Background and essential definitions

An initial data set consists of a spatial manifold with a Riemannian metric $g$, which measures lengths, and a symmetric tensor $K$, which describes how the slice bends in spacetime. Time-symmetric data have $K=0$. Maximal data satisfy the weaker condition $\operatorname{tr}_gK=0$; their full tensor can be nonzero.

The dominant energy condition says that the matter energy density $\mu$ is at least the magnitude of its momentum density $j$: $\mu\geq |j|_g$. This is a local condition. For charged data, the manuscripts specify separately whether electromagnetic energy and momentum have been removed from these densities.

On a boundary surface $S$, take the unit normal toward the exterior. Let $H$ be its mean curvature, the sum of its principal curvatures. The future outward light expansion is

$$\theta_+=H+\operatorname{tr}_S K.$$

The condition $\theta_+\leq0$ means outward-directed light initially fails to expand. Equality defines a marginally outer trapped surface, or MOTS. This is information on the slice; an event horizon is defined using the global spacetime and is a different object.

An asymptotically flat end approaches Euclidean space at large distance. Its ADM energy $E$ and linear momentum $P$ are fluxes measured at spatial infinity. When $E>|P|$, the invariant mass is

$$m=\sqrt{E^2-|P|^2}.$$

Using $E$ alone would make the bound depend on the asymptotic observer’s motion.

The relevant area is usually $A_{\min}$, the infimum of the areas of full enclosing cuts. Such a cut is the entire compact boundary of a connected exterior region containing the distant end. Every component counts, including portions coinciding with $S$. Cuts need not be trapped or connected. A smooth minimizer need not exist. “Outer area-minimizing” means that every such cut has area at least that of $S$; only then can one replace $A_{\min}$ by the original boundary area without further argument.

3. Development history and previous partial results

Penrose’s 1973 proposal combined gravitational collapse, the expected hiding of singularities, and black-hole area increase. If a configuration settles to a black hole while radiation removes energy, its initial mass should already cover the final black hole’s area cost. This physical argument motivates the inequality; an initial-data theorem must establish its own geometric statement. See Penrose’s Naked singularities.

The time-symmetric problem became a theorem through two different geometric flows. Huisken and Ilmanen developed weak inverse mean curvature flow, allowing the evolving surface to jump across troublesome regions. Their three-dimensional bound uses the area of each connected component of an outermost minimal boundary. Bray’s conformal flow changes the ambient metric and handles the total area of a possibly disconnected horizon. Both papers appeared in 2001.

Bray and Lee extended the Riemannian method to dimensions below eight. In higher dimensions, area-minimizing boundaries can have singularities. M9 and M10 explicitly place their treatment of such frontiers in the line of Bray, Bray–Lee, and Bi–Zhu; they do not present conformal flow or the high-dimensional program as originating here.

For general $K$, Jang-type graph deformations offered a route back to scalar-curvature inequalities. The problem is to preserve enough area and control mass while repairing curvature. Bray and Khuri developed this perspective. Counterexamples by Ben-Dov, and by Carrasco and Mars for a different horizon definition, also showed why the area convention cannot be ignored; M1, M9 and M12 discuss those precedents.

Charge introduced another branch condition. Khuri, Weinstein and Yamada established a charged Riemannian inequality for multiple black holes. Rotation required electromagnetic corrections to angular momentum and earlier harmonic-map comparisons. For negative cosmological constant, prior work treated graphs and perturbations of Schwarzschild–anti-de Sitter geometry. The local result of Khuri and Kopiński imposed a quantitative condition comparing a seed tensor’s interior and boundary behavior. M8 claims to remove that additional condition in its fixed-seed setting.

4. What the new manuscripts claim

Neutral data and original-data equality

M9 claims, in every spatial dimension $n\geq3$,

$$m\geq\frac12\left(\frac{A_{\min}}{\omega_{n-1}}\right)^{\frac{n-2}{n-1}},$$

where $\omega_{n-1}$ is the area of the unit sphere in $\mathbb R^n$. The exterior is smooth, complete with its compact boundary included, and has one asymptotically flat end. It obeys the dominant energy condition, finite ADM fluxes, integrable constraint densities, and weak future trapping.

Its all-dimensional version assumes positive enclosing area and differentiated decay through six metric derivatives and five derivatives of $K$, with decay exponent $(n-2)/2<q<n-2$. Future timelikeness is obtained in a corollary. In dimensions three and four, M9 also claims a weaker-decay version requiring only two metric derivatives and one tensor derivative, with $q>(n-2)/2$; positive enclosing area and timelikeness are then conclusions. These are bounds at infinity on otherwise smooth data.

M10 supplies the Riemannian comparison for locally perimeter-minimizing, outer-minimizing frontiers, including the permitted high-dimensional singular sets. M12 supplies boundary graph deformations in dimensions three and four. M3 includes weak-decay neutral results and designated-end versions for finitely many ends, whose enclosing-cut conventions must be kept separate.

