Gravitational Radiation and Waveforms

Gravitational radiation is where geometry becomes a propagating, measurable field. A useful account must connect several descriptions without confusing them: metric perturbations in a wave zone, curvature acting on freely falling bodies, shear and news at null infinity, multipolar fields generated by isolated sources, numerical modes in a chosen frame, and finite detector-facing waveforms. This volume builds those connections explicitly and then asks the practical question that every waveform model must answer: under what assumptions is it accurate enough for its intended scientific use?

Helpful background. Einstein Equations supplies the nonlinear field law that will be linearized and expanded, while Waves, Green Functions, and Geometric Optics supplies retarded propagation and the short-wavelength limit. The opening chapters revisit the pieces needed for radiation, but do not replace those broader treatments.

For a first encounter, follow one physical question: what changes when a wave passes? Begin with plane waves and freely falling separations, then see how a binary makes a chirp and how a burst can leave permanent displacement. Return to null infinity and multipolar generation once these concrete pictures are familiar. Readers comparing waveform models can instead begin with the inner product and keep its declared weighting in view throughout the volume.

One radiative field, several useful representatives

Far from a weak, isolated source, write

$$g_{\mu\nu}=\eta_{\mu\nu}+h_{\mu\nu}, \qquad \lvert h_{\mu\nu}\rvert\ll1.$$

The ten components of $h_{\mu\nu}$ are not ten radiative observables. Four coordinate functions are gauge freedom, four combinations are constrained rather than freely propagating, and two tensor polarizations remain. In a suitable radiation-zone transverse–traceless gauge, a wave moving toward increasing $z$ has

$$h^{\mathrm{TT}}_{ij} =h_+ e^+_{ij}+h_\times e^\times_{ij}.$$

Yet $h^{\mathrm{TT}}_{ij}$ is only one representative. A small freely falling separation $\xi^i$ responds to the electric curvature,

$$\frac{d^2\xi^i}{dt^2} =-R^i{}_{0j0}\,\xi^j, \qquad R_{0i0j}=-\frac12\ddot h^{\mathrm{TT}}_{ij},$$

so curvature, not a coordinate displacement, is the local observable. At future null infinity the radiative data are naturally expressed through Bondi shear and news. Numerical simulations often provide the leading Newman–Penrose curvature scalar $\Psi_4$ or harmonic modes of strain. Translating among these objects requires a declared tetrad, retarded time, sphere dyad, Fourier sign, integration constants, and asymptotic frame. The underlying relations are simple only after those choices have been made.

This volume uses

$$h=h_+-i h_\times, \qquad h(u,\theta,\phi) =\sum_{\ell=2}^{\infty}\sum_{m=-\ell}^{\ell} h_{\ell m}(u)\,{}_{-2}Y_{\ell m}(\theta,\phi),$$

with unit-normalized spin-weight $-2$ harmonics. Time-domain modes use the site convention $e^{-i\omega t}$. For an asymptotically outgoing wave, the volume fixes the tetrad and Bondi normalization before using the corresponding relation among $\Psi_4$, news, and the time derivatives of $h$; a formula copied from another convention is never changed one sign at a time. Flanagan and Hughes derive the linearized, tidal, retarded, and short-wavelength links in a single convention-explicit treatment Flanagan and Hughes 2005, §§2–5, pp. 4–29, PDF. Bondi–Sachs radiative data and mass loss are developed geometrically by Mädler and Winicour 2018, §§2–4, pp. 3–20, PDF.

Generation, transport, and observation are different questions

A waveform calculation has three logically separate stages:

1. Generation: specified stress–energy and gravitational fields produce radiative multipoles and outgoing boundary data.
2. Transport: those data propagate through curved or cosmological spacetime, possibly acquiring tails, lensing, diffraction, or frame changes.
3. Representation and use: strain, curvature, modes, spectra, or reduced models are supplied with enough metadata to be compared for a stated purpose.

The separation matters. A correct quadrupole formula does not determine the source trajectory. A high-accuracy numerical evolution does not by itself fix the Bondi frame of its extracted waveform. A small mismatch in one noise curve is not a universal model-error bound.

For slow internal motion and source size small compared with the wavelength, the leading far-zone field is controlled by the second derivative of the trace-free mass quadrupole $I_{ij}$,

$$\left.h^{\mathrm{TT}}_{ij}(u,\boldsymbol n)\right|_{\rm leading\ quadrupole} =\frac{2G}{r}\, \Lambda_{ij,kl}(\boldsymbol n)\,\ddot I_{kl}(u),$$

where $u=t-r$ and $\Lambda_{ij,kl}$ is the transverse–traceless projector. This is the leading term in the source slow-motion and multipole expansion, not the complete coefficient of $1/r$: higher radiative multipoles and post-Newtonian amplitude corrections also fall as $1/r$, while near-zone corrections introduce additional inverse powers of $r$. The leading luminosity is

$$\mathcal F_E =\frac{G}{5}\left\langle \dddot I_{ij}\dddot I_{ij} \right\rangle.$$

