smtg-bham / thermoparser

A tool for streamlining data analysis and visualisation for thermoelectrics and charge carrier transport in computational materials science.

________

///// \\ ________/________________________________________________________________ | | :| / | | ___ ___/ ___ ___ | | | | / | | | | / :| / | | / / | / | | | | | | | | | | | ___ _____ | | | | ___ | | | | ___/:| ___ ___ | :3.1.4|____________________________________:_______________________/____________:_____/

ThermoParser is a toolkit used to simplify the analysis of data produced by specialist materials science codes, centred around thermoelectrics, but also useful for anything pertaining to electronic and/ or phononic transport. ThermoParser is a Python library which contains functions for data retrieval, manipulation and plotting, which can be easily used with a little Python knowlege to generate a wide array of high-quality plots in only a few lines of code. ThermoParser also contains a suite of command-line tools which can retrieve specific data, save derived properties and plot graphs in a single command.

Click on the image to go to the gallery!

Installation

ThermoParser can easily be installed with git and pip:

git clone https://github.com/smtg-bham/ThermoParser.git
cd ThermoParser
pip install .

After installing, you may want to copy ThermoParser/tprc.yaml to ~/.config/tprc.yaml, if you want to set your own default axis labels, unit conversions, default style sheets (two are provided), other aesthetic alterations and more!

Mac

If installing on an m1 mac, you can't currently pip install h5py, so a longer process is required:

0. Install brew
1. Install hdf5 with brew
2. python3 -m pip install cython numpy
3. brew info hdf5 to retrieve the path to your hdf5 install
4. HDF5_DIR=YOUR_HDF5_PATH --no-build-isolation h5py
5. git clone https://github.com/smtg-bham/ThermoParser.git
6. cd ThermoParser
7. pip install --user -e .

Using conda may circumvent this process.

Command Line Interface (CLI)

ThermoParser uses click, which has an easily navigable structure from the command-line, detailed in the tutorials. The most frequently useful commands are included in the CLI for maximum ease, including the tp get functions, which verbosely retrieve data from files which are normally tiresome and error-prone to navigate; and most of the simplest plot-types available through the Python interface.

Python Interface

ThermoParser is designed to have four main stages:

1. Axes:

   Pick an axis layout from tp.axes.

2. Load:

   Use the functions is tp.data.load to load the relevant data.

3. Add:

   Use functions in tp.plot modules to add graphs to the axes.

4. Save:

   Use plt.savefig or equivalent to produce the figure.

As ThermoParser is dependent on matplotlib, each stage can be substituted with bespoke code, e.g. using matplotlib.pyplot.subplots or matplotlib.axes.Axes.scatter. These can still be supplemented with ThermoParser helper functions, such as default labels which the user can set in tp.settings, colourmap generators in tp.plot.colour or legend helpers such as tp.axes.legend.alphabetise.

The best way to get a feel for ThermoParser is to see it in action: Take a look at our examples and tutorials. Currently supported codes are:

• Phononic properties:
  • Phonopy
  • Phono3py
• Electronic properties:
  • AMSET
  • BoltzTraP

Current plotting modes are split into four areas.

• tp.plot.phonons contains plots along a high-symmetry path, including phonon dispersions and plots which project other quantities onto these paths in various ways.
• tp.plot.frequency plots frequency on the x-axis, including density of states (DoS), cumulative kappa, "waterfall" and density plots. Each function has a main argument, which can be useful when plotting multiple quantities on the same set of axes; and an invert argument, which swaps the x and y axes to let you plot DoS-style next to a tp.plot.phonons plot.
• tp.plot.mfp contains a cumulative kappa against mean free path plot.
• tp.plot.heatmap contains a heatmap plotter, and wrappers which format appropriately for ZT against temperature and doping concentration; and one which plots the lattice thermal conductivity required to reach a target ZT, again against temperature and doping.

A set of example scripts is provided in the tp/examples folder and in our online examples, and there is documentation.

Contributing

We welcome any contributions, whether they be a feature request or a new piece of code (or anything else). Adding options is inteded to be straightforward; the modularity of the code means that each step is mostly independent of the others.

Bugs and feature requests can be submitted to the issue tracker, while contributions can be made using the fork and pull approach. Contributions should include comprehensive docstrings, and where appropriate examples, further documentation and tests are greatly appreciated. Documentation uses the sphinx package, and can be built from the docs directory with sphinx-build -b html src/ .. In order to build the docs, download the extra dependencies with, e.g., pip install .[docs] from the ThermoParser directory.

Testing

Tests use the unittest package, and can be run from the test directory with python3 -m unittest. If you don't already have unittest installed, it can be directly with pip install unitest or, e.g., pip install .[tests] from the ThermoParser directory.

Contributors

Many thanks to all those who contributed code or ideas to ThermoParser! Roughly chronologically, they are so far:

• Kieran B. Spooner
• Maud Einhorn
• David O. Scanlon
• Daniel W. Davies
• Bonan Zhu
• Sean R. Kavanagh
• Warda Rahim
• Katarina Brlec
• Joe Willis

Thanks also to the JOSS reviewers, Evan Walter Clarke Spotte-Smith, Enric Tomás Grau-Luque, and Francesco Nattino; and the editor Mojtaba Barzegari. An unintimidating and productive review process, which I would recommend if the opportunity arises!

License

ThermoParser is licensed under the GNU Affero General Public License v3 (AGPLv3).

Requirements

ThermoParser uses the following open-source packages:

• click
• h5py
• json
• matplotlib
• numpy
• pymatgen
• scipy
• sphinx
• yaml