A High Accuracy Computing Reduction Algorithm Based on Data Reuse for Direct Dynamics Simulations

摘要

This paper proposes a high accuracy Computing Reduction Algorithm Based on Data Reuse. Hassian-Based Predictor-Corrector integration approach uses the current points and external interpolation to predict the location and energy of the next point based on Hessian update scheme. We tested the new approach on CO2 and H2O simulations with an approximated potential energy approximated solution method and an approximated potential energy true solution approach. The results establish that the new approach is better than APAS and APTS. Because the Hessian is calculated once every K time steps, the new approach greatly reduces the computation time without affecting the accuracy of the calculation.