These compact expressions hide stringent hypotheses: weak fields in the wave zone, slow source motion for the quadrupole truncation, retarded boundary conditions, and an asymptotically inertial frame. The displayed brackets denote a cycle average, useful for periodic or adiabatic sources; the leading instantaneous Bondi luminosity has the same contraction without brackets. Averaging is not a prerequisite for the null-infinity flux law. The multipolar post-Minkowskian construction systematically relaxes the leading approximation and distinguishes source, canonical, and radiative moments, including hereditary tails and memory Blanchet 2024, §§1.1.2 and 2.1–2.4, pp. 10–16 and 34–103.

Transport also has regimes. If the wavelength is much shorter than the background curvature scale, an eikonal phase generates null rays and amplitude transport along them. If a lens produces several stationary paths, geometric optics yields an image sum with time delays and Morse phases. If the wavelength resolves the lens or curvature structure, a diffraction integral—not a ray picture—is the controlling object. The propagation chapters make each limiting operation visible instead of calling every frequency-dependent effect “lensing.”

The route through the volume

The chapters form four cumulative arcs.

Radiation as geometry

Gravitational Radiation in Linearized Theory isolates the two propagating degrees of freedom, constructs TT fields, derives tidal response and retarded far-zone behavior, and explains the limited sense in which a rapidly varying wave carries an effective stress–energy.

Null Infinity and the Bondi–Sachs Framework replaces an informal “large radius” by conformal null infinity, Bondi shear and news, asymptotic constraints, BMS transformations, and radiative Newman–Penrose curvature. The spacetime geometry of null infinity is developed in Global Structure; this volume uses it to define radiation data and waveforms.

Multipolar Wave Generation develops symmetric trace-free multipoles, post-Minkowskian iteration, near/far matching, and the dictionary from source moments to radiative moments. Radiative Energy, Momentum, and Angular Momentum then derives outward fluxes and balance laws at null infinity and across horizons.

Memory, Soft Structure, and Persistent Observables integrates radiative data across a burst. It keeps ordinary displacement memory, nonlinear null memory, spin memory, center-of-mass memory, and soft-theorem statements distinct while explaining the assumptions under which their relations hold.

Propagation and observables

Propagation, Lensing, and Wave Optics starts with geometric-optics transport on curved and cosmological backgrounds, then identifies curvature scattering, multiple imaging, interference, and diffraction. Polarizations and Observable Wave Content defines polarization through the measurable tidal tensor, relates screen and null-tetrad descriptions, and separates a basis rotation from an additional physical mode.

Modes, Harmonics, Frames, and Conventions supplies the common dictionary for spin-weighted harmonics, strain and curvature modes, Wigner rotations, source and radiation frames, and spherical–spheroidal mixing. It is the practical seam between analytic formulas and numerical waveform products.

Source classes and waveform structure

Compact-Binary Waveforms organizes inspiral amplitude and phase, higher modes, precession, eccentricity, merger, and source-to-detector parameters. It accepts the orbital dynamics from Relativistic Two-Body Dynamics and develops the radiative signal assembled from them.

Ringdown Waveforms and Spectroscopy turns black-hole response into a finite-time waveform model with explicit start time, mode excitation, overtones, nonlinear coupling, mixing, and completeness limits. Quasinormal spectra and Green functions remain in Black-Hole Perturbation Theory; detector-level spectroscopy claims remain in Gravitational-Wave Astronomy.

Bursts, Continuous Waves, and Stochastic Backgrounds develops physical signal morphologies and spectra outside the compact-binary template sequence. It does not duplicate source microphysics, search pipelines, or current observational verdicts.

Building and judging models

Approximants, Hybrids, and Surrogates explains how analytic, effective-one-body, numerical-relativity, hybrid, reduced-order, and surrogate information is combined. A model is always paired with its versions, training or calibration domain, alignment choices, and independent validation set.

Waveform Accuracy and Scientific Error Budgets closes the volume with the noise-weighted geometry of waveform space. It distinguishes norm, match, faithfulness, and effectualness; separates numerical, analytic, calibration, and statistical errors; and states when a waveform discrepancy can or cannot support a claim about scientific bias. Lindblom, Owen, and Brown give the foundational measurement and distinguishability standards in Lindblom, Owen, and Brown 2008, §II.A–C, pp. 2–7, PDF.

