py4vasp

py4vasp is a python interface to extract data from VASP calculations. It is intended mainly to get a quick look at the data and provide the functionality to export it into common formats that can be used by other more sophisticated postprocessing tools. The second domain of application is for people that want to write python scripts based on the data calculated by VASP. This tool interfaces directly with the new HDF5 file format and thereby avoids parsing issues associated with the XML or OUTCAR files.

For these two groups of users, we provide a different level of access. The simple routines used in the tutorials will read the data from the file directly and then generate the requested plot. For script developers, we provide an expert interface where the data is lazily loaded as needed with some greater flexibility when the data file is opened and closed.

Installation

While this is not required to be able to run py4vasp, you may want to consider creating a separate environment for installation to avoid interference with other installed packages. 1 You can then install py4vasp from PyPI using the pip package installer

pip install py4vasp

This will automatically download py4vasp as well as all the required dependencies. However, we noticed that this approach is not fail-safe, because the installation of the mdtraj dependency does not work on all operating systems. So in case the simple installation above fails, you may need to use conda to install mdtraj

conda install -c conda-forge mdtraj
pip install py4vasp

If these commands succeed, you should be able to use py4vasp. You can make a quick test of your installation running the following command

python -c "import py4vasp; print(py4vasp.__version__)"

This should print the version of py4vasp that you installed.

Important

py4vasp extracts all information from the HDF5 output so you need to make sure to compile VASP adding -DVASP_HDF5 to the CPP_OPTIONS in the makefile.include. You will also need to add the HDF5 library to the include (INCS) and linking (LLIBS) instructions. py4vasp also requires a VASP version > 6.2 and because py4vasp is developed alongside VASP, we recommend that you use versions of these two codes released about at the same time for maximum compatibility.

Quick start

The user interface of py4vasp is optimized for usage inside a Jupyter environment (Jupyter notebook or Jupyter lab), though it can be used in regular Python scripts as well. To give you an illustrative example of what py4vasp can do, we assume that you created a Jupyter notebook inside the directory of your VASP calculation. Then you access all the results of this calculation with

>>> from py4vasp import Calculation
>>> calc = Calculation.from_path(".")

Naturally, if you created the notebook outside of the calculation directory, you would replace the path . with the directory of the calculation.

The attributes of the calculation correspond to different physical quantities that you could have calculated with VASP. If you have an interactive session you can type calc. and then hit Tab to get a list of all possible quantities. However only the ones that you computed with VASP will give you any meaningful result.

In the following, we will assume that you computed the density of states (DOS) with orbital projections (LORBIT = 11). You may now want to read the data from your VASP calculation to post-process it further with a script. This can be achieved in a single line of code

>>> dos = calc.dos.read()

Under the hood, this will access the vaspout.h5 file, because py4vasp knows where the output is stored after you ran VASP. It will read the relevant tags from the file and store them all in a Python dictionary. If you want to access particular orbital projections, let’s say the p orbitals, you can pass a select = "p" as an argument to the routine. More generally, you can check how to use a function with

>>> help(calc.dos.read)

The most common use case for the DOS data may be to prepare a plot to get some insight into the system of interest. Because of this, we provide an easy wrapper for this particular functionality

>>> calc.dos.plot()

This will return an interactive figure that you can use to investigate the DOS. Note that this requires a browser to work, which means it will open one if you execute this inside a script instead of a Jupyter notebook. The plot command takes the same arguments as the read command.

The interface for the other quantities is very similar. Every quantity provides a read function to get the raw data into Python and where it makes sense a plot function visualizes the data. However, note that in particular, all data visualized inside the structure require a Jupyter notebook to work.

If you want to experience more features of py4vasp, we highly recommend taking a look at the tutorials for VASP. Many of them use py4vasp to plot or analyze the data produced by VASP, so this may give you an excellent starting point to learn how you can apply py4vasp in your research.


1

To create a separate py4vasp from the rest of your packages you can create an environment with venv or conda. The advantage of the former is that it comes with your python installation, the latter requires the installation of Miniconda or Anaconda. Nevertheless, it may be a good idea to go with conda in particular on Windows and macOS, because it can help install dependencies of py4vasp. Below you find the instructions on how to create the environment depending on your environment managing tool and your operating system.

venv (Linux / MacOS)
python3 -m venv py4vasp-env
source py4vasp-env/bin/activate.sh
venv (Windows)
python3 -m venv py4vasp-env
py4vasp-env\Scripts\activate.bat
conda (Linux / MacOS / Windows)
conda create --name py4vasp-env python
conda activate py4vasp-env

Note

You will need to run the activation part of the command again if you open a new shell.