Skip to content

Radiative, Reflective, and Dissipative Boundaries

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 boundary condition can be mathematically stable yet physically reflective, or physically suggestive yet fail to control the boundary energy. These are separate tests. Maximally dissipative conditions control the incoming characteristic fields with a contraction bound. Their reflection coefficient depends on which fields were diagonalized, the location and curvature of the boundary, background scattering, and the asymptotic expansion used to model radiation. No local condition at a generic finite radius is an exact substitute for the missing exterior Einstein solution.

Required background. The Gibbons–Hawking–York Boundary Term supplies timelike-boundary geometry, while PDEs and Green Functions for Gravitation supplies characteristic energy flux, Sommerfeld conditions, and stability estimates.

Helpful background. The Einstein Equation supplies gravitational radiation and constraints; Hypersurface Normals and Tangential Projectors supplies outward orientation; Formulations of General Relativity orients the volume; and Causality and Global Structure supplies the exterior domain omitted by a finite boundary.

Boundary energy and reflection coefficient

Section titled “Boundary energy and reflection coefficient”

Consider the scalar wave equation on x≤0x\le0 with boundary at x=0x=0 and outward normal toward increasing xx:

utt−uxx=0.u_{tt}-u_{xx}=0.

Define the outward and inward characteristic derivatives

wout=ut−ux,win=ut+ux.w_{\mathrm{out}}=u_t-u_x, \qquad w_{\mathrm{in}}=u_t+u_x.

The energy

E=12∫−∞0(ut2+ux2) dxE=\frac12\int_{-\infty}^0(u_t^2+u_x^2)\,dx

satisfies

dEdt=utux∣0=14(win2−wout2).\frac{dE}{dt} =u_tu_x\big|_{0} =\frac14 \left(w_{\mathrm{in}}^2-w_{\mathrm{out}}^2\right).

For

win=R wout+q,w_{\mathrm{in}}=R\,w_{\mathrm{out}}+q,

homogeneous data with ∣R∣≤1\lvert R\rvert\le1 do not inject energy. The cases R=−1R=-1 and R=1R=1 are perfectly reflective Dirichlet and Neumann conditions; R=0R=0 is exactly absorbing for this one-dimensional outgoing mode; and 0<∣R∣<10\lt \lvert R\rvert\lt 1 is partially dissipative. Thus “maximally dissipative” describes an energy inequality, not necessarily zero reflection.

Kreiss and collaborators derive boundary and interior estimates for wave equations with Sommerfeld and more general tangential-derivative conditions in Kreiss et al. 2007, Theorems 1–3, pp. 3–10, PDF. Their hypotheses include the sign and spectral restrictions on the boundary operators; a visually similar condition outside those restrictions has no automatic estimate.

For a three-dimensional outgoing scalar monopole,

Φ=f(t−r)r,\Phi=\frac{f(t-r)}{r},

the Sommerfeld condition

(∂t+∂r)(rΦ)=0(\partial_t+\partial_r)(r\Phi)=0

is exact. Higher multipoles contain additional inverse powers of rr, curvature backscatter creates tails, and nonlinear gravity couples modes. Consequently a homogeneous first-order Sommerfeld condition at r=Rr=R reflects part of a generic outgoing solution. To specify both phase and normalization, consider the flat-space dipole

Φω=∂z[e−iω(t−r)r+κωe−iω(t+r−2R)r],\Phi_\omega =\partial_z\left[ \frac{e^{-i\omega(t-r)}}r +\kappa_\omega\frac{e^{-i\omega(t+r-2R)}}r \right],

where zz is a Cartesian coordinate, r=x2+y2+z2r=\sqrt{x^2+y^2+z^2}, and ω>0\omega\gt0. Both terms solve the three-dimensional wave equation away from the origin. Imposing (∂t+∂r)(rΦω)=0(\partial_t+\partial_r)(r\Phi_\omega)=0 at r=Rr=R gives

κω=12ω2R2−2iωR−1=O((ωR)−2)\kappa_\omega =\frac{1}{2\omega^2R^2-2i\omega R-1} =O((\omega R)^{-2})

as ωR→∞\omega R\to\infty. This is the complex conjugate of Winicour 2012, §5.1, equations (62)–(63), pp. 48–49, whose time factor is e+iωte^{+i\omega t}. The sign of the imaginary term must change with the Fourier convention.

