Focusing Theorem and Conjugate Points

The Focusing Theorem is the geometric heart of Lorentzian global analysis — the hinge on which local curvature is converted into global fate. It rests on a single kinematic identity, the Raychaudhuri equation, which tracks how the cross-sectional volume of a bundle of nearby free-falling particles evolves along the flow. Three numbers organize it: the expansion $\theta$ (the fractional rate of change of its volume), the shear $\sigma_{ab}$ (anisotropic distortion), and the twist $\omega_{ab}$ (rotation). The identity holds in any spacetime, before a single field equation is imposed — pure differential geometry of a geodesic congruence. Physics enters through one term: the Ricci contraction $R_{ab}u^au^b$ . Einstein's equation ties this to the local energy-momentum, and the strong and null energy conditions — obeyed by classical matter — force it non-negative. With twist switched off, every term on the right is then non-positive, so once a bundle starts converging it can only converge harder. Gravity, in this precise sense, is universally attractive: it focuses geodesics and never defocuses them. The payoff is the conjugate point. The $-\theta^2$ self-interaction drives $\theta\to-\infty$ within a bounded affine length once convergence begins — a caustic where nearby geodesics recross, equivalently where a non-trivial Jacobi field vanishes twice. Past it, a timelike geodesic is no longer the longest curve to its endpoint, and a null generator drops off the boundary $\partial J^{+}$ into the chronological interior. This is the exact mechanism behind the Penrose and Hawking singularity theorems and the black-hole area theorem.

Key takeaways

The Raychaudhuri equation governs how a congruence of timelike or null geodesics expands, shears, and twists; for twist-free families $\mathrm{d}\theta/\mathrm{d}\tau=-\theta^2/(n-1)-\sigma_{ab}\sigma^{ab}-R_{ab}u^au^b$ with $n-1=3$ (timelike) or $n-1=2$ (null).
The Focusing Theorem says that if an energy condition makes $R_{ab}u^au^b\ge0$ , an initially converging congruence ( $\theta_0<0$ ) is driven to $\theta\to-\infty$ within finite affine parameter — a caustic where neighbouring geodesics cross.
A conjugate point is where the expansion blows up, equivalently where a non-trivial Jacobi field vanishes at two points; nearby geodesics from one source refocus, exactly as great circles from a pole meet at the antipode.
Past a conjugate point a timelike geodesic stops maximizing proper time, and a null geodesic leaves $\partial J^{+}$ and enters the chronological future $I^{+}$ ; this is the engine of the singularity theorems and the horizon area law $\mathrm{d}A/\mathrm{d}v\ge0$ .
Sharp bounds: if $\theta_0<0$ a conjugate point occurs within affine length $\le 3/|\theta_0|$ (timelike) or $\le 2/|\theta_0|$ (null); twist and a positive $\Lambda$ can defeat it.
Key idea: Raychaudhuri's equation plus an energy condition makes $\mathrm{d}\theta/\mathrm{d}\tau\le-\theta^2/(n-1)$ ; integrating $\theta^{-1}$ shows any initial convergence drives $\theta\to-\infty$ within finite affine length, producing a conjugate point past which the geodesic stops being causally extremal.

The core insight

Raychaudhuri's equation plus an energy condition makes $\mathrm{d}\theta/\mathrm{d}\tau\le-\theta^2/(n-1)$ ; integrating $\theta^{-1}$ shows that any initial convergence drives $\theta\to-\infty$ within finite affine length, producing a conjugate point past which the geodesic stops being causally extremal.

Where this lives today

Focusing is the working engine of several live frontiers. The 2020 Nobel Prize to Roger Penrose recognized the argument built on null focusing — that black-hole formation is a robust prediction of general relativity — and the trapped surfaces at its core are now located inside the numerical-relativity simulations behind gravitational-wave templates, as apparent horizons ( $\theta=0$ ) on each time slice. The same Jacobi–Raychaudhuri machinery, repackaged as the Sachs optical equations, propagates light bundles through an inhomogeneous universe: their caustics are the critical curves of gravitational lensing, seen as the giant arcs and Einstein rings mapped by Hubble and JWST and used to weigh dark matter. And the semiclassical quantum focusing conjecture of Bousso and collaborators — that a generalized entropy obeys a focusing inequality even where the classical energy conditions fail — extends the theorem into the quantum regime, underpinning modern proofs of the averaged and quantum null energy conditions. Wherever gravity is asked whether it must produce a singularity, an area increase, or a caustic, the answer still comes from Raychaudhuri's one line.

