Rigorous force field optimization principles based on statistical distance minimization.

Area of Science:

• Statistical mechanics
• Computational chemistry
• Molecular modeling

Background:

• Developing accurate molecular force fields is crucial for simulating complex systems.
• Current methods like force matching have limitations in robustness and predictive accuracy.
• Integrating experimental data into model development is challenging.

Purpose of the Study:

• To define a rigorous basis for developing accurate and robust effective molecular force fields.
• To ensure compatibility with coarse-grained experimental data.
• To improve the predictive accuracy of molecular models.

Main Methods:

• Utilizing statistical distance to quantify distinguishability between statistical mechanical systems (model vs. target).
• Minimizing this statistical distance to achieve convergence of model properties to target properties.
• Developing and implementing new model optimization principles.

Main Results:

• Demonstrated minimization of statistical distance leads to convergence of static measurable properties.
• Achieved higher robustness and predictive accuracy compared to force matching and relative entropy minimization.
• Validated the approach through selected examples and efficient implementation.

Conclusions:

• The proposed method provides a rigorous foundation for effective molecular force field development.
• The approach enhances model accuracy and robustness, aligning with experimental data.
• Established connections to fundamental thermodynamic concepts like Gibbs-Bogoliubov inequality and thermodynamic length.