Time Recovery for a Complex Process Using Accelerated Dynamics.

Area of Science:

• Computational Chemistry
• Materials Science
• Statistical Mechanics

Background:

• The hyperdynamics (HD) method accelerates simulations by leveraging transition state theory, but requires pseudoequilibrium conditions (PEC) that are not always met.
• Conventional molecular dynamics (MD) lacks acceleration capabilities, while HD requires frequent recalibration of parameters as systems evolve through various trapped states.

Purpose of the Study:

• To develop a general, automatic scheme for switching between HD and conventional MD during simulations.
• To overcome limitations of the standard HD method, particularly the need for constant PEC assumptions and parameter recalibration.
• To enable efficient simulation of complex systems by combining the strengths of both HD and MD.

Main Methods:

• A hybrid HD-MD approach is proposed, featuring automatic switching between methods based on system energetics.
• A simplified bias function is introduced for straightforward, on-the-fly parameter setup.
• Criteria based on system energetic properties are used to determine HD initiation and termination.
• A method for assessing simulation quality is suggested.

Main Results:

• The hybrid HD-MD method demonstrated efficiency in simulations of a 2D model potential and nanoparticle coalescence.
• Relevant mechanistic properties of complex systems, such as nanoparticle coalescence (165 degrees of freedom), were successfully recovered.
• The automatic switching mechanism and simplified bias function facilitate robust and efficient accelerated simulations.

Conclusions:

• The hybrid HD-MD method provides a robust and efficient approach for accelerated simulations, overcoming limitations of standalone HD.
• This method enables the recovery of crucial mechanistic insights in complex systems that are computationally prohibitive with conventional MD.
• The automatic nature and simplified parameterization make the hybrid method broadly applicable for advanced molecular simulations.