How do gravitational multipoles prove the logarithmic soft graviton theorem?

What the logarithmic soft theorem actually states

A soft graviton theorem relates the gravitational field produced by a scattering event to the low-frequency limit of the emitted graviton. The logarithmic version concerns a specific correction: in four dimensions, infrared divergences and one-loop effects generate a soft factor that carries a logarithm of the graviton frequency, and its classical counterpart was originally established by Laddha and Sen in 2018. The paper's target is that classical statement, not the full quantum one.

The key move is to translate the frequency-space logarithm into position space. Because the Fourier transform of a logarithm in frequency involves inverse times, the logarithmic soft theorem is tied to the so-called 1/u tails of gravitational waveforms in the early or late limits u going to plus or minus infinity. Separately, interactions at order G squared, the field-theory one-loop level, break the peeling property of the asymptotic Weyl tensor and produce logarithmic divergences. The theorem is the statement that a particular linear combination of these two asymptotic quantities obeys a matching relation across spatial infinity.

Antipodal matching turns the theorem into a geometric identity

Spatial infinity is the region approached at fixed time as the radial coordinate grows, and it sits between past and future null infinity. The antipodal map sends a direction on the sphere to the opposite direction, so a relation across spatial infinity links the field at future null infinity in the limit u to minus infinity with the field at past null infinity in the limit v to plus infinity. Boschetti and Campiglia showed that the logarithmic soft theorem can be rewritten as such an antipodal map once all quantities at spatial infinity are expressed in terms of scattering data.

That rewriting is what makes a position-space proof possible, but the earlier version used the Laddha-Sen results as input to evaluate the asymptotic fields. This paper instead derives those fields from first principles, assuming only that a gravitational scattering event takes place. The antipodal identity itself had already been formally proven under a polyhomogeneity hypothesis; the new content is the independent derivation of the fields that feed it.

Multipoles in harmonic gauge give an infinite tower of relations

The calculation starts from the Iyer-Damour multipole solution, which expresses the most general retarded linear metric in canonical harmonic gauge through symmetric and trace-free mass and current multipole moments. Working in the post-Minkowskian formalism, the metric deviation is expanded as G times a first-order piece plus G squared times a second-order piece. At second order the source splits into a matter part and a quadratic gravitational part, and the paper keeps the matter-induced contribution while treating the gravitational part separately.

From this expansion the authors obtain an infinite tower of antipodal matching relations across spatial infinity for all five Newman-Penrose Weyl scalars. The five scalars are complex projections of the Weyl tensor onto a null tetrad, labelled Psi zero through Psi four, and they encode the gravitational field's different polarization and fall-off sectors. The relations hold at linear order in G and for the leading matter-induced logarithms at order G squared. The paper also connects the one relation relevant to the logarithmic soft theorem to the radiative-gauge formulation of Boschetti-Campiglia and Compere-Robert, and notes that the antipodal relationships look simpler in harmonic gauge than in ingoing or outgoing radiative gauges.

Asymptotic fields computed directly from scattering data

Having the matching relations is not enough; the proof needs the actual asymptotic fields. The paper computes the required Weyl coefficients directly from the scattering data of massive particles, including the matter contributions and the nonlinear effects responsible for graviton drag. Graviton drag here refers to the gravitational backreaction on the asymptotic trajectories, and it is one of the sources of the logarithmic behaviour that the theorem must reproduce.

The computation is organised as a hierarchy of equations for the metric perturbation, with the second-order matter and gravitational sources treated separately. The authors also discuss the transformation from harmonic gauge to radiative gauge, which is needed to compare with the radiative-gauge formulation of the matching condition. The result is a set of explicit expressions for the logarithmic Weyl coefficients in terms of particle masses, velocities, momenta and the total momentum of the scattering event.

Newman-Penrose evolution equations close the proof

The final step combines the matching relations with the Newman-Penrose evolution equations, which relate the different Weyl scalars to one another along the null directions. Feeding the computed asymptotic fields into these equations yields position-space proofs of both the classical leading and the classical logarithmic soft graviton theorems. The logarithmic result is written as an explicit expression for a quantity built from the news tensor's leading tail, matching the form given in the Boschetti-Campiglia formulation.

The proof is independent of the original frequency-space derivation: it does not use the Laddha-Sen results as input, and it recovers the same classical logarithmic soft factor. The authors emphasise that the operator appearing in the final matching equation has a trivial kernel on smooth symmetric trace-free tensors on the sphere, which is what allows the tensor relation to be inverted and the theorem to be stated in its final form.

Where the classical proof stops

The derivation is purely classical, and the paper says so explicitly: it retrieves only the classical version of the logarithmic soft graviton theorem. For quantum gravitational scattering based on Feynman diagrams, Sahoo and Sen obtained additional quantum contributions beyond the classical result, which would require Feynman propagators instead of retarded ones and would modify the antipodal matching relations. A systematic understanding of those quantum terms within the multipolar formalism remains open.

There are further scope conditions. The all-order structure of the antipodal maps relies on the validity of the polyhomogeneous expansion of the gravitational field at spatial infinity, which the paper notes still needs confirmation from an all-order gravitational perturbation analysis. The treatment is limited to order G squared. The paper also restricts to the case where all finite-energy scattering states are massive particles, and it points out that in compact-binary settings the asymptotic trajectories of hard particles do not acquire logarithmic corrections from gravitational binding, so the soft theorems there encompass only the graviton drag contribution. Extending the result to an asymptotically free scattering scenario would require an all-order treatment of gravitational corrections to the asymptotic trajectories.

Harmonic gauge makes the matching relations simpler

One of the paper's more forward-looking observations is a comparison of gauges. The antipodal relationships take a simpler form in harmonic gauge than in ingoing and outgoing radiative gauges, as a comparison of the relevant equations shows. This suggests an alternative to the gravitational scattering framework that maps incoming Bondi-Sachs gauge to outgoing Bondi-Sachs gauge through the intermediate Beig-Schmidt gauge: there might exist a nonlinear expansion of the gravitational field in harmonic gauge that exhibits a more natural map of fields at spatial infinity.

That direction is explicitly left unexplored. For a reader weighing the paper's contribution, the practical priority is therefore the combination of two things: the infinite tower of antipodal matching relations for all five Weyl scalars, and the first-principles evaluation of the asymptotic fields that turns the logarithmic soft theorem into a position-space identity rather than a frequency-space integral. The gauge comparison is a pointer to future work, not a completed result.

Sources used in this answer