Raychaudhuri Equation

Take a small ball of freely-falling test particles — dust with no pressure, each grain tracing a geodesic — and watch its shadow-volume as it drifts through curved spacetime. Does the ball grow or shrink? Does it twist into an ellipsoid, or swirl? The Raychaudhuri equation is the exact bookkeeping for that volume. It follows the expansion scalar $\theta=\nabla_a u^a$ , the fractional rate at which the ball's volume changes, and states precisely how $\theta$ evolves along the flow in terms of three competing effects: the congruence's own shear $\sigma_{ab}$ and rotation $\omega_{ab}$ , and the tidal pull of spacetime through the Ricci contraction $R_{ab}u^a u^b$ . What makes the equation the single most important inequality in gravitation is that it is, in the first instance, pure geometry. It is derived by decomposing the velocity gradient $\nabla_b u_a$ into its trace, symmetric-traceless, and antisymmetric parts and commuting two covariant derivatives with the Ricci identity — no field equations, no matter, no physics. Physics enters through one door only: Einstein's equations convert the curvature term into stress-energy, and an energy condition fixes its sign. The moment that sign is non-negative and rotation is absent, the right-hand side of the equation is forced to be non-positive, $\theta$ can only fall, and once it is negative the nonlinear $-\theta^2$ term drags it to $-\infty$ in finite parameter. Neighbouring geodesics cross: a caustic. That runaway is the analytic engine behind the Penrose–Hawking singularity theorems, the Hawking area theorem, the Friedmann acceleration equation, and the focusing of light in gravitational lensing. This article builds the equation from its kinematic decomposition, proves the focusing theorem, and follows the caustic wherever it leads.

Key takeaways

The Raychaudhuri equation governs how the expansion $\theta$ of a geodesic congruence evolves along the flow, splitting the change into shear, rotation, and the Ricci curvature contracted with the tangent field.
Timelike form: $\mathrm{d}\theta/\mathrm{d}\tau=-\tfrac13\theta^2-\sigma_{ab}\sigma^{ab}+\omega_{ab}\omega^{ab}-R_{ab}u^a u^b$ ; the null form replaces $\tfrac13$ by $\tfrac12$ .
The equation is a pure geometric identity — physics enters only when Einstein's equations turn $R_{ab}u^a u^b$ into matter and an energy condition fixes its sign ( $\rho+3p\ge0$ for the strong energy condition).
When rotation vanishes and curvature focuses ( $R_{ab}u^a u^b\ge0$ ), every term on the right is non-positive: an initially converging congruence ( $\theta_0<0$ ) is dragged to $\theta\to-\infty$ within proper time $\tau\le 3/|\theta_0|$ — a caustic.
That caustic bound is the lever the Penrose–Hawking singularity theorems use to turn "geodesics focus" into "spacetime is geodesically incomplete."
Key idea: the expansion of an irrotational geodesic congruence obeys $\mathrm{d}\theta/\mathrm{d}\tau \le -\tfrac13\theta^2 - R_{ab}u^a u^b$ ; whenever curvature focuses the right side is non-positive, so $1/\theta$ grows at least linearly and must cross zero in finite parameter — gravity always attracts, and this is its precise statement.

The core insight

The expansion of an irrotational geodesic congruence obeys $\mathrm{d}\theta/\mathrm{d}\tau \le -\tfrac13\theta^2 - R_{ab}u^a u^b$ ; whenever curvature focuses ( $R_{ab}u^a u^b\ge 0$ ) the right side is non-positive, so $1/\theta$ grows at least linearly and forces a caustic in finite parameter — gravity always attracts, and this is the precise statement.

Where this lives today

Raychaudhuri's identity is not a museum piece; it is a daily tool. In numerical relativity, the codes that model merging black holes and neutron stars locate horizons by tracking null expansion: an apparent horizon is a marginally outer-trapped surface where the outgoing null expansion $\theta=0$ , and horizon finders solve exactly this condition on each time slice, while the null Raychaudhuri equation guarantees that once $\theta\le0$ the generators focus and the surface caps off — the machinery behind every LIGO–Virgo–KAGRA waveform template built from simulated coalescences. In cosmology, the "averaging problem" — whether the small-scale lumpiness of the real universe biases its large-scale expansion (cosmic backreaction) — is posed and fought entirely in the language of a shear-and-vorticity-corrected Raychaudhuri equation, a live question as surveys like DESI and Euclid pin down the expansion history to percent precision. In gravitational lensing, the two-dimensional null form is the focusing theorem that underlies the magnification of distant galaxies and the interpretation of weak-lensing shear maps. And the equation remains the entry point to the modern singularity theorems of Hawking, Penrose, and their successors, including recent quantum-corrected versions in which a semiclassical energy flux replaces the classical energy condition — the frontier where Raychaudhuri's 1955 line of geometry meets the physics of the very early universe and the interiors of black holes.

