Anatomy of a Singularity Theorem

A classical singularity theorem is a contradiction argument with several independent engines. Initial or trapped geometry starts convergence; a Ricci-curvature condition sustains it; Jacobi or Raychaudhuri analysis produces focal or conjugate points; and global causality guarantees maximizing geodesics that are not allowed to contain those points. The escape from the contradiction is that an appropriate causal geodesic does not extend to arbitrarily large affine or proper parameter. No theorem in this chapter thereby manufactures a “singular point” inside the spacetime.

Required background. Congruences and Optical Scalars supplies expansion, shear, twist, and focusing; Einstein Equations relates stress–energy assumptions to Ricci convergence. Helpful background. Lorentzian Metrics and Causal Cones fixes causal signs; Geodesics and Jacobi Fields supplies affine completeness and conjugate points; Causality and Global Structure supplies global causal hypotheses; Curvature fixes the tidal operator.

The five-part theorem pattern

Work first in a connected, time-oriented spacetime $(M,g)$ of dimension $n\geq3$ with a $C^2$ Lorentzian metric; smoothness may be assumed for the classical proofs. A useful theorem statement separates five components.

1. Regularity and causal direction. One specifies whether timelike, null, or either kind of geodesic is considered; whether the conclusion concerns the future, the past, or an unspecified direction; and which differentiability class supports curvature, Jacobi fields, and the exponential map.
2. Curvature or energy input. Timelike convergence is $R_{ab}u^au^b\geq0$ for every timelike $u$; null convergence is $R_{ab}k^ak^b\geq0$ for every null $k$. Einstein’s equation can translate these into matter inequalities, but the proof uses the Ricci statements.
3. Initial or boundary geometry. A uniformly converging spacelike hypersurface, a closed trapped surface, a compact achronal edgeless set, or a trapped point supplies the seed for focusing.
4. Causality and topology. Global hyperbolicity, chronology, noncompactness of a Cauchy surface, or a trapped-set lemma turns local focusing into an impossible global configuration.
5. Conclusion. At least one inextendible causal geodesic has a finite endpoint of its affine or proper-time parameter interval in the specified direction.

This is the “pattern singularity theorem” presented in Steinbauer 2023, §3.1, pp. 9–10. It is a template, not one theorem with interchangeable hypotheses. Penrose uses null convergence, global hyperbolicity, and a noncompact Cauchy surface; Hawking uses timelike convergence and uniform hypersurface convergence; Hawking–Penrose replaces global hyperbolicity by chronology but adds generic curvature and a disjunction of initial conditions.

The focusing engine

For a twist-free affinely parametrized causal geodesic congruence, the site curvature convention gives

$$\frac{d\theta}{d\lambda} =-R_{ab}v^av^b-\sigma_{ab}\sigma^{ab}-\frac{\theta^2}{d},$$

where $d=n-1$ for unit timelike tangents and $d=n-2$ for null screen space. If $R_{ab}v^av^b\geq0$, then

$$\frac{d\theta}{d\lambda}\leq-\frac{\theta^2}{d}.$$

Once $\theta(\lambda_0)=\theta_0<0$, integration yields

$$\frac{1}{\theta(\lambda)} \geq \frac{1}{\theta_0}+\frac{\lambda-\lambda_0}{d}$$

for as long as the congruence remains regular. Hence $\theta$ diverges to $-\infty$ no later than

$$\lambda-\lambda_0\leq\frac{d}{\lvert\theta_0\rvert}.$$

The divergence signals a zero of the Jacobi determinant and thus a focal or conjugate point. This calculation, including the distinct timelike and null screen dimensions, is proved in Steinbauer 2023, §§3.2–3.3, pp. 10–15.

The curvature assumption sustains focusing; it does not usually start it. Negative initial expansion comes from the initial geometry. A future trapped codimension-two surface has both future null expansions negative. A contracting spacelike hypersurface, with the time direction chosen appropriately, gives negative expansion to its orthogonal timelike congruence. In the Hawking–Penrose argument, generic tidal curvature supplies focusing without assuming a uniform negative expansion at the start.

The global contradiction

A conjugate point has a variational consequence. A causal geodesic from a point ceases to maximize Lorentzian length after its first conjugate point. A geodesic normal to a spacelike submanifold ceases to maximize distance from that submanifold after its first focal point. Focusing therefore produces a finite upper bound on how long a particular class of geodesics can remain maximizing.