The physical picture

Release a cloud of test particles at rest relative to one another and let them fall freely. The expansion $\theta$ is the percentage growth rate of the cloud's volume. Tidal gravity — the Ricci focusing term — squeezes the cloud; classical matter never inflates it. The subtlety that makes the theorem inevitable is the $-\theta^2$ term: convergence feeds on itself, so a cloud that has begun to shrink shrinks faster and faster, reaching a focal point in finite time exactly as a converging lens brings parallel rays to a focus. The first figure draws such a congruence launched from a spacelike surface, converging to a conjugate point where the bundle collapses to a caustic.

Figure

A twist-free congruence launched from a spacelike surface Σ with negative initial expansion θ₀ focuses to a conjugate point — a caustic where neighbouring geodesics cross — within affine length at most 3/θ₀ under the strong energy condition

A conjugate point is the relativistic version of two everyday facts. On a sphere every great circle leaving the north pole reconverges at the south pole: the antipode is conjugate to the pole, and beyond it a great-circle arc is no longer the shortest route. A lens gathers parallel rays to a focus and, past it, the rays cross and spread. Lorentzian signature flips the extremal principle — timelike geodesics maximize proper time rather than minimize length — but the geometry of recrossing Jacobi fields is identical, and maximality fails past the conjugate point for the same reason a great-circle arc stops being shortest past the antipode. The second figure turns the theorem into a straight line: because $\mathrm{d}\theta/\mathrm{d}\tau\le-\theta^2/(n-1)$ , the quantity $1/\theta$ climbs with minimum slope $1/(n-1)$ , so a negative $1/\theta$ is dragged up through zero in bounded parameter — and $1/\theta\to0^{-}$ means $\theta\to-\infty$ .

Figure

The focusing bound as a straight line: 1/θ rises with slope at least 1/3, so from 1/θ₀ = −0.33 s at θ₀ = −3 s⁻¹ it must reach zero by τ⋆ = 3/θ₀ = 1 s, where θ diverges to −∞ and a conjugate point is guaranteed

The core equations

The whole subject flows from one evolution equation. Decompose the gradient of the congruence's four-velocity into its trace (expansion), symmetric trace-free part (shear), and antisymmetric part (twist), then differentiate the trace along the flow using the Ricci identity — the result is Raychaudhuri's equation, an exact identity with no field equations and no approximations.

\frac{\mathrm{d}\theta}{\mathrm{d}\tau}=-\frac{1}{3}\theta^2-\sigma_{ab}\sigma^{ab}+\omega_{ab}\omega^{ab}-R_{ab}u^au^b \qquad (1)

where: $\theta=\nabla_a u^a$ — expansion scalar, the fractional volume rate of the congruence; $\sigma_{ab}$ — shear (symmetric, trace-free, transverse); $\omega_{ab}$ — vorticity/twist (antisymmetric, transverse); $u^a$ — unit tangent, $u^au_a=-1$ ; $R_{ab}u^au^b$ — the curvature focusing term; $\tau$ — proper time. Reading it: the first two terms always focus ( $-\theta^2/3\le0$ , $-\sigma^2\le0$ ), twist defocuses ( $+\omega^2\ge0$ ), and the matter term focuses under the strong energy condition. For light the transverse geometry is two-dimensional rather than three, so the self-focusing coefficient changes from $1/3$ to $1/2$ . The expansion lives on the two-dimensional "screen space" orthogonal to the null generator, parametrized affinely — proper time is meaningless on a null curve.

\frac{\mathrm{d}\theta}{\mathrm{d}\lambda}=-\frac{1}{2}\theta^2-\sigma_{ab}\sigma^{ab}+\omega_{ab}\omega^{ab}-R_{ab}k^ak^b \qquad (2)

where: $k^a$ — null tangent, $k^ak_a=0$ , affinely parametrized; $\lambda$ — affine parameter along the generators; $\theta$ — expansion of the two-dimensional screen; $\sigma_{ab},\omega_{ab}$ — shear and twist on the screen. Reading it: the divisor is now $2=n-1$ , the transverse dimension, which is why the null bound carries a $2$ where the timelike bound carries a $3$ . Now impose an energy condition and set the twist to zero. Every retained term on the right is then non-positive, weakening the equation to an inequality; dividing by $-\theta^2$ (which flips the sense while $\theta<0$ ) turns it into a linear lower bound on $1/\theta$ .

