citrineinformatics / pif-dft

Tools for converting from DFT codes into PIF objects

dfttopif

A Python library for extracting the input settings and results from Density Functional Theory calculations, and then storing that data in pif format.

Requirements

Python 2.7 or >=3.5, with dependencies listed in requirements.txt

Installation

dfttopif is published on PyPI, so it can be installed with pip:

$ pip install dfttopif

Usage

Option 1: Call the command line tool dfttopif provided in the binary directory, which takes the directory containing DFT results as its only option

./bin/dfttopif /path/to/calculation/

Option 2: Generate the pif object via the python API

from dfttopif import directory_to_pif
data = directory_to_pif('/path/to/calculation/')

Currently supported DFT codes

• VASP (versions tested: 5.2.11, 5.3.2, 5.3.5)
• PWSCF (Quantum Espresso) (versions tested: 4.3.2, 5.0, 5.4.0, 6.0, 6.4.1)

Development

dfttopif is a collaborative effort. Contributions are welcome, both issues and pull requests. This project follows gitflow workflow, so please make PRs to the develop branch.

API documentation is maintained in the source code and hosted on github pages.

pif-dft's Issues

Data de-dedup

The generated PIFs duplicate a large condition block for every property. In the common case, these duplicated conditions could make up the majority of the PIF. We should dedup them somehow, maybe with top-level macros.

Switch to Function Decorators to Note Parsing Operations

I was thinking about how we approach marking which functions in a parser are results or settings, and think we can use decorators instead of the current (complicated, hard-to-maintain) function based-approach.

The decorators would be similar to partial, in that they add some attributes to the functions that store the name of the property, units etc.

Then, we could write a general "get all result function" that introspects the parser object and gets all functions with the functions that have the special fields added by the decorator.

Support python 3

It is the future.

Make DOSCAR optional

Right now, DOSCAR files are a hard req. It would be nice to handle their absence gracefully.