Curvature Blowup Criteria in Singularity Formation

Geodesic incompleteness says that the spacetime, as given, does not contain the whole affine history of at least one freely falling curve. It does not identify the obstruction. Curvature blowup, destructive tidal distortion, crushing of spatial slices, and failure of a $C^k$ extension are stronger but inequivalent conclusions. A rigorous singularity-formation result must say which quantity diverges, along which approach, in what frame or norm, and which extension regularity that divergence excludes.

Required background. Analysis for Mathematical Relativity supplies norm and extension arguments; timelike congruences and spatial projection supplies Jacobi fields and expansion; and the Einstein equation relates Ricci curvature to matter.

Helpful background. Causality and Global Structure supplies completeness and causal-boundary definitions, and PDEs and Green Functions for Gravitation clarifies continuation norms.

A hierarchy of singular conclusions

The following implications should not be silently reversed.

1. Incomplete curve. An inextendible causal geodesic has finite affine or proper-time length. Deleting a point from Minkowski space gives this behavior with zero curvature, showing that incompleteness alone need not represent a physical curvature singularity. A maximal development rules out that simple artificial deletion but does not identify its boundary regularity.
2. Invariant scalar blowup. A scalar such as $R_{\mu\nu\rho\sigma}R^{\mu\nu\rho\sigma}$ diverges along an approaching sequence. This is coordinate independent and immediately obstructs a $C^2$ extension through a compact boundary neighborhood.
3. Parallelly propagated blowup. Components $R(E_a,u,E_b,u)$ diverge in an orthonormal frame parallelly propagated along a causal geodesic. This can occur even when all polynomial scalar invariants stay bounded; algebraically special plane-wave geometries show why scalars are not a complete diagnostic.
4. Strong tidal singularity. Jacobi fields spanning a freely falling body are crushed to zero volume, or the appropriate integrated tidal distortion diverges. Tipler and Królak conditions use different integral strengths and are not synonymous with pointwise curvature divergence.
5. Weak null singularity. A null boundary can admit a continuous metric extension while transverse derivatives, Christoffel symbols, Hawking mass, or curvature fail at stronger regularity. “Weak” refers to metric extendibility or finite tidal deformation, not to a small curvature value.
6. Crushing singularity. The mean curvature of a family of spacelike slices tends uniformly to an infinite magnitude. This foliation-based condition is useful for cosmological singularities but is not an invariant description of every black-hole interior.
7. $C^k$ inextendibility. No extension with a Lorentzian metric of the named differentiability exists. The answer may change between $C^0$, $C^1$, $C^{1,1}$, $C^2$, and weak Christoffel-$L^2$ classes.

The adjacent Global Structure page gives the physical taxonomy of these possibilities; the present page supplies the analytic tests that justify its curvature-and-strength distinctions.

Curvature along an observer

Let $u$ be the unit tangent to a timelike geodesic and $J$ an orthogonal Jacobi field connecting it to a neighboring geodesic. Then

$$\frac{D^2J^\mu}{d\tau^2} =-R^\mu{}_{\nu\rho\sigma} u^\nu J^\rho u^\sigma.$$

In a parallel orthonormal frame, the tidal matrix is $\mathcal E_{ab}=R(E_a,u,E_b,u)$. Pointwise divergence of $\mathcal E_{ab}$ does not by itself determine the accumulated distortion: an integrable spike can leave finite Jacobi volume. Sufficient strong-curvature criteria therefore examine one or two proper-time integrals of selected tidal components. Tipler’s definition concerns the vanishing limit of volumes formed by admissible independent Jacobi fields that vanish at an earlier point, with the analogous transverse-area definition for null geodesics. It is not a statement about every possible initial configuration of an extended object. Królak’s limiting-focusing condition is weaker; integral tests for the two definitions require their own sign and curvature hypotheses. See Tipler 1977, pp. 8–10 and Clarke and Królak 1985, pp. 127–143.

This observer-based formulation catches non-scalar tidal singularities, but it has its own hypotheses: the frame, curve, and limiting endpoint must be fixed. A divergence along one exceptional geodesic does not automatically establish a uniform singular boundary.

Pointwise blowup and accumulated tidal effects

An elementary integral test explains why a blowup exponent matters. Let $s=\tau_*-\tau$ and suppose the magnitude of one tidal component behaves as $C s^{-p}$, where $C\gt0$ has the required dimensions. A single integral near the endpoint converges for $p\lt1$. The endpoint value of the double integral has the convergence test

$$\int^{\tau_*}(\tau_*-\tau)\,C(\tau_*-\tau)^{-p}d\tau \sim C\int_0^\epsilon s^{1-p}ds,$$

which is finite for $p\lt2$ and logarithmically divergent at $p=2$. Thus $1\le p\lt2$ already permits unbounded curvature and a divergent single integral while the double integral stays finite. This is a calculation of integrability, not a universal classification of singularities: the Jacobi equation is a matrix equation, and curvature signs and the volume of its solutions still matter.

