Constraint and Gauge Propagation

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.

The local existence proof solves a reduced wave system, not Einstein’s equation directly. The gap is closed by a subsidiary theorem: the harmonic-gauge constraint obeys a homogeneous wave equation, and the Hamiltonian and momentum constraints provide its missing initial derivative. Zero subsidiary Cauchy data therefore remain zero. If stress-energy is not conserved, the equation is forced and the conclusion fails.

Required background. Analysis for Mathematical Relativity supplies uniqueness at the required regularity, The Einstein Equation supplies the Bianchi identity, and PDEs and Green Functions for Gravitation supplies wave propagation.

Helpful background. Causality and Global Structure supplies the domain-of-dependence statement.

Reduced and full equations

Let

E_{\mu\nu} =G_{\mu\nu}+\Lambda g_{\mu\nu}-8\pi G T_{\mu\nu}

and let $C^\mu$ denote the generalized harmonic-gauge constraint. A convenient trace-adjusted reduced equation is

E^{\mathrm{red}}_{\mu\nu} =E_{\mu\nu}-\nabla_{(\mu}C_{\nu)} +\frac12g_{\mu\nu}\nabla_\alpha C^\alpha=0.

Its trace reverse is the diagonal reduced Ricci wave system, so Wave-Gauge Local Existence constructs $g$ . Until $C=0$ is established, the residual $E_{\mu\nu}$ is only known to equal a differential expression in $C$ ; solving the reduced equations alone does not establish the full field equation.

Assume that the matter equations imply

\nabla^\mu T_{\mu\nu}=0.

Taking the divergence of $E^{\mathrm{red}}_{\mu\nu}=0$ , using $\nabla^\mu G_{\mu\nu}=0$ , and commuting covariant derivatives gives

\Box_g C_\nu+R_\nu{}^\mu C_\mu=0,

exactly for the displayed reduced equation. The gauge source has already been included in the definition of $C$ ; merely choosing a nonzero prescribed source does not add another term. Constraint-damping modifications to the reduced equations would change the subsidiary equation. Once the reduced metric is known, the displayed equation is linear and homogeneous. See Friedrich and Rendall 2000, §2.4, immediately after equation (2.39).

How the initial constraints enter

Gauge initialization is chosen so that

C^\mu|_\Sigma=0.

That condition alone is insufficient for a second-order wave equation. The normal derivative $\nabla_nC$ is related by the reduced equations to the normal projections of $E_{\mu\nu}$ . Those projections are exactly the Hamiltonian and momentum constraints. With

K_{ij}=-h_i{}^\mu h_j{}^\nu\nabla_\mu n_\nu,

they read

R(h)+K^2-K_{ij}K^{ij}=2\Lambda+16\pi G\rho,

D^j(K_{ij}-Kh_{ij})=8\pi G j_i.

One can see the missing derivative explicitly. At a point of $\Sigma$ , choose an orthonormal frame $(n,e_i)$ and set $V_\nu=(\nabla_nC)_\nu$ . Since $C$ vanishes all along the slice, its tangential covariant derivatives vanish there. The reduced equation gives

E(n,n)=\frac12V_n,\qquad E(n,e_i)=\frac12V_i.

The four constraint equations set these four projections of $E$ to zero. Thus, if the geometric constraints and initial gauge condition hold,

\nabla_nC^\mu|_\Sigma=0.

Uniqueness for the subsidiary wave equation yields $C=0$ in $D(\Sigma)$ . Substitution into the reduced equation gives $E_{\mu\nu}=0$ . This proves constraint and gauge propagation simultaneously at the continuum level. The reduction, vanishing initial subsidiary data, and propagation argument are developed together in Friedrich and Rendall 2000, §2.4, equations (2.34)–(2.39) and the discussion immediately following.

