Near-Zone and Far-Zone Matching

Near-zone and exterior descriptions become one physical solution only when their asymptotic expansions agree in a common region. For a compact, slowly moving source of size $a$ and characteristic reduced gravitational wavelength $\lambda\sim\omega^{-1}$, a controlled overlap $a\ll r\ll\lambda$ lies outside the matter but inside one wavelength. Matching there determines the source moments of the exterior field and imports the retarded boundary condition into the near-zone solution.

Required background. Weak-Field Gravity supplies the perturbative metric and gauge expansion, while Ordinary Differential Equations supplies the asymptotic-series and boundary-matching logic.

Helpful background. Einstein Equations supplies the nonlinear source equation; Waves, Green Functions, and Geometric Optics explains retarded versus advanced solutions; and Relativistic Two-Body Dynamics supplies possible near-zone source data. The matching procedure consumes such data but does not derive detailed binary equations of motion or detector inference.

Two expansions and one overlap

Let $\mathfrak h$ denote the gothic metric deviation in harmonic coordinates. The inner description is a post-Newtonian expansion $\overline{\mathfrak h}$, obtained by expanding retardation and the field equations in the slow-motion parameter $\epsilon\sim a/\lambda\sim v\ll1$. It covers the material source and is accurate throughout the near zone $r\ll\lambda$, but each coefficient generally develops a large-$r$ expansion.

The outer description $\mathcal M(\mathfrak h)$ is a multipolar post-Minkowskian vacuum field. It is physically valid outside the source, $r\gt a$, and is singular if formally extended to $r=0$. Its multipole expansion is not the same thing as its radiation-zone expansion: $\mathcal M(\mathfrak h)$ applies throughout the exterior, whereas a leading $1/r$ far-zone truncation additionally requires $r\gg\lambda$.

When $a\ll\lambda$, both full asymptotic descriptions are valid in

$$a\ll r\ll\lambda.$$

The hierarchy is essential. If $a$ is not small compared with $\lambda$, there is no parametrically controlled overlap and this post-Newtonian matching argument does not apply. Blanchet 2024, §2.2.1, pp. 54–56 states the matching hypothesis and carefully distinguishes the exterior multipole expansion from a radiation-zone expansion.

The matching equation

Place an overbar on a field to denote its formal near-zone re-expansion, and let $\mathcal M$ denote a formal multipole expansion. Equality in the overlap is converted into the term-by-term identity

$$\overline{\mathcal M(\mathfrak h)} =\mathcal M(\overline{\mathfrak h}).$$

The left side expands the exterior retarded multipolar field for $r/\lambda\to0$. The right side expands every post-Newtonian coefficient for $a/r\to0$. Both take the common STF form

$$\sum_{L,m,p}\widehat n_L\,r^m \left[\ln\left(\frac{r}{r_\star}\right)\right]^p F_{Lmp}(t),$$

where $m$ can be negative and $r_\star$ is an auxiliary length making every logarithm dimensionless. Matching means equating the coefficient of each independent $\widehat n_L r^m[\ln(r/r_\star)]^p$ at each approximation order. It is not merely equality at one arbitrarily chosen matching radius.

This identity performs two jobs. First, it expresses the exterior source moments in terms of near-zone matter and gravitational fields. Second, when supplemented by no incoming radiation, it fixes the time-antisymmetric near-zone pieces responsible for radiation reaction. Only the first interface is needed for waveform generation here; the detailed dynamics remains with the two-body or matter-source treatment.

The equality is formal. It assumes the existence of an exact solution admitting both asymptotic expansions and does not prove convergence of either infinite series. At finite order, one should carry separate remainders for the post-Newtonian, post-Minkowskian, and multipole truncations.