M11 treats equality on the original data. Under its additional horizon and decay assumptions, equality identifies both $g$ and $K$ with a spacelike slice of Schwarzschild–Tangherlini spacetime, smoothly attached to a horizon. The all-dimensional strong-decay theorem has connected boundary. Its three-dimensional finite-component theorem uses an ambient complete-data hypothesis and stronger outermostness allowing partial coincidence with existing components; equality then forces connectedness. M3 also presents neutral rigidity in its stated three- and four-dimensional classes.

Charge and rotation

For M1’s three-dimensional source-free electric and magnetic fields, write $Q=\sqrt{Q_E^2+Q_B^2}$ and $r_A=\sqrt{A_{\min}/(4\pi)}$. It claims

$$m\geq Q,\qquad r_A\leq m+\sqrt{m^2-Q^2}.$$

The boundary can be disconnected, $K$ arbitrary, and ADM momentum nonzero. The non-electromagnetic matter obeys its charged dominant energy condition. Fields are divergence-free on the exterior, with the stated decay and finite fluxes. On each boundary component the theorem even permits a fixed future or past trapping sign.

The familiar rearrangement

$$m\geq\frac12\left(r_A+\frac{Q^2}{r_A}\right)$$

is asserted on the outer branch $r_A>Q$, not for every positive radius. Under connected, outermost, outer-area-minimizing future-horizon assumptions and $m>Q$, equality in M1 recovers a slice of dyonic Reissner–Nordström, including both original electromagnetic fields. “Dyonic” means that both charge types are allowed.

M2 gives a purely electric higher-dimensional reduction using the companion neutral theorem. Replace $r_A$ above by $r_A^{n-2}$, with the paper’s normalized signed charge taken in absolute value. Its nonextremal rigidity conclusion is Reissner–Nordström–Tangherlini and includes only slices whose induced magnetic two-form vanishes. M3 supplies related three-dimensional charged numerical statements, without itself asserting charged rigidity.

M4 claims the rotating bound

$$m^2\geq\frac{A}{16\pi}+\frac{Q^2}{2} +\frac{\pi(Q^4+4J^2)}{A}.$$

This is for its one-ended axisymmetric electrovacuum exterior class, a connected outermost and outer-area-minimizing future MOTS, zero ADM momentum, and Coulomb leading electromagnetic tails. The physical branch $A\geq4\pi\sqrt{Q^4+4J^2}$ is an explicit assumption. The angular momentum $J$ is the conserved total, including its electromagnetic contribution. Strict-branch equality identifies admissible Kerr–Newman slices. Maximality is not assumed.

M5 supplies a complementary warning: if one instead uses bare gravitational ADM angular momentum and merely permits fields of order $r^{-2}$ with angularly varying leading terms, it constructs violations and nonrigid equality examples. Their total electric and magnetic charges vanish, although their fields do not. These counterexamples do not contradict M4’s stronger conventions.

Curved infinity and Bondi mass

With cosmological constant normalized to $-3$, M6 claims

$$m_{\mathrm{AH}}\geq\frac{r_A+r_A^3}{2}.$$

The data have one spherical asymptotically anti-de Sitter end, weighted differentiable decay with exponent between $3/2$ and $3$, weighted integrable constraint sources, the dominant energy condition, and the same future trapping sign on all boundary components. The hyperbolic metric mass four-flux is assumed future timelike. Its Lorentz norm is $m_{\mathrm{AH}}$; its spatial entries are not ADM linear momentum. No maximality is required. The connected horizon subclass has original-data Schwarzschild–anti-de Sitter rigidity, with explicit matching of the slice’s metric mass.

M7 gives a maximal-data theorem and a time-symmetric time-component comparison. M8, also reflected in M3, treats small maximal vacuum conformal perturbations of a positive-mass Schwarzschild–anti-de Sitter exterior. For each fixed decaying transverse-traceless seed and prescribed solution branch, its bound uses the boundary’s own area. Small positive parameters give equality for radial seeds and strictness otherwise. This is a local branch statement, without a uniform neighborhood over all seeds.

Finally, M13 replaces an end in the precise Cha–Khuri–Sakovich hyperboloidal class by asymptotically flat ends while protecting every enclosing area. Using M9, it obtains the three-dimensional bound with invariant Bondi initial-data mass. It assumes strictly future-timelike charge and positive enclosing area, and gives sharp Schwarzschild examples. Identification with a Bondi–Sachs charge at null infinity requires the compatible spacetime development described in its scope discussion.

5. Why the advance matters

Provisional importance: 5/5. The claimed general spacetime inequalities address a central connection between black-hole physics and geometric analysis. Moving beyond time symmetry means retaining the tensor that records motion of the slice, rather than treating spatial geometry alone.

