Post-Newtonian Dynamics

Post-Newtonian dynamics is the controlled expansion of relativistic two-body motion about Newtonian gravity when orbital speeds are small compared with $c$ and gravitational potentials are weak. Its importance is broader than a list of corrections to Kepler’s law: the method connects Einstein’s field equations to explicit equations of motion, conserved quantities, precessing orbits, and the analytic ingredients of inspiral waveforms. This chapter develops that chain while keeping gauge choices, singular-source regularization, and finite-order uncertainty visible.

Helpful background. The Einstein Equation supplies the nonlinear field equations being expanded; Weak Metric Perturbations and Gauge Freedom supplies perturbative gauge control; Presymplectic Potentials and Currents supplies the variational and canonical language used to compare formulations; Gravitational Radiation and Waveforms supplies the radiative observables and balance laws that extend conservative motion to inspiral.

The slow-motion two-body problem

Let the component masses be $m_1$ and $m_2$, and define

$$M=m_1+m_2, \qquad \mu=\frac{m_1m_2}{M}, \qquad \nu=\frac{\mu}{M}.$$

The symmetric mass ratio satisfies $0\lt \nu\leq1/4$. The limit $\nu\to0$ connects the comparable-mass problem to test-body motion and black-hole perturbation theory, while $\nu=1/4$ describes equal masses. For a separation scale $r$ and relative speed $v$, a virialized bound orbit obeys

$$\epsilon^2\equiv\frac{v^2}{c^2} \sim\frac{GM}{rc^2}\ll1.$$

A term proportional to $\epsilon^{2n}$ beyond Newtonian gravity is called $n$PN. Odd half orders are not forbidden: time-antisymmetric radiation reaction first enters the orbital equations at 2.5PN, and hereditary propagation effects generate further half-integer powers and logarithms. This chapter concentrates on the conservative mechanics and its reliability; the radiative multipoles and detector-facing waveform construction belong to the gravitational-radiation volume.

For quasi-circular motion, the frequency variable

$$x=\left(\frac{GM\Omega}{c^3}\right)^{2/3}$$

is especially valuable because $\Omega$ can be defined invariantly for a helical circular orbit. In a chosen coordinate system, $x=GM/(rc^2)+O(c^{-4})$, but the higher-order relation between frequency and coordinate separation is gauge dependent. This distinction recurs throughout the chapter: coordinates and canonical variables organize calculations, whereas invariant relations such as $E(\Omega)$, periastron advance, scattering angle, or redshift observables support physical comparison.

Modern PN theory reaches high orders only by combining methods. Near-zone expansion converts retardation into a hierarchy of instantaneous potentials; matching supplies the homogeneous solution and hereditary terms; point-particle models require ultraviolet regularization; reduced actions and canonical transformations expose equivalent descriptions. A comprehensive derivation of the near-zone and compact-binary construction is given in Blanchet 2024, §§ 2.2–2.3 and 3.1–3.3, while the Hamiltonian route is developed in Schäfer and Jaranowski 2024, §§ 4–7.

The chapter sequence

The six pages form one derivational and interpretive sequence.

1. Post-Newtonian Power Counting and Near-Zone Expansion identifies the small parameter, distinguishes near and radiation zones, and shows how retarded fields become a controlled expansion. It also separates PN counting from the post-Minkowskian expansion in powers of $G$.
2. Conservative PN Equations of Motion derives the relative acceleration through the first relativistic correction, constructs the conserved energy, and tests the result in circular, test-mass, and dimensional limits.
3. PN Hamiltonians, Lagrangians, and Contact Transformations explains why different coordinate accelerations or reduced actions can represent the same physics. Contact and canonical transformations connect formulations; invariant observables decide agreement.
4. Regularization of Point-Particle Singularities treats the ultraviolet singularities created by idealized compact objects. Hadamard finite parts, dimensional regularization, effective field theory, worldline shifts, and scale cancellation are compared rather than conflated.
5. Spinning and Eccentric PN Orbits adds spin supplementary conditions, spin-orbit and spin-spin couplings, quasi-Keplerian eccentric motion, and the hierarchy of orbital and precession timescales.
6. PN Convergence, Resummation, and Error Estimates asks what finite-order accuracy actually means. It separates mathematical convergence from asymptotic and phenomenological usefulness, compares Taylor and resummed representations, and gives a dated procedure for uncertainty estimates.

The order matters. Error estimates are meaningful only after the expansion parameter and observable are specified; equivalence claims require the transformation theory of the third page; spin and eccentric formulas inherit the regularization and canonical choices made earlier.

