A curved line search algorithm for atomic structure relaxation

TL;DR

This paper introduces a novel curved line search algorithm that leverages on-the-flight force learning to significantly speed up atomic structure relaxation in ab initio calculations, outperforming traditional methods.

Contribution

The paper presents a new curved line minimization algorithm based on on-the-flight force learning, offering a substantial acceleration in atomic relaxation processes.

Findings

Demonstrates significant speedup over conjugate-gradient method

Effective for metal cluster relaxations

Reduces number of steps for convergence

Abstract

Ab initio atomic relaxations often take large numbers of steps and long times to converge. An atomic relaxation method based on on-the-flight force learning and a corresponding new curved line minimization algorithm is presented to dramatically accelerate this process. Results for metal clusters demonstrate the significant speedup of this method compared with conventional conjugate-gradient method.