Interatomic potential parameterization using particle swarm optimization: Case study of glassy silica

Area of Science:

• Materials Science
• Computational Chemistry
• Condensed Matter Physics

Background:

• Accurate interatomic force fields are crucial for classical molecular dynamics simulations of glassy materials.
• Parameterizing these potentials is challenging due to complex, non-convex optimization problems that often lead to inefficient or biased results with traditional methods.

Purpose of the Study:

• To present a novel, efficient parameterization method for interatomic potentials using particle swarm optimization (PSO).
• To demonstrate the method's efficacy by developing and validating potentials for glassy silica.

Main Methods:

• Employed particle swarm optimization (PSO), a stochastic population-based algorithm, for potential parameterization.
• Utilized ab initio simulations and experimental neutron diffraction data for glassy silica to guide and validate the parameterization process.
• Developed two distinct interatomic potentials for glassy silica using the PSO approach.

Main Results:

• The PSO algorithm proved highly efficient in searching for and identifying viable potential parameters.
• The parameterized potentials accurately reproduced the target structural features derived from ab initio simulations and neutron diffraction data.
• The developed potentials demonstrate the effectiveness of PSO in interatomic potential parameterization.

Conclusions:

• The presented PSO-based method offers a robust and efficient solution for parameterizing interatomic potentials for glassy materials.
• This generalizable approach can be readily adapted to various interatomic potential forms and material systems.
• The findings significantly advance the capability to generate accurate and reliable force fields for molecular dynamics simulations.