Benchmark of Schemes for Multiscale Molecular Dynamics Simulations.

Area of Science:

• Computational Chemistry
• Materials Science
• Statistical Mechanics

Background:

• Multiscale molecular dynamics (MD) simulations combine detailed and reduced representations for efficiency.
• Hybrid approaches integrating fine-grained (FG) and coarse-grained (CG) models are desirable for sampling enhancement.
• An adjustable mixing parameter is key for tuning FG/CG integration.

Purpose of the Study:

• To benchmark three algorithms for hybrid FG/CG MD simulations.
• To evaluate their performance in sampling enhancement.
• To assess the impact of different FG particle coupling methods.

Main Methods:

• Developed and benchmarked three FG/CG hybrid MD algorithms using a Lagrangian formalism.
• Employed three distinct methods for maintaining FG particle cohesion: addition of forces, mass scaling, and temperature scaling.
• Applied the benchmark to liquid hexadecane, evaluating configurational entropy of FG and CG subsystems.

Main Results:

• Temperature-scaling achieved a 3-fold sampling speedup with minimal deviation in FG properties.
• Addition-of-forces best preserved FG properties but offered limited speedup.
• Mass-scaling provided the highest speedup (5-fold) but resulted in the most significant deviation from FG properties.

Conclusions:

• The choice of FG/CG coupling method impacts the trade-off between simulation speedup and accuracy.
• Temperature scaling presents a promising approach for efficient sampling in multiscale simulations.
• Further research can optimize these hybrid methods for specific applications.