An analytical benchmark for MD codes: testing LAMMPS on the 2nd generation Brenner potential

TL;DR

This paper introduces an analytical benchmark for testing MD codes on graphene and carbon nanotubes using REBO potentials, revealing significant discrepancies in LAMMPS results with the second generation Brenner potential.

Contribution

It provides a new benchmark for MD simulations of carbon nanostructures and evaluates LAMMPS performance against this standard.

Findings

LAMMPS results deviate significantly from the benchmark

Identified reasons for discrepancies in LAMMPS implementation

Benchmark can be used to improve MD code accuracy

Abstract

An analytical benchmark is proposed for graphene and carbon nanotubes, that may serve to test whatsoever molecular dynamics code implemented with REBO potentials. By exploiting the benchmark, we checked results produced by LAMMPS (Large-scale Atomic/Molecular Massively Parallel Simulator) when adopting the second generation Brenner potential, we made evident that the code in its current implementation produces results which are offset from those of the benchmark by a significant amount, and provide evidence of the reason.