DFT automatization using VASP and ASE

This repo contains a library that aims at automatizing some Density Functional Theory (DFT) workflows in VASP by using the ASE toolkit. The workflows are adapted from the wonderful ressources provided by Prof. John Kitchin. I am not an expert in DFT but I hope that these scripts might be helpful to some people starting up with DFT and management of DFT workflows.

Installation

• You need to have a properly working and licensed VASP installation on your local computer or a high performance cluster.
• You need to have ASE properly installed and set up to work with VASP (check out the instructions here). The workflow has been tested with version 3.20.1.

Structure

The repo consists of two scripts and a function library. The script create_atoms.py creates a structure and determines the number of bands and cores that should be allocated for a calculation. The script job.py describes the details of a job, which is a workflow consisting of several consecutive minimization procedures. In the example, some convergence tests and minimization procedures are performed on a Copper structure.

Scripts

create_atoms.py

This script is used to create/load a structure and determine the number of bands and cores for a calculation based on some common formulae. This is useful to estimate the parallelization options when optimizing large super-cells. Created structures are saved to and loaded from the subfolder ạtoms.

job.py

The job script shows an example workflow on a Copper structure. Here the parameters e-cut and kpoints are determined by convergence testing. Then the structure is optimized with different conditions for structure relaxation. All results are saved in a subfolder results which is created upon the first execution of the program. The results contain the raw input and output files from VASP as well as the results from convergence and optimization runs in the form of ASCII files and images of the plots.

Parallelization

The scripts can be run as-is on a personal computer, by default it will do so with kpar = 1 and ncore = 1. If you have VASP and ASE installed on a High Performance Cluster (HPC), or if you have a parallelized version of VASP compiled on your personal computer, you can supply the arguments kpar and ncore on the command line. As an example, you can run the calculations with kpar = 2 and ncore = 16 by executing: job.py 2 16

If you want to run VASP in a parallel environment, you will have to set the ASE environment variable ASE_VASP_COMMAND to "mpirun vasp_std".