Gravitational Radiation and Compact Binaries

This course follows a gravitational-wave claim from Einstein’s equation to radiation at infinity, source dynamics, waveform construction, detector statistics, and parameter inference. Its organizing question is not “which method is best?” but “which approximation controls this regime, how is it matched to the others, and what uncertainty reaches the final claim?”

The framework can support independent study or local teaching and implies no institution, instructor, schedule, enrollment, or credential.

Required background. Einstein’s equation is used to linearize the geometric dynamics, invoke Bianchi consistency, and track matter sources; ordinary differential equations are used for driven systems, boundary data, and approximation checks. Review The Einstein Equation for the first capability and Differential Equations for Relativity for the second. Use the broader Einstein Field Equations sequence if the Einstein-equation diagnostic exposes a larger gap.

Helpful background. Hyperbolic PDEs, Green functions, probability, and inference. Use PDEs and Green Functions for Gravitation and Probability and Inference for Gravity for focused review.

Gravitational-wave and compact-binary capabilities

By the end, you should be able to:

1. derive a radiation or waveform result within a declared approximation regime;
2. compare post-Newtonian, post-Minkowskian, self-force, effective-one-body, numerical, hybrid, and surrogate methods without erasing their control parameters;
3. evaluate detection and parameter-inference results with noise, calibration, waveform uncertainty, priors, and selection effects exposed; and
4. complete a cross-method waveform study with analytic and numerical checks.

Workload and pacing

Treat each of the four modules and its numbered assignment as one assessed unit. Finish the radiation calculation before building the two-body regime map, finish that map before waveform matching, and keep waveform-systematics work separate from the inference assignment so that their uncertainties are not conflated. Revise each assignment after consulting its feedback, then complete the cross-regime transfer task and capstone as two further synthesis stages. Preparation reviews add only the work needed before the dependent module.

Check your preparation

Attempt these before Module 1. Use an equivalent accessible response mode or a declared reference aid when needed; the check concerns the reasoning you can reconstruct, not speed or presentation medium.

1. Linearize a nonlinear equation $F[g]=0$ about $g=\bar g+h$ and explain what it means to retain first order in $h$.
2. Solve a driven linear oscillator with a retarded Green function and state how the boundary condition selects the solution.
3. Explain why $\nabla^\mu T_{\mu\nu}=0$ follows from Einstein’s equation.
4. Distinguish likelihood, prior, posterior, and selection function in a population inference.

Response criteria

1. A secure response expands $F[\bar g+h]$ as $F[\bar g]+DF_{\bar g}[h]+O(h^2)$, identifies the background equation, and states exactly which quadratic and higher terms are neglected.
2. A secure response writes the solution as a source integral against a retarded Green function, connects its support or pole prescription to the causal boundary condition, and checks the homogeneous initial data.
3. A secure response uses the contracted Bianchi identity and metric compatibility to derive covariant stress-energy conservation.
4. A secure response identifies the likelihood as the data model conditional on parameters, the prior as pre-data parameter information, the posterior as their normalized update, and the selection function as the probability that a source enters the observed sample.

Mark each task secure, uncertain, or not yet. Any not-yet result routes to its preparation review before the dependent module; with no not-yet result, review only the uncertain task; all secure means begin Module 1. Tasks 1–3 use the Einstein, differential-equations, or PDE reviews above, while task 4 uses the inference module. A familiar formula without an approximation parameter or boundary condition is uncertain rather than secure. Direct entry remains available.

Module 1 — Radiation, infinity, and source multipoles

Study in order:

1. Gravitational Radiation in Linearized Theory
2. Radiation at Null Infinity and Bondi–Sachs Structure
3. Multipolar and Post-Minkowskian Wave Generation

Keep four objects distinct: a metric perturbation in a chosen gauge, the gauge-invariant tidal response of a detector, asymptotic radiative data, and a flux or balance law. At null infinity, record the falloff, conformal or Bondi structure, retarded-time convention, and residual gauge freedom before using “waveform” or “energy.”

