Kinetic Activation–Relaxation Technique and Self-Evolving Atomistic Kinetic Monte Carlo: Comparison of on-the-fly Kinetic Monte Carlo algorithms

•We compare two adaptive Kinetic Monte Carlo (aKMC) algorithms.•We contrast the approximations at the core of each algorithm.•We assess the performance of different saddle-search methods in the context of aKMC.•Tests are performed for point defects and defect clusters in Fe and FeCr.•The two aKMCs lead to similar kinetics.

We present a comparison of the Kinetic Activation–Relaxation Technique (k-ART) and the Self-Evolving Atomistic Kinetic Monte Carlo (SEAKMC), two off-lattice, on-the-fly Kinetic Monte Carlo (KMC) techniques that were recently used to solve several materials science problems. We show that if the initial displacements are localized the dimer method and the Activation–Relaxation Technique nouveau provide similar performance. We also show that k-ART and SEAKMC, although based on different approximations, are in agreement with each other, as demonstrated by the examples of 50 vacancies in a 1950-atom Fe box and of interstitial loops in 16,000-atom boxes. Generally speaking, k-ART’s treatment of geometry and flickers is more flexible, e.g. it can handle amorphous systems, and rigorous than SEAKMC’s, while the later’s concept of active volumes permits a significant speedup of simulations for the systems under consideration and therefore allows investigations of processes requiring large systems that are not accessible if not localizing calculations.

Graphical abstractFigure optionsDownload full-size imageDownload as PowerPoint slide