The physical picture

Picture a small spherical cloud of freely-falling test particles — dust — released at rest relative to one another and left to fall. Each grain follows a geodesic, and together they form a congruence: a smooth family of worldlines filling a tube of spacetime, one through every point. Attach to the central grain a set of connecting vectors reaching its neighbours; as the cloud falls, tidal gravity reshapes that little sphere. Three independent things can happen to it, and the velocity gradient $\nabla_b u_a$ encodes all three at once. Its trace is the expansion $\theta$ — the fractional rate of volume growth. Its symmetric trace-free part is the shear $\sigma_{ab}$ — distortion into an equal-volume ellipsoid. Its antisymmetric part is the rotation $\omega_{ab}$ — rigid swirl. The first figure separates these three motions.

Figure

A small ball of freely-falling test particles evolves in three independent ways — expansion θ grows its volume, shear σ distorts it into an equal-volume ellipsoid, and rotation ω swirls it — the irreducible split ∇ᵦuₐ = ⅓ θ hₐᵦ + σₐᵦ + ωₐᵦ that Raychaudhuri's equation tracks

Raychaudhuri's equation is the ledger for just one of these — the volume, $\theta$ — and the entries tell a decisive story. Tidal gravity, carried by $R_{ab}u^a u^b$ , squeezes the volume whenever the matter the ball surrounds is gravitationally attractive. Shear always squeezes it further: $\sigma_{ab}\sigma^{ab}$ enters with a minus sign, so distortion feeds collapse. Rotation alone fights back — $\omega_{ab}\omega^{ab}$ enters with a plus sign, the centrifugal resistance that lets a spinning cloud avoid crunching. And the trace term $-\tfrac13\theta^2$ is the nonlinear heart: convergence breeds convergence. Once the ball is shrinking, that quadratic term accelerates the shrinking, and nothing short of rotation can arrest it. The consequence is a runaway. Writing $s\equiv 1/\theta$ , the focusing inequality becomes $\mathrm{d}s/\mathrm{d}\tau\ge \tfrac13$ , so $s$ climbs along a straight line and — if the congruence started converging, $\theta_0<0$ , hence $s_0<0$ — it must reach zero, meaning $\theta\to-\infty$ , within a bounded proper time. That is a caustic: neighbouring geodesics cross and the ball's volume collapses to zero. The second figure is the punchline, drawn with real numbers.

Figure

The caustic engine — for an irrotational geodesic congruence obeying the focusing condition, the reciprocal 1/θ runs along the straight line 1/θ(τ) = 1/θ₀ + τ/3 of slope ⅓, so from θ₀ = −1 it reaches zero at τ = 3 (that is 3 over the magnitude of θ₀), exactly where the expansion θ plunges to −∞ and neighbouring geodesics cross

Everything the equation predicts flows from these two pictures. The energy condition is the bridge: $R_{ab}u^a u^b$ equals $8\pi(T_{ab}-\tfrac12 T g_{ab})u^a u^b$ by Einstein's equations, and demanding this be non-negative — the statement that ordinary matter gravitates attractively — is exactly what tips the ledger toward collapse and turns a geometric identity into a theorem about the fate of the universe.

The core equations

The starting point is not the evolution equation itself but the object it evolves. The velocity gradient $B_{ab}\equiv\nabla_b u_a$ of a geodesic congruence is purely spatial (orthogonal to $u^a$ on both indices), and any spatial tensor splits uniquely into a trace, a symmetric trace-free piece, and an antisymmetric piece. That irreducible decomposition names the three motions of the falling ball.

\nabla_b u_a = \tfrac{1}{3}\theta\, h_{ab} + \sigma_{ab} + \omega_{ab} - u_b a_a \qquad (1)

where: $u^a$ — unit tangent to the congruence, $u^a u_a=-1$ ; $h_{ab}=g_{ab}+u_a u_b$ — projector onto the 3-space orthogonal to $u^a$ ; $\theta=h^{ab}\nabla_b u_a=\nabla_a u^a$ — expansion (trace part), the fractional volume growth rate; $\sigma_{ab}$ — shear, symmetric, trace-free, orthogonal to $u^a$ ; $\omega_{ab}$ — rotation/vorticity, antisymmetric, orthogonal to $u^a$ ; $a_a=u^b\nabla_b u_a$ — acceleration, which vanishes for geodesics. Reading it: $\theta$ tracks volume, $\sigma_{ab}$ tracks shape at fixed volume, $\omega_{ab}$ tracks swirl, and for a geodesic congruence the last term is absent. Evolving the trace of $B_{ab}$ along the flow — differentiating, commuting derivatives with the Ricci identity, and using the geodesic equation — yields the central result. It is an exact identity in any spacetime, valid before a single assumption about matter is made.