Assignment 1. Starting from a transverse-traceless plane perturbation, derive the relative acceleration of neighboring freely falling test masses. Explain why the observable is curvature even though a convenient calculation uses $h_{ij}^{\mathrm{TT}}$. Then use dimensions and source timescales to recover the leading quadrupolar scaling of the far-zone strain.

Conceptual hint

The detector responds through geodesic deviation. Treat $h_{ij}^{\mathrm{TT}}$ as a convenient representation of the curvature that enters that equation.

Setup hint

Relate $R_{0i0j}$ to two time derivatives of the TT perturbation, and estimate the source quadrupole as $Q\sim ML^2$.

Next-step hint

Combine the retarded $1/r$ falloff with $\ddot Q$ and $v\sim L/T$. State the weak-field, slow-source, and far-zone assumptions before simplifying the scaling.

Model solution

In a local detector frame, geodesic deviation relates the separation $\xi^i$ to $R^i{}_{0j0}\xi^j$. For a weak transverse-traceless wave, $R_{0i0j}$ is proportional, with the sign fixed by the site’s curvature convention, to the second time derivative of $h_{ij}^{\mathrm{TT}}$. Thus the changing separation is invariant even though the perturbation components and gauge are not. A localized slow source has mass quadrupole $Q\sim ML^2$; two time derivatives and a retarded $1/r$ field give $h\sim (G/r)\ddot Q$, hence $h\sim GMv^2/r$ in units with $c=1$, up to the trace-free projection and numerical factors. The derivation is valid only in the weak, slow-source, far-zone regime.

Module 2 — The two-body hierarchy

Continue through:

4. Scales, Invariants, and Regimes of the Two-Body Problem
5. Post-Newtonian Dynamics
6. Post-Minkowskian Dynamics and Scattering
7. Gravitational Self-Force and Regularization

For each method, record what is small: velocity or compactness in a post-Newtonian expansion, powers of $G$ at unrestricted velocity in a post-Minkowskian expansion, or mass ratio in self-force theory. Numerical relativity supplies strong-field solutions over a finite parameter and resolution domain. Effective-one-body theory reorganizes analytic and calibrated information; it does not erase the provenance of its ingredients.

Assignment 2. Build a regime map with axes for characteristic velocity, field strength, and mass ratio. Place post-Newtonian, post-Minkowskian, self-force, effective-one-body, and numerical methods on the map, showing overlap regions. Choose one gauge-invariant quantity—binding energy versus frequency, scattering angle, or redshift invariant—and design a comparison in an overlap region.

Conceptual hint

The methods are organized by different small parameters, so no single axis orders all of them.

Setup hint

Label post-Newtonian control by velocity or compactness, post-Minkowskian control by powers of $G$, and self-force control by mass ratio. Mark numerical resolution separately.

Next-step hint

Translate masses, spins, and normalizations before comparing one invariant. Then vary analytic order or numerical resolution within the overlap region.

Solution criteria

There is no single correct drawing, but each placement must name its expansion or numerical control. The comparison must use the same masses, spin conventions, normalization, and invariant parameter; coordinate separation is not a valid universal horizontal axis. Agreement should be tested as a function of truncation order and numerical resolution. A discrepancy is not assigned to a method until convention translation, extrapolation, and overlap validity have been checked.

Module 3 — Waveforms, hybrids, and surrogates

Study:

8. Compact-Binary Waveform Structure
9. Approximants, Hybridization, and Surrogate Waveforms

Separate source-frame modes, polarization, detector response, and sampled data. A hybrid requires alignment freedoms, a matching window, compatible conventions, overlap of validity, and uncertainty under reasonable changes of each choice. A surrogate adds a training domain and interpolation or regression error; it cannot be trusted silently outside that domain.

Assignment 3. Hybridize two supplied waveforms over three admissible matching windows. Optimize only the declared time and phase freedoms, report the mismatch in a stated inner product, and propagate truncation and numerical errors separately. Predeclare a tolerance above which the hybrid is rejected.

