The Timelike Focusing Theorem

The timelike focusing theorem turns the Raychaudhuri identity into a finite-proper-time statement. An initially contracting, hypersurface-orthogonal timelike geodesic congruence subject to nonnegative timelike Ricci curvature cannot remain smooth indefinitely: in four dimensions it develops a focal point within proper time at most $3/|\theta_0|$. The result is local and conditional; it is not by itself a singularity theorem.

Required background. We use timelike congruence kinematics, Einstein’s equation, and the metric definition of stress-energy.

Helpful background. The causal normalization uses special relativity and smooth tensor fields. Global incompleteness consequences require the separate theory of causality and global structure.

Theorem and hypotheses

Timelike focusing theorem. Let $u^a$ be the future-directed unit tangent to a smooth timelike geodesic congruence in $D\ge2$ dimensions, normal to an initial spacelike hypersurface. Suppose, along a chosen geodesic and up to its first focal point,

1. proper time $\tau$ is used and $u^b\nabla_bu^a=0$;
2. hypersurface orthogonality gives $\omega_{ab}=0$;
3. $R_{ab}u^au^b\ge0$;
4. the initial expansion is $\theta(0)=\theta_0<0$; and
5. the spacetime and normal Jacobi fields remain regular along the geodesic, and the expansion is differentiable before the first focal point.

If the geodesic extends for proper time $(D-1)/|\theta_0|$, then a focal point occurs no later than

$$\tau_*\le\frac{D-1}{|\theta_0|}.$$

In four dimensions, $\tau_*\le3/|\theta_0|$. MIT derives the Raychaudhuri identity and the required differential inequality Guth–Turner 2017, Lecture 14, pp. 59–61, Eqs. (14.14)–(14.18), PDF; Kontou and Sanders state the integrated focal-time bound Kontou and Sanders 2020, § 5.1, p. 40, Eqs. (105)–(106), PDF.

Normality to a hypersurface is stronger than choosing zero vorticity at one arbitrary point; it supplies a twist-free congruence in the smooth domain. The theorem requires that the geodesic extend to the comparison time unless it has already focused. It does not assume that the congruence coordinates stay regular there: their failure at a focal point is the conclusion. This distinction matters when the result is later combined with completeness assumptions.

The elementary proof also assumes enough regularity for the expansion to be continuously differentiable and the curvature contraction to be integrable. If the congruence has already formed a caustic, $u^a$ is not a single smooth vector field there and the same differential equation cannot simply be continued through the crossing.

Proof by a scalar comparison inequality

For a timelike geodesic congruence, Raychaudhuri gives

$$\frac{d\theta}{d\tau} =-\frac{\theta^2}{D-1} -\sigma_{ab}\sigma^{ab} +\omega_{ab}\omega^{ab} -R_{ab}u^au^b.$$

The spatial shear norm is nonnegative. With $\omega_{ab}=0$ and timelike convergence,

$$\frac{d\theta}{d\tau}\le-\frac{\theta^2}{D-1}.$$

As long as $\theta$ remains finite and negative, define $z=1/\theta$. Then

$$\frac{dz}{d\tau} =-\frac{1}{\theta^2}\frac{d\theta}{d\tau} \ge\frac{1}{D-1}.$$

Integration yields

$$\frac1{\theta(\tau)} \ge\frac1{\theta_0}+\frac{\tau}{D-1}.$$

The right side reaches zero at $(D-1)/|\theta_0|$. A continuous finite negative $\theta$ cannot persist through that time because its reciprocal cannot pass through zero. Hence $\theta\to-\infty$ at or before the bound. The equality case solves

$$\theta(\tau)=\frac{\theta_0} {1+\theta_0\tau/(D-1)},$$

confirming the time and its dimensions.

Kontou and Sanders present the comparison logic and distinguish focal points from the extra hypotheses in Hawking- and Penrose-type results Kontou and Sanders 2020, § 5.1, pp. 39–41, Eqs. (105)–(110), PDF.

