quippy-ase 0.9.14

ASE-compatible Python bindings for the QUIP and GAP codes

QUIP - QUantum mechanics and Interatomic Potentials

The QUIP package 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.

For more details, see the online documentation. There is separate documentation for SOAP and GAP.

Long term support of the package is ensured by:

• Noam Bernstein (@bernstei, Naval Research Laboratory)
• Gabor Csanyi (@gabor1, University of Cambridge)
• James Kermode (@jameskermode, 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, Tamas K Stenczel, Adam Fekete.

Copyright 2006-2021.

Most of the publicly available version is released under the GNU General Public license, version 2, with some portions in the public domain. The GAP code, included as a submodule, is distributed under a non-commerical academic source license

Citing QUIP, quippy and GAP

Please cite the following publication if you use QUIP:

@ARTICLE{Csanyi2007-py,
  title   = "Expressive Programming for Computational Physics in Fortran 95+",
  author  = "Cs{\'a}nyi, G{\'a}bor and Winfield, Steven and Kermode, J R and De
             Vita, A and Comisso, Alessio and Bernstein, Noam and Payne,
             Michael C",
  journal = "IoP Comput. Phys. Newsletter",
  pages   = "Spring 2007",
  year    =  2007
}

If you use the quippy Python interface, please cite:

@ARTICLE{Kermode2020-wu,
  title    = "f90wrap: an automated tool for constructing deep Python
              interfaces to modern Fortran codes",
  author   = "Kermode, James R",
  journal  = "J. Phys. Condens. Matter",
  month    =  mar,
  year     =  2020,
  keywords = "Fortran; Interfacing; Interoperability; Python; Wrapping codes;
              f2py",
  language = "en",
  issn     = "0953-8984, 1361-648X",
  pmid     = "32209737",
  doi      = "10.1088/1361-648X/ab82d2"
}

If you use the GAP code please cite

@ARTICLE{Bartok2010-pw,
  title    = "Gaussian approximation potentials: the accuracy of quantum
              mechanics, without the electrons",
  author   = "Bart{\'o}k, Albert P and Payne, Mike C and Kondor, Risi and
              Cs{\'a}nyi, G{\'a}bor",
  journal  = "Phys. Rev. Lett.",
  volume   =  104,
  number   =  13,
  pages    = "136403",
  month    =  apr,
  year     =  2010,
  issn     = "0031-9007, 1079-7114",
  pmid     = "20481899",
  doi      = "10.1103/PhysRevLett.104.136403"
}

Features

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

• BKS (van Beest, Kremer and van Santen) (silica)
• EAM (fcc metals)
• Fanourgakis-Xantheas (water)
• Finnis-Sinclair (bcc metals)
• Flikkema-Bromley
• GAP (Gaussian Approximation Potentials) with (growing...) online documentation
• Guggenheim-McGlashan
• Brenner (carbon)
• OpenKIM (general interface)
• Lennard-Jones
• MBD (many-body dispersion correction)
• Morse
• Partridge-Schwenke (water monomer)
• Stillinger-Weber (carbon, silicon, germanium)
• SiMEAM (silicon)
• Sutton-Chen
• Tangney-Scandolo (silica, titania etc)
• Tersoff (silicon, carbon)
• Tkatchenko-Sheffler pairwise dispersion correction

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
• Molpro
• ASE (required if using quippy Python interface; latest version recommended)

Code philosophy and goals

QUIP was born because of the need to efficiently tie together a wide variety of different models, both empirical and quantum mechanical. It will not be competitive in terms of performance with codes such as LAMMPS and Gromacs. The Atomic Simulation Environment also does this, and is much more widely used, but QUIP has a number of unique features:

• Access to Fortran types and routines from Python via the quippy package
• Support for Gaussian Approximation Potentials (GAP) - online docs
• Does not assume minimum image convention, so interatomic potentials can have cutoffs that are larger than the periodic unit cell size

Binary Installation of QUIP and quippy

Binary for QUIP and the associated quippy Python bindings that provide interopability with the Atomic Simulation Environment (ASE) are available from the Python package index (PyPI) under the package name quippy-ase. This means you can install the latest release with:

pip install quippy-ase

Installing via pip also makes the quip and gap_fit command line programs available (providing the directory that pip installs scripts to is on your PATH).

