Well-Posedness of the Einstein Initial-Boundary Value Problem

An Einstein initial-boundary value problem supplements Cauchy data on a spacelike hypersurface with data on a timelike boundary. A successful formulation must do more than count incoming fields: it must be locally well posed, preserve the Einstein constraints, control gauge freedom, and reconcile the initial and boundary data at their corner. These requirements are coupled, so a boundary condition that looks physically natural can still make the PDE problem ill posed.

Required background. Analysis for Mathematical Relativity supplies energy estimates and trace theory; The Einstein Equation supplies constraints and Bianchi propagation; Hypersurface Normals and Tangential Projectors supplies boundary geometry.

Helpful background. Causality and Global Structure clarifies domains of dependence, and PDEs and Green Functions for Gravitation reviews hyperbolic boundary problems.

The spacetime corner and the data

Let $\Sigma$ be a compact spacelike three-manifold with boundary $S=\partial\Sigma$. Evolution sweeps $S$ into a timelike hypersurface $\mathcal T$, and the spacetime region has corner

$$S=\Sigma\cap\mathcal T.$$

Initial data $(h,K)$ on $\Sigma$ satisfy the Hamiltonian and momentum constraints. Here

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

for the future unit normal. Boundary data on $\mathcal T$ include some combination of gauge information, constraint-preserving fields, and the incoming gravitational degrees of freedom. The exact variables depend on the chosen hyperbolic reduction.

The initial metric determines the value of the evolved fields at $S$; the boundary prescription determines the same value from $\mathcal T$. Equality is the zeroth-order corner condition. Differentiating the evolution and boundary equations gives first- and higher-order compatibility conditions. Smooth initial and boundary functions that fail this jet hierarchy cannot produce a smooth solution at the corner.

What well posedness requires

After gauge fixing, write a first-order reduced system schematically as

$$A^0(U)\partial_tU+A^i(U)\partial_iU=F(U),$$

where $A^0$ is positive definite and the system is symmetric hyperbolic. At a timelike boundary with outward spatial conormal $s_i$, the eigenvalues of $(A^0)^{-1}A^is_i$ divide the fields into incoming, outgoing, and zero-speed sectors. Negative speeds are incoming in this sign convention. Boundary conditions determine incoming fields from prescribed data and, possibly, outgoing fields; they do not independently prescribe the outgoing fields already determined by the interior. A dissipativity or uniform boundary-stability estimate must control the energy flux.

For quasilinear systems, local well posedness means that compatible data in specified Sobolev spaces give a solution for a common short time, that the solution is unique in the chosen gauge, and that it depends continuously on the data. It does not follow merely from strong hyperbolicity of the interior equations. The boundary operator must satisfy its own stability condition; frozen-coefficient analysis is necessary but not always sufficient for the variable-coefficient problem.

Kreiss and collaborators establish energy estimates for a hierarchy of constraint-preserving harmonic Einstein boundary conditions in Kreiss et al. 2007, §§2–3, Theorems 1–3, and Appendix 1. Sections 2–3 treat the frozen constant-coefficient wave system; Appendix 1 explains the variable-coefficient and local quasilinear extension. Their harmonic theorem and the Friedrich–Nagy theorem use different reductions and boundary variables; their conclusions should not be blended into a formulation-independent list of freely prescribable geometric data.

The Friedrich–Nagy vacuum theorem

Friedrich and Nagy formulate the four-dimensional vacuum equations on a compact spatial manifold with boundary using an orthonormal frame, connection and curvature variables, gauge source functions, and a symmetric-hyperbolic subsidiary system. The hypotheses are smooth vacuum Cauchy data satisfying the constraints, smooth admissible timelike-boundary data, and compatibility to all required orders at $S$. These determine a smooth local vacuum development. The reduced fields are unique in the fixed gauge, and the construction controls the constraints and boundary flux. The hyperbolic boundary analysis is in Friedrich–Nagy 1999, §3, pp. 625–629, the gauge and reduced/subsidiary systems in §§4–6, pp. 629–644, the data in §7, pp. 644–650, and the existence statement in Theorem 8.1, pp. 650–652.

Its proof architecture is:

1. choose a frame and gauge adapted to $\mathcal T$;
2. isolate a reduced symmetric-hyperbolic evolution system;
3. impose maximally dissipative conditions on its incoming fields;
4. construct a homogeneous boundary problem for the subsidiary constraints;
5. use energy uniqueness to keep those constraints zero;
6. reconstruct a vacuum metric from the evolved frame and connection.

The theorem’s smoothness, compactness, timelike-boundary, gauge-source, and corner assumptions are part of the result. It is not a theorem for arbitrary finite-regularity data, a null outer boundary, or conformal infinity. Nor does it give a canonical physical meaning to every boundary gauge function.

Three boundary sectors

A helpful conceptual split is:

• physical sector: two incoming radiative degrees of freedom, often encoded through suitable Weyl-curvature combinations;
• constraint sector: combinations whose boundary values must make the subsidiary constraint system homogeneous or controlled;
• gauge sector: lapse, shift, coordinates, frame rotation, or gauge-source information needed to locate and parametrize the boundary.

The split is reduction-dependent. Prescribing only the physical sector does not prevent incoming constraint violations. Conversely, setting all incoming characteristic fields to zero generally fixes gauge and physical information more strongly than “no incoming radiation.” The next page treats constraint-preserving and radiation-controlling conditions directly.

Solved diagnostic: boundary flux for a wave

For a wave on $x\geq0$, determine which sign in $(\partial_t\pm\partial_x)u=0$ at $x=0$ removes incoming waves. Assume finite energy and sufficient decay to discard the flux at infinity.

Solution

Consider $u_{tt}-u_{xx}=0$ on $x\ge0$, with energy

$$E(t)=\frac12\int_0^\infty\left(u_t^2+u_x^2\right)dx.$$

Integration by parts gives

$$\frac{dE}{dt}=-u_tu_x\big|_{x=0}.$$

The outward normal points toward decreasing $x$, so the no-incoming Sommerfeld condition is $(\partial_t+\partial_n)u=u_t-u_x=0$. It yields $dE/dt=-u_t^2\le0$. Reversing the sign gives $dE/dt=+u_t^2$ and permits boundary energy injection. This scalar calculation is the model for checking the sign of a maximally dissipative Einstein boundary condition.

References

• Friedrich, Helmut, and Gabriel Nagy. “The Initial Boundary Value Problem for Einstein’s Vacuum Field Equation.” Communications in Mathematical Physics 201 (1999): 619–655. Open PDF.
• Kreiss, Heinz-Otto, Oscar Reula, Olivier Sarbach, and Jeffrey Winicour. “Well-Posed Initial-Boundary Value Problem for the Harmonic Einstein Equations Using Energy Estimates.” Classical and Quantum Gravity 24 (2007): 5973–5984. DOI. Open PDF.