Implicit Time Integration for Multiscale Molecular Dynamics Using Transcendental Padé Approximants.

Area of Science:

• Computational chemistry
• Biophysics
• Molecular modeling

Background:

• Molecular dynamics simulations face time step limitations due to diverse time scales.
• Multiscale factorization aims to overcome these limitations by coevolving coarse-grained and atom-resolved states.
• Developing stable time-marching schemes for multiscale dynamics is challenging due to the microstate-dependent nature of coarse-grained equations.

Purpose of the Study:

• To develop an accurate and stable implicit time integration scheme for multiscale molecular dynamics.
• To enable long-timescale simulations of large molecular systems with high fidelity.
• To address the challenge of closed-form equations for coarse-grained dynamics.

Main Methods:

• Utilized Padé approximants to capture stochastic and deterministic features of coarse-grained dynamics.
• Developed an implicit time integration scheme for coevolving coarse-grained and atom-resolved states.
• Applied Trotter factorization for multiscale factorization in molecular dynamics.

Main Results:

• The proposed scheme successfully simulates protein conformational changes and migration under external forces.
• Demonstrated acceleration of multiscale molecular dynamics simulations.
• Achieved high accuracy without loss of atomic precision.

Conclusions:

• The developed implicit time integration scheme effectively enhances multiscale molecular dynamics simulations.
• The method accelerates simulations of large systems over long durations without compromising accuracy.
• This approach avoids the need to conjecture the form of coarse-grained governing equations.