A reversible numerical integrator of the isokinetic equations of motion

Abstract

An explicit second-order numerical method to integrate the isokinetic equations of motion is derived by fitting circular arcs through every three consecutive points of the discretized trajectory, so that the tangent and the curvature satisfy the equations exactly at every central point. This scheme is reversible and robust, and allows an adaptive step size control. Its performance is tested by computing the thermodynamic properties of simple pair-potential models, and its chemical application is shown for the global search for stable structures, using canonical sampling and energy minimization, of hydrogen-bonded molecular clusters.