The Raychaudhuri Identity

The Raychaudhuri identity is the trace evolution equation for a congruence. It says exactly how isotropic expansion changes because of expansion itself, shear, vorticity, curvature, and nongeodesic acceleration. The equation is geometric; converting it into a focusing or singularity result requires additional local and global hypotheses that are not imposed here.

Its usefulness comes from compressing a matrix evolution equation into one scalar while keeping every source term invariant. Its limitation comes from the same compression: the trace cannot reconstruct the full deformation tensor, locate a caustic by itself, or determine global continuation. The identity should therefore be read together with initial data for the congruence and, when focal behavior matters, with the underlying Jacobi map.

Required background. The Riemann Curvature Tensor fixes the commutator sign and Ricci contraction; Affine and Variational Geodesics supplies affine geodesics, affine parameters, and the variation setup behind geodesic congruences; Jacobi Fields and Geodesic Deviation supplies the deviation equation and Jacobi-map interpretation used to distinguish directional focal behavior from divergence of the scalar expansion. Helpful background. Calculus and Linear Algebra for Relativity supplies trace identities and first-order differential inequalities; Frames, Coframes, and Anholonomy supplies spatial components for independent checks.

Commuting the derivatives

Let $u^\mu u_\mu=-1$, define $a^\mu=u^\nu\nabla_\nu u^\mu$, and set $\theta=\nabla_\mu u^\mu$. Along the congruence,

$$\dot\theta\equiv u^\nu\nabla_\nu\theta =u^\nu\nabla_\nu\nabla_\mu u^\mu.$$

Apply the product rule to $\nabla_\mu(u^\nu\nabla_\nu u^\mu)$ and use the curvature convention

$$[\nabla_\mu,\nabla_\nu]V^\rho =R^\rho{}_{\sigma\mu\nu}V^\sigma.$$

The result is

$$\dot\theta =\nabla_\mu a^\mu -(\nabla_\mu u^\nu)(\nabla_\nu u^\mu) -R_{\mu\nu}u^\mu u^\nu.$$

The minus sign on the Ricci term is fixed by the order of the commuted derivatives. Checking it from the stated commutator is safer than importing an equation from a source that may use the opposite curvature convention.

More explicitly, the product rule isolates

$$\nabla_\mu a^\mu =(\nabla_\mu u^\nu)(\nabla_\nu u^\mu) +u^\nu\nabla_\mu\nabla_\nu u^\mu.$$

The commutator replaces the last double derivative by the derivative of $\theta$ plus the Ricci contraction. Rearranging gives the stated intermediate identity. This route also shows why the acceleration contribution is a spacetime divergence: it is generated before the derivative of $u^\mu$ is decomposed into spatial pieces.

Tracing the deformation map

Decompose the derivative as

$$\nabla_\nu u_\mu =\frac{\theta}{n-1}h_{\mu\nu} +\sigma_{\mu\nu} +\omega_{\mu\nu} -u_\nu a_\mu.$$

Orthogonality makes all cross terms with $a_\mu$ vanish in the contraction above. Symmetry and antisymmetry make the shear–vorticity cross term vanish. Finally,

$$B^\nu{}_\mu B^\mu{}_\nu =\frac{\theta^2}{n-1} +\sigma_{\mu\nu}\sigma^{\mu\nu} -\omega_{\mu\nu}\omega^{\mu\nu}.$$

The vorticity enters with the opposite sign because $A^\nu{}_\mu A^\mu{}_\nu=-A_{\mu\nu}A^{\mu\nu}$ for an antisymmetric spatial tensor. Substitution yields

$$\dot\theta =-\frac{\theta^2}{n-1} -\sigma_{\mu\nu}\sigma^{\mu\nu} +\omega_{\mu\nu}\omega^{\mu\nu} -R_{\mu\nu}u^\mu u^\nu +\nabla_\mu a^\mu.$$

The geodesic trace equation is discussed in Kar and SenGupta 2007, §II.C, pp. 10–12, PDF, and the acceleration-divergence term is added in Kar and SenGupta 2007, §II.E, Eq. (18), p. 14, PDF. The commutator calculation above fixes the complete nongeodesic expression in the conventions used here.

