Neural network-assisted model of interfacial fluids with explicit coarse-grained molecular structures.

Area of Science:

• Computational chemistry
• Materials science
• Biophysics

Background:

• Interfacial fluids are complex, exhibiting nonlinear, multiphase, and multicomponent behaviors.
• Accurate coarse-grained (CG) models are challenging to develop due to intricate many-body interactions and interfacial phenomena.
• Existing models struggle to capture diverse density distributions and fluctuations at interfaces.

Purpose of the Study:

• To construct a robust coarse-grained (CG) model for interfacial fluids.
• To effectively capture many-body interactions and interfacial properties using advanced machine learning.
• To validate the model's efficacy across different CG mapping strategies and simulation settings.

Main Methods:

• Utilized advanced machine learning techniques, including force matching and diffusion probabilistic models.
• Developed a novel coarse-grained (CG) model for simulating interfacial fluid systems.
• Evaluated the model through simulations of water-air interfaces, bulk decane, and lipid membranes.

Main Results:

• The developed CG model accurately reproduces essential many-body and interfacial properties.
• The model demonstrates effectiveness across various CG mapping strategies.
• The simulations validated the model's capability in diverse interfacial fluid systems.

Conclusions:

• The advanced machine learning-based CG model is a robust tool for simulating interfacial fluids.
• This work validates the model's utility for multiscale simulations.
• The study provides a foundation for future advancements in simulating complex interfacial systems.