Exact hypotheses and proof closure

Assume the reduced metric belongs to the Sobolev class of the local theorem, the gauge source is prescribed with compatible regularity, and the initial hypersurface is spacelike. Assume the matter evolution is solved—not independently prescribed—and implies stress-energy conservation. These hypotheses ensure that the curvature term in the subsidiary equation has the regularity needed for its energy estimate.

The proof closes in a specific order. First, $C|_\Sigma=0$ fixes the contracted connection or wave-map tension on the slice. Second, the reduced equations express the normal derivatives of $C$ in terms of $E(n,n)$ and $E(n,h_i)$ ; the Hamiltonian and momentum constraints set those projections to zero. Third, a wave-energy estimate gives

E_C(t)\le E_C(0) \exp\left(\int_0^t a(\tau)d\tau\right),

where $a$ depends on controlled metric and curvature coefficients. Since $E_C(0)=0$ , the gauge constraint vanishes. The argument uses uniqueness, not a sign cancellation special to vacuum.

One can perform an equivalent check after a $3+1$ reduction. Let $\mathcal H$ and $\mathcal M_i$ be the Hamiltonian and momentum residuals. The normal and tangential projections of $\nabla^\mu E_{\mu\nu}=0$ form a homogeneous first-order system for $(\mathcal H,\mathcal M_i)$ once the evolution equations hold. An energy estimate for that system again makes zero residuals propagate. Agreement of the wave-gauge and projected derivations is an independent consistency check on the signs.

Wald 1984, §10.2, pp. 259–262, equations (10.2.31) and (10.2.35)–(10.2.36) presents both the contracted-Bianchi reasoning and its harmonic-coordinate subsidiary version. The coordinate gauge functions there differ from the covariant wave-map constraint used here, so their lower-order terms should not be identified component by component without translating definitions.

The conclusion is confined to the domain of dependence covered by the local solution. It does not imply that constraints remain small under arbitrary discretization, nor does it supply boundary conditions when subsidiary characteristics enter through a timelike boundary.

As a final check, the zero solution of the subsidiary equation must be admissible in exactly the same coefficient class as the uniqueness theorem. If a derivation demands more derivatives of $g$ than local existence supplies, it has not closed at the claimed regularity.

Matter conservation is structural

If the prescribed source has

J_\nu=8\pi G\nabla^\mu T_{\mu\nu}\ne0,

the subsidiary equation becomes, with this definition of $J$ and the displayed sign of the reduction,

\Box_gC_\nu+R_\nu{}^\mu C_\mu=-2J_\nu.

Even zero initial gauge data then generate nonzero $C$ . A physically consistent matter model avoids this by coupling matter evolution to the metric so that its equations imply stress-energy conservation. Merely prescribing a time-dependent tensor $T_{\mu\nu}$ is generally incompatible with Einstein evolution.

At finite numerical resolution, truncation error also produces nonzero constraint residuals. The continuum theorem predicts homogeneous propagation only for exact equations. Constraint damping alters the subsidiary system so small violations decay in selected regimes; it does not turn an inconsistent source into a solution of Einstein’s equation.

Another limitation is regularity. The divergence, commutator, and uniqueness argument must make sense in the solution class. A formal Bianchi calculation on a metric below the theorem’s threshold does not establish propagation without a weak formulation and a corresponding uniqueness result.

Check your understanding

Suppose $C|_\Sigma=0$ and $\nabla_nC|_\Sigma=0$ , but $\nabla^\mu T_{\mu\nu}=J_\nu/(8\pi G)$ with $J\ne0$ . Can the reduced solution remain in harmonic gauge?

Solution

No, wherever $J$ is nonzero the identically zero field $C$ fails the forced equation $\Box_gC_\nu+R_\nu{}^\mu C_\mu=-2J_\nu$ . Zero initial data therefore do not preserve the gauge constraint. In fact no exact Einstein solution with that prescribed nonconserved stress tensor can exist there: taking a divergence of Einstein’s equation would give $J=0$ .

References

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.
Wald, Robert M. General Relativity. University of Chicago Press, 1984. Publisher record.