Express: nonstop calculations with Quantum ESPRESSO
ORAL
Abstract
Ab initio mineral physics studies' intrinsic complexity inspired the development of workflows to automate long and extensive sequences of the ab initio calculations [1]. Here we introduce Express, a new generation of workflows designed to facilitate calculations of the thermodynamic properties of materials. These workflows are exquisitely designed and well-tested, inspired by their predecessor that is known as VLab [1].
Various materials properties can be computed in Express, e.g., static and thermal equations of state, phonon density of states, and thermodynamic properties. Each workflow consists of three steps:
1) pre-processing: dynamic generation and validation of input files;
2) processing: submitting, monitoring jobs, and retrieving output from the computing resources;
3) post-processing: analyzing results and handling errors, such as rerunning failed jobs.
The workflows are highly modularized that can be separated, chained, and customized according to practical needs. It is achieved by tracking the provenance of each step.
Written in language Julia, Express is thus born to be performant, extensible, and run on numerous high-performance platforms.
[1] da Silveira PRC, da Silva CRS, and Wentzcovitch RM (2008) DOI:10.1016/j.cpc.2007.09.001
Various materials properties can be computed in Express, e.g., static and thermal equations of state, phonon density of states, and thermodynamic properties. Each workflow consists of three steps:
1) pre-processing: dynamic generation and validation of input files;
2) processing: submitting, monitoring jobs, and retrieving output from the computing resources;
3) post-processing: analyzing results and handling errors, such as rerunning failed jobs.
The workflows are highly modularized that can be separated, chained, and customized according to practical needs. It is achieved by tracking the provenance of each step.
Written in language Julia, Express is thus born to be performant, extensible, and run on numerous high-performance platforms.
[1] da Silveira PRC, da Silva CRS, and Wentzcovitch RM (2008) DOI:10.1016/j.cpc.2007.09.001
–
Presenters
-
Qi Zhang
Department of Applied Physics and Applied Mathematics, Columbia University
Authors
-
Qi Zhang
Department of Applied Physics and Applied Mathematics, Columbia University
-
Hongjin Wang
Department of Computer Science, Columbia University, Department of Computer Sciences, Columbia University
-
Jingyi Zhuang
Department of Earth and Environmental Sciences, Columbia University, Lamont-Doherty Earth Observatory, Columbia University
-
Pedro da Silveira
Department of Applied Physics and Applied Mathematics, Columbia University
-
Renata M Wentzcovitch
Columbia Univ, Department of Applied Physics and Applied Mathematics, Columbia University, Applied Physics and Applied Mathematics, Columbia University, Columbia University