Geometric Uniqueness Modulo Diffeomorphism

Scrollable formula. Use Left and Right arrow keys to reveal hidden content.

Scrollable table. Use Left and Right arrow keys to reveal additional columns.

Scrollable figure. Use Left and Right arrow keys to inspect hidden content.

A gauge-fixed hyperbolic system gives uniqueness of coordinate components, but Einstein’s equation asks for uniqueness of geometry. The correct statement is that two developments of the same initial data are locally isometric by a map that identifies their embedded initial hypersurfaces. Wave-map gauge converts this geometric comparison into the analytic uniqueness theorem without identifying arbitrary coordinate labels.

Required background. Analysis for Mathematical Relativity supplies uniqueness for hyperbolic systems, The Einstein Equation supplies diffeomorphism covariance, PDEs and Green Functions for Gravitation supplies local domains of dependence, and Tensor Fields and Natural Operations supplies pullbacks and isometries.

Helpful background. Causality and Global Structure supplies Cauchy developments and causal domains.

Developments of one data set

Let $(\Sigma,h,K)$ satisfy the vacuum constraints, or the compatible constraints for a specified matter model. A development is a time-oriented Lorentzian spacetime $(M,g)$ together with an embedding $\iota:\Sigma\to M$ such that

\iota^*g=h, \qquad K_{ij}=-h_i{}^\mu h_j{}^\nu\nabla_\mu n_\nu,

and $(M,g)$ solves the field equations. A globally hyperbolic development additionally has $\iota(\Sigma)$ as a Cauchy hypersurface.

Two developments $(M_1,g_1,\iota_1)$ and $(M_2,g_2,\iota_2)$ represent the same local solution when neighborhoods $U_a$ of $\iota_a(\Sigma)$ admit a diffeomorphism

\Phi:U_1\longrightarrow U_2, \qquad \Phi^*g_2=g_1, \qquad \Phi\circ\iota_1=\iota_2.

This is geometric uniqueness. Requiring $g_{1,\mu\nu}(x)=g_{2,\mu\nu}(x)$ in coordinates chosen independently on the two spacetimes would be meaningless.

The wave-map comparison

Choose the same gauge data on $\Sigma$ and use wave coordinates, or more invariantly wave maps, to compare both developments with a common target geometry. A map $\psi:(M,g)\to(N,\widehat g)$ is a wave map when its tension field vanishes:

\tau(\psi)^A =g^{\mu\nu} \left( \partial_\mu\partial_\nu\psi^A -\Gamma^\rho{}_{\mu\nu}(g)\partial_\rho\psi^A +\widehat\Gamma^A{}_{BC}(\psi) \partial_\mu\psi^B\partial_\nu\psi^C \right)=0.

Prescribe identical Cauchy data for $\psi_1$ and $\psi_2$ relative to the given identifications of $\Sigma$ , with an initially invertible differential. The inverse-function theorem makes them local coordinate maps for a short time. On the common target, compare $\widetilde g_a=(\psi_a^{-1})^*g_a$ ; these metrics satisfy the same reduced Einstein system with the same initial metric and normal derivative. Hyperbolic uniqueness gives $\widetilde g_1=\widetilde g_2$ . Undoing the gauge produces $\Phi=\psi_2^{-1}\circ\psi_1$ and hence $\Phi^*g_2=g_1$ .

There are three nontrivial steps:

solve the wave-map equations with compatible initial data;
show that each reduced metric is a solution of the full Einstein equation using the subsidiary system;
translate equality in the common gauge into an isometry fixing the initial data.

The local geometric-uniqueness argument and the coordinate-regularity issue are developed in Friedrich and Rendall 2000, §5.2. At finite Sobolev regularity one must use a theorem that controls the differentiability of the comparison maps; formal substitution into their equations is not enough. The global result additionally requires the patching construction of Choquet-Bruhat and Geroch 1969, pp. 329–335, treated on The Maximal Globally Hyperbolic Development.