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

where: $\mathrm{d}/\mathrm{d}\tau=u^a\nabla_a$ — derivative along the congruence (proper time for timelike geodesics); $\sigma_{ab}\sigma^{ab}\ge0$ — shear scalar, a positive-definite contraction of a spatial tensor; $\omega_{ab}\omega^{ab}\ge0$ — vorticity scalar; $R_{ab}u^a u^b$ — Ricci curvature along the flow, the focusing term. Reading it: three of the four terms drive $\theta$ down (they focus); only vorticity pushes $\theta$ up (it defocuses). For a bundle of light rays — the generators of a light cone or an event horizon — the tangent is null, $k^a k_a=0$ , and one transverse direction coincides with the ray itself, leaving a two-dimensional "screen." The trace coefficient changes accordingly.

\frac{\mathrm{d}\theta}{\mathrm{d}\lambda}=-\tfrac{1}{2}\theta^2-\sigma_{ab}\sigma^{ab}+\omega_{ab}\omega^{ab}-R_{ab}k^a k^b \qquad (3)

where: $k^a$ — null tangent, affinely parametrised so that $k^b\nabla_b k^a=0$ ; $\lambda$ — affine parameter along the null generator; the factor $\tfrac12$ reflects the 2-dimensional transverse screen space instead of the 3-dimensional space orthogonal to a timelike $u^a$ . Reading it: light focuses under the weaker null energy condition, which is why this form governs horizons and lensing. So far there is no physics. It enters by trace-reversing Einstein's equations to express the curvature term through the stress-energy tensor, and then demanding a sign.

R_{ab}u^a u^b = 8\pi\left(T_{ab}-\tfrac{1}{2}T\,g_{ab}\right)u^a u^b = 4\pi(\rho+3p)\ \ge\ 0 \qquad (4)

where: $T_{ab}$ — stress-energy tensor; $T=g^{ab}T_{ab}$ — its trace; $\rho,p$ — energy density and pressure of a perfect fluid; $G=c=1$ — geometrised units. The final equality holds for $T_{ab}=(\rho+p)u_a u_b+p g_{ab}$ ; the non-negativity is the strong energy condition, $\rho+3p\ge0$ and $\rho+p\ge0$ . Reading it: "gravity is attractive for ordinary matter" is exactly the inequality $R_{ab}u^a u^b\ge0$ . Feeding the sign into the identity, dropping the two manifestly non-negative terms $\sigma^2$ and $R_{ab}u^a u^b$ , and setting $\omega=0$ for a hypersurface-orthogonal congruence produces the focusing inequality and, on integrating the reciprocal, the caustic bound.

\frac{\mathrm{d}\theta}{\mathrm{d}\tau}\le -\tfrac{1}{3}\theta^2 \quad\Longrightarrow\quad \frac{1}{\theta(\tau)}\ge \frac{1}{\theta_0}+\tfrac{1}{3}\tau \qquad (5)

where: $\theta_0=\theta(0)$ — initial expansion; $\theta(\tau)$ — expansion a proper time $\tau$ later. Reading it: if $\theta_0<0$ then $1/\theta$ starts negative and, growing with slope at least $\tfrac13$ , reaches zero — where $\theta\to-\infty$ — no later than $\tau=3/|\theta_0|$ . Applied to the comoving matter of a homogeneous, isotropic universe, where shear, vorticity, and acceleration all vanish and $\theta=3\dot a/a$ , the same equation collapses to a cornerstone of cosmology.

\frac{\ddot a}{a}=-\frac{4\pi}{3}\left(\rho+3p\right) \qquad (6)

where: $a(t)$ — FLRW scale factor; $\rho,p$ — energy density and pressure of the cosmic fluid. Reading it: for ordinary matter ( $\rho+3p>0$ ) the expansion always decelerates — Raychaudhuri's own 1955 application, and the first hint that a cosmic singularity is unavoidable.

Derivation ADeriving the timelike Raychaudhuri equation from the Ricci identity

The goal is to obtain equation (2) as an exact geometric identity, using nothing but the definition of the velocity gradient, the geodesic condition, and the Ricci identity that commutes two covariant derivatives at the cost of a Riemann tensor. We evolve the tensor $B_{ab}=\nabla_b u_a$ along the flow, take its trace to isolate $\theta$ , and finally insert the irreducible decomposition so that the shear and vorticity scalars appear with their characteristic signs. Throughout, the congruence is geodesic ( $u^b\nabla_b u_a=0$ ) and normalised ( $u^a u_a=-1$ ), and no field equation is used anywhere in the derivation.

Step 1. Differentiate $B_{ab}=\nabla_b u_a$ along the flow: $u^c\nabla_c B_{ab}=u^c\nabla_c\nabla_b u_a$ . This is just the rate of change of the velocity gradient carried along each worldline. Why: the left side is exactly $\mathrm{d}B_{ab}/\mathrm{d}\tau$ , the quantity whose trace we want.