\frac{\mathrm{d}\theta}{\mathrm{d}\tau}\le-\frac{1}{3}\theta^2\quad\Longrightarrow\quad \frac{1}{\theta(\tau)}\ge\frac{1}{\theta_0}+\frac{\tau}{3} \qquad (3)

where: $\theta_0<0$ — initial expansion at $\tau=0$ . Reading it: the right-hand side reaches zero at $\tau=3/|\theta_0|$ , so $1/\theta$ is forced up through zero — meaning $\theta\to-\infty$ — at or before that proper time. This is the Focusing Theorem in analytic form; its null twin replaces $3$ by $2$ . The blow-up of $\theta$ is a genuine geometric event named by the Jacobi equation: a deviation vector connecting neighbouring geodesics obeys the equation of geodesic deviation, driven entirely by curvature.

\frac{\mathrm{D}^2 \xi^a}{\mathrm{d}\tau^2}=-R^{a}{}_{cbd}\,u^c\xi^b u^d \qquad (4)

where: $\xi^a$ — deviation (Jacobi) field; $\mathrm{D}/\mathrm{d}\tau=u^c\nabla_c$ — covariant derivative along the geodesic; $R^{a}{}_{cbd}$ — Riemann tensor. Reading it: a Jacobi field measures the tidal separation of nearby geodesics, so a solution vanishing at two parameters marks geodesics that leave one point and meet again. That coincidence is exactly the definition of conjugacy, and it links back to the expansion through the Jacobi matrix whose determinant measures the transverse volume of the bundle.

p=\gamma(\tau_1)\ \text{conjugate to}\ q=\gamma(\tau_0)\iff \exists\,\xi^a\neq0:\ \xi^a(\tau_0)=\xi^a(\tau_1)=0,\qquad \theta=\frac{\mathrm{d}}{\mathrm{d}\tau}\ln\det A^{i}{}_{j} \qquad (5)

where: $\gamma$ — the geodesic; $\xi^a$ — non-trivial Jacobi field vanishing at both endpoints; $A^{i}{}_{j}$ — Jacobi matrix mapping initial deviations to deviations at $\tau$ , with transverse volume $V\propto\det A$ . Reading it: a conjugate point is where $\det A\to0$ , and since $\theta$ is the logarithmic derivative of $\det A$ , the vanishing determinant is precisely the $\theta\to-\infty$ of the focusing inequality — analytic blow-up and geometric caustic are one event. Applied to a black-hole horizon, focusing runs in reverse: a generator can never leave the horizon to the future, so its expansion can never go negative, and non-negative expansion means non-decreasing area.

\theta_{H}\ge0\ \text{on a future event horizon}\ \Longrightarrow\ \frac{\mathrm{d}A}{\mathrm{d}v}\ge0 \qquad (6)

where: $\theta_H$ — expansion of the horizon's null generators; $A$ — cross-sectional area; $v$ — advanced time along the generators. Reading it: this is Hawking's area theorem — the classical second law of black-hole mechanics — nothing but the Focusing Theorem read as a prohibition.

Derivation AFrom the velocity gradient to the Raychaudhuri equation

The goal is to obtain equation (1) as a pure geometric identity, valid in any spacetime and independent of the Einstein equations, by decomposing the gradient of the congruence's four-velocity into expansion, shear, and twist and evolving its trace along the flow. Every ingredient is kinematic — the geodesic equation, the Ricci identity, and the algebraic split of a transverse tensor into trace, symmetric-traceless, and antisymmetric parts — the bedrock on which all the physics is later hung.

Step 1. Define the transverse velocity gradient $B_{ab}=\nabla_b u_a$ and split it as $B_{ab}=\tfrac13\theta h_{ab}+\sigma_{ab}+\omega_{ab}$ , where $h_{ab}=g_{ab}+u_au_b$ projects orthogonal to $u^a$ , $\theta=h^{ab}B_{ab}$ , $\sigma_{ab}=B_{(ab)}-\tfrac13\theta h_{ab}$ , and $\omega_{ab}=B_{[ab]}$ . Why: for a geodesic $u^c\nabla_c u_a=0$ forces $B_{ab}u^a=B_{ab}u^b=0$ , so $B_{ab}$ is transverse and the split is exhaustive.

