| Abstract Scope |
Efficiently sampling the potential energy surface (PES) is pivotal across materials science, chemistry, physics, and biology, necessitating advanced saddle point searches (SPS) techniques. In the study, we enhance SPS processes by integrating artificial intelligence through the concept of the dynamic active volume (DAV) in self-evolving atomistic kinetic Monte Carlo (SEAKMC). The DAV method strategically reduces PES dimensionality during the elevation phase of SPS, streamlining the initial search process. As the search progresses to convergence, DAV's constraints are relaxed, guiding the SPS to accurately pinpoint saddle points. Coupled with the dimer method, the DAV not only significantly reduces the time cost for a given search attempt, but also dramatically increases the probability of finding the relevant saddle points in the high-dimensional PES. A Python software package within the framework SEAKMC (SEAKMC_py) with the DAV has been developed.
|