Journal of Chemical Theory and Computation, Год журнала: 2025, Номер unknown

Опубликована: Фев. 25, 2025

This work presents an efficient approach to optimizing force field parameters for sulfone molecules using a combination of genetic algorithms (GA) and Gaussian process regression (GPR). Sulfone-based electrolytes are significant interest in energy storage applications, where accurate modeling their structural transport properties is essential. Traditional parametrization methods often computationally expensive require extensive manual intervention. By integrating GA GPR, our active learning framework addresses these challenges by achieving optimized 12 iterations only 300 data points, significantly outperforming previous attempts requiring thousands parameters. We demonstrate the efficiency method through comparison with state-of-the-art techniques, including Bayesian Optimization. The GA-GPR was validated against experimental reference data, density, viscosity, diffusion coefficients, surface tension. results demonstrated excellent agreement between predictions values, widely used OPLS field. model accurately captured both bulk interfacial properties, effectively describing molecular mobility, caging effects, arrangements. Furthermore, transferability across different temperatures structures underscores its robustness versatility. Our study provides reliable transferable molecules, enhancing accuracy simulations. establishes strong foundation future machine learning-driven development, applicable complex systems.

Язык: Английский