Step 2. Evolve $B_{ab}$ along the flow using the Ricci identity: $u^c\nabla_c B_{ab}=\nabla_b(u^c\nabla_c u_a)-(\nabla_b u^c)(\nabla_c u_a)-R_{acbd}u^cu^d=-B_{cb}B^{c}{}_{a}-R_{acbd}u^cu^d$ . Why: the geodesic equation kills the first term, the second is $-B$ contracted with itself, and the third is the curvature that commuting two covariant derivatives always produces — this is where Riemann enters.

Step 3. Take the trace with $g^{ab}$ , using $g^{ab}R_{acbd}u^cu^d=R_{cd}u^cu^d$ , to get $\mathrm{d}\theta/\mathrm{d}\tau=-B_{ab}B^{ba}-R_{ab}u^au^b$ . Why: contracting collapses the tensor evolution to a scalar equation for the expansion, isolating the curvature scalar the energy conditions later control.

Step 4. Substitute the split of Step 1 into the quadratic: $B_{ab}B^{ba}=\tfrac13\theta^2+\sigma_{ab}\sigma^{ab}-\omega_{ab}\omega^{ab}$ , since the three pieces are mutually orthogonal and $\omega_{ab}\omega^{ba}=-\omega_{ab}\omega^{ab}$ . Inserting this yields equation (1). Why: the antisymmetric part contracts with a relative minus sign — the algebraic origin of the defocusing $+\omega^2$ , the single sign that makes rotation resist collapse.

Go deeper · Derivation B · 1 minThe focusing inequality and the conjugate-point bound

Derivation BThe focusing inequality and the conjugate-point bound

The goal is to convert the exact identity (1) into the sharp statement that an initially converging twist-free congruence must reach a conjugate point within affine length $3/|\theta_0|$ (timelike) or $2/|\theta_0|$ (null), given only an energy condition. The strategy: discard the manifestly focusing terms for a one-line differential inequality, pass to the reciprocal variable where it becomes linear, and integrate — a Riccati-to-linear trick that makes the finite-time blow-up transparent.

Step 1. Set $\omega_{ab}=0$ (hypersurface-orthogonal congruence) and invoke the energy condition $R_{ab}u^au^b\ge0$ (strong for timelike, null for null). Since $\sigma_{ab}\sigma^{ab}\ge0$ too, equation (1) gives $\mathrm{d}\theta/\mathrm{d}\tau\le-\tfrac13\theta^2$ . Why: dropping two non-positive terms only strengthens the focusing, so the bound is conservative — the real congruence focuses at least this fast.

Step 2. While $\theta<0$ the quantity $1/\theta$ is well defined; differentiate it: $\dfrac{\mathrm{d}}{\mathrm{d}\tau}\!\left(\dfrac1\theta\right)=-\dfrac{1}{\theta^2}\dfrac{\mathrm{d}\theta}{\mathrm{d}\tau}\ge\dfrac13$ . Why: dividing by $-\theta^2<0$ reverses the inequality and cancels the quadratic exactly, leaving a constant lower bound on a linear rate — the point of the reciprocal substitution.

Step 3. Integrate from $\tau=0$ : $\dfrac{1}{\theta(\tau)}\ge\dfrac{1}{\theta_0}+\dfrac{\tau}{3}$ . With $\theta_0<0$ the starting value $1/\theta_0$ is negative, and the right-hand side rises to zero at $\tau=3/|\theta_0|$ . Why: $1/\theta$ is squeezed up through zero at or before this parameter value, and $1/\theta\to0^{-}$ can only mean $\theta\to-\infty$ — a caustic.

Step 4. Identify the blow-up as a conjugate point and repeat for null. By equation (5) the divergence $\theta\to-\infty$ is $\det A\to0$ , a Jacobi field vanishing a second time — a conjugate point to the initial surface. For a null congruence the coefficient is $1/2$ , so the identical algebra gives $\lambda^\ast\le2/|\theta_0|$ . Why: the only change from timelike to null is the transverse dimension $n-1$ (3 versus 2), which sets the divisor and hence the affine bound; everything else is signature-independent.

