An efficient and universal parallel algorithm for high-dimensional quantum dynamics in poly-atomic reactions.

This study presents a parallel algorithm for high-dimensional quantum dynamics simulations in poly atomic reactions, integrating distributed- and shared-memory models. The distributions of the wave function and potential energy matrix across message passing interface processes are based on bundled radial and angular dimensions, with implementations featuring either two- or one-sided communication schemes. Using realistic parameters for the H + NH₃ reaction, performance assessment reveals linear scalability, exceeding 90% efficiency with up to 600 processors. In addition, owing to the universal and concise structure, the algorithm demonstrates remarkable extensibility to diverse reaction systems, as demonstrated by successes with six-atom and four-atom reactions. This work establishes a robust foundation for high-dimensional dynamics studies, showcasing the algorithm's efficiency, scalability, and adaptability. The algorithm's potential as a valuable tool for unraveling quantum dynamics complexities is underscored, paving the way for future advancements in the field. [ABSTRACT FROM AUTHOR]