Coarse master equation from Bayesian analysis of replica molecular dynamics simulations

Area of Science:

• Computational Chemistry
• Statistical Mechanics
• Chemical Kinetics

Background:

• Molecular dynamics (MD) simulations generate trajectory data.
• Coarse master equations (CMEs) model system dynamics.
• Deriving accurate rate coefficients for CMEs is challenging.

Purpose of the Study:

• To develop a Bayesian inference framework for deriving CME rate coefficients from MD simulations.
• To apply the framework to model transitions in carbon nanotubes.
• To demonstrate the accurate computation of thermodynamic and kinetic properties.

Main Methods:

• Utilizing Bayesian inference to estimate CME rate coefficients.
• Employing propagators from multiple short simulation trajectories.
• Constructing a likelihood function from propagators for posterior distribution analysis.
• Discussing extensions to non-Markovian dynamics.

Main Results:

• Accurate rate coefficients for Markovian dynamics were derived.
• Thermodynamic properties (free energy surfaces) were computed.
• Kinetic properties (rate coefficients) were determined for carbon nanotube transitions.
• The framework successfully linked MD simulations to CME parameters.

Conclusions:

• Bayesian inference provides a robust method for CME parameterization from MD data.
• The approach enables accurate calculation of crucial thermodynamic and kinetic properties.
• This methodology enhances the predictive power of coarse-grained models in chemical systems.