Accelerating Variable Cell Shape Molecular Dynamics with a Position-Dependent Mass Matrix.

Area of Science:

• Computational Chemistry
• Materials Science
• Molecular Dynamics

Background:

• Molecular dynamics (MD) simulations are limited by small time steps due to fast molecular vibrations.
• Efficient simulation of molecular systems requires overcoming these time scale limitations.

Purpose of the Study:

• To enhance the time step in molecular dynamics simulations using Mass Tensor Molecular Dynamics (MTMD).
• To adapt MTMD for variable cell shape systems and develop a suitable symplectic integrator.

Main Methods:

• Augmenting the Hamiltonian with a position-dependent mass matrix based on an approximate Hessian.
• Applying MTMD to molecular crystals (N-(4-Methylbenzylidene)-4-methylaniline) and liquid water simulations at the density functional theory level.
• Developing a symplectic integrator for variable cell shape systems.

Main Results:

• Achieved a 4.4-fold increase in time step for simulating transitions between polymorphs of N-(4-Methylbenzylidene)-4-methylaniline.
• Obtained a 2.8-fold acceleration for simulating liquid water at the density functional theory level.
• Demonstrated the efficiency of the adapted MTMD method for molecular systems without bonding changes.

Conclusions:

• The adapted MTMD method effectively increases simulation time steps for molecular systems.
• This approach significantly accelerates the sampling of molecular configurations and transitions.
• MTMD offers a viable strategy for enhancing the efficiency of molecular dynamics simulations.