A convention sheet for the whole volume

The site uses metric signature $(-,+,+,+)$, retains $G$ explicitly by default, and sets $c=\hbar=k_B=1$. The following choices are especially consequential here:

• retarded time is $u=t-r$ in an asymptotically inertial Minkowski frame;
• complex strain is $h_+-i h_\times$ and therefore has spin weight $-2$ under a rotation of the screen dyad;
• spin-weighted spherical harmonics are unit normalized on the unit sphere;
• Fourier modes use $e^{-i\omega t}$, so decaying quasinormal frequencies lie in the lower half of the complex $\omega$ plane;
• outward flux at future null infinity is positive, while the source’s energy change is its negative after every additional channel is included;
• source, coprecessing, inertial, radiation, Bondi, and detector frames are named rather than silently identified;
• one-sided power spectral densities and positive frequencies are used when defining detector-weighted waveform norms.

Some relations cannot be global defaults. A page using $\Psi_4$ states the null-tetrad normalization and curvature contraction. A Bondi page states the sphere dyad and news normalization. A numerical waveform states its extraction radius or null-infinity method, retarded-time construction, extrapolation, mode normalization, and BMS-frame treatment. A detector-facing mass or frequency is labeled as source-frame or redshifted. Bishop and Rezzolla show why coordinate-radius extraction, curvature integration, extrapolation, and characteristic extraction carry different systematics Bishop and Rezzolla 2016, §§2–7.

Scientific boundaries

This volume treats gravitational radiation as a field and waveform. It does not duplicate every calculation that supplies or uses one.

Relativistic Two-Body Dynamics develops compact-body equations of motion, radiation reaction, self-force evolution, and orbital resummation. Computational Relativity treats production initial data, gauge evolution, discretization, mesh refinement, and numerical spacetime solution. Source microphysics and population or transfer histories are covered in Relativistic Matter and Compact Objects and Relativistic Cosmology. Those results enter here as specified inputs to a radiation calculation.

At the other end, calibration, detector-response implementation, searches, likelihoods, event catalogs, and population inference are treated in Gravitational-Wave Astronomy. This volume may use a declared detector noise spectrum to define waveform-space accuracy, but it does not turn that definition into an event claim. Keeping the interfaces explicit prevents two common category errors: an accurate orbit is not yet a complete waveform, and an accurate waveform model is not yet an observational conclusion.

Check the dictionary

1. Gauge and measurement. Let a plane perturbation be pure gauge,

$$h_{\mu\nu}=-2\partial_{(\mu}\xi_{\nu)},$$

using the volume’s passive-coordinate convention.

What should a freely falling detector measure at linear order?

Solution

The linearized Riemann tensor of a smooth pure-gauge perturbation vanishes because its terms reduce to commuting third derivatives of $\xi_\mu$. Geodesic deviation therefore has no wave-induced tidal acceleration at this order. Coordinate positions can change, but the measurable curvature does not. This is why a calculation that reads physical displacement directly from arbitrary components of $h_{\mu\nu}$ is incomplete.

2. Flux sign. Suppose the outward energy flux at future null infinity is $\mathcal F_\infty\gt0$ and the absorbed horizon flux is $\mathcal F_H\gt0$. Neglecting all other channels, what is the exterior source balance law?

Solution

With outward or absorbed fluxes defined as positive, the exterior mechanical/source budget—defined to exclude energy stored in the evolving black holes, including their mass and spin changes—decreases:

$$\frac{dE_{\mathrm{source}}}{dt} =-\mathcal F_\infty-\mathcal F_H.$$

Dropping the horizon term is justified only when it is absent or below a stated approximation order. Changing the orientation convention changes both the flux definition and the balance sign; it is not a physical disagreement.

3. Representation versus physics. A rotation of the transverse screen by angle $\psi$ changes $h=h_+-i h_\times$ by a phase. Has the wave acquired a new polarization?

Solution

No. With the dyad rotation $m\mapsto e^{i\psi}m$, where $m=(p+iq)/\sqrt2$, the complex components transform as $h\mapsto e^{-2i\psi}h$ because the field has spin weight $-2$. The curvature tensor and the tidal pattern are unchanged; only their components in the rotated screen basis differ. An additional physical polarization would add an independent tidal pattern that no screen rotation can remove.

References

• Bishop, Nigel T., and Luciano Rezzolla. “Extraction of Gravitational Waves in Numerical Relativity.” Living Reviews in Relativity 19 (2016): 2. DOI. Open PDF.
• Blanchet, Luc. “Post-Newtonian Theory for Gravitational Waves.” Living Reviews in Relativity 27 (2024): 4. DOI. Open PDF.
• Flanagan, Éanna É., and Scott A. Hughes. “The Basics of Gravitational Wave Theory.” New Journal of Physics 7 (2005): 204. DOI. Open PDF.
• Lindblom, Lee, Benjamin J. Owen, and Duncan A. Brown. “Model Waveform Accuracy Standards for Gravitational Wave Data Analysis.” Physical Review D 78 (2008): 124020. DOI. Open PDF.
• Mädler, Thomas, and Jeffrey Winicour. “Bondi–Sachs Formalism.” Scholarpedia 11 (2016): 33528; revised arXiv version (2018). DOI. Open PDF.