PSO-Assisted Development of New Transferable Coarse-Grained Water Models

Area of Science:

• Computational chemistry
• Materials science
• Physical chemistry

Background:

• Developing accurate and efficient coarse-grained (CG) models is crucial for simulating large-scale molecular systems.
• Existing CG water models often face trade-offs between accuracy and computational cost.

Purpose of the Study:

• To develop novel CG water models with enhanced accuracy and computational efficiency.
• To optimize force-field parameters using a combination of particle swarm optimization (PSO) and gradient descent.
• To evaluate the predictive capabilities of these models for various macroscopic properties of water.

Main Methods:

• Employed a two-to-one mapping scheme to create 1-, 2-, and 3-site CG water models.
• Utilized coupled PSO and gradient descent for force-field parameter optimization.
• Performed coarse-grained molecular dynamics (CG MD) simulations to assess model performance.
• Investigated mixture simulations combining different CG water models.

Main Results:

• Optimized CG models accurately reproduce key water properties like density, self-diffusion coefficient, and dielectric constant at 300 K.
• Models demonstrate excellent predictive accuracy for surface tension, heat of vaporization, hydration free energy, and isothermal compressibility.
• The 1-site model offers substantial computational speedup (3-4.5x) over 2- and 3-site models.
• Mixture simulations of different CG water models also yield accurate predictions.
• Developed new CG benzene models and confirmed the water models' solvation capacity.

Conclusions:

• The developed CG water models provide a robust and efficient approach for simulating water.
• The optimization strategy effectively balances accuracy and computational cost.
• These models show promise for studying complex systems involving water and hydrophobic solutes like benzene.