Thermostatting of Active Hamiltonian Systems via Symplectic Algorithms

Abstract

We consider a class of non-standard, two-dimensional (2D) Hamiltonian models that may show features of active particle dynamics, and therefore, we refer to these models as active Hamiltonian (AH) systems. The idea is to consider a spin fluid where -- on top of spin-spin and particle-particle interactions -- spins are coupled to the particle's velocities via a vector potential. Continuous spin variables interact with each other as in a standard $XY$ model. Typically, the AH models exhibit non-standard thermodynamic properties (e.g., for temperature and pressure) and equations of motion with non-standard forces. This implies that the derivation of symplectic algorithms to solve Hamilton's equations of motion numerically, as well as the thermostatting for these systems, is not straightforward. Here, we derive a symplectic integration scheme and propose a Nos\'e-Poincar\'e thermostat, providing a correct sampling in the canonical ensemble. The expressions for AH systems that we find for temperature and pressure might have parallels with the ongoing debate about the definition of pressure and the equation of state in active matter systems. For a specific AH model, recently proposed by Casiulis et al. [Phys. Rev. Lett. {\bf 124}, 198001 (2020)], we rationalize the symplectic algorithm and the proposed thermostatting, and investigate the transition from a fluid at high temperature to a cluster phase at low temperature where, due to the coupling of velocities and spins, the cluster phase shows a collective motion that is reminiscent to that observed in a variety of active systems.