A numerical approach to large deviations in continuous time

Abstract

We present an algorithm to evaluate large deviation functions associated to history-dependent observables. Instead of relying on a time discretization procedure to approximate the dynamics, we provide a direct continuous-time algorithm valuable for systems with multiple timescales, thus extending the work of Giardinà, Kurchan and Peliti (2006 Phys. Rev. Lett. 96 120603 ). The procedure is supplemented with a thermodynamic-integration scheme which improves its efficiency. We also show how the method can be used to probe large deviation functions in systems with a dynamical phase transition—revealed in our context through the appearance of a non-analyticity in the large deviation functions.