A graph neural network-state predictive information bottleneck (GNN-SPIB) approach for learning molecular thermodynamics and kinetics
Ziyue Zou, Dedi Wang, Pratyush Tiwary
Abstract
We present a graph-based differentiable representation learning method from atomic coordinates for enhanced sampling methods to learn both thermodynamic and kinetic properties of a system.
Topics & Concepts
BottleneckArtificial neural networkInformation bottleneck methodArtificial intelligenceKineticsComputer scienceGraphMachine learningChemistryThermodynamicsStatistical physicsPhysicsTheoretical computer scienceCluster analysisQuantum mechanicsEmbedded systemMachine Learning in Materials ScienceComputational Drug Discovery MethodsProtein Structure and Dynamics