Curvature Blowup and Measures of Singularity Strength

Curvature can signal a singular endpoint in several inequivalent ways. A scalar invariant may diverge; a freely falling orthonormal frame may measure unbounded tidal components even when every scalar stays finite; or the integrated tidal distortion may crush an extended body. None of these conditions is identical to geodesic incompleteness or to inextendibility at a chosen metric regularity.

Required background. Geodesics and Jacobi Fields supplies parallel frames and deviation fields; Curvature supplies the Riemann tensor and invariants. Helpful background. Lorentzian Metrics and Causal Cones supplies observer frames; Causality and Global Structure orients endpoint questions; Exact Solutions of Einstein’s Equations supplies contrasting singular geometries.

Scalar and frame curvature blowup

A scalar polynomial curvature invariant is a complete contraction of polynomials in the Riemann tensor and, when specified, its covariant derivatives. Familiar examples are

$$R, \qquad R_{ab}R^{ab}, \qquad K=R_{abcd}R^{abcd}.$$

If such a scalar diverges along a curve approaching a finite-parameter endpoint, the divergence is coordinate independent. For a polynomial in the Riemann tensor alone, convergence to a candidate boundary point in a $C^2$ extension would force the scalar to have a finite limit there; its divergence obstructs that extension. If the invariant contains up to $m$ covariant derivatives of curvature, the analogous elementary obstruction is to a $C^{m+2}$ extension. Divergence of a curvature derivative alone is not a $C^2$ obstruction. Schwarzschild’s Kretschmann scalar,

$$K=\frac{48G^2M^2}{r^6},$$

is finite at $r=2GM$ and diverges as $r\to0$. It therefore separates the removable static-coordinate horizon from the central scalar-curvature boundary.

Scalar tests are not exhaustive in Lorentzian geometry. Along a timelike geodesic with parallelly propagated orthonormal frame $e_{\hat0}=u,e_{\hat i}$, physical tidal components are

$$\mathcal E_{\hat i\hat j} =R_{\hat0\hat i\hat0\hat j}.$$

A component can diverge while all scalar polynomial invariants remain bounded or even vanish. Plane-wave spacetimes provide the classic mechanism: the algebraic form of curvature makes complete contractions vanish although selected frame components are nonzero and can grow. Parallelly propagated curvature blowup is therefore stronger than a bad coordinate component and complementary to scalar divergence.

The frame condition matters. An artificially boosted tetrad can make components large in a regular region. Parallel transport ties the frame to a freely falling observer and removes that arbitrary acceleration, though the initial finite Lorentz transformation still changes components by finite amounts.

Geodesic deviation and extended bodies

Let $\gamma:[0,s_*)\to M$ be an incomplete causal geodesic whose endpoint is reached at finite affine parameter $s_*$. For a unit-speed timelike geodesic take $s=\tau$, its proper time, and $v^a=u^a$; for a null geodesic take an affine parameter $s=\lambda$ and $v^a=k^a$. A transverse Jacobi field satisfies

$$\frac{D^2Z^a}{ds^2} =-R^a{}_{bcd}v^bZ^cv^d$$

in the site’s curvature convention. For each $b\in[0,s_*)$, let $J_b(\gamma)$ denote the transverse Jacobi fields that vanish at $s=b$. In four spacetime dimensions, an independent triad in $J_b(\gamma)$ along a timelike geodesic defines the transverse 3-volume magnitude

$$V_b(s)=\left\lVert Z_1\wedge Z_2\wedge Z_3\right\rVert.$$

Along a null geodesic, two independent spacelike classes in the screen $k^\perp/\langle k\rangle$ define the transverse 2-area magnitude

$$A_b(s)=\left\lVert \widehat Z_1\wedge\widehat Z_2\right\rVert.$$

The Tipler strong-curvature condition along $\gamma$ requires, for every $b$ and every such independent family,

$$\liminf_{s\to s_*}V_b(s)=0$$

in the timelike case, or $\liminf_{s\to s_*}A_b(s)=0$ in the null case. An ordinary limit need not exist. The weaker Królak limiting-focusing condition requires, for every $b$ and every such family, some $c\in[b,s_*)$ at which $dV_b/ds<0$, or $dA_b/ds<0$ in the null case. Where the corresponding magnitude is positive, this is equivalent to negative expansion. Królak’s condition requires neither vanishing volume nor expansion tending to $-\infty$.

These conditions use an oriented affine parameter. Proper time supplies that parameter for a unit-speed timelike geodesic; a null geodesic has no proper-time parametrization. The quantifiers, the timelike 3-volume, and the null screen 2-area follow Ashley and Scott 2003, Definitions 4.1–4.3, pp. 6–7, which records the Tipler and Królak definitions.

The geometric definitions are primary. Integral criteria involving Ricci or frame Riemann components provide useful sufficient or necessary conditions under additional hypotheses. Roughly, a single integral of a divergent tidal component controls accumulated relative velocity, while a double integral controls accumulated separation. This motivates the schematic tests

$$\int^{\tau_*}d\tau\,\lvert\mathcal E(\tau)\rvert, \qquad \int^{\tau_*}d\tau \int^{\tau}d\tau'\,\lvert\mathcal E(\tau')\rvert,$$

but absolute values can erase cancellations and one component need not control the full Jacobi determinant. Clarke and Królak give the precise curvature conditions and qualifications Clarke and Królak 1985, §§2–4, pp. 127–143. Their one- and two-fold integral criteria are distinct necessary or sufficient theorems with stated causal-type and curvature hypotheses, not a general if-and-only-if test for an arbitrarily selected tidal component. Because the original article contains misleading typographical errors, Ashley and Scott 2003, p. 7 recommend the corrected proofs in Clarke 1993, especially p. 164.

