Ridge-based bias potentials to accelerate molecular dynamics.

Area of Science:

• Computational Chemistry
• Molecular Dynamics Simulations
• Biophysics

Background:

• Hyperdynamics accelerates rare events in molecular dynamics by applying bias potentials.
• Current methods struggle with scaling due to diminishing reactant basin volumes with increasing system size.

Purpose of the Study:

• To develop novel bias potentials that mitigate dimensionality scaling problems in hyperdynamics.
• To improve the efficiency of rare event simulations in large systems.

Main Methods:

• Constructing bias potentials based on the distance to the reactant basin boundary.
• Quantifying this distance by following the minimum mode direction.
• Utilizing machine learning to derive analytic expressions for the boundary distance.

Main Results:

• Ridge-based bias potentials demonstrate qualitatively better scaling with dimensionality compared to existing methods.
• These new potentials achieve a greater filling fraction of the reactant state.

Conclusions:

• Ridge-based bias potentials offer a more scalable and effective approach for accelerating rare events in molecular dynamics.
• This work provides a significant advancement for simulating complex systems where rare events are crucial.