A monochromatic matching calculation

A scalar outgoing wave displays the coefficient matching without tensor clutter. With the global time convention $e^{-i\omega t}$ and $\omega\gt0$,

$$\Phi_{\mathrm{out}}(t,r) =\frac{A}{r}e^{-i\omega(t-r)} =A e^{-i\omega t} \left( \frac1r+i\omega-\frac{\omega^2r}{2} -\frac{i\omega^3r^2}{6}+\cdots \right).$$

Suppose an inner solution has the exterior expansion

$$\Phi_{\mathrm{near}}(t,r) =e^{-i\omega t} \left( \frac{C_{-1}}r+C_0+C_1r+C_2r^2+\cdots \right).$$

Matching fixes $C_{-1}=A$, $C_0=i\omega A$, $C_1=-\omega^2A/2$, and so on. The constant term $i\omega A$ is recovered by re-expanding the outgoing phase; dropping that phase and matching only a Coulombic $A/r$ term would lose the propagation information. For this monopolar scalar example the outgoing $e^{i\omega r}/r$ expression is already the complete exterior solution. Higher multipoles and gravity additionally have subleading radial terms that must be retained before the near-zone re-expansion.

An incoming solution $B e^{-i\omega(t+r)}/r$ would instead contribute $C_0=-i\omega B$. Local coefficient matching alone can accommodate a linear combination of incoming and outgoing fields. The no-incoming-radiation condition at past null infinity selects the outgoing sign. Retardedness is therefore boundary data, not a consequence of the STF basis.

Scales, stationarity, and consistency checks

Analytic finite-part regularization introduces lengths such as $r_0$, while logarithmic coordinate transformations can introduce another scale in radiative time. A matching calculation may contain these scales in its inner and outer pieces. Their dependence must cancel in a radiative observable once both pieces are included at the same order. Changing a scale on only one side is an inconsistent comparison.

The monochromatic example supplies several independent checks:

• expanding the outgoing exponential reproduces the sign $+i\omega$ dictated by $e^{-i\omega t}$;
• every coefficient has the dimension required to multiply its power of $r$;
• the stationary limit $\omega\to0$ leaves the Coulombic $A/r$ term and removes the retardation corrections;
• replacing the retarded solution by an advanced one flips every odd power of $\omega$.

For tensor gravity, the same logic is applied separately to each STF rank and each logarithmic power. Thorne’s STF expansion provides the irreducible angular basis needed for this separation Thorne 1980, §§II–III and Table I.

Common pitfalls

Matching a near-zone field directly to a leading far-zone waveform. There is no radius satisfying both $r\ll\lambda$ and $r\gg\lambda$. One instead re-expands the full exterior multipolar solution into the near zone.

Choosing a single matching radius. Matched asymptotics equates coefficient families in an overlap, not numerical values at one surface. A residual dependence on a chosen surface is a truncation diagnostic, not part of the exact construction.

Hiding logarithmic units. Expressions such as $\ln r$ are incomplete. The argument must be $r/r_\star$, and the arbitrary scale must disappear from the final observable or be identified with a genuine physical reference scale.

Exercise

Add an incoming component to the scalar example,

$$\Phi=\frac{1}{r}\left[ A e^{-i\omega(t-r)}+B e^{-i\omega(t+r)} \right].$$

Find the $1/r$, constant, and linear-in-$r$ near-zone coefficients. Which coefficient distinguishes outgoing from incoming propagation?

Solution

Expanding both exponentials gives $C_{-1}=A+B$, $C_0=i\omega(A-B)$, and $C_1=-\omega^2(A+B)/2$. Even powers of $\omega r$ depend on the sum $A+B$, whereas odd powers depend on the difference $A-B$. Thus the constant coefficient is the first term that distinguishes propagation direction. A purely outgoing retarded field has $B=0$ and $C_0=i\omega C_{-1}$; a purely incoming field has $A=0$ and $C_0=-i\omega C_{-1}$. This sign check fails immediately if the adopted Fourier convention is not carried through the matching.

References

• Blanchet, Luc. “Post-Newtonian Theory for Gravitational Waves.” Living Reviews in Relativity 27 (2024) 4. DOI. Open PDF.
• Thorne, Kip S. “Multipole Expansions of Gravitational Radiation.” Reviews of Modern Physics 52 (1980) 299–339. DOI.