Hamiltonian Monte Carlo with Constrained Molecular Dynamics as Gibbs Sampling

Area of Science:

• Computational chemistry
• Molecular dynamics simulations
• Statistical mechanics

Background:

• Constrained molecular dynamics (MD) offers advantages like larger time steps over fully flexible MD.
• However, achieving ergodic sampling from the Boltzmann distribution in constrained systems remains a challenge.
• Existing methods struggle to efficiently explore the conformational space of complex molecular systems.

Purpose of the Study:

• To develop an enhanced sampling method for molecular simulations using constrained dynamics.
• To improve the efficiency and accuracy of free energy calculations and conformational sampling.
• To address the limitations of traditional molecular dynamics in simulating complex systems like macrocycles.

Main Methods:

• Implementation of Hamiltonian Monte Carlo (HMC) using constrained molecular dynamics as a Gibbs sampling move.
• Leveraging recent generalizations of the equipartition principle and Fixman potential.
• Combining HMC based on fully flexible and torsional dynamics for hybrid sampling.

Main Results:

• Successfully reproduced free energy landscapes of simple model systems.
• Demonstrated enhanced sampling efficiency for macrocycles.
• Validated the effectiveness of the hybrid HMC approach in achieving ergodic sampling.

Conclusions:

• The proposed HMC method based on constrained dynamics provides an effective strategy for improving molecular simulation sampling.
• This approach overcomes challenges in achieving Boltzmann distribution sampling for constrained systems.
• The method shows significant potential for applications in computational chemistry, particularly for large and flexible molecules.