Numerical simulations of generalized Langevin equations with deeply asymptotic parameters

A unified algorithm for solving Langevin equations with deeply asymptotic parameters is proposed and tested. The method consists of identifying solvable linear friction and implementing the force evaluations by use of the Runge–Kutta method. We apply the present scheme to the periodic motion of an overdamped particle subjected to a multiplicative white noise. The accurate calculations for the temporal velocity of the particle and its correlation function can be realized by introducing an inertial term. It is shown that the fluctuation around the steady quantity increases with decreasing time step in the overdamped white-noise algorithm, however, a massive white-noise technique greatly reduces this spurious drift, and the result can converge to the correct value if the added inertia approaches zero. The other application is the simulation of generalized Langevin equation with an exponential memory friction, this allows us to treat a weak non-Markovian process.