Adaptive Neural Quantum States: A Recurrent Neural Network Perspective

TL;DR

This paper introduces an adaptive recurrent neural network approach to optimize neural-network quantum states, significantly reducing computational costs and improving the accuracy of ground state calculations in quantum many-body systems.

Contribution

It presents a novel adaptive scheme for training neural-network quantum states using RNNs, enabling efficient scaling and resource optimization in large-scale quantum simulations.

Findings

Reduces training fluctuations and improves variational energy estimates.

Decreases computational cost by training small RNNs and reusing them for larger models.

Enhances scalability of neural-network quantum state simulations.

Abstract

Neural-network quantum states (NQS) are powerful neural-network ans\"atzes that have emerged as promising tools for studying quantum many-body physics through the lens of the variational principle. These architectures are known to be systematically improvable by increasing the number of parameters. Here we demonstrate an Adaptive scheme to optimize NQSs, through the example of recurrent neural networks (RNN), using a fraction of the computation cost while reducing training fluctuations and improving the quality of variational calculations targeting ground states of prototypical models in one- and two-spatial dimensions. This Adaptive technique reduces the computational cost through training small RNNs and reusing them to initialize larger RNNs. This work opens up the possibility for optimizing graphical processing unit (GPU) resources deployed in large-scale NQS simulations.