materialstheory/triqs repository overview

⁠Dockerfiles for DFT codes

This repository contains the dockerfiles we use to build DFT codes for use on supercomputers.

⁠Applications

Dockerfiles for the following codes are available:

• VASP: 5.4.4 and 5.4.4 with GPU support (Note that you need a valid license to build the image.)
• Quantum ESPRESSO: 6.2.0
• triqs 2.0 (https://triqs.github.io/triqs/master/⁠) built with MPICH and MKL from base-bionic-sci-python

⁠base images

The base images provide a reproducible environment to compile applications with specific versions of compilers, libraries and MPI. Currently, we have:

• base-gnu: Ubuntu 16.04 with gcc/gfortran 5.4.1, MKL 2018.3, MPICH 3.1.4
• base-cuda: Same as above, but with nvidia/cuda:9.0-ubuntu16.04 as base image
• base-bionic-gnu: Ubuntu 18.04 with gcc/gfortran 7.3.0, MKL 2018.3, MPICH 3.1.4, boost 1.68
• base-bionic-sci-python: same Ubuntu 18.04 as above plus numpy, scipy, mpi4py, h5py (all built manually with MKL and MPICH)

These images can be pulled from Dockerhub at https://hub.docker.com/u/materialstheory/⁠.