Causality supplies the other side. In a globally hyperbolic spacetime, relevant causally related points are joined by maximizing causal geodesics. Points in the future Cauchy development of a spacelike hypersurface are reached by normal maximizing timelike geodesics without focal points before the endpoint. The Penrose proof instead studies null generators of $E^+(P)=J^+(P)\setminus I^+(P)$: while a generator remains on this achronal boundary, it cannot have a focal point to $P$.

Now assume the required completeness. The geodesics last long enough for the focusing estimate to force focal or conjugate points. Yet the causal construction requires an indefinitely extendible family of maximizers without such points. The remaining topological or compactness step prevents those maximizers from simply disappearing. The hypotheses jointly contradict completeness; no single line of Raychaudhuri does so.

This division diagnoses failed arguments. Without convergence, the inequality may reverse. Without initial convergence or genericity, $\theta$ can remain zero. Without sufficient causality, a maximizing geodesic need not exist. Without the Penrose noncompact-Cauchy hypothesis, a compact achronal boundary need not contradict the topology of a slice.

Energy conditions and Einstein’s equation

In four-dimensional Einstein gravity,

$$R_{ab}-\frac12Rg_{ab}+\Lambda g_{ab}=8\pi G T_{ab}.$$

For a null vector $k$, contraction removes both terms proportional to $g_{ab}$:

$$R_{ab}k^ak^b=8\pi G\,T_{ab}k^ak^b.$$

Thus the null energy condition implies null convergence, independently of $\Lambda$. For a unit timelike vector $u$, the analogous contraction contains the trace of $T_{ab}$ and the cosmological term. The strong energy condition on matter is arranged to give timelike convergence when $\Lambda=0$; positive $\Lambda$ can oppose it. A page must therefore state the geometric convergence condition actually used rather than say only that “reasonable matter attracts.”

Averaged energy conditions can support other theorems, and quantum fields can violate pointwise classical conditions. Those facts do not modify the hypotheses of the three classical statements here. They motivate distinct generalized theorems with their own sampling, genericity, and causal assumptions.

Exactly what escapes the contradiction

The conclusion is geodesic incompleteness. There exists an inextendible timelike or null geodesic, as specified, whose affine or proper parameter covers only a finite interval in the relevant direction. “Inextendible” is evaluated within $(M,g)$; the theorem does not add an endpoint to the manifold.

This conclusion does not imply that curvature diverges, that every observer is incomplete, that an event horizon forms, that the incomplete curve ends at finite coordinate time, or that no extension of lower regularity exists. Even flat Minkowski spacetime with a point deleted is geodesically incomplete without curvature blowup. Conversely, curvature components can become unbounded along complete curves in contrived geometries. These distinctions are developed on What Singularity Theorems Conclude.

Classical proofs rely on pointwise curvature and Jacobi fields. Extensions to $C^{1,1}$ and $C^1$ metrics exist, but they replace important ingredients with regularization, distributional convergence conditions, and—in some $C^1$ results—nonbranching assumptions. They are separate theorems, summarized in Steinbauer 2023, §4, pp. 21–40, not evidence that differentiability can be omitted from the classical statement.

Exercise

A proposed proof assumes a future trapped surface and null convergence, integrates Raychaudhuri to obtain focal points along its future null normals, and immediately announces future null incompleteness. Identify two missing global ingredients in the Penrose route and explain their roles.

Solution

First, one needs a causal framework—classically global hyperbolicity—so that the future horismos of the trapped surface is an achronal boundary generated by null geodesics and can be compared with a Cauchy surface. Focal points force generators to leave that boundary, but focusing alone does not prohibit the geodesics from continuing inside $I^+(P)$.

Second, one needs the noncompactness of a Cauchy surface. Under assumed null completeness, uniform focusing makes $E^+(P)$ compact. Projecting this compact achronal hypersurface along a timelike flow to the Cauchy surface produces a subset that is both open and closed; connectedness would make it the whole slice, contradicting noncompactness. Without that topology step, compact focusing may coexist with completeness, as closed cosmological examples show.

References

• Hawking, Stephen W., and Roger Penrose, “The Singularities of Gravitational Collapse and Cosmology”, Proceedings of the Royal Society A 314, pp. 529–548, 1970.
• Penrose, Roger, “Gravitational Collapse and Space-Time Singularities”, Physical Review Letters 14, pp. 57–59, 1965.
• Steinbauer, Roland, “The Singularity Theorems of General Relativity and Their Low Regularity Extensions”, Jahresbericht der Deutschen Mathematiker-Vereinigung 125, §§3.1–3.5, pp. 9–20, 2023.