Predicting Properties of Cyclohexene with Electronic Structure Methods through Adaptive Force Matching.

Area of Science:

• Computational chemistry
• Materials science
• Molecular dynamics

Background:

• Electronic structure calculations are vital for molecular properties but computationally expensive for large systems.
• Calculating finite-temperature properties requires ensemble averages, often limited by computational cost.

Purpose of the Study:

• To demonstrate the Adaptive Force Matching (AFM) method for predicting macroscopic molecular properties using only electronic structure information.
• To validate the accuracy of AFM by comparing predictions with experimental data for cyclohexene.

Main Methods:

• Fitting AFM models to reference forces obtained from MP2 and B3LYP-D3 electronic structure calculations.
• Applying AFM-based molecular dynamics simulations to predict various macroscopic properties of cyclohexene.

Main Results:

• AFM successfully predicted 16 distinct properties of cyclohexene with high accuracy compared to experimental values.
• Predictions from AFM models filled gaps in experimentally missing data.
• Separate models for neat and hydrated phases accurately predicted solubility, while a mixed-phase model was needed for interfacial tension.

Conclusions:

• AFM-based molecular dynamics accurately predicts diverse macroscopic properties directly from electronic structure data.
• The method bridges the gap between electronic structure calculations and macroscopic property prediction.
• AFM opens new avenues in computational materials science and accelerates the discovery of novel chemicals.