Accelerating Quantum Eigensolver Algorithms With Machine Learning

2024-09-20

arXiv.org

AI Summary

To address the low accuracy in ground-state energy estimation and inefficient hyperparameter tuning of the Variational Quantum Eigensolver (VQE) on Noisy Intermediate-Scale Quantum (NISQ) devices, this work proposes a machine learning–guided hybrid optimization framework. The method integrates classical pretraining (up to 16 qubits), Bayesian hyperparameter search, and quantum-classical co-processing. Key contributions include: (i) the first application of an XGBoost regression model to predict critical VQE hyperparameters, substantially improving initialization quality; and (ii) a novel data augmentation paradigm leveraging Hamiltonian spectral features and symmetry constraints to mitigate scarcity of small-scale training data. Experiments on a 28-qubit system demonstrate a 0.13–0.15% reduction in ground-state energy estimation error relative to baseline methods, validating both the efficacy and scalability of ML-driven automated optimization for quantum algorithms.

Technology Category

• Machine Learning: Quantum Machine Learning
• Search and Optimization: Learning to Search
• Computer Vision: Learning & Optimization for CV

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications
• Economics, Online Markets and Human Computation: LLM based quality controls for crowd work

Abstract

In this paper, we explore accelerating Hamiltonian ground state energy calculation on NISQ devices. We suggest using search-based methods together with machine learning to accelerate quantum algorithms, exemplified in the Quantum Eigensolver use case. We trained two small models on classically mined data from systems with up to 16 qubits, using XGBoost's Python regressor. We evaluated our preliminary approach on 20-, 24- and 28-qubit systems by optimising the Eigensolver's hyperparameters. These models predict hyperparameter values, leading to a 0.13%-0.15% reduction in error when tested on 28-qubit systems. However, due to inconclusive results with 20- and 24-qubit systems, we suggest further examination of the training data based on Hamiltonian characteristics. In future work, we plan to train machine learning models to optimise other aspects or subroutines of quantum algorithm execution beyond its hyperparameters.

Problem

• Accelerate quantum eigensolver algorithms with machine learning
• Optimize hyperparameters for Hamiltonian ground state calculations
• Improve accuracy in NISQ device quantum computations

Innovation

• Machine learning accelerates quantum eigensolver algorithms
• XGBoost predicts hyperparameters for energy calculation
• Training on classical data improves qubit system accuracy