QUIP and quippy documentation

The QUIP package (GitHub) is a collection of software tools to carry out molecular dynamics simulations. It implements a variety of interatomic potentials and tight binding quantum mechanics, and is also able to call external packages, and serve as plugins to other software such as LAMMPS, CP2K and also the python framework ASE. Various hybrid combinations are also supported in the style of QM/MM, with a particular focus on materials systems such as metals and semiconductors.

quippy is a Python interface to QUIP that provides deep access to most of the Fortran types and routines. The quippy interface is principally maintained by James Kermode.

Long term support of the package is ensured by:

• Noam Bernstein (Naval Research Laboratory)
• Gabor Csanyi (University of Cambridge)
• James Kermode (University of Warwick)

Portions of this code were written by: Albert Bartok-Partay, Livia Bartok-Partay, Federico Bianchini, Anke Butenuth, Marco Caccin, Silvia Cereda, Gabor Csanyi, Alessio Comisso, Tom Daff, ST John, Chiara Gattinoni, Gianpietro Moras, James Kermode, Letif Mones, Alan Nichol, David Packwood, Lars Pastewka, Giovanni Peralta, Ivan Solt, Oliver Strickson, Wojciech Szlachta, Csilla Varnai, Steven Winfield.

Copyright 2006-2016.

Most of the publicly available version is released under the GNU General Public license, version 2, with some portions in the public domain.

Overview of libAtoms and QUIP

• The libAtoms package is a software library written in Fortran 95 for the purposes of carrying out molecular dynamics simulations.
• The QUIP (QUantum mechanics and Interatomic Potentials) package, built on top of libAtoms, implements a wide variety of interatomic potentials and tight binding quantum mechanics, and is also able to call external packages.
• Various hybrid combinations are also supported in the style of QM/MM, including Learn on the Fly scheme [LOTF]
• quippy is a Python interface to libAtoms and QUIP.

Features

The following interatomic potentials are presently coded or linked in QUIP:

• EAM (fcc metals)
• Fanourgakis-Xantheas (water)
• Finnis-Sinclair (bcc metals)
• Flikkema-Bromley
• GAP (Gaussian Approximation Potentials: general many-body)
• Guggenheim-!McGlashan
• Brenner (carbon)
• OpenKIM (general interface)
• Lennard-Jones
• Morse
• Partridge-Schwenke (water monomer)
• Stillinger-Weber (carbon, silicon, germanium)
• SiMEAM (silicon)
• Sutton-Chen
• Tangney-Scandolo (silica, titania etc)
• Tersoff (silicon, carbon)

The following tight-binding functional forms and parametrisations are implemented:

• Bowler
• DFTB
• GSP
• NRL-TB

The following external packages can be called:

• CASTEP
• VASP
• CP2K
• ASAP
• ASE (recent version, 3.11+, recommended)
• Molpro

Contents:

• Introduction
  • Overview
  • First exmaple
• Installation
  • Precompiled Containers
  • Compilation Instructions
    • Quick start
    • Step by step
    • Installing the Jupyter notebook
  • Custom settings
  • Common Problems
    • Permission errors when installing
    • Installing on Mac OS X with macports
    • RuntimeError when importing
    • ImportError when importing
    • Possible problems installing atomeye module
    • Error compiling IPModel_GAP
    • Warning about quippy.castep when importing quippy
    • Internal compiler error with ifort
    • Linking error on Mac OS X
    • Segmentation Faults with OpenBLAS
• Learning GAP
  • Gaussian Approximation Potentials
  • Additional helper
• Descriptors
  • List of descriptors
• Using Potentials
  • QUIP
    • Potentials implemented in QUIP
    • Performing calculations
    • Creating a Potential
  • GAP
    • XML functions
• Tutorials
  • Using Quippy to compute descriptors of atomic environments
    • A simple descriptor: pairwise distances
    • A many-body descriptor: SOAP
    • General_monomer example: benzene
  • How to train a GAP model from scratch
    • steps
    • here we will fit twice, to see the difference between a 2b-only and a 2b+3b fit
    • generating data only with ASE, using the EMT calculator
    • get e0 for H and O - energies of the isolated atoms
    • separate the dataset into a training and a validation set
    • train our GAP model from the command line
    • notice, that the script is running in parallel, using all 8 cores of the current machine
    • use the potential with QUIP on trani.xyz and validate.xyz
    • make simple plots of the energies and forces on the EMT and GAP datas
    • train our GAP_3b model from the command line
    • both training and quip takes significantly more time than the last one!!!
    • use the potential with QUIP on trani.xyz and validate.xyz
    • look at the outputs - clear improvement

Indices and tables

• Index
• Module Index
• Search Page

References

[LOTF]

Csányi, G., Albaret, T., Payne, M., & De Vita, A. ‘Learn on the Fly’: A Hybrid Classical and Quantum-Mechanical Molecular Dynamics Simulation. Physical Review Letters, 93(17), 175503. (2004) http://prl.aps.org/abstract/PRL/v93/i17/e175503>