稳健性(进化)
惯性参考系
缩放比例
统计物理学
材料科学
算法
物理
机械
计算机科学
数学
经典力学
化学
几何学
生物化学
基因
作者
Sebastián Echeverri Restrepo,Predrag Andric
标识
DOI:10.1016/j.commatsci.2022.111978
摘要
The Fast Inertial Relaxation Engine (FIRE) is a pseudo-dynamical minimisation algorithm commonly used for the structural relaxation of atomistic systems. We introduce an additional factor to FIRE that modifies the bias and the scaling of the velocities of the atoms during the “mixing” step. The robustness of the modified algorithm has been tested in a number of representative configurations having different levels of complexity. Reductions of 9% (Σ5(310) grain boundary in aluminium with 3184 atoms) to 65% (fracture simulation in Fe25Ni25Cr50 with ∼1.6 million atoms) were obtained in the number of iterations needed to reach convergence. We dub this variation of the original algorithm ABC-FIRE, where ABC stands for Accelerated Bias-Corrected.
科研通智能强力驱动
Strongly Powered by AbleSci AI