Step 2. Commute the covariant derivatives with the Ricci identity, $\nabla_c\nabla_b u_a-\nabla_b\nabla_c u_a=R_{abcd}\,u^d$ (with the sign convention of the boxed decomposition), giving $u^c\nabla_c\nabla_b u_a=u^c\nabla_b\nabla_c u_a - R_{acbd}\,u^c u^d$ after contracting. Why: this is the only place curvature can enter — it is the price of exchanging the order of two derivatives in a curved space.

Step 3. Rewrite the first term on the right: $u^c\nabla_b\nabla_c u_a=\nabla_b(u^c\nabla_c u_a)-(\nabla_b u^c)(\nabla_c u_a)=-B_{ac}B^{c}{}_{b}$ . The first bracket is the covariant derivative of the geodesic acceleration and vanishes. Why: the geodesic equation $u^c\nabla_c u_a=0$ kills the total-derivative piece, leaving only the quadratic $B$ -product.

Step 4. Assemble the matrix Raychaudhuri equation, $u^c\nabla_c B_{ab}=-B_{ac}B^{c}{}_{b}-R_{acbd}\,u^c u^d$ . Everything scalar now follows by tracing. Why: this tensor equation contains the evolution of expansion, shear, and rotation all at once; taking its trace projects out the volume.

Step 5. Contract with $g^{ab}$ . Using $g^{ab}B_{ac}B^{c}{}_{b}=B_{ab}B^{ba}$ and $g^{ab}R_{acbd}u^c u^d=R_{cd}u^c u^d$ , the trace gives $\mathrm{d}\theta/\mathrm{d}\tau=-B_{ab}B^{ba}-R_{ab}u^a u^b$ , since $g^{ab}B_{ab}=\theta$ . Why: the trace of the velocity gradient is precisely the expansion, so its evolution is what we are after.

Step 6. Insert $B_{ab}=\tfrac13\theta h_{ab}+\sigma_{ab}+\omega_{ab}$ . Cross terms vanish because $\sigma$ is trace-free, $\omega$ is antisymmetric, and $\sigma_{ab}\omega^{ab}=0$ ; with $h_{ab}h^{ab}=3$ one finds $B_{ab}B^{ba}=\tfrac13\theta^2+\sigma_{ab}\sigma^{ab}-\omega_{ab}\omega^{ab}$ . The minus sign on $\omega^2$ is the signature of contracting an antisymmetric tensor with its transpose. Why: this single algebraic step is what converts $-B_{ab}B^{ba}$ into the focusing $-\tfrac13\theta^2-\sigma^2$ and the defocusing $+\omega^2$ .

Step 7. Substitute to reach $\mathrm{d}\theta/\mathrm{d}\tau=-\tfrac13\theta^2-\sigma_{ab}\sigma^{ab}+\omega_{ab}\omega^{ab}-R_{ab}u^a u^b$ , equation (2). Why: the signs are now fixed by geometry alone — shear and expansion focus, vorticity defocuses, and curvature is still an unsigned spectator until an energy condition speaks.

Go deeper · Derivation B · 2 minThe focusing theorem and conjugate points

Derivation BThe focusing theorem and conjugate points

The goal is to show that the bare identity, once armed with two independent physical inputs, forces a caustic in finite proper time. The inputs are Frobenius' theorem — a hypersurface-orthogonal congruence has $\omega_{ab}=0$ — and the strong energy condition, $R_{ab}u^a u^b\ge0$ . With rotation gone, the only term that could resist collapse is removed, and the remaining terms conspire to make $1/\theta$ increase without limit. We turn the differential inequality into an integral bound and read off the moment the expansion diverges.

Step 1. With $\omega_{ab}=0$ set aside $\sigma_{ab}\sigma^{ab}\ge0$ and $R_{ab}u^a u^b\ge0$ as the two non-negative terms in equation (2); dropping both leaves $\mathrm{d}\theta/\mathrm{d}\tau\le-\tfrac13\theta^2$ . Why: an inequality is all we need — the exact evolution can only focus faster than this bound, never slower.

Step 2. Change variable to $s=1/\theta$ . Then $\mathrm{d}s/\mathrm{d}\tau=-\theta^{-2}\,\mathrm{d}\theta/\mathrm{d}\tau\ge\tfrac13$ wherever $\theta\ne0$ . Why: the reciprocal linearises the nonlinear $\theta^2$ term — the quadratic that made focusing run away becomes a constant lower bound on a slope.

Step 3. Integrate from $0$ to $\tau$ : $s(\tau)\ge s_0+\tfrac13\tau$ , i.e. $1/\theta(\tau)\ge 1/\theta_0+\tfrac13\tau$ . Why: a function whose derivative is at least $\tfrac13$ grows at least as fast as a line of slope $\tfrac13$ .

