Recurrent neural network wave functions
Mohamed Hibat-Allah, Martin Ganahl, Lauren E. Hayward Sierens, Roger G. Melko, Juan Carrasquilla
Abstract
This paper introduces a new class of computationally tractable wavefunctions, called recurrent neural network wavefunctions, based on recurrent neural network technology. The authors show that these wavefunctions outperform optimization methods for strongly correlated many-body systems with less variational parameters.
Topics & Concepts
Wave functionArtificial neural networkMathematicsPhysicsComputer scienceArtificial intelligenceQuantum mechanicsNeural Networks and Applications