Currently, wheels are available for x86_64 architectures with Python 3.6+ on macOS and glibc-based Linux distributions (e.g. Ubuntu, CentOS) and for macOS arm64. The wheels are updated periodically using GitHub Actions CI. Please open issues here if you have problems installing with pip.

Precompiled Containers

If you have access to Docker or Singularity, you can try one of the precompiled images to get up and running quickly.

Compilation Instructions

1. To compile QUIP the minimum requirements are:

   • A working Fortran compiler. QUIP is tested with gfortran 4.4 and later, and ifort 11.1.

   • Linear algebra libraries BLAS and LAPACK. QUIP is tested with reference versions libblas-dev and liblapack-dev on Ubuntu 12.04, and mkl 11.1 with ifort.

   • MPI: To use the MPI parallelisatin of gap_fit, you need a ScaLAPACK library, e.g. libscalapack-openmpi on Ubuntu, or as part of the MKL.

2. Clone the QUIP repository from GitHub. The --recursive option brings in submodules automatically (If you don't do this, then you will need to run git submodule update --init --recursive from the top-level QUIP directory after cloning) ::

   git clone --recursive https://github.com/libAtoms/QUIP.git
   

   One submodule is the GAP code, which can be found in src/GAP. Note that GAP is distributed under a diferent license.

   GAP is a machine learning method that uses Gaussian process regression, and needs large data files to run. You can find potentials that have been published as well as training data in our data repository, see also the online docs.

3. Decide your architecture by looking in the arch/ directory, and define an environmental variable QUIP_ARCH, e.g.::

   export QUIP_ARCH=linux_x86_64_gfortran
   

   for standard gfortran on Linux. Here is where you can adjust which compiler is being used, if you do not like the defaults. You may need to create your own arch/Makefile.${QUIP_ARCH} file based on an existing file for more exotic systems.

   MPI: Some arch files already include adjustments for MPI use. Those usually have mpi in their name, e.g. linux_x86_64_gfortran_openmpi+openmp.

4. Customise QUIP, set the maths libraries and provide linking options::

   make config
   

   Makefile.config will create a build directory, build/${QUIP_ARCH}, and all the building happen there. First it will ask you some questions about where you keep libraries and other stuff, if you don't use something it is asking for, just leave it blank. The answers will be stored in Makefile.inc in the build/${QUIP_ARCH} directory, and you can edit them later (e.g. to change compiler, optimisation or debug options).

   If you later make significant changes to the configuration such as enabling or disabling tight-binding support you should force a full rebuild by doing a make deepclean; make.

   MPI: To use the MPI parallelisation of gap_fit, you have to add your system library to the linking options, e.g. -lscalapack or -lscalapack-openmpi, enable GAP support, enable QR decomposition, and enable ScaLAPACK.

5. Compile all programs, modules and libraries::

   make
   

   From the top-level QUIP directory. All programs are built in build/${QUIP_ARCH}/. You can also find compiled object files and libraries (libquip.a) in that directory. Programs can be called directly from that directory.

   Other useful make targets include:

   • make install : copies all compiled programs it can find to QUIP_INSTALLDIR, if it's defined and is a directory (full path required), and copies bundled structures to QUIP_STRUCTS_DIR if it is defined.

   • make libquip: Compile QUIP as a library and link to it. This will make all the various libraries and combine them into one: build/${QUIP_ARCH}/libquip.a, which is what you need to link with (as well as LAPACK).

