Radiation Fields and Scattering

A radiation field is the finite asymptotic trace left after the leading radial decay of a wave has been removed. A scattering theorem goes further: it proves that finite-energy Cauchy data determine radiation data, that admissible radiation data reconstruct a unique solution, and that the two maps are continuous in named norms. These conclusions are established in important model and perturbative regimes; they are not a single theorem for arbitrary four-dimensional nonlinear vacuum spacetimes.

Required background. Analysis for Mathematical Relativity supplies trace and weighted Sobolev estimates, and exterior algebra and differential forms supplies conserved currents and flux forms.

Helpful background. The Einstein equation supplies the nonlinear system; Causality and Global Structure supplies horizon and null-infinity boundaries; PDEs and Green Functions for Gravitation supplies wave scattering; and Gravitational Radiation and Waveforms supplies the physical radiation data.

The model radiation trace

For a scalar wave $\Box_\eta\phi=0$ on $3+1$ dimensional Minkowski space, set $u=t-r$ and define

$$\Phi_+(u,\omega)=\lim_{r\to\infty}r\phi(u+r,r\omega),$$

whenever the limit exists. The outgoing energy flux is, up to the standard normalization,

$$E_{\mathscr I^+}[\phi] =\int_{\mathbb R\times S^2} \lvert\partial_u\Phi_+\rvert^2\,du\,d\omega.$$

For smooth compactly supported Cauchy data, an energy identity obtained from the stress tensor and the time-translation multiplier gives

$$E_{\Sigma_0}[\phi]=E_{\mathscr I^+}[\phi].$$

The proof integrates the divergence-free current over a truncated domain, passes $r\to\infty$, and shows that the caps at timelike infinity carry no residual energy. Density then extends the trace map from smooth data to the energy completion. Surjectivity is proved by solving backward from suitable null data. Thus the radiation transform is an isometry after constants and zero modes are treated consistently. This flat-space radiation-field construction, its energy inequalities, and its translation representation are developed in Friedlander 1980, pp. 483–515. In the present elementary case the Fourier–Radon representation supplies an independent proof of surjectivity; the divergence theorem alone supplies only the forward isometry.

This four-step pattern—forward estimate, asymptotic trace, no-lost-energy limit, inverse construction—is the architecture to look for in a geometric scattering theorem. Existence of the pointwise limit alone proves neither norm continuity nor completeness.

A regular radial pulse

Let $F$ be a smooth compactly supported function of one real variable. The radial solution

$$\phi(t,r)=\frac{F(t-r)-F(t+r)}{r}$$

is regular at the center: its limit is $-2F'(t)$, and its radial Taylor series contains only even powers of $r$. The second term is essential for that regularity. At fixed $u=t-r$, compact support gives $\Phi_+(u)=F(u)$. Thus a reader can extract a radiation field by taking one explicit limit, rather than interpreting $r\phi$ as the field at a finite detector radius.

With $E=\tfrac12\int_{\mathbb R^3}[(\partial_t\phi)^2+|\nabla\phi|^2]d^3x$, write $\psi=r\phi$ and integrate the radial cross term by parts. Smoothness at zero and compact support remove the boundary term, giving

$$E=2\pi\int_0^\infty[(\partial_t\psi)^2+(\partial_r\psi)^2]dr =4\pi\int_{\mathbb R}|F'(u)|^2du.$$

The last expression equals the sphere-integrated radiation flux above. The calculation verifies the normalization and absence of a center flux in this example; surjectivity for general angular data is still a separate step.

