|FIELD||Colloquium: Comp. Sciences|
|DATE||May 22 (Wed), 2019|
|TITLE||Finding multiple transition pathways via efficient global optimization of Onsager-Machlup action|
Finding multiple transition pathways between different states is one of the most challenging problems in computational chemistry and biology. In this talk, a new computational approach, Action-CSA, to find multiple transition pathways of diffusive systems is presented. The method searches multiple pathways between fixed initial and final states through global optimization of the Onsager-Machlup action using the conformational space annealing method1. This approach successfully finds all possible pathways of small systems without initial guesses on pathways. Pathway space searching is efficiently carried out by crossover and mutation operations of pathways and preserving the diversity of the set. The search efficiency of the approach is assessed by finding pathways for the conformational changes of alanine dipeptide, hexane and the folding dynamics of a fast-folding protein, FSD-1. A comparison with long Langevin dynamics simulations demonstrates that action optimization by Action-CSA not only identifies the most dominant pathway, but also correctly determines the rank order of less dominant pathways. It is also observed that the lowest action folding pathways of the mini-protein FSD-1 identified by Action-CSA is showing a good agreement with and experimental results. These results show that efficient optimization of Onsager-Machlup action is a general and powerful approach to investigate transition behaviors of various diffusive systems.