Step 4. Suppose the congruence starts converging, $\theta_0<0$ , so $s_0=1/\theta_0<0$ . The right-hand side of Step 3 reaches zero at $\tau_\star=3/|\theta_0|$ at the latest, and $s\to0^-$ means $\theta\to-\infty$ . A caustic — a conjugate point where neighbouring geodesics cross and the transverse volume $V\propto\exp\!\int\theta\,\mathrm{d}\tau$ collapses to zero — forms no later than $\tau_\star$ . Why: this finite, computable bound is the entire content of "gravity focuses" — collapse cannot be postponed indefinitely once it begins.

Step 5. Translate to singularities. A timelike geodesic stops being locally length-maximising past a conjugate point; if global hyperbolicity guarantees a maximal geodesic between a Cauchy surface and any point, the unavoidable caustic contradicts maximality — unless the geodesic is incomplete. Incompleteness is what "spacetime is singular" means. Why: this is the exact hinge on which the Penrose–Hawking theorems turn, converting a local focusing bound into a global statement about the existence of an edge to spacetime.

Go deeper · Derivation C · 1 minReducing to the Friedmann acceleration equation

Derivation CReducing to the Friedmann acceleration equation

The goal is to recover the Friedmann acceleration equation as the simplest non-trivial application of Raychaudhuri's identity — the very reduction Raychaudhuri performed in his 1955 paper. The congruence is the comoving matter of a homogeneous, isotropic (FLRW) universe: symmetry forces shear, vorticity, and acceleration all to vanish, so the equation reduces to a relation purely between the expansion and the Ricci term, which the perfect-fluid stress-energy makes explicit. The payoff is a direct statement that ordinary matter decelerates cosmic expansion.

Step 1. By homogeneity and isotropy, $\sigma_{ab}=0$ , $\omega_{ab}=0$ , and $a_a=0$ ; the expansion of the comoving congruence is $\theta=3H=3\dot a/a$ , three times the Hubble rate because the volume grows as $a^3$ . Why: only the trace survives — an isotropic universe cannot shear, swirl, or accelerate its own dust.

Step 2. Differentiate $\theta=3\dot a/a$ along the flow: $\mathrm{d}\theta/\mathrm{d}t=3(\ddot a/a-\dot a^2/a^2)=3(\ddot a/a-H^2)$ . Why: this expresses the left-hand side of Raychaudhuri directly in terms of the scale factor's acceleration.

Step 3. Insert into equation (2) with only the surviving terms: $3(\ddot a/a-H^2)=-\tfrac13(3H)^2-R_{ab}u^a u^b=-3H^2-4\pi(\rho+3p)$ , using equation (4). Why: the shear and vorticity terms are gone, and the perfect-fluid form of the Ricci contraction is exactly $4\pi(\rho+3p)$ .

Step 4. The $-3H^2$ terms cancel from both sides, leaving $3\ddot a/a=-4\pi(\rho+3p)$ , hence $\ddot a/a=-\tfrac{4\pi}{3}(\rho+3p)$ , equation (6). Why: for ordinary matter the right side is negative, so $\ddot a<0$ — the expansion always decelerates, and a converging past congruence must have emerged from $a=0$ a finite time ago.

Limiting cases

The trace coefficient and the sign of the curvature term shift with the type of congruence and the matter it threads. Each row is the same identity read in a different regime.

Case	Result	Meaning
Timelike geodesics, irrotational	$\dfrac{\mathrm{d}\theta}{\mathrm{d}\tau}=-\tfrac13\theta^2-\sigma_{ab}\sigma^{ab}-R_{ab}u^a u^b$	Transverse space is 3-dimensional, hence the $\tfrac13$ . Under the SEC every term is non-positive: unconditional focusing.
Null geodesics	$\dfrac{\mathrm{d}\theta}{\mathrm{d}\lambda}=-\tfrac12\theta^2-\sigma_{ab}\sigma^{ab}-R_{ab}k^a k^b$	Transverse "screen" is 2-dimensional, giving $\tfrac12$ . Only the weaker NEC, $R_{ab}k^a k^b\ge0$ , is needed to focus light and horizon generators.
Rotating congruence	$+\,\omega_{ab}\omega^{ab}\ge0$	Vorticity defocuses — centrifugal resistance. A rigidly rotating cloud can avoid caustics, which is why singularity theorems demand $\omega=0$ (Frobenius).
Accelerated (non-geodesic) flow	$\dfrac{\mathrm{d}\theta}{\mathrm{d}\tau}=-\tfrac13\theta^2-\sigma^2+\omega^2-R_{ab}u^a u^b+\nabla_a a^a$	A pressure-supported fluid has $a^a\ne0$ ; the divergence $\nabla_a a^a$ can counter focusing — how a static star holds itself up.
FLRW cosmology	$\sigma=\omega=0,\ \theta=3\dot a/a\ \Rightarrow\ \dfrac{\ddot a}{a}=-\tfrac{4\pi}{3}(\rho+3p)$	The comoving dust congruence reduces the equation to the Friedmann acceleration equation — Raychaudhuri's original use.