A common mathematical spine

The field-theory starting point is a weak-field expansion of the gothic inverse metric or an equivalent perturbation variable. In harmonic coordinates, the relaxed Einstein equations have the schematic form

$$\Box h^{\alpha\beta} =\frac{16\pi G}{c^4}\tau^{\alpha\beta}, \qquad \partial_\beta h^{\alpha\beta}=0,$$

where $\tau^{\alpha\beta}$ contains matter and nonlinear gravitational terms. In the near zone, the inverse wave operator is expanded in time derivatives. This operation is not merely “setting propagation speed to infinity”: boundary conditions and matching retain information about radiation reaction and tails that a naive Poisson solution would lose.

After solving for the field and eliminating it from the action, one obtains reduced particle dynamics. At first PN order, an ordinary Lagrangian $L(\mathbf r,\mathbf v)$ suffices; at higher orders, a convenient harmonic-coordinate Lagrangian can depend on accelerations. Order reduction and contact transformations then lead to an ordinary canonical description. In center-of-mass variables, the conservative Hamiltonian takes the pattern

$$H =H_{\mathrm N} +\frac{1}{c^2}H_{\mathrm{1PN}} +\frac{1}{c^4}H_{\mathrm{2PN}} +\cdots.$$

Hamilton’s equations,

$$\dot{\mathbf r} =\frac{\partial H}{\partial\mathbf p}, \qquad \dot{\mathbf p} =-\frac{\partial H}{\partial\mathbf r},$$

must reproduce the reduced equations of motion after the appropriate variable transformation. The comparison is therefore three-layered:

• algebraic checks test exchange symmetry, time-reversal parity, rotational covariance, and dimensions term by term;
• dynamical checks verify the Euler–Lagrange or Hamilton equations, conserved energy and angular momentum, and the Poincaré algebra to the stated PN order;
• invariant checks compare $E(\Omega)$, periastron advance, scattering information, or test-mass limits after gauge-dependent variables have been eliminated.

The last layer is decisive. Two Hamiltonians may look different yet be related by a canonical transformation generated by $g(\mathbf r,\mathbf p)$. To first order in a bookkeeping parameter $\lambda$,

$$K=H-\lambda\{H,g\}+O(\lambda^2), \qquad Z=z+\lambda\{z,g\}.$$

Here $K$ is the Hamiltonian written in the new canonical coordinates $Z$ and $g$ is time independent. The inverse coordinate map accounts for the minus sign; the transformation page checks it with an explicit example. Coordinate trajectories change, but consistently transformed invariant relations do not. Conversely, agreement of one gauge-dependent coefficient is weak evidence because a coordinate transformation can alter it.

Point-particle regularization fits into the same spine. The local field of each idealized body diverges, so intermediate integrals acquire poles or arbitrary scales. Dimensional regularization continues spatial dimension to $d=3+\varepsilon$ and organizes singular pieces as

$$I(d)=\frac{A_{-1}}{\varepsilon}+A_0+O(\varepsilon).$$

The pole and scale dependence must cancel after all contributions, counterterms, and worldline redefinitions are included. A remaining dependence on an arbitrary regularization scale signals an incomplete calculation or an inconsistent comparison; it is not a new observable.

Three routes through the material

Readers interested in deriving formulas from the field equations should follow the chapter in order through the regularization page. The essential chain is power counting, near-zone iteration, reduced equations, variational formulation, and removal of singular-source artifacts. Re-deriving the 1PN acceleration and energy before approaching higher orders is worthwhile because nearly every later consistency check is already visible there.

Readers focused on orbital modeling can begin with the equations of motion, proceed to the Hamiltonian page, and then study spinning and eccentric solutions. The quasi-Keplerian representation illustrates a recurring strategy: replace a coordinate trajectory by constants and frequencies with controlled transformation properties. Spin adds a second lesson—the phase space itself depends on the spin supplementary condition and variable choice, so a precession equation is meaningful only with those choices stated.

Readers focused on waveform accuracy should still read the power-counting and transformation pages before the convergence page. “3PN versus 4PN” does not determine error without naming the quantity, approximant, and parameter domain. The most informative comparisons combine an invariant PN series with self-force or numerical-relativity data in an overlap regime and then propagate conservative and flux errors into phase.

These routes meet at the same standard: every displayed approximation needs assumptions, a counting rule, checks that survive changes of variables, and a breakdown statement.

