Using Diffusion Maps to Analyze Reaction Dynamics for a Hydrogen Combustion Benchmark Dataset.

Area of Science:

• Computational Chemistry
• Chemical Physics
• Reaction Dynamics

Background:

• Assessing the quality of collective variables (CVs) is crucial for analyzing molecular dynamics (MD) simulations.
• Diffusion maps offer a promising approach for CV assessment and reaction analysis, previously applied to classical MD.

Purpose of the Study:

• To evaluate the effectiveness of local diffusion maps for assessing CVs in ab initio MD simulations.
• To apply global diffusion maps for committor analysis in quantum MD simulations of hydrogen combustion.

Main Methods:

• Utilized local diffusion maps to analyze collective variables (CVs) from ab initio MD trajectories.
• Employed global diffusion maps to perform committor analysis on hydrogen combustion datasets.
• Leveraged normal modes along minimum energy paths for analysis.

Main Results:

• Demonstrated the efficacy of local diffusion maps for CV quality assessment in quantum ab initio MD.
• Successfully applied global diffusion maps to identify transition regions in hydrogen combustion reaction channels.
• The committor function derived from global diffusion maps accurately pinpoints critical areas in reaction pathways.

Conclusions:

• Local diffusion maps are effective for evaluating CVs in ab initio MD simulations.
• Global diffusion maps provide a robust method for committor analysis and transition region identification in complex chemical reactions.
• This study validates diffusion map approaches for quantum MD simulations, advancing reaction mechanism studies.