Common trapThe equation itself does not prove that gravity focuses

It is tempting to read equation (2) as "gravity always attracts," but the bare equation is a curvature identity with no physical input and no preferred sign for $R_{ab}u^a u^b$ . Focusing needs two additional, independent assumptions bolted on. First, rotation must vanish: a congruence with $\omega_{ab}\omega^{ab}>0$ carries a defocusing term and can orbit forever without ever forming a caustic — Gödel's rotating universe is the classic escape hatch. Second, an energy condition must fix the sign of the curvature term; without it, matter with $\rho+3p<0$ (a cosmological constant, an inflaton, dark energy) makes $R_{ab}u^a u^b<0$ , and the term defocuses, driving accelerated expansion rather than collapse. This is exactly why the singularity theorems state their energy and rotation hypotheses so carefully, and why the observed accelerating universe is no contradiction: it simply violates the strong energy condition.

Worked examples

1 · A past caustic within one Hubble time (fully worked)

Model the late universe crudely as pressureless dust obeying the strong energy condition, expanding today at the Hubble rate $H_0=70\ \mathrm{km\,s^{-1}\,Mpc^{-1}}$ . Run the comoving congruence backward in time and use the focusing bound to show that a caustic — a cosmic singularity — must lie within a finite past, and estimate how far back.

Step 1. Convert the Hubble rate to inverse seconds. With $1\ \mathrm{Mpc}=3.086\times10^{19}\ \mathrm{km}$ , $H_0=70/(3.086\times10^{19})\ \mathrm{s^{-1}}=\mathbf{2.27\times10^{-18}\ s^{-1}}$ . The comoving expansion is $\theta=3H_0=6.80\times10^{-18}\ \mathrm{s^{-1}}$ , positive because the universe expands.

Step 2. Reverse the time direction. Under $t\to-t$ the tangent reverses and the expansion flips sign, so the time-reversed congruence has $\theta_0=-3H_0=-6.80\times10^{-18}\ \mathrm{s^{-1}}$ — it is converging, precisely the setup of the focusing theorem.

Step 3. Apply the caustic bound $\tau_\star\le 3/|\theta_0|$ from equation (5): $\tau_\star\le 3/(3H_0)=1/H_0$ . The factors of three cancel, leaving the Hubble time. Numerically $1/H_0=1/(2.27\times10^{-18})\ \mathrm{s}=4.40\times10^{17}\ \mathrm{s}$ .

Step 4. Convert to years: $4.40\times10^{17}/(3.156\times10^{7})=\mathbf{1.40\times10^{10}\ yr}$ , about 14 billion years. So a matter-dominated congruence traced backward reaches $\theta\to-\infty$ within one Hubble time: with no symmetry assumption, the comoving congruence cannot be extended smoothly beyond a finite past. Turning that caustic into a genuine past singularity — geodesic incompleteness — takes the further global input of Derivation B, and it is that chain, not the bound alone, that delivers the Big Bang.

Step 5. Reality check. The measured age of the universe is $\approx13.8$ Gyr, comfortably inside the bound $1/H_0\approx14$ Gyr, as it must be. The bound is not saturated because matter-dominated deceleration made the expansion faster in the past than it is today: a pure-dust Einstein–de Sitter universe with this $H_0$ would be only $t_0=(2/3)/H_0\approx9.3$ Gyr old. The recent dark-energy phase, which violates the SEC and defocuses, works the other way — at fixed $H_0$ it lengthens the age, pushing it from that $9.3$ Gyr up to the observed $13.8$ Gyr, close to but still inside $1/H_0$ . The lesson: as long as matter dominated ( $\rho+3p>0$ ), the universe had a beginning within a Hubble time, independent of exact composition.

2 · Caustic time for a converging dust ball, and why shear only hastens it

An irrotational, pressureless dust ball is released and found to be converging with initial expansion $\theta_0=-2.0\times10^{-8}\ \mathrm{s^{-1}}$ . Assuming the strong energy condition holds, find the latest proper time at which a caustic must form, and show that any nonzero shear can only shorten it.

Step 1. The focusing bound gives the latest caustic time directly: $\tau_\star=3/|\theta_0|=3/(2.0\times10^{-8}\ \mathrm{s^{-1}})=\mathbf{1.5\times10^{8}\ s}$ , about 4.8 years of proper time. This is the worst case — shear-free and with $R_{ab}u^a u^b=0$ , the equality case $\mathrm{d}\theta/\mathrm{d}\tau=-\tfrac13\theta^2$ integrates exactly to $\theta(\tau)=\theta_0/(1+\tfrac13\theta_0\tau)$ , which diverges precisely at $\tau=3/|\theta_0|$ .

Step 2. Now restore shear. Equation (2) with $\omega=0$ reads $\mathrm{d}\theta/\mathrm{d}\tau=-\tfrac13\theta^2-\sigma_{ab}\sigma^{ab}-R_{ab}u^a u^b$ , and both extra terms are non-negative, so $\mathrm{d}\theta/\mathrm{d}\tau$ is more negative than $-\tfrac13\theta^2$ at every instant. The expansion therefore falls faster than the shear-free curve.

