libAtoms/QUIP

What it solves

QUIP provides a unified software environment for molecular dynamics simulations, allowing researchers to efficiently combine and switch between various empirical interatomic potentials and quantum mechanical models. It specifically addresses the need to tie together diverse models for materials systems like metals and semiconductors without being locked into a single simulation method.

How it works

QUIP implements a wide array of interatomic potentials (such as Lennard-Jones, EAM, and Tersoff) and tight-binding functional forms. It can act as a standalone tool, a plugin for other software like LAMMPS and CP2K, or be accessed via the quippy Python interface for interoperability with the Atomic Simulation Environment (ASE). A key component is the GAP (Gaussian Approximation Potentials) submodule, which uses Gaussian process regression—a machine learning technique—to achieve quantum-level accuracy without the computational cost of simulating electrons.

Who it’s for

It is designed for computational physicists and materials scientists focusing on the simulation of metals, semiconductors, and other materials systems.

Highlights

• Broad Model Support: Includes numerous interatomic potentials and tight-binding parametrisations.
• ML Integration: Supports Gaussian Approximation Potentials (GAP) for high-accuracy machine learning-based potentials.
• Extensibility: Integrates with major simulation packages like LAMMPS, CP2K, and ASE.
• Flexible Boundary Conditions: Does not assume the minimum image convention, allowing for cutoffs larger than the periodic unit cell size.
• Python Interop: Provides deep access to Fortran routines through the quippy package.