Go deeper · Derivation C · 2 minLoss of maximality past a conjugate point

Derivation CLoss of maximality past a conjugate point

The goal is the variational punchline: a timelike geodesic from $p$ ceases to be a local maximum of proper time the instant it passes the first point conjugate to $p$ , so points beyond it can be reached by strictly longer neighbouring curves. The method is the second variation of the length functional — the index form — using the fact that a Jacobi field vanishing at both ends is a null direction of it, then smoothing a corner to push the index positive. This is the Morse-theoretic bridge from the local caustic to the global statement that a null generator leaves $\partial J^{+}$ .

Step 1. Write the second variation of proper time about the geodesic for a transverse variation $\xi_\perp$ vanishing at the endpoints, $\delta^2\tau=-\int_0^{\tau_1}\!\big[(\mathrm{D}\xi_\perp/\mathrm{d}\tau)^2-R_{abcd}u^a\xi_\perp^b u^c\xi_\perp^d\big]\,\mathrm{d}\tau\equiv-I(\xi_\perp,\xi_\perp)$ . Why: for timelike curves "longer" means larger $\tau$ , so the geodesic maximizes proper time exactly when the index form $I$ is positive-definite on admissible variations — maximality is a statement about the sign of $I$ .

Step 2. Note that a Jacobi field is a zero mode of the index form. At the first conjugate point $q=\gamma(\tau_c)$ there is a non-trivial Jacobi field $J$ with $J(0)=J(\tau_c)=0$ , and an integration by parts using equation (4) gives $I(J,J)=0$ on $[0,\tau_c]$ . Why: the Jacobi equation is the Euler–Lagrange equation of $I$ , so its solutions annihilate the index form — the conjugate point is where $I$ first develops a null direction.

Step 3. Extend $J$ past the conjugate point by zero, defining $\tilde\xi=J$ on $[0,\tau_c]$ and $\tilde\xi=0$ on $[\tau_c,\tau_1]$ with $\tau_1>\tau_c$ . This continuous field reaches the later endpoint $q'=\gamma(\tau_1)$ and still has $I(\tilde\xi,\tilde\xi)=0$ , but it has a corner at $\tau_c$ where $\mathrm{D}\tilde\xi/\mathrm{d}\tau$ jumps. Why: the corner is the geometric signature that two geodesics of the family meet at $q$ ; a smooth extremal would not kink, and that kink is what we exploit.

Step 4. Round the corner: for a suitable field $W$ localized at $\tau_c$ , the smoothed variation satisfies $I(\tilde\xi+\varepsilon W,\tilde\xi+\varepsilon W)<0$ for small $\varepsilon$ , hence $\delta^2\tau>0$ . A neighbouring curve to $q'$ is strictly longer, so the geodesic is not maximal past $q$ ; in the null version $q'$ then lies in $I^{+}(p)$ rather than on $\partial J^{+}(p)$ . Why: this corner-smoothing lemma is exactly what the singularity theorems need — a boundary-generating null geodesic cannot contain a conjugate point, yet focusing forces one.

Limiting cases

The focusing inequality specializes sharply by congruence type and matter content; the table collects the four cases that matter most.

Case	Result	Meaning
Timelike, strong energy condition	$\dfrac{\mathrm{d}\theta}{\mathrm{d}\tau}\le-\dfrac{\theta^2}{3}$	Irrotational timelike geodesics with $R_{ab}u^au^b\ge0$ ; a conjugate point occurs within proper time $3/\lvert\theta_0\rvert$ if $\theta_0<0$ (divisor $n-1=3$ ).
Null, null energy condition	$\dfrac{\mathrm{d}\theta}{\mathrm{d}\lambda}\le-\dfrac{\theta^2}{2}$	Null generators with $R_{ab}k^ak^b\ge0$ ; the screen is two-dimensional, so the divisor is $2$ and a caustic occurs within $2/\lvert\theta_0\rvert$ .
Vacuum with $\Lambda$ only	$R_{ab}u^au^b=-\Lambda$ (timelike, $u\cdot u=-1$ )	With $R_{ab}=\Lambda g_{ab}$ , a positive $\Lambda$ adds a defocusing $+\Lambda$ to $\mathrm{d}\theta/\mathrm{d}\tau$ ; the strong energy condition fails and de Sitter geodesics never refocus.
Rotating (non-zero twist)	$\dfrac{\mathrm{d}\theta}{\mathrm{d}\tau}=-\dfrac{\theta^2}{3}-\sigma^2+2\omega^2-R_{ab}u^au^b$	The $+2\omega^2$ vorticity term opposes focusing; sufficient rotation averts the caustic — centrifugal support against collapse.

