Computational Implementation of Nudged Elastic Band, Rigid Rotation, and Corresponding Force Optimization

Area of Science:

• Computational Chemistry
• Chemical Physics

Background:

• The nudged elastic band (NEB) algorithm is widely used for calculating chemical transition states.
• Existing literature lacks comprehensive guidance on implementing NEB optimization methods.

Purpose of the Study:

• To provide detailed implementation guidance for various gradient descent algorithms within the NEB framework.
• To introduce and validate a novel, accelerated backtracking line search method combined with partial Procrustes superimposition.

Main Methods:

• Implementation of six gradient descent algorithms: steepest descent, quick-min Verlet, FIRE, conjugate gradient, BFGS, and LBFGS.
• Development and integration of an accelerated backtracking line search with partial Procrustes superimposition.
• Benchmark calculations for chemical system transition states.

Main Results:

• Successful implementation and validation of multiple optimization methods for NEB.
• Demonstrated performance improvements with the new accelerated line search and superimposition method.
• Comparative analysis against the established Atomic Simulation Environment (ASE) codebase.

Conclusions:

• The study provides essential practical guidance for NEB method implementation in computational chemistry.
• The novel line search method offers enhanced efficiency for NEB calculations.
• The findings contribute to more robust and accurate transition state calculations in chemical systems.