Gibbs–Duhem-informed neural networks for binary activity coefficient prediction
Jan G. Rittig, Kobi Felton, Alexei A. Lapkin, Alexander Mitsos
Abstract
Gibbs–Duhem-informed neural networks provide a flexible hybrid approach to predicting binary activity coefficients with both high accuracy and thermodynamic consistency.
Topics & Concepts
Binary numberConsistency (knowledge bases)Artificial neural networkActivity coefficientGibbs free energyComputer scienceThermodynamicsStatistical physicsMathematicsArtificial intelligencePhysicsChemistryPhysical chemistryArithmeticAqueous solutionMachine Learning in Materials ScienceComputational Drug Discovery MethodsAdvanced Thermodynamics and Statistical Mechanics