Interpolating moving least-squares methods for fitting potential energy surfaces: applications to classical dynamics...

Area of Science:

• Computational Chemistry
• Chemical Dynamics
• Theoretical Chemistry

Background:

• Accurate potential energy surfaces (PES) are crucial for chemical dynamics simulations.
• Interpolating Moving Least-Squares (IMLS) methods offer a promising approach for PES generation.
• Previous IMLS methods require further refinement for enhanced accuracy and efficiency.

Purpose of the Study:

• To develop and validate advanced IMLS methods for constructing accurate PES.
• To investigate the impact of coordinate systems, cutoff radii, and data selection on IMLS fitting.
• To assess the reliability of IMLS-generated PES for chemical kinetics calculations.

Main Methods:

• Classical trajectory calculations were performed on higher-degree IMLS surfaces.
• Systematic investigation of fitting parameters: coordinate systems, cutoff radii, data point selection, and basis elements.
• Application and testing of the refined IMLS method using hydrogen peroxide (HOOH) as a model system.

Main Results:

• The study evaluated the influence of various fitting parameters on IMLS surface accuracy.
• Reaction rates for O-O bond breaking in HOOH were computed using classical trajectories on IMLS-fitted PES.
• The accuracy of IMLS-generated potentials for dynamics calculations was demonstrated.

Conclusions:

• The refined IMLS methods provide accurate potential energy surfaces for dynamics simulations.
• The choice of fitting parameters significantly impacts the quality of the generated PES.
• IMLS surfaces are reliable for calculating chemical kinetics properties, as shown for HOOH dissociation.