Extended global area search algorithm for unimolecular reactions and multistep conformational transitions

Abstract

In this study, we propose an extended Global Area Search (GLAS) algorithm for the automatic exploration of multistep transitions without specifying reaction coordinates in advance. The key concept of the original GLAS method is to introduce repulsive interactions among walkers representing molecular structures and to promote their dispersion by increasing these interactions, thereby enabling the system to overcome activation barriers and identify surrounding equilibrium structures (EQs). The proposed method inherits this core concept while introducing two important improvements to the original GLAS framework: (1) iterative cycles of local optimisation and exploration of surrounding EQs to trace multistep transitions, and (2) precise regulation of the walkers’ energies to guide the exploration toward elementary transitions with activation barriers below a specified target energy. We demonstrate the effectiveness of this algorithm by applying it to the conformational changes of alanine dipeptide and the Cope rearrangement of 3-methyl-1,5-hexadiene. Specifically, all conformers of alanine dipeptide were located in a single exploration run, and the multistep pathway of the Cope rearrangement was successfully elucidated. Furthermore, the walkers automatically explored the chemical space, selectively identifying elementary transitions with activation barriers below the target energy for each step. These results indicate that the extended GLAS algorithm can efficiently traverse complex potential energy surfaces involving specific activation barriers in a stepwise manner, providing a powerful and versatile tool for exploring molecular structural changes.