Hypotheses of the comparison

Assume both spacetimes solve the same vacuum or Einstein–matter system in the same Sobolev solution class, their initial embeddings induce the same $(h,K)$ and matter data, and their time orientations agree. The wave maps must exist with enough regularity to pull back the metrics, and their initial values and normal derivatives must be normalized identically. For matter with its own gauge, such as a Maxwell potential, that gauge must be aligned as well. These are the comparison hypotheses behind the standard local uniqueness treatment in Friedrich and Rendall 2000, §§2–4.

The proof then has a reversible structure. Pull both metrics to the common wave-map target; apply uniqueness to the full reduced metric–matter system; use subsidiary propagation to ensure both representatives solve the unreduced equations; and push the equality back to an isometry. If the target maps cease to be injective farther away, the argument still proves uniqueness on their common local domain. Local wave-map failure does not imply that the geometry has become singular.

The initial-data condition on the isometry is also stronger than setwise preservation of $\Sigma$ . It fixes the induced fields and the chosen future normal. Without this normalization, a symmetry of the initial data can be composed with the comparison map, so the isometry need not be unique as a bare map even though the geometric development is unique in the initial-data-preserving category.

What uniqueness does not say

First, the isometry need not preserve coordinates selected after evolution. Start with Minkowski spacetime in inertial coordinates and replace $t$ by $t'=t+f(x)$ in a region while preserving spacelikeness. The component arrays differ, but the identity geometry is unchanged.

Second, uniqueness is relative to the initial data and solution class. Lowering regularity can invalidate the theorem used to construct the wave maps or compare coefficients. Adding a timelike boundary requires boundary data and compatibility conditions; identical Cauchy data alone then do not determine a solution.

Third, local uniqueness does not prove geodesic completeness or global extendibility. It says that wherever two globally hyperbolic developments of the same data overlap, they agree geometrically. Constructing a largest such development requires compatible unions and control of topological pathologies.

Finally, residual gauge transformations are not counterexamples. For the linearized covariant harmonic condition on a fixed background, the gauge generator satisfies

\Box_g\xi_\mu+R_\mu{}^\nu\xi_\nu=0.

Here $\Box_g$ is the connection wave operator on a covector, not a scalar wave operator applied independently to its coordinate components. The Ricci term vanishes on a Ricci-flat background, and in inertial Minkowski coordinates this becomes $\Box_\eta\xi_\mu=0$ . Fixing both $\xi|_\Sigma$ and $\nabla_n\xi|_\Sigma$ removes this freedom in the comparison problem. If those data are not fixed, coordinate uniqueness was never claimed.

Check your understanding

Suppose $g_2=\chi^*g_1$ for a diffeomorphism $\chi$ that equals the identity on $\iota(\Sigma)$ but has nontrivial second derivatives away from it. Are $g_1$ and $g_2$ distinct solutions of the same geometric initial-value problem?

Solution

They can have different coordinate components, but $\chi:(M,g_2)\to(M,g_1)$ is an isometry by construction. Give $g_2$ the pulled-back time orientation. Because $\chi$ fixes the embedding pointwise, it fixes its tangential differential, and its isometry property takes the future $g_2$ -unit normal to the future $g_1$ -unit normal. Thus both $h$ and $K$ agree automatically. Requiring the coordinate normal vector itself to remain unchanged would be an unnecessary gauge restriction. The example refutes coordinate uniqueness, not geometric uniqueness.

References

Choquet-Bruhat, Yvonne, and Robert Geroch. “Global Aspects of the Cauchy Problem in General Relativity.” Communications in Mathematical Physics 14 (1969): 329–335. DOI.
Friedrich, Helmut, and Alan D. Rendall. “The Cauchy Problem for the Einstein Equations.” In Einstein’s Field Equations and Their Physical Implications, Lecture Notes in Physics 540, 127–224. Springer, 2000. arXiv:gr-qc/0002074.