Common trapa conjugate point is a singularity

A conjugate point is not a singularity, and twist is not optional. First, a conjugate point is a property of a congruence, not of the manifold: it is where neighbouring geodesics recross, like the antipode of a sphere or the focus of a lens, and the metric and curvature are perfectly regular there. Conjugate points prove singularities only when a global argument forbids the boundary-generating geodesic from containing one, yet focusing forces one — the contradiction is geodesic incompleteness, which is the definition. Second, the focusing inequality requires vanishing vorticity: the identity is $B_{ab}B^{ba}=\tfrac13\theta^2+\sigma^2-\omega^2$ , so twist enters Raychaudhuri as $+2\omega^2$ , a defocusing term. A rotating congruence resists collapse, which is why every focusing theorem is stated for hypersurface-orthogonal families — geodesics normal to a spacelike surface, or the generators of a null hypersurface.

Worked examples

1 · The Hubble time as a focusing bound (fully worked)

Model the matter of the universe as a congruence of comoving observers. In an expanding FRW universe its expansion is $\theta=3H$ , with $H=\dot a/a$ the Hubble rate; take $H_0=70\ \mathrm{km\,s^{-1}\,Mpc^{-1}}$ today. Time-reverse the congruence and use the timelike focusing bound to cap the proper time back to a conjugate point of today's slice — Hawking's cosmological singularity in miniature.

Step 1. Convert $H_0$ to inverse seconds. With $1\ \mathrm{Mpc}=3.086\times10^{19}\ \mathrm{km}$ , $H_0=70/(3.086\times10^{19})=\mathbf{2.27\times10^{-18}\ s^{-1}}$ . The comoving congruence has expansion $\theta=3H_0$ ; running it into the past reverses the affine parameter, so the past-directed expansion is $\theta_0=-3H_0=-6.81\times10^{-18}\ \mathrm{s^{-1}}$ — a converging congruence, the hypothesis of the theorem.

Step 2. Apply the strong-energy-condition bound $\tau\le 3/|\theta_0|$ . Here $|\theta_0|=3H_0$ , so the 3's cancel: $\tau\le 3/(3H_0)=1/H_0$ . The bound is exactly the Hubble time $1/H_0=4.41\times10^{17}\ \mathrm{s}=\mathbf{14.0\ Gyr}$ .

Step 3. Interpret. Every past-directed timelike geodesic in today's slice must hit a conjugate point within one Hubble time into the past, provided ordinary (strong-energy-condition) matter fills the intervening spacetime. Past that focal point the geodesics cannot be extended while remaining maximal: the theorem forces past timelike incompleteness, a Big Bang. The measured age of the universe, $13.8\ \mathrm{Gyr}$ , sits just under this $14.0\ \mathrm{Gyr}$ ceiling, as it must. The focusing bound reproduces the Hubble time as an upper limit on the age of an SEC-obeying universe.

2 · A caustic in flat space saturates the null bound

Take a sphere of radius $r_0$ in Minkowski spacetime and fire the ingoing radial light rays inward: no matter ( $R_{ab}=0$ ), and by spherical symmetry no shear ( $\sigma_{ab}=0$ ). Show that the null bound $\lambda\le2/|\theta_0|$ is saturated exactly, and the caustic is a genuine focal point, not a singularity.

Step 1. Compute the initial expansion. For a sphere of area-radius $r$ the ingoing null congruence has $\theta=-2/r$ , so at the launch sphere $\theta_0=-2/r_0<0$ . Affinely parametrize the ingoing radial rays by $\lambda$ with $r=r_0-\lambda$ , so $\lambda$ runs from $0$ at the sphere to $r_0$ at the centre.

