Statistical mechanics of gravitational systems with regular orbits: rigid body model of vector resonant relaxation (original) (raw)

2019, Journal of Physics A: Mathematical and Theoretical

I consider a self-gravitating, N-body system assuming that the N constituents follow regular orbits about the center of mass of the cluster, where a central massive object may be present. I calculate the average over a characteristic timescale of the full, N-body Hamiltonian including all kinetic and potential energy terms. The resulting effective system allows for the identification of the orbital planes with N rigid, disk-shaped tops, that can rotate about their fixed common centre and are subject to mutual gravitational torques. The time-averaging imposes boundaries on the canonical generalized momenta of the resulting canonical phase space. I investigate the statistical mechanics induced by the effective Hamiltonian on this bounded phase space and calculate the thermal equilibrium states. These are a result of the relaxation of spins' directions, identified with orbital planes' orientations, which is called vector resonant relaxation. I calculate the dependence of spins' angular velocity dispersion on temperature and calculate the velocity distribution functions. I argue that the range of validity of the gravitational phase transitions, identified in the special case of zero kinetic term by Roupas, Kocsis & Tremaine, is expanded to non-zero values of the ratio of masses between the cluster of N-bodies and the central massive object. The relevance with astrophysics is discussed focusing on stellar clusters. The same analysis performed on an unbounded phase space accounts for continuous rigid tops. timescales. The orbital angular momentum's vectors' directions (the orbital planes' orientations) relax in several, realistic circumstances independently from their magnitudes, in which case the process is called Vector Resonant Relaxation (VRR). The relaxation of orbital angular momentum's magnitudes is called Scalar Resonant Relaxation. Resonant Relaxation has been studied in astrophysical settings [32-36] especially with numerical simulations [37-40], but also on a kinetic theory basis [41-45]. The method of time-averaging of gravitational orbits and their approximation with rigid wires was introduced by Gauss and has been extensively used in planetary dynamics [46]. In Ref. [47], the time-averaging was applied in a VRR system without any reference to a kinetic energy term. A dynamics of non-canonical variables (the components of orbital planes' direction vector) satisfying the SO(3) algebra on a non-canonical phase space is induced by solely the effective potential energy term of VRR. For this dynamics, Roupas, Kocsis & Tremaine [48] identified gravitational phase transitions in VRR. They calculated the spacial distribution of orbital planes' orientation vectors at thermodynamic equilibrium. In this work, I will again apply the time-averaging method over the apsidal precession's timescale , but now on the full N-body Hamiltonian, with all kinetic terms consistently included. The resulting "rigid-body decomposition" of the effective energy accounts for three terms determining the evolution; namely, a rotational, normal kinetic term accounting for the orbital planes' precession and nutation, a spin kinetic term accounting for the in-plane rotation and the gravitational interaction term at quadrapole and higher order. This effective Hamiltonian describes rigid, disk-shaped, spinning tops allowed to rotate about any of their diameters crossing the common fixed centre, in direct analogy with rigid body dynamics [49] Torques on each disk develop due to mutual gravitational attraction. The general dynamical equations of motion of VRR are calculated in the rigid-body decomposition. They naturally induce new physical parameters, which connect the physical properties of the effective system (rigid annular disks) with these of the implicit system (orbiting point masses). These parameters are the moments of inertia and spin magnitudes of the effective rigid disks. They are connected with the averaging timescale and the ratio ε of the mass of the cluster to that of the central object. The gravitational couplings mediate the two views-implicit and effective-of the system and allow for such relations to emerge. The aforementioned SO(3) evolution induced by a zero kinetic term turns out to be the approximation of the special limit ε = 0 at zeroth order. More importantly, the identified relations between properties of the implicit and effective systems allow for the generalization of the dynamics and the validation and further generalization of the gravitational phase transitions in the cases that the clusters' mass is comparable to that of the central massive object. I specify the dynamical conditions for which such generalization may be valid. Last, but not least, I calculate the dependence of the dispersion of disks' precession and nutation on temperature. It depends on ε and moments of inertia in a non-trivial way. Due to the later dependence, it is possible that different families of bodies acquire different dispersions, even at orders of magnitude. Note that VRR resembles mathematically in certain aspects the Hamiltonian mean-field model (HMF) [50, 51] and the interested reader might find instructive the analogy. In the next section 2 I time-average the self-gravitating N-body Hamiltonian, demonstrate the equations of motion that emerge and calculate the boundaries of the effective, canonical phase space. In section 3 I develop in detail the statistical mechanics of the system. I formally define the microcanonical, the canonical and the Gibbs-canonical ensembles and consider a thermodynamic limit. In section 4, I discuss the inequivalence of ensmbles. In section 5 I review, validate and generalize the VRR gravitational phase transitions. In section 6 I inspect the kinetic energy term and calculate the dependence of the velocity dispersion on temperature. In section 7 I briefly modify the analysis to account for continuous rigid bodies. In the final section 8 I discuss the results.