From divergent expansion to a focal point

Let $\mathcal V$ be the positive infinitesimal $(D-1)$-volume spanned by independent Jacobi fields orthogonal to $u^a$, and put $\tau_0=(D-1)/|\theta_0|$. Before a focal point,

$$\theta=\frac{d}{d\tau}\ln\mathcal V.$$

For $0\le\tau<\tau_0$, the reciprocal inequality is equivalent, while $\theta<0$, to

$$\theta(\tau) \le\frac{\theta_0}{1+\theta_0\tau/(D-1)} =-\frac{D-1}{\tau_0-\tau}.$$

Integrating the volume equation therefore gives

$$0<\mathcal V(\tau) \le\mathcal V(0)\left(1-\frac{\tau}{\tau_0}\right)^{D-1}.$$

If no focal point occurred earlier, the upper bound would force the Jacobi determinant to vanish as $\tau\uparrow\tau_0$. Thus a focal point occurs at or before $\tau_0$. At that event the exponential map normal to the initial hypersurface loses rank: neighboring normal geodesics focus, and the event is focal to the hypersurface. The geodesics themselves can continue smoothly; it is the congruence chart and the maximizing property of the segment that change.

This geometric step is why a divergence of $\theta$ is not automatically a curvature singularity. Even flat spacetime geodesics aimed inward can meet at a focal point with every curvature invariant zero.

An exact flat example makes this concrete. In the past light cone of the Minkowski origin, start from the spacelike hyperboloid $t=-\sqrt{T^2+r^2}$ with $T\gt0$. Its future unit normal geodesics are straight lines aimed at the origin. Every one reaches it after proper time $T$, and their transverse Jacobi lengths scale as $1-\tau/T$. Therefore

$$\mathcal V(\tau)=\mathcal V(0)(1-\tau/T)^{D-1}, \qquad \theta(\tau)=-\frac{D-1}{T-\tau}.$$

The congruence has zero shear and vorticity, and the spacetime has zero curvature. It saturates the bound exactly. The straight geodesics pass smoothly through the origin even though their normal congruence has focused there.

Before its first focal point, a normal geodesic segment has the relevant local proper-time extremizing property relative to the initial hypersurface. Beyond a focal point that property fails. This supplies a bridge used in global singularity arguments, but proving the bridge requires the second variation and causal geometry in addition to the scalar comparison above.

Why each hypothesis matters

If $\theta_0\ge0$, the comparison gives no future focusing bound. Positive shear assists focusing, but vorticity appears with the opposite sign and can oppose it; a rotating flat-spacetime congruence shows why twist cannot simply be omitted. Acceleration adds $\nabla_aa^a$, so supported worldlines are not covered by the geodesic theorem.

Positive cosmological constant provides another discriminator. With $R_{ab}=\Lambda g_{ab}$ in four-dimensional vacuum,

$$R_{ab}u^au^b=-\Lambda<0$$

for $\Lambda>0$, violating timelike convergence and allowing defocusing. Matter SEC implies convergence only after the cosmological term is assigned consistently.

Dropping the shear term is safe for this one-sided bound because $\sigma_{ab}\sigma^{ab}\ge0$: its presence only makes $d\theta/d\tau$ more negative and can focus sooner. Dropping vorticity would not be safe because its sign is positive. This sign comparison explains why hypersurface orthogonality is a real theorem hypothesis rather than a convenient simplification.

What the theorem does not conclude

A focal point is not geodesic incompleteness. A singularity theorem adds global causal hypotheses, a trapped or contracting initial set, and arguments about maximizing geodesics or compactness to turn focusing into a contradiction with completeness. The theorem here supplies the local differential engine only. It makes no claim that curvature blows up, topology changes, or every geodesic ends.

The numerical bound is stated in proper time along unit timelike geodesics. It should not be copied to an arbitrary curve parameter. Proper-time normalization fixes the scale physically, unlike the remaining constant rescaling freedom of an affine null parameter.

Exercises

In four dimensions take $\theta_0=-6\,\mathrm{s}^{-1}$ and assume the equality case. Find the focusing time and $\theta(0.4\,\mathrm{s})$.

Solution

The bound is $3/|\theta_0|=0.5\,\mathrm{s}$. The equality solution is

$$\theta(\tau) =\frac{-6\,\mathrm{s}^{-1}} {1-(2\,\mathrm{s}^{-1})\tau}.$$

At $0.4\,\mathrm{s}$, $\theta=-30\,\mathrm{s}^{-1}$. It diverges to $-\infty$ as $\tau\uparrow0.5\,\mathrm{s}$.

References

• Guth, Alan H., lectures with notes by Andrew P. Turner. 8.962 General Relativity, Spring 2017. MIT (2017). PDF.
• Kontou, Eleni-Alexandra, and Ko Sanders. “Energy Conditions in General Relativity and Quantum Field Theory.” Classical and Quantum Gravity 37 (2020): 193001. DOI; arXiv.