What the identity does and does not say

For geodesic curves, $a^\mu=0$. For a hypersurface-orthogonal congruence, $\omega_{\mu\nu}=0$. Those are independent conditions: Rindler observers are hypersurface orthogonal but accelerated, while a rotating Born-rigid congruence is accelerated and vortical.

The signs of the remaining terms do not follow from kinematics alone. The spatial norm $\sigma_{\mu\nu}\sigma^{\mu\nu}$ is nonnegative, but the Ricci contraction can have either sign in a general Lorentzian geometry. The identity therefore supplies an evolution law, not an automatic conclusion that the congruence must converge.

Even when additional assumptions force $\theta$ toward negative infinity, the immediate geometric conclusion concerns the degeneration of the congruence’s separation map. Establishing incomplete geodesics, inextendibility, or singular curvature requires further hypotheses and global arguments. No Einstein equation or matter inequality has been used in this derivation.

The dimension is also consequential. The coefficient $1/(n-1)$ counts the directions transverse to a timelike curve. For a null congruence it becomes $1/(n-2)$ because the propagation direction must first be quotiented out. Merely replacing a timelike unit tangent by a null vector in the displayed equation would retain the wrong projector, trace dimension, normalization, and acceleration concept.

An infinitesimal volume formed from projected material separations obeys $\dot{\mathcal V}=\theta\mathcal V$, even when vorticity is nonzero; orthogonal hypersurfaces are unnecessary for this local statement. Divergence of $\theta$ can accompany $\mathcal V\to0$, but the determinant is the decisive object. A scalar differential equation for $\theta$ loses directional information about which connecting vectors vanish and cannot distinguish every multiplicity of a focal point.

Independent flat-space checks

For the Milne congruence in four-dimensional Minkowski spacetime,

$$\theta=\frac3\tau, \qquad \sigma_{\mu\nu}=\omega_{\mu\nu}=a^\mu=0, \qquad R_{\mu\nu}=0.$$

The two sides agree because $\dot\theta=-3/\tau^2=-\theta^2/3$. This is a stringent coefficient check: replacing $n-1$ by $n$ would fail even in flat spacetime.

For Rindler observers with metric

$$ds^2=-\rho^2d\eta^2+d\rho^2+dy^2+dz^2,$$

On $\rho\gt0$, $u=\rho^{-1}\partial_\eta$ has $\theta=\sigma=\omega=0$ and $a^\rho=1/\rho$. Since $\sqrt{-g}=\rho$,

$$\nabla_\mu a^\mu =\frac1\rho\partial_\rho(\rho a^\rho)=0.$$

All terms vanish. The nonzero magnitude $a_\mu a^\mu=1/\rho^2$ is not itself a term in this form of the identity; replacing the divergence by $a^2$ would give a false result.

Common pitfalls

Curvature signs travel together. Changing the definition of $R^\rho{}_{\sigma\mu\nu}$ changes the commutator and the Ricci term. Translate the whole convention set, not one displayed sign.

Affine and nonaffine tangents differ. The timelike equation above uses a unit tangent and proper time. Null generators require a separate screen construction, and a nonaffine null parameter adds an inaffinity term.

A caustic is not a curvature singularity. The Milne congruence traced backward meets at one event in flat spacetime. Its divergent expansion records the failure of the congruence chart there, while the Riemann tensor remains zero.

Exercises

Derive the contraction identity

$$B^\nu{}_\mu B^\mu{}_\nu =\frac{\theta^2}{n-1}+\sigma^2-\omega^2$$

from the irreducible decomposition of $B_{\mu\nu}$.

Solution

The trace part is orthogonal under contraction to the trace-free shear and to the antisymmetric vorticity. Symmetric–antisymmetric contraction also vanishes. The trace part contributes

$$\frac{\theta^2}{(n-1)^2}h^\nu{}_\mu h^\mu{}_\nu =\frac{\theta^2}{n-1}.$$

For the antisymmetric part, $\omega^\nu{}_\mu=-\omega_\mu{}^\nu$, so

$$\omega^\nu{}_\mu\omega^\mu{}_\nu =-\omega_{\mu\nu}\omega^{\mu\nu}.$$

Adding the shear norm gives the stated identity and fixes the positive vorticity term in Raychaudhuri’s equation.

References

• Kar, Sayan, and Soumitra SenGupta. “The Raychaudhuri Equations: A Brief Review.” Pramana 69 (2007), 49–76. arXiv:gr-qc/0611123.