Optimization of Empirical Force Fields by Parameter Space Mapping: A Single-Step Perturbation Approach.

Area of Science:

• Computational Chemistry
• Molecular Modeling
• Physical Chemistry

Background:

• Accurate parametrization of atomic interaction functions is crucial for molecular simulations.
• Current methods often struggle with complex parameter spaces and combining diverse datasets.
• Understanding parameter relationships is key to developing robust and transferable force fields.

Purpose of the Study:

• To present a general method for parametrizing atomic interaction functions using parameter space mapping.
• To improve the understanding of parameter-data relationships and identify optimal parameter regions.
• To refine Lennard-Jones parameters for chlorine using experimental data from multiple compounds.

Main Methods:

• Parameter space mapping by analyzing surfaces of calculated vs. target data differences.
• Simultaneous refinement of chlorine 6-12 Lennard-Jones parameters against experimental data (solvation free enthalpies, density, heat of vaporization).
• Utilizing single-step perturbation for efficient calculation of solvation free enthalpies across various parameter combinations.

Main Results:

• Demonstrated a robust method for combining target data from multiple molecules.
• Successfully identified optimal parameter regions in the parameter space.
• Achieved refinement of chlorine parameters against a comprehensive set of experimental data for 10 aromatic-chloro compounds.

Conclusions:

• The presented method offers a more intuitive understanding of parameter optimization compared to local value or gradient-based approaches.
• This approach facilitates the creation of accurate and transferable force fields for molecular simulations.
• The technique is effective for refining parameters against diverse experimental data, improving predictive power.