Regime, scope, and breakdown

The orbital PN expansion is controlled when orbital velocities and interbody potentials are small, the bodies remain well separated compared with their compact-object scales, and the multipolar and near-zone expansions overlap. The bodies may have strong internal gravity: it is their mutual orbital field that is expanded. For compact binaries this describes a substantial inspiral interval, but it is not uniform through plunge and merger. Approaching the strong-field regime, $x$ grows, successive terms need not decrease, and orbit-averaged descriptions can fail near separatrices or during rapid precession.

The point-particle approximation encodes structure only through selected worldline operators. For nonspinning black holes, effacement postpones many finite-size effects, but neutron-star tidal deformability and spin-induced multipoles require additional coefficients. Those coefficients are not determined by point-particle regularization. They must be obtained by matching to the internal solution or supplied as physical response parameters.

This chapter also does not identify a coordinate-dependent PN trajectory with an observable waveform. Radiation-zone matching, radiative multipoles, propagation to null infinity, detector response, and data-analysis likelihoods add distinct layers. Energy balance,

$$\frac{dE}{dt}=-\mathcal F,$$

connects the conservative energy $E$ to the gravitational-wave luminosity $\mathcal F$ only after both are computed consistently for the same physical system and truncation scheme.

Finally, no universal statement that “PN converges up to a given radius” is justified. Accuracy depends on mass ratio, spins, eccentricity, observable, PN order, and resummation. The chapter’s final page therefore treats uncertainty as a function over parameter space and makes clear which conclusions depend on current evidence.

Review the chapter

Consider a nonspinning compact binary with $\nu=1/4$ on an approximately circular orbit at $x=0.04$.

1. Estimate $v/c$, $GM/(rc^2)$ at leading order, and the relative scale of a 1PN correction.
2. Explain why replacing the retarded inverse wave operator by a single Poisson solve is insufficient beyond the instantaneous conservative approximation.
3. Suppose two 1PN Hamiltonians differ by $\delta H=-\{H_{\mathrm N},g\}/c^2$. What must be transformed before comparing their coordinate trajectories, and what kind of quantity provides a clean equivalence test?
4. A dimensional-regularization calculation leaves $\ln(r/\ell_0)$ in the final circular binding energy, where $\ell_0$ was introduced only by the regularization prescription. What does this imply?
5. Successive predictions for an invariant are $Q_2=0.980$, $Q_3=0.987$, and $Q_4=0.988$. Is $\lvert Q_4-Q_3\rvert=0.001$ by itself a reliable error bar?

Solution

From $x\simeq v^2/c^2\simeq GM/(rc^2)$ at leading order,

$$\frac{v}{c}\simeq\sqrt{x}=0.2, \qquad \frac{GM}{rc^2}\simeq0.04.$$

A nominal 1PN correction is therefore about four percent of the Newtonian term, multiplied by an observable-dependent coefficient. The coefficient can be larger or smaller than unity, so this is power counting rather than an accuracy guarantee.

The near-zone expansion of the retarded propagator generates a hierarchy of time-derivative corrections. Moreover, a particular Poisson solution does not by itself impose the global no-incoming-radiation condition. Matching to the exterior solution supplies homogeneous pieces, radiation reaction, and hereditary information. A lone instantaneous Poisson solve omits that structure.

If $\delta H=-\{H_{\mathrm N},g\}/c^2$, use the corresponding new coordinates $Z=z+c^{-2}\{z,g\}$. Comparing $\mathbf r(t)$ while holding the old coordinates fixed mixes two descriptions. A relation such as the circular binding energy $E(\Omega)$ or the scattering angle at fixed invariant energy and angular momentum gives a clean equivalence test.

The surviving $\ell_0$ dependence is unacceptable because $\ell_0$ has no physical definition. Some term, counterterm, worldline shift, tail contribution, or conversion between variables is missing or inconsistent. Only a physical matching scale associated with a specified body could remain in an observable.

The final difference $0.001$ is useful evidence but not a reliable error bar by itself. It may be accidentally small, and the sequence contains too little information about later coefficients. One should compare additional orders or representations, examine parameter dependence, and use an independent benchmark if available. The appropriate uncertainty is tied to this invariant and this point in parameter space, not automatically to the entire inspiral.

References

• Blanchet, Luc. “Post-Newtonian Theory for Gravitational Waves.” Living Reviews in Relativity 27, 4 (2024). doi:10.1007/s41114-024-00050-z.
• Schäfer, Gerhard, and Piotr Jaranowski. “Hamiltonian Formulation of General Relativity and Post-Newtonian Dynamics of Compact Binaries.” Living Reviews in Relativity 27, 2 (2024). doi:10.1007/s41114-024-00048-7.