Accurate Free Energies for Complex Condensed-Phase Reactions Using an Artificial Neural Network Corrected DFTB/MM Methodology

This paper presents a methodology combining density functional tight binding (DFTB) with molecular mechanics (MM) and artificial neural networks to accurately calculate free energies for complex condensed-phase reactions. The neural network corrections enable DFTB/MM simulations to achieve accuracy comparable to higher-level quantum mechanical methods at a fraction of the computational cost.

Journal of Chemical Theory and Computation, 18(2)