Neural network encoded variational quantum algorithms

TL;DR

This paper introduces NN-encoded variational quantum algorithms that leverage neural networks to enhance the efficiency and scalability of VQAs on NISQ devices, demonstrated through ground state energy estimation of parameterized spin models.

Contribution

The paper presents a novel hybrid quantum-classical framework that uses neural networks to parameterize quantum circuits, significantly accelerating training and enabling handling of diverse problem inputs.

Findings

High-precision ground state energy estimation without fine-tuning.

Reduced training cost for parameterized Hamiltonians.

Active learning improves training efficiency while maintaining accuracy.

Abstract

We introduce a general framework called neural network (NN) encoded variational quantum algorithms (VQAs), or NN-VQA for short, to address the challenges of implementing VQAs on noisy intermediate-scale quantum (NISQ) computers. Specifically, NN-VQA feeds input (such as parameters of a Hamiltonian) from a given problem to a neural network and uses its outputs to parameterize an ansatz circuit for the standard VQA. Combining the strengths of NN and parameterized quantum circuits, NN-VQA can dramatically accelerate the training process of VQAs and handle a broad family of related problems with varying input parameters with the pre-trained NN. To concretely illustrate the merits of NN-VQA, we present results on NN-variational quantum eigensolver (VQE) for solving the ground state of parameterized XXZ spin models. Our results demonstrate that NN-VQE is able to estimate the ground-state energies of parameterized Hamiltonians with high precision without fine-tuning, and significantly reduce the overall training cost to estimate ground-state properties across the phases of XXZ Hamiltonian. We also employ an active-learning strategy to further increase the training efficiency while maintaining prediction accuracy. These encouraging results demonstrate that NN-VQAs offer a new hybrid quantum-classical paradigm to utilize NISQ resources for solving more realistic and challenging computational problems.