Step 2. Check the exact Raychaudhuri evolution. With $R_{ab}k^ak^b=0$ and $\sigma=0$ , equation (2) reads $\mathrm{d}\theta/\mathrm{d}\lambda=-\tfrac12\theta^2$ . Test the candidate $\theta=-2/(r_0-\lambda)$ : its derivative is $-2/(r_0-\lambda)^2$ , and $-\tfrac12\theta^2=-\tfrac12\cdot4/(r_0-\lambda)^2=-2/(r_0-\lambda)^2$ . The two sides agree identically — the bare $-\tfrac12\theta^2$ term drives the focusing all by itself.

Step 3. Locate the caustic and compare with the bound. As $\lambda\to r_0$ , $\theta=-2/(r_0-\lambda)\to-\infty$ : all the rays cross at the centre, $\lambda^\ast=r_0$ , while the bound predicts $\lambda\le2/|\theta_0|=2/(2/r_0)=r_0$ . The bound is saturated exactly, $\lambda^\ast=r_0=2/|\theta_0|$ . Yet Minkowski space is flat and regular at the centre — the caustic is a focal point, not a singularity, the cleanest illustration that focusing is driven by convergence itself, matter and shear only hastening it.

Physics of the portrait

The Focusing Theorem carries one name above all others. Its central identity was written down by a young physicist working largely in isolation, and only years later did Penrose and Hawking see it was the missing lever for turning Einstein's local equations into theorems about the global fate of spacetime.

Origins · 1 minAmal Kumar Raychaudhuri and how this was discovered

Amal Kumar Raychaudhuri

1923–2005 · Indian relativist who found the focusing equation

Amal Kumar Raychaudhuri derived his equation in 1955, in a short Physical Review paper written at the Indian Association for the Cultivation of Science in Calcutta, without a research group or senior mentor in relativity. His aim was concrete and, at the time, unfashionable: to decide whether the singularity of cosmological models was an artefact of their perfect symmetry. Working with the geodesics of a general dust distribution — no assumption of homogeneity or isotropy — he obtained the exact evolution equation for the expansion and read off its lesson: with ordinary attractive matter and no rotation, an initially contracting flow must focus to a caustic in finite proper time. The singularity was not an accident of symmetry but a robust consequence of attraction. For a decade the result sat unexploited, until Penrose in 1965 recast it for null geodesics and married it to trapped surfaces and global causal structure, and Hawking applied the timelike version to cosmology — the two combining in 1970 for the general Hawking–Penrose theorem. Raychaudhuri's equation is now the common root of the singularity theorems, the horizon area law, and gravitational lensing — one of the most consequential single lines in general relativity.

History

1955

— Raychaudhuri publishes his equation for the expansion of a geodesic congruence

1957

— Landau and, independently, Komar note the same focusing relation

1965

— Penrose proves the first singularity theorem using null focusing and trapped surfaces

1966

— Hawking applies timelike focusing to force a cosmological past singularity

1970

— Hawking and Penrose combine both into the general singularity theorem

1971

— Hawking derives the black-hole area theorem from null focusing

Going further

Texts. Wald, General Relativity (1984), §9.2–9.3 (Raychaudhuri, conjugate points) and §9.5, §12.2 · Hawking & Ellis, The Large Scale Structure of Space-Time (1973), §4.4–4.5, §6.2, §8 · Misner, Thorne & Wheeler, Gravitation (1973), §22.5 · Carroll, Spacetime and Geometry (2004), §3.4, §9.5 · Poisson, A Relativist's Toolkit (2004), §2.3–2.4 (Raychaudhuri and the optical scalars). Key checkpoints for the results in this article: the $-\omega^2$ sign in $B_{ab}B^{ba}=\tfrac13\theta^2+\sigma^2-\omega^2$ ; the reciprocal integration of the focusing inequality and the divisor $n-1$ (3 timelike, 2 null); the cosmology bound $1/H_0=14.0\ \mathrm{Gyr}$ at $H_0=70\ \mathrm{km\,s^{-1}\,Mpc^{-1}}$ ; the flat-space null caustic $\theta=-2/(r_0-\lambda)$ saturating $\lambda^\ast=r_0=2/|\theta_0|$ ; and the de Sitter sign $R_{ab}u^au^b=-\Lambda$ . Metric signature $(-+++)$ , $G=c=1$ unless units are shown.