The equality results add substantial content. A numerical bound can survive a deformation even when much of the original geometry is lost. Recovering the original metric, second fundamental form, fields, and horizon attachment says which configurations exhaust the mass budget.

The family also clarifies the mathematical question itself. Enclosing area, charge branches, conserved angular momentum, and the choice of mass at infinity are structural ingredients. Its positive and negative statements together explain why a single unqualified “Penrose inequality” cannot represent all thirteen papers.

6. The main idea

The numerical arguments build comparison geometries. A graph height and conformal factors change the original data until scalar curvature has a useful sign. Their estimates must simultaneously protect every enclosing cut and prevent an excessive increase in mass. An artificial filling or carefully controlled boundary allows the elliptic equations to be solved before a minimizing enclosure is selected.

One elementary principle explains the area bookkeeping. If a comparison metric $h$ satisfies $h\geq c g$ as quadratic forms, then every hypersurface has $h$-area at least $c^{(n-1)/2}$ times its $g$-area. Thus a single metric estimate controls all competing cuts, even when the minimizing surface changes or is singular. The papers combine such estimates with sharper constructions where a pointwise bound alone is insufficient.

Charged reductions turn conserved flux into a controlled energy decrease, then invoke a neutral comparison. Rotating arguments retain electric, magnetic and twist potentials, which encode the angular momentum constraints. Anti-de Sitter arguments must also account for the cubic area-radius term and for all four hyperbolic mass fluxes. M13 instead changes the distant end while leaving each fixed compact interior intact.

Equality is handled separately. Varying the original mass-minus-area deficit produces stationary equations through constraint multipliers. These identify a static or stationary model and then reconstruct the original spacelike slice. The claim is therefore stronger than observing that some auxiliary comparison metric resembles a black hole.

7. Scope, limitations, and open questions

These are initial-data claims, not proofs of cosmic censorship or of convergence of gravitational evolution to a stationary black hole. They classify exteriors in specified equality classes and generally say nothing about the region behind the boundary.

There is no universal substitution of apparent-horizon area for enclosing area, no unrestricted polynomial charged bound below the charge threshold, and no rotating theorem here for arbitrary asymptotic electromagnetic tails. The all-dimensional neutral result, weak-decay low-dimensional results, and multi-end results have different assumptions. The charged higher-dimensional paper explicitly depends on the stated neutral companion input.

Extremal charged or rotating equality is not covered by the stated nondegenerate classifications. The local conformal anti-de Sitter result does not provide one parameter bound valid for every seed. Natural further questions include quantitative stability near equality and extensions beyond the stated end, topology, and asymptotic classes. None follows merely from the sharp inequalities presented here.

8. Sources and further reading

The linked historical papers in section 3 explain the original proposal and the two principal Riemannian flow methods. For precise earlier charged and deformation results, see the references to Khuri–Weinstein–Yamada and Bray–Khuri in M1, and the detailed historical discussions in M6, M8 and M9. M4 and M5 should be read together when comparing angular-momentum conventions. All thirteen source manuscripts are listed below; their theorem-level hypotheses take precedence over abbreviated descriptions.

Source manuscripts

• M1. OpenAI, The spacetime Penrose inequality with charge and original-data rigidity (October 5, 2026).
• M2. OpenAI, A Charged Reduction of the Spacetime Penrose Inequality in Spatial Dimensions at Least Four (October 5, 2026).
• M3. OpenAI, Spacetime Penrose inequalities: enclosing area, charge, and rigidity (October 5, 2026).
• M4. OpenAI, The Kerr–Newman Penrose Inequality for Axisymmetric Electrovacuum Exteriors (October 5, 2026).
• M5. OpenAI, Electromagnetic tails and the Kerr–Newman Penrose inequality (October 5, 2026).
• M6. OpenAI, The nonmaximal anti-de Sitter Penrose Inequality and original-data rigidity (October 5, 2026).
• M7. OpenAI, The Penrose inequality for maximal asymptotically hyperbolic initial data (October 5, 2026).
• M8. OpenAI, A local Penrose inequality for conformal perturbations of Schwarzschild–anti-de Sitter data (October 5, 2026).
• M9. OpenAI, The spacetime Penrose inequality and enclosing area (September 27, 2026).
• M10. OpenAI, Conformal flow and the Riemannian Penrose inequality with minimizing frontiers (September 27, 2026).
• M11. OpenAI, Equality and rigidity in the spacetime Penrose inequality (September 27, 2026).
• M12. OpenAI, Boundary graph deformations for the spacetime Penrose inequality (September 27, 2026).
• M13. OpenAI, Area-controlled end replacement and the Bondi Penrose inequality in the CKS class (September 27, 2026).

Sources consulted on 25 October 2026 at repository revision 1581d4c. Download the manuscript citations.