Step 3. Since $1/\theta$ then grows with slope strictly greater than $\tfrac13$ , it reaches zero earlier: the true caustic time obeys $\tau_{\text{caustic}}\le 3/|\theta_0|$ , with equality only when $\sigma=R_{ab}u^a u^b=0$ . Physically, shear stretches the ball along one axis while pinching the others, and the pinch always wins the volume accounting. Result: the caustic forms at or before 4.8 years; shear and attractive curvature can only bring it sooner, never later.

Physics of the portrait

The Raychaudhuri equation is unusual among the great results of general relativity: it carries the name of a man who worked in near-isolation, on a modest salary, far from the field's centres of gravity, and who lived to see his single-line identity become the load-bearing beam of the entire singularity-theorem edifice.

Origins · 1 minAmal Kumar Raychaudhuri and how this was discovered

Amal Kumar Raychaudhuri

1923–2005 · the geometry of gravitational focusing and the inevitability of cosmic singularities

Amal Kumar Raychaudhuri worked out his equation in Calcutta in the early 1950s, in a research culture then dominated by the belief — encouraged by Einstein himself — that the singularities appearing in cosmological solutions were mere artefacts of their exact symmetry, doomed to vanish in any realistic, lumpy universe. Raychaudhuri set out to test that belief head-on. Rather than solve Einstein's equations for a special metric, he asked a purely geometric question about an arbitrary congruence of matter worldlines: how does the volume they enclose evolve? The answer, published as a terse paper in Physical Review in 1955, was the focusing equation now bearing his name (a closely related relation appears in Landau and Lifshitz, so the result is sometimes called the Landau–Raychaudhuri equation). Its power lay in what it did not assume — no symmetry, no special solution — so that when he showed a converging congruence of ordinary matter must reach a caustic, the conclusion could not be dismissed as a symmetry accident. A decade later Roger Penrose and Stephen Hawking recognised that this humble identity, married to global causal structure and the topology of trapped surfaces, was the missing analytic engine, and built upon it the singularity theorems that reshaped cosmology and black-hole physics. Raychaudhuri, who spent much of his career teaching at Presidency College, remained a devoted teacher and, characteristically, wrote a general-relativity textbook rather than chasing the fame his equation had earned.

History

1955

— Raychaudhuri publishes the geodesic focusing equation in Physical Review 98, 1123

1957

— an equivalent relation appears in Landau and Lifshitz, giving the name Landau–Raychaudhuri equation

1959

— Komar argues that focusing implies unavoidable singularities in gravitational collapse

1965

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

1970

— Hawking and Penrose combine the timelike and null equations into the general singularity theorem

1971

— Hawking's black-hole area theorem follows from null focusing and the null energy condition

Going further

Texts. Wald, General Relativity (1984), §9.2 and §9.5 · Misner, Thorne & Wheeler, Gravitation (1973), §22.2 · Carroll, Spacetime and Geometry (2004), §3.10 and App. F · Hawking & Ellis, The Large Scale Structure of Space-Time (1973), §4.1 and §4.3 · Poisson, A Relativist's Toolkit (2004), §2.3 · Raychaudhuri, Physical Review 98, 1123 (1955). The key identifications used throughout are: the irreducible decomposition $B_{ab}=\tfrac13\theta h_{ab}+\sigma_{ab}+\omega_{ab}$ fixes the coefficients ( $\tfrac13$ timelike, $\tfrac12$ null) and signs (shear $-$ , vorticity $+$ ); the caustic bound $\Delta\tau\le3/|\theta_0|$ follows from integrating $\mathrm{d}(1/\theta)/\mathrm{d}\tau\ge\tfrac13$ ; the FLRW reduction $\ddot a/a=-\tfrac{4\pi}{3}(\rho+3p)$ uses $\theta=3\dot a/a$ and the perfect-fluid identity $R_{ab}u^a u^b=4\pi(\rho+3p)$ from the trace-reversed Einstein equation; and the Hubble-time estimate gives $1/H_0\approx14$ Gyr. All results are in geometrised units $G=c=1$ .

Prerequisites. Geodesic Deviation and the Riemann Tensor · Covariant Derivative and the Ricci Identity · Energy Conditions · Null Geodesics and Light Cones. Leads to. Penrose–Hawking Singularity Theorems · Causal Structure of Spacetime · Global Hyperbolicity · Domain of Dependence and Cauchy Horizons.

Ask the tutor: ask me to walk through the trace-and-decompose derivation line by line, or to show exactly how the caustic bound becomes a singularity theorem via conjugate points and global hyperbolicity.

