Coupling the time-warp algorithm with the graph-theoretical kinetic Monte Carlo framework for distributed simulations of heterogeneous catalysts

Abstract

Despite the successful and ever widening adoption of kinetic Monte Carlo (KMC) simulations in the area of surface science and heterogeneous catalysis, the accessible length scales are still limited by the inherently sequential nature of the KMC framework. Simulating long-range surface phenomena, such as catalytic reconstruction and pattern formation, requires consideration of large surfaces/lattices, at the μm scale and beyond. However, handling such lattices with the sequential KMC framework is extremely challenging due to the heavy memory footprint and computational demand. The Time-Warp algorithm proposed by Jefferson (1985) [43] offers a way to enable distributed parallelization of discrete event simulations. Thus, to enable high-fidelity simulations of challenging systems in heterogeneous catalysis, we have coupled the Time-Warp algorithm with the Graph-Theoretical KMC framework of Stamatakis and Vlachos (2011) [17]; Nielsen et al. (2013) [18] and implemented the approach in the general-purpose KMC code Zacros. We have further developed a “parallel-emulation” serial algorithm, which produces identical results to those obtained from the distributed runs (with the Time-Warp algorithm) thereby validating the correctness of our implementation. These advancements make Zacros the first-of-its-kind general-purpose KMC code with distributed computing capabilities, thereby opening up opportunities for detailed meso-scale studies of heterogeneous catalysts and closer-than-ever comparisons of theory with experiments.