Vqezy Dataset Enables Variational Quantum Eigensolver Parameter Initialization across 12,110 Instances and Seven Tasks

Variational Quantum Eigensolvers represent a promising approach to solving complex problems with near-term quantum computers, but their effectiveness hinges on carefully chosen starting parameters. Chi Zhang, Mengxin Zheng, and Qian Lou from the University of Central Florida, alongside Hui Min Leung and Fan Chen from Indiana University, addressed a critical bottleneck in this field by creating a comprehensive, open-source dataset called VQEzy. This new resource overcomes the limitations of existing datasets, which are often small and restricted to specific problem types, by providing over 12,000 instances spanning multiple domains and complete optimisation pathways. VQEzy empowers researchers to develop and benchmark machine learning-based methods for initialising these algorithms, ultimately accelerating progress towards practical quantum computation and enabling more reliable results from noisy quantum hardware.

Achieved state-of-the-art performance, their progress has been limited by the lack of comprehensive datasets. Existing resources are typically restricted to a single domain and lack complete coverage of Hamiltonians, ansatz circuits, and optimisation trajectories. To overcome these limitations, scientists developed VQEzy, the first large-scale dataset for VQE parameter initialisation, comprising 12,110 instances spanning three major application domains, quantum many-body physics, quantum chemistry, and random benchmarking. The dataset is available at https://github. com/chizhang24/VQEzy, and will be continuously refined and expanded to facilitate advancements in machine learning-based initialization methods.

VQEzy Dataset For Parameter Initialization Studies

Scientists established VQEzy, a comprehensive large-scale dataset designed to facilitate advancements in Variational Quantum Eigensolver (VQE) parameter initialization. To construct VQEzy, researchers systematically generated problem Hamiltonians, selected appropriate ansatz circuits, and performed VQE optimization for each instance. The resulting dataset includes rich attributes for each instance, encompassing the problem Hamiltonian, detailed circuit specifications, and the optimized VQE parameter vector. For quantum many-body physics tasks, the team generated instances for the 1D Heisenberg XYZ model, the 1D Fermi-Hubbard model, and the 2D Transverse-Field Ising model, varying parameters like coupling constants and magnetic fields. Quantum chemistry tasks focused on the hydrogen molecule (H2), utilizing different basis sets and molecular geometries. This comprehensive approach enables researchers to explore diverse VQE configurations and develop more effective initialization strategies.

VQEzy Dataset Enables Large-Scale Parameter Initialisation

The research team has established VQEzy, a comprehensive large-scale dataset designed to facilitate advancements in Variational Quantum Eigensolver (VQE) parameter initialization. VQEzy comprises 12,110 instances spanning three major application areas, quantum many-body physics, quantum chemistry, and random VQE, and provides complete optimization trajectories for each instance. Within the quantum many-body physics domain, the team generated data for the one-dimensional Heisenberg XYZ model, the one-dimensional Fermi-Hubbard model, and the two-dimensional Transverse-Field Ising model. For quantum chemistry applications, the dataset includes configurations of H2, HeH+, and NH3, generated by varying bond lengths.

The random VQE domain contributed four-qubit Hamiltonians with randomly generated coefficients. The team employed t-distributed stochastic neighbor embedding (t-SNE) and multidimensional scaling (MDS) to visualize the optimized VQE parameters, revealing distinct patterns within each application area. This dataset, openly available to the research community, is designed to be continuously refined and expanded, establishing a foundation for future VQE research.

VQEzy Dataset Accelerates Algorithm Development

This paper introduces VQEzy, a publicly available dataset designed to accelerate research in the field of Variational Quantum Eigensolver (VQE) algorithms. The authors identified a lack of large, diverse datasets hindering progress in VQE research, particularly in areas like algorithm optimization and architecture design. VQEzy comprises 12,110 instances spanning three major VQE application domains, quantum chemistry, materials science, and physics. Researchers can use VQEzy to improve the efficiency and accuracy of VQE algorithms, develop algorithms that can generalize across different quantum systems, and evaluate the performance of different VQE algorithms. The authors plan to expand the dataset to include larger molecules and more complex systems, and data from real quantum hardware to improve practical relevance.