Parts | Days | Selection | Search | Updates | Downloads | Help
O: Fachverband Oberflächenphysik
O 16: Heterogeneous Catalysis: Theory
O 16.10: Talk
Monday, March 12, 2018, 17:30–17:45, MA 141
Tensor-train approximations for catalytic reaction systems — Patrick Gelß1, •Sebastian Matera1, and Christof Schütte1,2 — 1Freie Universität Berlin, Germany — 2Zuse Institute Berlin, Germany
For modeling heterogeneous catalytic processes, kinetic Monte Carlo (kMC) simulations have become an important tool since direct molecular dynamics simulations of rare-event systems are prohibitive. However, the drawback of kMC is the large number of simulations needed to capture the relevant dynamics. We present an alternative approach, which mitigates the curse of dimensionality and directly solves a Markovian master equation corresponding to a given microkinetic mechanism by exploiting the tensor-train (TT) format [1]. Different numerical integration methods and step-size adaptation techniques enable a tunable accuracy and linear scaling in the system size for a large range of input parameters. We benchmark the TT approach against highly accurate kMC simulations for catalytic processes based on nearest-neighbor interactions [2], e.g. a model for the CO2 fixation into methanol at a Cu/ZrO2 interface [3].
[1] P. Gelß, S. Matera, C. Schütte, J. Comput. Phys., 314, pp. 489-502 (2016)
[2] P. Gelß, S. Klus, S. Matera, C. Schütte, J. Comput. Phys., 341, pp. 140-162 (2017)
[3] Q.-L. Tang, Q.-J. Hong, Z.-P. Liu, J. Catal., 263, pp. 114-122 (2009)