Prerequisites. Null Geodesics and Light Cones · Causal Structure of Spacetime · Chronological and Causal Futures and Pasts. Leads to. Achronal Sets and Edges · Global Hyperbolicity · Domain of Dependence and Cauchy Horizons.

Ask the tutor: Ask me to walk you from Raychaudhuri's one-line identity to a trapped surface and a Big Bang — or to show why rotation, and only rotation, can save a collapsing star from a conjugate point.

Is the Raychaudhuri equation itself a consequence of Einstein's equations?

No, and that is the whole point. Raychaudhuri's equation is a pure geometric identity for any congruence of geodesics in any spacetime — it follows from the definition of the expansion, the geodesic equation, and the Ricci identity, nothing more. Einstein's equations enter only through one term, $R_{ab}u^au^b$ , which they relate to the local energy-momentum. So the kinematics of focusing is fixed by geometry alone, and the physics — whether that term focuses or defocuses — by the matter content through an energy condition. Separating the two is what makes the theorem robust: it applies to any theory sharing Einstein's geodesic structure.

Why is the divisor 3 for timelike geodesics but 2 for null ones?

The divisor is the transverse dimension — the directions in which the bundle can spread. A timelike worldline has three orthogonal spatial directions, so $n-1=3$ . A null generator is trickier: the tangent $k^a$ is orthogonal to itself ( $k^ak_a=0$ ), so one transverse direction is "used up", leaving a genuinely two-dimensional screen space, $n-1=2$ . The self-focusing term $-\theta^2/(n-1)$ then propagates this difference into the bounds $3/|\theta_0|$ and $2/|\theta_0|$ .

If a conjugate point is not a singularity, how do the singularity theorems use it?

By contradiction. The theorems first show, from global causal structure, that certain geodesics must be free of conjugate points — a null geodesic generating the boundary $\partial J^{+}$ of a causal future cannot contain one, because past a conjugate point it would drop into the chronological interior and stop being a boundary generator. Focusing then shows those same geodesics must contain a conjugate point within bounded affine length, given an energy condition and an initial convergence such as a trapped surface. The only escape is that the geodesics cannot be extended that far — they are incomplete, and geodesic incompleteness is the actual definition of a singularity. So the caustic never has to form; its forced impossibility is the theorem.

What exactly is a trapped surface, and why does it trigger focusing?

A trapped surface is a compact two-dimensional spacelike surface whose both families of orthogonal null geodesics — outgoing and ingoing — have negative expansion, $\theta<0$ . Normally an outgoing wavefront expands; strong enough gravity focuses even the outgoing light so hard that its area shrinks. Because both expansions start negative, null focusing guarantees a caustic on every generator within finite affine parameter. Penrose's insight was that this local condition, once it appears in a globally hyperbolic spacetime with non-compact Cauchy surface, is incompatible with complete boundary-generating null geodesics, forcing incompleteness. Trapped surfaces are now found numerically as the apparent horizons of colliding black holes.

Does the shear help or hurt focusing?

It always helps — it can only make a bundle focus sooner. Shear enters Raychaudhuri as $-\sigma_{ab}\sigma^{ab}$ , and since $\sigma_{ab}$ is real, $\sigma_{ab}\sigma^{ab}\ge0$ , so the term is non-positive. Physically, shear is anisotropic squeezing at fixed volume, but the equation shows any such distortion also drives the volume down, accelerating collapse — which is why highly aspherical collapse still forms singularities. Only vorticity, entering with the opposite sign as $+2\omega^2$ , can oppose focusing: the centrifugal support a rotating star or an orbiting congruence enjoys.

Do quantum fields, which violate the classical energy conditions, break the theorem?

They break the hypotheses, so the classical theorem does not directly apply — quantum fields can make $R_{ab}k^ak^b<0$ locally, as the Casimir effect and Hawking radiation both do. But the mechanism is resilient. Averaged energy conditions, which bound the energy density integrated along a geodesic rather than pointwise, restore focusing and are provable from quantum field theory. More strikingly, the quantum focusing conjecture replaces the classical expansion with the derivative of a generalized entropy and asserts that focuses monotonically, implying the quantum null energy condition. So focusing does not break under quantization so much as get promoted — from area to entropy.