Gibbs–Helmholtz graph neural network: capturing the temperature dependency of activity coefficients at infinite dilution
Edgar Iván Sánchez Medina, Steffen Linke, Martin Stoll, Kai Sundmacher
Abstract
A hybrid model that combines the Gibbs–Helmholtz equation with Graph Neural Networks for predicting limiting activity coefficients.
Topics & Concepts
DilutionExtrapolationActivity coefficientHelmholtz free energyThermodynamicsArtificial neural networkMathematicsSeries (stratigraphy)ChemistryBiological systemAlgorithmComputer scienceArtificial intelligencePhysicsPhysical chemistryMathematical analysisAqueous solutionBiologyPaleontologyMachine Learning in Materials ScienceComputational Drug Discovery MethodsProcess Optimization and Integration