6. A good starting point is to use the quip program, which can calculate the properties of an atomic configuration using a variety of models. For example::

   quip atoms_filename=test.xyz init_args='IP LJ' \
       param_filename=share/Parameters/ip.parms.LJ.xml E
   

   assuming that you have a file called test.xyz with the following data in it representing Cu atoms in a cubic fcc lattice::

   4
   Lattice="3.61 0 0 0 3.61 0 0 0 3.61" Properties=species:S:1:pos:R:3
   Cu     0.000 0.000 0.000
   Cu     0.000 1.805 1.805
   Cu     1.805 0.000 1.805
   Cu     1.805 1.805 0.000
   

   The Lennard-Jones parameters in the above example are defined in the ip.parms.LJ.xml file under share/Parameters (ensure the path to this file is correct). The format of the atomic configuration is given in Extended XYZ format, in which the first line is the number of atoms, the second line is a series of key=value pairs, which must at least contain the Lattice key giving the periodic bounding box and the Properties key that describes the remaining lines. The value of Properties is a sequence of triplets separated by a colon (:), that give the name, type and number of columns, with the type given by I for integers, R for reals, S for strings.

   Most string arguments can be replaced by --help and QUIP programs will then print a list of allowable keywords with brief help messages as to their usage, so e.g. init_args=--help will give a list of potential model types (and some combinations). The parsing is recursive, so init_args="IP --help" will then proceed to list the types of interatomic potentials (IP) that are available.

7. To compile the Python wrappers (quippy), the minimum requirements are as follows. f90wrap will be installed automatically by the build process, but you might need to check that the directory where pip installs executuable scripts to is on your path (e.g. by setting PATH=~/.local/bin:$PATH).

   • Python 3
   • NumPy (numpy>=1.5.0)
   • Atomic Simulation Environment (ase>=3.17.0)
   • f90wrap
   • (optional) SciPy
   • (optional) matscipy.

   Note: If you are using a Python virtual environment (virtualenv) and would like to install quippy into it, ensure the environment is activated (source <env_dir>/bin/activate, where <env_dir> is the root of your virtual environment) before building quippy (otherwise library versions may cause unexpected conflicts).

8. To compile the Python wrappers (quippy), run::

   make quippy
   

   Quippy can be used by adding the lib directory in quippy/build/${QUIP_ARCH} to your $PYTHONPATH, however it can be more convenient to install into a specific Python distribution::

   make install-quippy
   

   will either install into the current virtualenv or attempt to install systemwide (usually fails without sudo). To install only for the current user (into ~/.local), execute the command QUIPPY_INSTALL_OPTS=--user make install-quippy, or use QUIPPY_INSTALL_OPTS=--prefix=<directory> to install into a specific directory. QUIPPY_INSTALL_OPTS can also be set in the file build/${QUIP_ARCH}/Makefile.inc.

9. More details on the quippy installation process and troubleshooting for common build problems are available in the online documentation.

10. To run the unit and regression tests, which depend on quippy:: bash make test

11. To get back to a state near to a fresh clone, use bash make distclean

12. Some functionality is only available if you check out other modules within the QUIP/src/ directories, e.g. the ThirdParty (DFTB parameters, TTM3f water model).

13. In order to run QUIP potentials via LAMMPS, make libquip to get QUIP into library form, and then follow the instructions in the LAMMPS documentation. You need at least 11 Aug 2017 version or later.

Developer notes:

Fixing/updating the version of GAP:

cd src/GAP

git checkout <commit>

OR

git checkout main

Updating the version in the QUIP repository:

cd ../..
git add src/GAP
git commit -m "updating the version of GAP"

Mac OS

We do not recommend Apple-shipped compilers and python, and we do not test compatibility with them. Either use MacPorts or Homebrew to obtain GNU compilers, and also use the python from there or Anaconda. As of this edit, gcc-8.1 produces as internal compiler error, but gcc-4.6 through to gcc-7 is fine.

Triggering the wheel build

Wheels are built on push and pull requests to public using cibuildwheel with this workflow.

To make a release candidate create a tag with a suffix such as -rc1 for the first attempt, push to trigger the build:

git commit -m 'release v0.x.z-rc1'
git tag v0.x.y-rc1
git push --tags

If all goes well, the .whl files will show up as assets within a new GitHub release. The installation process can now be tested locally.

Release wheels to PyPI

Once everything works correctly, make a full release (i.e. create a tag named just v0.x.y without the -rc1 suffix). This will trigger the upload of wheels and source distribution to PyPI.