atomman 1.3.7

Atomistic Manipulation Toolkit

Atomistic Manipulation Toolkit

AtomMan: the Atomistic Manipulation Toolkit is a Python library for creating, representing, manipulating, and analyzing large-scale atomic systems of atoms. The focus of the package is to facilitate the rapid design and development of simulations that are fully documented and easily adaptable to new potentials, configurations, etc. The code has no requirements that limit which systems it can be used on, i.e. it should work on Linux, Mac and Windows computers.

Features:

1. Allows for efficient and fast calculations on millions of atoms, each with many freely defined per-atom properties.

2. Built-in tools for generating and analyzing crystalline defects, such as point defects, stacking faults, and dislocations.

4. Call LAMMPS directly from Python and instantly retrieve the resulting data or LAMMPS error statement.

5. Easily convert systems to/from the other Python atomic representations, such as ase.Atoms and pymatgen.Structure.

6. Can read and dump crystal structure information from a number of formats, such as LAMMPS data and dump files, and POSCAR.

7. Built-in unit conversions.

Installation

The atomman package is compatible with Python 3.6+.

The latest release can be installed using pip:

pip install atomman

or using conda from the conda-forge channel:

conda config --add channels conda-forge
conda config --set channel_priority strict
conda install <package-name>

For Windows users, it is recommended to use an Anaconda distribution and use conda to install numpy, scipy, matplotlib, pandas and cython prior to installing atomman.

Alternatively, all code and documentation can be downloaded from GitHub.

• The stable releases are available at https://github.com/usnistgov/atomman.

• The working development versions are at https://github.com/lmhale99/atomman.

Library setup

Starting with version 1.3.3 atomman uses the potentials Python package to interact with the potentials.nist.gov database. This allows for users to easily search, discover and use the NIST-hosted interatomic potentials as well as some computed properties associated with the potentials. There are a few setup options that can help you get the most out of this feature.

Note: the Python packages potentials, atomman and iprPy all share the same settings file. Updating the library settings in one package will carry over to the other packages.

Changing the library directory

Potential files and database records associated with the potentials can be saved locally to a library directory. The default location of the library directory is <HOME>/.NISTPotentials/library. You can easily change the path to a more accessible location if you wish:

import atomman as am
settings = am.Settings()
settings.set_library_directory("NEWPATH")

Default behavior

The atomman.Library class serves as the central means of interacting with the records database. By default, the class will search for matching records first from the local library directory, then from the remote potentials.nist.gov. If records with the same name are found in both locations, the local copy will be taken over the remote. This makes it possible for users to modify existing records and add their own user-defined records.

All of the Library methods that retrieve/load records and the load function options that use the Library class have the following parameters

• remote (bool) indicates if the remote potentials.nist.gov will be searched.

• local (bool) indicates if the local library directory will be searched.

• localpath (str) allows for an alternate local library directory path

  to be searched.

Heavy-usage/offline behavior

For users who plan on extensive interactions with the database or who are running on systems with limited internet availability, the following steps are recommended:

1. Download/clone the github repository at https://github.com/lmhale99/potentials-library to the library directory. This github repository is a snapshot copy of all potential records at potentials.nist.gov and is kept (mostly) up to date.

2. In Python, load atomman.Settings() and set the default remote behavior to False:

   settings.set_remote(False)

   With this setting, the default value of remote parameter (above) for the Library class methods will be False. This eliminates the need for an internet connection (after step 1) and is typically much faster at retrieving records.

Documentation

Web-based documentation for the atomman package is available at https://www.ctcms.nist.gov/potentials/atomman.

Source code for the documentation can be found in the github doc directory. The doc directory contains the information both as the source RestructuredText files and as unformatted HTML. If you download a copy, you can view the HTML version offline by

cd {atomman_path}/doc/html python -m http.server

Then, opening localhost:8000 in a web browser.

The documentation consists of two main components:

1. Tutorial Jupyter Notebooks: Online html version, Downloadable Notebook version. The tutorials starting with ##. provide a general overview/example of the various capabilities. The tutorials starting with ##.#. give more detailed descriptions and list options available to the tools mentioned in the overview tutorials.

2. Code Documentation: Online html version. This provides a rendering of the Python docstrings for the included functions and classes.

Optional packages

This is a list of additional Python packages that are needed for some of the optional features of the package.

• diffpy.Structure: CIF reader. Required for loading systems from CIF files.

• ase: The Atomic Simulation Environment for interacting with small systems and DFT calculations. Required for converting to/from ase.Atoms objects.

• pymatgen: The Python Materials Genomics package used by the Materials Project for DFT calculations. Required for converting to/from pymatgen.Structure objects.

• spglib: A Python interface to the spglib spacegroup analysis code. spglib can be used to analyze and determine the spacegroup for an atomic system. Required for converting to/from spglib.cell objects.