Efficient parallelization of correlated quantum chemistry methods on HPC platforms
||Efficient parallelization of correlated quantum chemistry methods on HPC platforms|
||SNIC Medium Compute|
||Zilvinas Rinkevicius <email@example.com>|
||Kungliga Tekniska högskolan|
||2020-01-31 – 2021-02-01|
This proposal aims to develop and implement efficient parallelization strategies for correlated quantum chemistry methods on HPC platforms. We have developed the VeloxChem quantum chemistry software and demonstrated that the self-consistent field code in VeloxChem scales efficiently on up to 16 000 CPU cores on Beskow . In VeloxChem, we have also implemented the second order Møller-Plesset perturbation theory (MP2) approach in a direct fashion, which eliminates the memory bottleneck and enables computation of molecular systems containing thousands of contracted basis functions. This forms a natural starting point for the development of correlated quantum chemistry methods for large molecular systems embedded in complex environments. Efficient parallelization of reliable correlated methods, like coupled-cluster (CC) and algebraic-diagrammatic construction (ADC), will provide invaluable tools for characterization of experimental spectrum and rational design of novel materials.
 Zilvinas Rinkevicius, Xin Li, Olav Vahtras, Karan Ahmadzadeh, Manuel Brand, Magnus Ringholm, Nanna Holmgaard List, Maximilian Scheurer, Mikael Scott, Andreas Dreuw, and Patrick Norman. "VeloxChem: a Python-driven density-functional theory program for spectroscopy simulations in high-performance computing environments", WIREs Computational Molecular Science, 2019;e1457. https://doi.org/10.1002/wcms.1457