Reaction path methods based on holonomic constraints

COMP 62

Jhih-Wei Chu, jwchu@berkeley.edu, Department of Chemical Engineering, University of Carlifornia, Berkeley, 101A Gilman, UC Berkeley, Berkeley, CA 94720
The requirements of equal distances between replicas during reaction path optimization are formulated as holonomic constrains. This scheme thus eliminates the need of using penalty functions; as a result, the quality of convergence of path optimization can be greatly improved compared to conventional methods such as the nudged elastic band (NEB) method. In addition to minimum energy paths (MEP), we also derived a formulation for optimizing dynamic paths based on the least action principle. Moreover, this holonomic constraint formation also allows a straightforward implementation of super-linear minimization schemes such as the Adopted Basis Newton Raphson (ABNR) method. Therefore, MEPs and dynamic paths can now be obtained with much fewer energy and force evaluations. Performance of this method on the isomerization of alanine dipeptide and the structural transition of amyloid beta peptides will be presented. Combining this method with a planer constraint formulation in calculating minimum free energy pathways will also be illustrated. Using this method to develop effective force fields for describing complicated reactions will also be discussed.