Multi-interval trajectory recording for efficient analyses of time correlations

We proposed a general method for efficiently recording trajectories of states for analyzing the time correlation functions (TCFs) A(t) of arbitrary quantities from short to long times, for computer simulations, to reproduce dynamic behaviors. If the quantities for TCF are defined, then on-the-fly calculations are efficient; otherwise, we have to record the trajectories of the states of the system in study. Although the multiple-tau method is popular as an efficient method for error accumulation in calculation of TCFs of defined values for longer t, the problem of the large storage capacity required for recording the trajectories needed to calculate TCFs of arbitrary quantities for shorter t remains. To solve this problem, we propose a method to reduce data size and=or computing times to compute A(t) for shorter t by recording multiple trajectories with different intervals. We named the proposed method multi-interval trajectory recording (MITR). The combination of MITR and recurrence-averages was confirmed to be necessary for precise connection to the multiple-tau method for longer t. In the present paper, we demonstrate an example of a stress autocorrelation function of melts of Kremer–Grest chains.