GAEAM: a new package for optimizing the embedded atom method potentials of solids by using the genetic algorithm

Abstract

The embedded atom method (EAM) is a widely used interatomic potential model in molecular dynamics (MD) simulations in solids and alloys. However, the optimization of EAM potential parameters is a complex global optimization problem that traditional methods struggle to solve efficiently. In this work, we present GAEAM, a novel package developed for optimizing EAM potentials of solids using a genetic algorithm (GA) and global optimization. To validate the performance of GAEAM, five typical alloy systems were selected as test cases. MD simulations were performed using the optimized EAM potentials from GAEAM for 1.0 µs (1.0 × 10⁹ fs). Dominant results of simulations including radial distribution functions (RDFs), coordination numbers (CN), root-mean-squared displacements (RMSD), and energy evolution, were analyzed to evaluate the accuracy of the optimized potentials. Detailed MD simulation results revealed that the optimized EAM potentials from GAEAM can accurately reproduce the structural and dynamic properties of the selected alloys. This work demonstrates that GAEAM provides a robust and efficient tool for EAM potential optimization, which can be extended to a wide range of solid and alloy systems. This package also reduces the manual effort required for potential parameter tuning, facilitating progress in computational materials science research.