accounting for terms of mixed quadrupolar and 1PN order, deviations from adiabaticity must be taken
into account. In addition to this, backreaction from quickly oscillating terms that are suppressed in
amplitude will also affect the averaging, contrarily to what happened at lower orders. We address
these complications by following the method of near-identity transformations [38], which allows to
consistently implement the averaging procedure to any order of accuracy. While several authors
already studied quadrupolar couplings at 1PN order [9,31,32,34,39–42], we are aware of only three
which took into account these deviations from the adiabatic approximation [31–34]. However, we
believe that we give in this work the first complete expressions of quadrupolar 1PN terms. Indeed,
only the particular case of a circular outer orbit is considered in [31–33] neglecting some PN interactions
that we describe in this work. On the other hand, the derivation in [34] reports a puzzling result that
we will mention in Section 4.
Another source of complexity lies in finding the suitable dynamical variables to efficiently package
relativistic corrections in our results. While the idea at the core of our approach of identifying the
inner binary to a spinning point-particle with multipole moments is very intuitive, providing a definite
relation between the variables describing the inner binary and the parameters of the effective point-
particle is subtle in practice. For example, in our previous work [35] we showed how the choice of a Spin
Supplementary Condition (a gauge condition on the spin tensor of the effective point-particle [43,44]) is
related to the center-of-mass choice of the inner binary. In the present computations, two new similar
subtleties arise. The first one concerns the definition of osculating elements describing both inner
and outer orbits. In our previous work, we followed the usual convention and used the osculating
orbital elements defined as the parameters of the ellipse instantaneously tangent to the trajectory
(described with positions and velocities). However, at 1PN quadrupolar order we find that it becomes
more convenient to use osculating contact elements, which are defined through momenta rather than
velocities [45]. Since PN corrections induce a non-trivial relation between momentum and velocity,
these two sets of osculating elements will differ in general. We will give in Section 2.1 the precise
definition of contact elements, and we will elaborate more on their difference with respect to orbital
elements in Appendix G. One remarkable conclusion of the present analysis is that, while the slowly
evolving part of the contact semimajor axes are conserved throughout the evolution of the system (as
is common in long-timescale evolution of triple systems [18]), their orbital counterpart features small
variations over long time-scales, which offers a new point of view on earlier findings of [31–34].
The second subtle point in the matching between inner binary and point-particle is that the
quantities describing the inner binary system are inherently defined in the rest frame of its center
of mass, which is accelerating because of the presence of the third body. This entails non-trivial
relations between the absolute positions of the inner binary components, defined in the rest frame
of the three-body center-of-mass, and the relative quantities defined in the binary rest frame, as we
show in Section 3.2. As far as we know, this point went so far unnoticed in the relativistic three-body
literature. While this step just amounts to a redefinition of the osculating elements of the inner binary,
it proves to be crucial in order to perform correctly the matching procedure described in Section 3.4.
Let us now describe in more detail the organization of this article. We will begin by summarising
our ”effective two-body” approach in Section 2, which is defined by four main steps. Steps 1 and 2
will then be performed in Section 3, where the core of our matching procedure is explained. In order
to keep the discussion as simple as possible, we have deferred the computation of beyond-adiabatic
corrections to Appendix Cand use only the final result of this appendix in the main text. We then
perform the two final steps of our approach in Section 4, where we integrate out the gravitational field
due to the outer object. Finally, in Section 5we provide a numerical solution implementing the new
relativistic interaction derived in the present work and we show how it influences the long time-scale
dynamics in the case of a particular three-body system. The rest of the Appendices are devoted to: a
presentation of the averaging procedure that we employ (Appendix A); a discussion of the conservation
of the semimajor axis (Appendix B); a review of the Lagrange Planetary Equations (Appendix D);
the derivation of the expressions connecting the absolute coordinates of the two inner bodies in the
3