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

No, and this is the single most important thing to understand about it. The equation is a geometric identity that holds in any spacetime, curved by any means or by none — it follows purely from the definition of the expansion of a congruence, the geodesic condition, and the Ricci identity that commutes covariant derivatives. Einstein's equations never appear in its derivation. What they do is supply the physics afterward: they let you replace the abstract curvature term $R_{ab}u^a u^b$ by the matter content $8\pi(T_{ab}-\tfrac12 T g_{ab})u^a u^b$ . Only then, and only if an energy condition fixes the sign of that matter term, does the identity acquire predictive teeth. Keeping the geometry and the physics separate is exactly what makes the equation so powerful — the geometry is unconditional, and you can plug in whatever matter model you like.

Why does shear focus but rotation defocus?

Both $\sigma_{ab}$ and $\omega_{ab}$ are purely spatial tensors, but they enter the volume ledger with opposite signs, and the reason is algebraic. When you contract the velocity gradient with itself, $B_{ab}B^{ba}$ , the symmetric shear contributes $+\sigma_{ab}\sigma^{ab}$ while the antisymmetric vorticity contributes $-\omega_{ab}\omega^{ab}$ , because contracting an antisymmetric tensor with its own transpose flips a sign. That whole combination then enters $\mathrm{d}\theta/\mathrm{d}\tau=-B_{ab}B^{ba}-R_{ab}u^a u^b$ with an overall minus, so the net effect is $-\sigma^2$ (focusing) and $+\omega^2$ (defocusing). Physically it is intuitive: shear pinches the cloud along some directions while stretching others, and the pinch dominates the volume; rotation, by contrast, is centrifugal — a swirling cloud resists being crushed, exactly as a spinning skater resists being pulled inward.

What exactly is a conjugate point, and how does it relate to θ → −∞?

A conjugate point along a geodesic is a place where infinitesimally neighbouring geodesics of the congruence — say, all those emanating from a common point or from a common spacelike surface — reconverge and cross. It is a caustic, the geometric analogue of the bright cusp where sunlight focused by a wine glass piles up. Because the expansion is the logarithmic rate of change of the transverse cross-sectional volume, $\theta=\mathrm{d}\ln V/\mathrm{d}\tau$ , a crossing where $V\to0$ forces $\theta\to-\infty$ . Conjugate points matter because past one, a timelike geodesic is no longer locally length-maximising: there is always a nearby curve of greater proper time. That failure of maximality is the precise lever the singularity theorems pull, using it to derive a contradiction with the global existence of maximal geodesics — a contradiction resolved only by geodesic incompleteness.

Can a rotating congruence avoid singularities entirely?

Locally, yes — and that is exactly why the singularity theorems are so careful about rotation. The vorticity term $+\omega_{ab}\omega^{ab}$ is the one contribution to the Raychaudhuri equation that fights collapse, so a sufficiently rapidly rotating congruence can keep its expansion from ever running to $-\infty$ . Kurt Gödel's famous 1949 rotating cosmological model exploits precisely this: it is a solution of Einstein's equations with rotating dust and no initial singularity (though it has other pathologies, like closed timelike curves). This is why the singularity theorems restrict to hypersurface-orthogonal congruences, where Frobenius' theorem guarantees $\omega=0$ . The physical caveat is that irrotationality is generic for matter falling from rest and for the null generators of horizons, so the theorems still bite in the cases that matter most — realistic gravitational collapse and cosmology.

How does the null version connect to the Hawking area theorem?

The future event horizon of a black hole is generated by null geodesics, and their expansion $\theta$ measures how the horizon's cross-sectional area changes along them. Suppose, for contradiction, that $\theta<0$ somewhere on a generator. The null Raychaudhuri equation, plus the null energy condition ( $R_{ab}k^a k^b\ge0$ ), then forces $\theta\to-\infty$ within finite affine parameter — a caustic. But a caustic means the generator has left the horizon and entered its interior, which a future horizon's generators are forbidden to do (they have no future endpoints). The only way out is that $\theta\ge0$ everywhere on the horizon, so its area can never decrease. That is Hawking's area theorem, the classical forerunner of black-hole thermodynamics and the reason horizon area behaves like an entropy.

Why is the coefficient exactly 1/3 for timelike and 1/2 for null geodesics?

The coefficient is $1/n$ , where $n$ is the dimension of the transverse space over which the volume element is defined. For a timelike geodesic in four-dimensional spacetime, the space orthogonal to the tangent $u^a$ is three-dimensional, so the little ball of test particles lives in three dimensions and $n=3$ . For a null geodesic, something subtle happens: the tangent $k^a$ is orthogonal to itself (it is null), so one of the transverse directions collapses onto the ray itself and carries no independent volume. The physically meaningful cross-section — the "screen" onto which you would project a shadow — is only two-dimensional, so $n=2$ and the coefficient is $1/2$ . The change is not cosmetic: it is why light focuses under the weaker null energy condition and why the null equation, not the timelike one, governs light cones and horizons.