Exploring optimization strategies for improving explicit water models: Rigid n-point model and polarizable model...

Area of Science:

• Computational chemistry
• Molecular modeling
• Physical chemistry

Background:

• Rigid n-point water models are standard in atomistic simulations but have accuracy limitations.
• Improving accuracy often involves increasing point charges or adding electronic polarizability, but this increases computational cost.
• Existing strategies offer modest accuracy gains relative to their computational expense.

Purpose of the Study:

• To guide the development of more accurate and efficient water models.
• To explore the accuracy limits of "electrostatically globally optimal" n-point water models.
• To evaluate strategies for incorporating electronic polarizability into water models.

Main Methods:

• Developed "electrostatically globally optimal" n-point water models by optimizing charge distributions against reference multipole moments.
• Optimized models to reproduce the water dimer total dipole moment.
• Evaluated model accuracy by comparing reproduced water dimer geometry to ab initio references.
• Investigated the impact of adding electronic polarizability (Drude particle) to pre-optimized rigid models.
• Explored a coupled optimization strategy for rigid base parameters and polarizability.

Main Results:

• Global optimization of charge distribution alone yields high accuracy: n=4 and n=5 models achieve water dimer geometries within 50 of ab initio references.
• Accuracy gains are substantial from n=3 to n=4 but marginal from n=4 to n=5.
• Adding polarizability to a standard rigid model (n=3) provides minimal accuracy improvement.
• A 3-point polarizable model with globally optimized rigid base parameters significantly outperforms even optimal 5-point rigid models.

Conclusions:

• Global optimization of charge distribution is a highly effective strategy for developing accurate rigid water models.
• Standard methods of adding polarizability to unoptimized rigid models are insufficient.
• Coupled global optimization of rigid base parameters and polarizability offers a promising route to highly accurate, efficient polarizable water models.
• Future development should focus on 3- and 4-point polarizable models utilizing coupled global optimization.