Berlin 2024 – wissenschaftliches Programm
Bereiche | Tage | Auswahl | Suche | Aktualisierungen | Downloads | Hilfe
O: Fachverband Oberflächenphysik
O 108: Electronic Structure Theory II
O 108.7: Vortrag
Freitag, 22. März 2024, 12:00–12:15, MA 043
Benchmark Auxiliary-Field Quantum Monte Carlo Method — •Zoran Sukurma1, Martin Schlipf2, Moritz Humer1, Amir Taheridehkordi1, and Georg Kresse1,2 — 1Kolingasse 14-16, University of Vienna — 2Berggasse 21, VASP GmbH
Phaseless Auxiliary-Field Quantum Monte Carlo (ph-AFQMC) is a promising method to approximate the ground state properties of various quantum systems, including molecules, solids, and lattice models.
We present a scalable Fortran implementation of ph-AFQMC and demonstrate its excellent performance and favorable computational scaling. We apply the method to the HEAT set, the benzene molecule, water clusters, and a 2x2x2 diamond supercell. We observe a mean absolute deviation of the total energy of 1.15kcal/mol for the HEAT set. For water clusters, we find that ph-AFQMC yields binding energies that differ from CCSD(T) by typically less than 0.5kcal/mol. Excellent agreement with CCSD(T) is observed for diamond. In addition, we investigate modifications of the phaseless approximation that can help to reduce overcorrelation problems common to ph-AFQMC. Using the benzene molecule as an example, we demonstrate that the modified algorithm reproduces the same energy as the original algorithm with 400 Slater determinants.
Our results underscore the potential of ph-AFQMC to accurately predict ground-state properties in various systems.
Z. Sukurma, M. Schlipf, M. Humer, A. Taheridehkordi, and G. Kresse, J. Chem. Theory Comput. 19, 4921 (2023)
Keywords: ph-AFQMC; Quantum Monte Carlo; Many-body ground state; Correlated electrons; Chemical Accuracy