Machine Learning Wavefunction

Abstract

This chapter introduces the main ideas and the most important methods for representing the electronic wavefunction through machine learning models. The wavefunction of a N-electron system is an incredibly complicated mathematical object, and models thereof require enough flexibility to properly describe the complex interactions between the particles, but at the same time a sufficiently compact representation to be useful in practice. Machine learning techniques offer an ideal mathematical framework to satisfy these requirements, and provide algorithms for their optimization in both supervised and unsupervised fashions. In this chapter, various examples of machine learning wavefunctions are presented and their strengths and weaknesses with respect to traditional quantum chemical approaches are discussed; first in theory, and then in practice with two case studies.

To be published in the upcoming book "Quantum Chemistry in the Age of Machine Learning", edited by P. Dral

References in corpus (9)

• Restricted-Boltzmann-Machine Learning for Solving Strongly Correlated Quantum Systems
• Inhomogeneous backflow transformations in quantum Monte Carlo calculations
• Weak binding between two aromatic rings: feeling the van der Waals attraction by quantum Monte Carlo methods
• A deep neural network for molecular wave functions in quasi-atomic minimal basis representation
• Unbiased Monte Carlo Cluster Updates with Autoregressive Neural Networks
• Determinant-free fermionic wave function using feed-forward neural networks
• Boltzmann machines as two-dimensional tensor networks
• On Representing (Anti)Symmetric Functions
• A Bayesian Inference Framework for Compression and Prediction of Quantum States

Cited by in corpus (1)

• Ab-initio quantum chemistry with neural-network wavefunctions