For an explicit nonscalar diagnostic, consider the vacuum plane wave

$$g=-2\,du\,dv+dx^2+dy^2+F(u)(x^2-y^2)du^2, \qquad u\gt0.$$

Its only independent curvature components are $R_{xuxu}=-F(u)$ and $R_{yuyu}=F(u)$. They have opposite signs, so $R_{uu}=0$, and all algebraic polynomial curvature scalars vanish: raising a $u$ index pairs it with a $v$ index, for which the curvature is zero. Nevertheless, along the central timelike geodesic $x=y=0$, $v=u/2$, the parameter $u$ is proper time and $\partial_x,\partial_y$ are parallelly propagated unit vectors. They measure exactly those tidal components. Choosing $F(u)=L^{p-2}u^{-p}$ makes them diverge at finite proper time $u=0$ for $p\gt0$, while every polynomial scalar stays zero. The example is local and not an asymptotically flat collapse model; it isolates why a scalar test alone can miss a tidal obstruction.

From blowup to inextendibility

The cleanest implication uses regularity. If $g$ extended as a $C^2$ Lorentzian metric across a boundary point, its Riemann tensor would be continuous in a regular chart. Hence an invariant scalar tending to infinity along a sequence approaching that point excludes such an extension. Similarly, a $C^{1,1}$ extension has essentially bounded second weak derivatives and curvature in $L^\infty_{\mathrm{loc}}$, so suitably formulated essential blowup obstructs it.

The converse fails. A spacetime may be $C^0$-inextendible for global geometric reasons even when no convenient scalar has been shown to diverge. Sbierski’s Schwarzschild result uses the intrinsic spacelike diameter and causal geometry, not a scalar-curvature estimate, to rule out a continuous extension Sbierski 2018, Theorem 1.1 and §§3–5. At $C^0$ regularity the curvature is not defined, so a proof that begins by evaluating the extension’s Riemann tensor is circular.

Continuation criteria give another route. Suppose a local existence theorem continues a reduced Einstein solution whenever a geometric Sobolev norm, injectivity radius, lapse bound, and curvature flux remain controlled. A finite maximal time then forces at least one controlled quantity to fail. This is a breakdown alternative, not automatically pointwise curvature blowup; the failed quantity and gauge-to-geometry conversion must be identified.

Weak null boundaries and mass inflation

Charged or rotating black-hole interiors exhibit the sharpest separation of regularities. Radiation decaying along the event horizon is blue-shifted toward the Cauchy horizon. In spherical Einstein–Maxwell–scalar models, the Hawking mass and curvature may diverge while the area radius and metric extend continuously. For vacuum Kerr perturbations, Dafermos and Luk construct a continuous extension across a nonempty Cauchy-horizon segment from suitable spacelike Cauchy data already inside the black-hole interior Dafermos and Luk 2025, main theorem and §§1–3; 2017 preprint. The result is an existence theorem for a $C^0$ extension under its data hypotheses; it is not a claim that curvature remains bounded or that the extension satisfies vacuum Einstein equations classically.

To upgrade mass inflation to a strong-censorship statement one needs a lower bound—typically a Price-law tail or nonvanishing radiation coefficient—that forces blowup generically. An upper decay estimate alone cannot supply that lower bound. One must then translate the divergence into the selected extension class: Hawking-mass blowup may obstruct $C^2$, Christoffel nonintegrability may obstruct weak solutions, yet neither by itself disproves a $C^0$ extension.

A discriminating workflow

For any claimed singularity, ask in order:

• What is the approach: a causal geodesic, a sequence of spheres, a spacelike foliation, or a null boundary?
• Is the quantity scalar, frame dependent, foliation dependent, or a norm over a hypersurface?
• Is the estimate an upper bound, a lower bound, or a two-sided asymptotic?
• Does it hold for one solution, an open set, or a generic set in a named topology?
• Which extension regularity does it actually exclude?

That sequence prevents the common but invalid jump from “a singularity theorem applies” to “curvature blows up and the spacetime is $C^0$-inextendible.”

Check the regularity implication

Assume the Kretschmann scalar $K=R_{\mu\nu\rho\sigma}R^{\mu\nu\rho\sigma}$ tends to infinity along a curve approaching a putative boundary point. Which extension classes are immediately excluded, and why does the same fact not by itself exclude a $C^0$ extension?

Solution

A $C^2$ Lorentzian extension has continuous curvature in a regular coordinate chart, so $K$ must remain bounded on a compact neighborhood of the boundary point; the divergence is a contradiction. With the appropriate essential bound, it also rules out a $C^{1,1}$ extension. A merely continuous metric has no classically defined connection or Riemann tensor. Curvature from the original side may diverge while the metric coefficients converge continuously, as at a weak null singularity. Excluding $C^0$ extension requires a causal, distance, topological, or other genuinely low-regularity obstruction.

References

• Clarke, C. J. S., and Andrzej Królak. “Conditions for the Occurence of Strong Curvature Singularities.” Journal of Geometry and Physics 2 (1985): 127–143. DOI.
• Dafermos, Mihalis, and Jonathan Luk. “The Interior of Dynamical Vacuum Black Holes I: The $C^0$-Stability of the Kerr Cauchy Horizon.” Annals of Mathematics 202 (2025): 309–630. DOI. Open PDF.
• Sbierski, Jan. “The $C^0$-Inextendibility of the Schwarzschild Spacetime and the Spacelike Diameter in Lorentzian Geometry.” Journal of Differential Geometry 108 (2018): 319–378. arXiv.
• Tipler, Frank J. “Singularities in Conformally Flat Spacetimes.” Physics Letters A 64 (1977): 8–10. DOI.