Constraint-Preserving and Radiation-Controlling Boundaries

At a finite timelike boundary, Einstein evolution has incoming physical, constraint, and gauge characteristic fields. A mathematically controlled boundary prescription treats all three sectors: it closes the hyperbolic energy estimate, prevents the subsidiary constraints from entering, and assigns the desired incoming radiation without overconstraining outgoing waves. “Outgoing boundary condition” is therefore not a single geometric equation independent of formulation.

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

Helpful background. Causality and Global Structure clarifies boundary domains, and PDEs and Green Functions for Gravitation reviews characteristic boundary fields.

Incoming fields at a timelike boundary

Let $T$ be a future unit tangent to the timelike boundary $\mathcal T$, and let $N$ be its outward unit spacelike normal. The combinations $T+N$ and $T-N$ are null. For a first-order hyperbolic reduction, diagonalizing the principal boundary matrix separates fields according to the sign of their speed relative to $N$.

Outgoing fields are determined by the interior solution. Incoming fields require boundary data. Zero-speed fields evolve tangentially or obey boundary constraints. Prescribing an outgoing field independently would generally overdetermine the PDE, while leaving an incoming field uncontrolled prevents an energy estimate.

The exact characteristic variables change with harmonic, tetrad, generalized-harmonic, or other reductions. Their conceptual division is nevertheless stable:

1. two physical gravitational degrees of freedom;
2. fields associated with the four Einstein constraints or gauge constraints;
3. fields fixing coordinates, frame, or boundary location.

This is a division of boundary roles, not a claim that each sector is invariant under arbitrary changes of reduction.

Making the subsidiary problem homogeneous

In a chosen chart, use the same generalized harmonic constraint as in the Cauchy chapters:

$$C^\mu=g^{\alpha\beta}\Gamma^\mu{}_{\alpha\beta}-F^\mu(x,g).$$

The reduced Einstein equations imply, through the contracted Bianchi identity, a wave-type subsidiary system

$$\Box_g C^\mu+R^\mu{}_{\nu}C^\nu=0$$

in vacuum for the undamped reduction $R_{\mu\nu}-\nabla_{(\mu}C_{\nu)}=0$. The gauge-source functions $F^\mu$ are prescribed smooth functions of position and metric, with their coordinate transformation understood as part of the chosen gauge; equivalently one can use the tensorial wave-map formulation relative to a background connection. Vanishing initial values of $C$ and its normal derivative are not enough on a region with timelike boundary: incoming characteristic combinations of $C^\mu$ must also vanish or be controlled. Otherwise the unique solution of the subsidiary initial-boundary problem need not be $C^\mu=0$.

A constraint-preserving boundary condition is constructed by rewriting the incoming subsidiary fields in terms of incoming main-system fields and prescribing those combinations so that the subsidiary boundary data are homogeneous. Kreiss and collaborators carry out this hierarchical construction for harmonic Einstein wave equations in Kreiss et al. 2007, §3, Theorems 2–3, and Appendix 1. The main sections establish constant-coefficient estimates; the appendix explains their local quasilinear use.

Constraint preservation does not by itself control physical radiation. The constraint subsystem can be homogeneous while arbitrary gravitational waves enter through the two remaining physical modes.

Radiation control

After fixing an oriented null tetrad at $\mathcal T$, physical incoming radiation can be represented by a trace-free tidal or Weyl-curvature combination. It is often denoted by a Newman–Penrose scalar such as $\Psi_0$, but which scalar is “incoming” depends on the ordering and normalization of the null legs. A valid boundary statement must define that tetrad rather than rely on the symbol alone.

A schematic radiation condition has the form

$$\mathcal W_{\mathrm{in}} =q+\mathcal R\mathcal W_{\mathrm{out}},$$

where $q$ is prescribed incoming radiation and $\mathcal R$ is a reflection operator compatible with the boundary estimate. The choice $q=0$, $\mathcal R=0$ is an absorbing idealization, not an exact transparent boundary for generic curved spacetime. Curvature backscatter and the finite location of $\mathcal T$ make perfect absorption nonlocal in time in general.

The simplest energy calculation shows what “compatible” means. For $u_{tt}-u_{xx}=0$ on $x\geq0$, define $w_{\mathrm{in}}=u_t-u_x$ and $w_{\mathrm{out}}=u_t+u_x$. The boundary contribution to the standard wave energy is

$$\frac{dE}{dt} =\frac14\left(w_{\mathrm{in}}^2-w_{\mathrm{out}}^2\right)_{x=0}.$$

Thus a homogeneous condition $w_{\mathrm{in}}=R\,w_{\mathrm{out}}$ is dissipative for $\lvert R\rvert\leq1$. For an energy-normalized vector of characteristic fields the analogous sufficient condition is $\mathcal R^\dagger\mathcal R\leq I$. One must first identify the actual boundary energy and characteristic normalization; this matrix inequality cannot simply be imposed on arbitrarily normalized Weyl components.

Friedrich and Nagy’s frame formulation uses freely specifiable boundary curvature data while arranging a maximally dissipative reduced system; see Friedrich–Nagy 1999, §§3–7, pp. 625–650. The geometric meaning of these data is tied to the boundary gauge and frame. Winicour discusses the worldtube–nullcone alternative and its radiation data in Winicour 2012, §§6.1 and 6.6.

A hierarchical boundary design

For many hyperbolic reductions, robust boundary conditions are imposed in an order dictated by the principal system:

• prescribe boundary gauge data needed to locate $\mathcal T$ and fix coordinates;
• impose constraint combinations that make the subsidiary system homogeneous;
• prescribe the two physical incoming fields;
• obtain remaining incoming variables from boundary evolution equations;
• verify maximal dissipativity or a uniform Kreiss estimate;
• enforce initial-boundary compatibility at the corner.

This ordering prevents circular definitions. A condition involving tangential derivatives may still be admissible, but it requires an estimate that controls those derivatives. Counting the number of conditions is only the first check; their differential form and coupling determine stability.

Limitations and common confusions

Setting all incoming characteristic variables to zero is stronger than “no incoming gravitational radiation”: it also fixes incoming gauge and constraint fields. Conversely, specifying a Weyl scalar alone does not guarantee constraint preservation. Conditions derived for a frozen planar boundary do not automatically remain stable on a moving curved boundary. Finally, a boundary stable at the continuum level may be unstable after discretization; that numerical question belongs to Computational Relativity.

The well-posed Einstein IBVP supplies the existence framework. The remaining complication is that the boundary and initial surface meet, producing the corner and AdS compatibility problem.

Solved diagnostic: physical control does not control constraints

Use two independent half-line wave equations to show why a dissipative boundary condition on a physical field does not prevent an unconstrained mode from entering through the same boundary.

Solution

Consider two independent half-line waves, $u_{tt}-u_{xx}=0$ and $c_{tt}-c_{xx}=0$, where $u$ models a physical field and $c$ a constraint. Imposing the dissipative condition $u_t-u_x=0$ at $x=0$ controls the energy of $u$. If no boundary condition is imposed on $c$, the function

$$c(t,x)=F(t-x)$$

enters the domain from $x=0$ for any smooth $F$ supported at positive time, even when $c$ and $c_t$ initially vanish. Thus physical radiation control alone cannot prove $c=0$. Adding $c_t-c_x=0$ makes the constraint energy nonincreasing and, with zero initial data, forces $c=0$.

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.
• Winicour, Jeffrey. “Characteristic Evolution and Matching.” Living Reviews in Relativity 15 (2012): 2. arXiv:0810.1903.