Reweighting non-equilibrium steady-state dynamics along collective variables

Area of Science:

• Computational Chemistry
• Statistical Mechanics
• Molecular Dynamics

Background:

• Computer simulations generate microscopic trajectories for complex systems.
• Dynamical reweighting methods are crucial for analyzing simulation data.
• Existing methods face challenges with high-dimensional systems.

Purpose of the Study:

• To develop an efficient dynamical reweighting method for large, complex systems.
• To reduce the dimensionality of configurational space using collective variables (CVs).
• To apply the Maximum Caliber (MaxCal) approach with CVs for enhanced trajectory analysis.

Main Methods:

• Introduced a Maximum Caliber (MaxCal) approach for dynamical reweighting.
• Mapped trajectories to a Markovian description on configurational coordinates.
• Reduced dimensionality by defining collective variables (CVs) and local entropy productions.
• Used entropy production as a constraint for non-equilibrium steady states in CV space.

Main Results:

• The CV-based MaxCal approach successfully reweighted trajectory probabilities.
• Demonstrated the method's applicability to systems of increasing dimensionality.
• Validated the approach on a 2D potential and a coarse-grained peptide model.
• Showed entropy production is a suitable constraint for non-equilibrium steady states in CV space.

Conclusions:

• The CV-based MaxCal approach significantly expands dynamical reweighting capabilities for larger systems.
• Enables analysis of both static and dynamical properties across a wide range of driving forces.
• Offers a computationally feasible method for complex molecular simulations.