For a direct check, the outgoing term contributes R−2R^{-2} to the boundary residual after removing its common phase and cos⁡θ\cos\theta, while the incoming term contributes κω(−2ω2+2iω/R+R−2)\kappa_\omega(-2\omega^2+2i\omega/R+R^{-2}). Their sum vanishes for the displayed coefficient. The reflected-to-incident far-field power ratio is

∣κω∣2=11+4(ωR)4.|\kappa_\omega|^2=\frac{1}{1+4(\omega R)^4}.

It is 1/51/5 when ωR=1\omega R=1 and tends to one at low frequency. “Far away” must therefore be measured relative to wavelength, not only in coordinate units. This finite-radius error persists after grid refinement at fixed RR.

For Einstein evolution, two physical incoming characteristic combinations can be related to Weyl curvature or metric derivatives, but gauge and constraint incoming fields must be treated simultaneously. Setting a nominal “incoming radiation scalar” to zero does not by itself make the full Einstein boundary condition stable or constraint preserving.

A reflective boundary sends outgoing disturbances back into the domain. It can be conservative and analytically clean, which is appropriate for a physical wall or for common asymptotically AdS boundary conditions. It is inappropriate when a finite artificial boundary is meant to approximate an isolated system over times long enough for reflected radiation to return.

Reflectivity also depends on the variable. Dirichlet data for a metric representative need not mean a simple sign reversal of every gauge-invariant gravitational polarization. One should compute the frozen-coefficient scattering matrix for the physical characteristic fields and verify the energy flux in the formulation’s symmetrizer norm.

A strict contraction ∥H∥<1\lVert H\rVert\lt 1 in

Uin=HUout+qU_{\mathrm{in}}=H U_{\mathrm{out}}+q

removes energy from homogeneous perturbations in the controlled norm. With q≠0q\ne0, the estimate bounds injected energy by the prescribed boundary data. This is useful for an artificial outer boundary, but dissipation can absorb physical, gauge, and constraint fields at different rates. It is not a proof of accurate radiation transmission.

At the continuum level, stability, constraint preservation, and absorption must all pass. At the discrete level, a summation-by-parts or equivalent boundary treatment must reproduce the estimate; a stable continuum condition can be destabilized by its implementation.

ChoiceEnergy behaviorPhysical interpretationMain limitation
ReflectiveZero net homogeneous fluxClosed cavity or AdS-type wallReturning radiation
Homogeneous SommerfeldOutward flux for model modesApproximate isolationFinite-radius reflection and backscatter
Inhomogeneous radiativeControlled incoming signalExterior forcingBoundary data must represent missing exterior
Strictly dissipativeNorm decreases for homogeneous dataAbsorbing open boundaryMay damp the wrong characteristic content

Stable is not nonreflecting. Both Dirichlet and Neumann conditions can obey an energy estimate while reflecting everything.

Zero incoming data are not no incoming physics globally. Exterior backscatter can legitimately return radiation through a finite worldtube.

Moving the boundary and refining the grid test different errors. Only the former diagnoses finite-radius modeling error.

Exercise: classify three scalar conditions

Section titled “Exercise: classify three scalar conditions”

Using the half-line model, find RR for u=0u=0, ux=0u_x=0, and ut+ux=0u_t+u_x=0 at the boundary and classify their energy flux.

Solution

For time-preserved Dirichlet data, ut=0u_t=0, so win=−woutw_{\mathrm{in}}=-w_{\mathrm{out}} and R=−1R=-1: the flux vanishes and reflection is complete. Neumann data give ux=0u_x=0, hence win=woutw_{\mathrm{in}}=w_{\mathrm{out}} and R=1R=1, again complete reflection. The Sommerfeld condition is win=0w_{\mathrm{in}}=0, so R=0R=0 and dE/dt=−wout2/4dE/dt=-w_{\mathrm{out}}^2/4; it transmits the outgoing one-dimensional mode without reflection.

  • 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, no. 23 (2007): 5973–5984. DOI; Open PDF.
  • Winicour, Jeffrey. “Characteristic Evolution and Matching.” Living Reviews in Relativity 15, 2 (2012). DOI.