We present an algorithm for the rapid numerical integration of smooth, timeperiodic differential equations with small nonlinearity, particularly suited to problems with small dissipation. The emphasis is on speed without compromising accuracy and we envisage applications in problems where integration over long time scales is required; for instance, orbit probability estimation via Monte Carlo simulation. We demonstrate the effectiveness of our algorithm by applying it to the spin-orbit problem, for which we have derived analytical results for comparison with those that we obtain numerically. Among other tests, we carry out a careful comparison of our numerical results with the analytically predicted set of periodic orbits that exists for given parameters. Further tests concern the long-term behaviour of solutions moving towards the quasi-periodic attractor, and capture probabilities for the periodic attractors computed from the formula of Goldreich and Peale. We implement the algorithm in standard double precision arithmetic and show that this is adequate to obtain an excellent measure of agreement between analytical predictions and the proposed fast algorithm.Keywords Fast numerics for differential equations · Spin-orbit problem · Periodic attractors · Quasi-periodic attractors · Series solution · Perturbation theory
MotivationIn this paper, we discuss an algorithm for the rapid numerical solution of smooth, nonlinear, non-autonomous, time-periodic, dissipative differential equations, with reference to 123 234 M. V. Bartuccelli et al. a particular example, known as the spin-orbit equation. The spin-orbit ordinary differential equation (ODE) describes the coupling, in the presence of tidal friction, between the orbital and rotational motion of an ellipsoidal satellite orbiting a primary, and many authors have studied it since the original work of Danby (1962) and Goldreich and Peale (1966); see also Murray and Dermott (1999), Celletti (2010) and Correia and Laskar (2004). In cases of interest, both the nonlinear and the dissipative terms are multiplied by small parameters, and as the dissipation parameter decreases, the ODE possesses an ever-increasing number of co-existing periodic orbits, with the initial conditions selecting which one is observed. In this sense, the problem is not simple, despite the fact that the nonlinearity is small: small dissipation coupled with small nonlinearity leads here to non-trivial dynamics.One interesting application of the spin-orbit equation is as a model of the rotation of Mercury, whose primary is considered to be the Sun; other applications come to mind with the discovery of extra-solar planetary systems. Mercury appears to be unique in the Solar System, in that it rotates three times on its own axis for every two orbits of the Sun: all other regular satellites for which we have data are in a one-to-one resonance with their primaries. See for instance Noyelles et al. (2014) for a recent survey offering a new perspective on the problem.In order to estimate numerically the...