Skip to main content

VeloxChem Workshop

Roberto Di Remigio Eikås, ENCCS

VeloxChem is a clean-slate implementation of Hartree-Fock and Kohn-Sham density-functional theory, with a focus on the efficient calculation of molecular response properties. The code is designed to run on modern high-performance (HPC) architectures and, at the same time, to bridge the gap between ease of use and computational efficiency. More information about VeloxChem can be found in the cover article of the PDC Newsletter No. 2, 2019: www.pdc.kth.se/publications/pdc-newsletter-articles/2019-no-2/veloxchem-1.945304 .

The development of VeloxChem is a success story arising from an ongoing collaboration between researchers at the KTH Royal Institute of Technology (KTH) and research software engineers at the EuroCC National Competence Centre Sweden (ENCCS) and PDC. (ENCCS is the Swedish node of the EuroCC project: the European network of 33 competence hubs in HPC for academia, industry, and the public sector.) Together, we have reduced the complexity of obtaining a working copy of the software, through the use of Conda packages, and expanded its use as a foundation for modern courses in computational chemistry.

In the past year and a half, ENCCS and PDC have delivered two training workshops about VeloxChem: “VeloxChem: Quantum chemistry from laptop to HPC” and then “VeloxChem: quantum chemistry towards pre-exascale and beyond”. The training material from both of these workshops is freely available online at enccs.github.io/veloxchem-workshop  and enccs.github.io/veloxchem-hpc  respectively.

The previous workshop (which was held on 6-7 May 2021) focused on the use of VeloxChem as a platform for interactive quantum chemistry computing, through the use of Jupyter notebooks. An interactive platform is appealing both for beginner and experienced quantum chemists. The former can freely explore the relation between the pen-and-paper derivation of established methods and their practical implementation. The latter can be immediately productive when implementing new methods. In both cases, there is no need to pay a performance penalty for the high-level programming framework as all the numerical heavy lifting is done in C++ and parallelised with MPI and OpenMP. More information about the workshop can be found in the PDC Newsletter No. 1, 2021: www.pdc.kth.se/publications/pdc-newsletter-articles/2021-no-1/veloxchem-quantum-chemistry-from-laptop-to-hpc-1.1094940 .

The latest workshop (which was held on 3-4 March 2022) was geared towards high-performance quantum chemistry and the efficient usage of VeloxChem on the new PDC flagship system, Dardel. We decided to set the stage for our hands-on sessions by providing an overview of some foundational principles underlying HPC, quantum chemistry and performance evaluation. The training material discusses at length the characteristics of the modern CPUs found in newer computing clusters, how these may affect performance, and their influence on programmers’ choices and algorithmic decisions. We included a discussion of performance theory, thus providing an empirical framework for evaluating the performance of VeloxChem on Dardel using real-world workloads.

Overall, both of these events were well received, and thanks to the European network of ENCCS, the events have seen attendance from an international audience. Participants appreciated the ease of use of VeloxChem: the ease of installation, its seamless integration with Jupyter notebooks, and, last but not least, its excellent performance on large HPC clusters.