“Strong” therefore describes transverse 3-volume or null-screen 2-area behavior along a specified incomplete geodesic. Any use of “strong” for an entire boundary must state which approaching geodesics are being quantified over. It is not a claim that curvature scalars diverge, that no continuous extension exists, or that quantum matter is destroyed. Each of those is a different question.

Weak singularities and extension regularity

A singular boundary may be null even when the observer used to test its strength follows a timelike geodesic: boundary causal character and probe causal type are separate data. At some null singularities, model-specific estimates establish failure of Tipler strength together with finite deformation, while a separate argument gives a continuous metric extension despite divergent curvature Burko 1999, §§II–IV. None of those conclusions follows from failure of Tipler strength alone.

Failure of the Tipler-strong condition along $\gamma$ means only that the universal $\liminf$ condition above fails. It does not imply that every Jacobi field, the body’s shape distortion, or even its transverse volume remains bounded: unbounded stretching in one direction can be offset by crushing in another. Ori’s deformationally strong condition includes both Tipler-strong behavior and the case in which, for every initialization point $b$, some field in $J_b(\gamma)$ has unbounded norm at the endpoint Ori 2000, pp. 064016-1–064016-4. Finite deformation must therefore be established through deformational weakness or direct bounded-Jacobi estimates, and continuous extendibility requires a separate extension argument. Nolan gives explicit continuous-metric counterexamples to the inference that $C^0$ regularity forces gravitational weakness Nolan 1999, pp. 1–4.

This possibility breaks several tempting implications:

• unbounded curvature does not by itself imply $C^0$-inextendibility;
• finite tidal distortion does not imply a $C^2$ extension;
• bounded scalar invariants do not imply bounded freely falling tidal components; and
• geodesic incompleteness does not imply any curvature blowup.

Conversely, curvature can diverge only as infinite affine parameter is approached. Then the spacetime may remain geodesically complete despite unbounded curvature along a complete curve. Completeness asks whether the parameter ends; blowup asks what a field does along the limit.

The regularity threshold supplies a clean partial implication. If a curve approaches a point in a $C^2$ extension and its parallel frame has a finite limit there, the extended continuous Riemann tensor gives bounded frame components. Divergent parallelly propagated curvature contradicts that scenario. It says less about $C^0$ or $C^1$ extensions, for which classical curvature is not controlled.

Bounded-curvature incompleteness

Flat spacetime with a point deleted is geodesically incomplete and has identically zero curvature. A flat cone away from its apex combines local flatness with incomplete radial geodesics and nontrivial holonomy. Quotient constructions can add causal pathology while leaving every local invariant unchanged. These examples establish that incompleteness is fundamentally a global curve property.

They also discipline singularity language. The puncture is smoothly removable; the cone may be quasiregular rather than curvature divergent; a non-Hausdorff attempted filling can be topologically obstructed. One should report the measured facts—finite affine length, curvature behavior, holonomy, and extension class—instead of ranking examples by an undefined scalar “severity.”

Tipler’s original strong-curvature criterion was designed to capture destructive tidal behavior beyond mere incompleteness Tipler 1977, pp. 8–10. It is most informative when used alongside, not in place of, extension and causal classifications.

Exercise

Suppose that, along a unit-speed timelike geodesic with proper time $\tau$ and finite endpoint $\tau_*$, one parallelly propagated tidal eigenvalue behaves as $\mathcal E(\tau)\sim C(\tau_* -\tau)^{-p}$. Determine when its single and double absolute integrals diverge.

Solution

Set $x=\tau_* -\tau$. The single integral behaves as $\int_0 x^{-p}dx$, which diverges for $p\geq1$ and is finite for $p\lt1$. Integrating twice gives a leading integral $\int_0 x^{1-p}dx$ for $p\neq1$, which diverges for $p\geq2$ and is finite for $p\lt2$; at $p=1$ the first integral is logarithmic but the second remains finite.

Thus the range $1\leq p\lt2$ can produce divergent accumulated relative velocity while the twice-integrated displacement remains finite. This power count explains why one- and two-fold curvature integrals have different convergence thresholds; it does not determine Królak or Tipler strength of the full Jacobi family.

Abstract Boundaries and Ideal Singular Endpoints next asks how such limiting behavior can be attached and compared without declaring the endpoint a physical event.

References

• Ashley, Michael J. S. L., and Susan M. Scott, “Curvature Singularities and Abstract Boundary Singularity Theorems for Space-Time”, arXiv:gr-qc/0309070, 2003.
• Burko, Lior M., “Strength of the Null Singularity Inside Black Holes”, Physical Review D 60, 104033, 1999; Open PDF.
• Clarke, C. J. S., The Analysis of Space-Time Singularities, Cambridge Lecture Notes in Physics 1, Cambridge University Press, 1993.
• Clarke, C. J. S., and A. Królak, “Conditions for the Occurrence of Strong Curvature Singularities”, Journal of Geometry and Physics 2, pp. 127–143, 1985.
• Królak, Andrzej, “Towards the Proof of the Cosmic Censorship Hypothesis”, Classical and Quantum Gravity 3, pp. 267–280, 1986.
• Nolan, Brien C., “A Regular C⁰ Singularity Is Not Necessarily Weak”, arXiv:gr-qc/9902020, 1999.
• Ori, Amos, “Strength of Curvature Singularities”, Physical Review D 61, 064016, 2000.
• Tipler, Frank J., “Singularities in Conformally Flat Spacetimes”, Physics Letters A 64, pp. 8–10, 1977.