Black-hole exteriors

On a stationary black-hole exterior, future energy can leave through both $\mathscr I^+$ and the event horizon $\mathcal H^+$. The natural forward map is

$$\mathscr F^+: \text{Cauchy data on }\Sigma \longmapsto (\Phi_{\mathcal H^+},\Phi_{\mathscr I^+}).$$

Write $M_{\mathrm{geom}}=GM_{\mathrm{phys}}$ and $a=J/M_{\mathrm{phys}}$ (with $c=1$), so both Kerr parameters have dimensions of length. For the scalar wave equation on every subextremal Kerr exterior $\lvert a\rvert<M_{\mathrm{geom}}$, Dafermos, Rodnianski, and Shlapentokh-Rothman prove existence and uniqueness of finite-energy scattering states and a bounded scattering operator from past horizon-plus-null-infinity data to future data Dafermos, Rodnianski, and Shlapentokh-Rothman 2018, Theorems 1 and 2, pp. 379–391. Their energy spaces are deliberately degenerate at the horizon. The inverse map is not bounded on the corresponding nondegenerate finite-energy class, because the forward red-shift becomes a blue-shift when evolution is reversed.

Superradiance adds a second subtlety: the flux through $\mathscr I^+$ can exceed the incoming flux from $\mathscr I^-$ for selected modes, with the difference supplied by negative Killing-energy flux through the horizon. The scattering matrix is bounded, not necessarily contractive in every component. This example shows why “captures all asymptotic energy” must name the combined boundary and the coercive norm.

Nonlinear Einstein radiation fields

For the Einstein equation, gauge equivalence and nonlinear constraints enter the definition of both domain and range. In harmonic-type gauges, one can rescale the metric perturbation, compactify the spacetime, and extract its leading trace at $\mathscr I^+$. The scattering map must then be quotiented by admissible diffeomorphisms or tied to a fixed gauge and asymptotic frame.

Fang Wang proves an especially clear nonlinear result for vacuum solutions near Minkowski spacetime in spatial dimension $n\ge4$: in suitable weighted $b$-Sobolev spaces, the map from small Cauchy data to the radiation field is an isomorphism Wang 2014, Theorems 1.1–1.3 and §7. Its hypotheses include small constraint-satisfying data in the paper’s weighted spaces, a fixed wave-coordinate formulation, and the stated higher spatial dimension. The dimension restriction is part of the theorem. It does not establish the analogous isomorphism for physical $3+1$ dimensional vacuum gravity, where long-range terms and logarithmic asymptotics are more delicate.

Likewise, a global nonlinear stability result can prove geodesic completeness and decay without constructing a surjective scattering map. Conversely, a linear scattering theorem on a fixed background does not control backreaction. These limitations separate three levels:

1. a radiation trace exists for every solution in a data class;
2. its norm accounts for all flux through every asymptotic end;
3. every compatible radiation state is realized uniquely by a nonlinear solution, modulo gauge.

Only the third is nonlinear asymptotic completeness. The central limitation is that each established theorem above occupies only one cell of this hierarchy and one specified equation, dimension, and norm.

Check the boundary accounting

Suppose a scalar solution on a Schwarzschild exterior, using its stationary Killing-energy balance, has initial energy $E_0$, future horizon flux $E_{\mathcal H^+}$, null-infinity flux $E_{\mathscr I^+}$, and a nonnegative remainder $E_{\mathrm{tail}}$ on late spacelike slices. What additional estimate turns the balance identity into asymptotic completeness of the forward energy?

Solution

The identity $E_0=E_{\mathcal H^+}+E_{\mathscr I^+}+E_{\mathrm{tail}}$ only proves boundedness of the trace. One needs integrated decay strong enough to show $E_{\mathrm{tail}}\to0$ as the slices approach future infinity. Then the combined boundary norm equals the initial norm. An inverse construction and uniqueness are still required for surjectivity; energy exhaustion alone does not provide them.

References

• Dafermos, Mihalis, Igor Rodnianski, and Yakov Shlapentokh-Rothman. “A Scattering Theory for the Wave Equation on Kerr Black Hole Exteriors.” Annales Scientifiques de l’École Normale Supérieure 51 (2018): 371–486. DOI.
• Friedlander, F. G. “Radiation Fields and Hyperbolic Scattering Theory.” Mathematical Proceedings of the Cambridge Philosophical Society 88 (1980): 483–515. DOI.
• Wang, Fang. “Radiation Field for Einstein Vacuum Equations with Spacial Dimension $n\ge4$.” arXiv:1304.0407v2 (2014). arXiv.