Conceptual hint

Alignment, blending, and error estimation are three distinct operations; a good match after tuning does not establish a controlled hybrid.

Setup hint

Declare the time and phase freedoms, matching window, smooth blend, sampling, and waveform inner product before optimizing anything.

Next-step hint

Vary the window, analytic order, and numerical resolution separately. Compare the resulting mismatch envelope with the tolerance chosen before inspection.

Model solution architecture

First translate mass, spin, mode, phase, polarization, and time conventions. Align by minimizing a declared phase or waveform discrepancy over each window, blend with a smooth function whose endpoint derivatives do not create artificial structure, and compare the resulting hybrids after identical sampling and tapering. Window dependence estimates matching systematics; analytic-order variation estimates truncation error; numerical-resolution and extraction variation estimate simulation error. The result passes only if the combined envelope remains below the predeclared tolerance in the intended detector-weighted or unweighted norm.

Module 4 — Detection and parameter inference

Finish with:

10. Detection Statistics and Signal Searches
11. Compact-Binary Parameter Estimation

A detection statistic is calibrated under a noise and trials model. A posterior is conditional on waveform, calibration, noise, prior, and cosmological assumptions. A population statement additionally requires the selection function. Keep these layers separate when translating a waveform calculation into an astrophysical or gravity claim.

Assignment 4. Consider a parameter whose waveform effect is partially degenerate with spin. Describe how a detection analysis and a parameter analysis answer different questions. Then predict how a restrictive spin prior, a waveform systematic, and a selection effect could each move the reported result.

Conceptual hint

Locate detection, event-level parameter inference, and population selection at different stages of the analysis rather than merging them into one error.

Setup hint

Place the spin prior, waveform model, and selection function respectively in the prior, likelihood model, and detected-sample construction.

Next-step hint

Trace each ingredient through the spin degeneracy. Predict whether it shifts posterior support, produces coherent model bias, or changes the observed population relative to the underlying one.

Solution

Detection compares a signal-plus-noise hypothesis or ranking statistic with a noise background; it need not identify the physical parameters sharply. Parameter inference conditions on a signal and maps likelihood and prior into a posterior. A restrictive spin prior can transfer support to the correlated parameter. A missing waveform effect can produce coherent residual bias that shrinks neither with more posterior samples nor with a narrower numerical error bar. Selection favors systems that are easier to detect and therefore changes the observed population relative to the underlying one. These effects enter different parts of the analysis and must not be combined into a generic “uncertainty.”

Cross-regime transfer task

Begin with an equal-mass, nearly circular, nonprecessing binary and transfer the reasoning to one changed system: an extreme mass ratio, appreciable eccentricity, strong precession, or an unbound encounter. State:

1. which expansion or numerical method gains or loses control;
2. which invariants remain suitable for cross-method comparison;
3. which waveform modes or timescales become important;
4. how hybridization or surrogate coverage changes; and
5. which inference degeneracies or selection effects appear.

The task is complete when the changed system leads to a revised method chain, not merely a list of additional parameters.

Capstone — a carefully tested waveform calculation

Choose one binary configuration and one observable claim. Build a compact, rerunnable chain from dynamics to inference containing:

• parameters, units, conventions, approximation regime, and target accuracy;
• one analytic result and its truncation check;
• one numerical or supplied-waveform comparison and its resolution or extraction check;
• waveform construction, alignment, and uncertainty budget;
• detector/noise model and statistic or likelihood;
• a changed-setting stress test; and
• a precise conclusion with a failure criterion.

Assess correctness, regime control, convention translation, numerical verification, statistical interpretation, and reproducibility separately. A capstone may conclude that the target claim is not identifiable at the stated accuracy. That is a successful scientific result when the negative conclusion is supported by a sensitive, controlled test.

Continue

Deepen the source dynamics with Compact Binaries and Gravitational Waves, specialize in Black-Hole Perturbation Theory, or develop production numerical evidence in Computational Relativity. For contested or changing interpretations, continue to the Research Seminar.