Quantum VQE Benchmark Dataset

Quantum Circuit Parameter Initialization Database for VQE Energy Solutions and GNN-based Machine LearningThis database contains data from 380 VQE experiments conducted on quantum chemistry calculations involving molecules such as H2, LiH, and BeH2. The focus of this database is to record optimal quantum circuit parameters and the VQE-solved energy for each molecule. This data can be used for parameter initialization in future quantum optimization tasks and for training Graph Neural Networks (GNNs) to predict optimal quantum circuit parameters more efficiently.Key FeaturesMolecule Type: Information about the molecule being studied:H2 (Hydrogen molecule)LiH (Lithium Hydride)BeH2 (Beryllium Hydride)Optimizers: The optimization algorithms used in the VQE process:COBYLA (Constrained Optimization BY Linear Approximations): A commonly used gradient-free optimizer.L-BFGS-B (Limited-memory Broyden-Fletcher-Goldfarb-Shanno with Box constraints): A quasi-Newton optimization algorithm that is efficient for larger problems with constraints.VQE-Solved Energy: The energy obtained from solving the molecule's Hamiltonian using VQE, recorded for each experiment.Quantum Circuit Parameters: The optimal quantum circuit parameters after the optimization process, stored as final_params and best_params.Energy Values: Includes initial energy, final energy, and the best energy found during the optimization.Optimization Process: Records details about the number of optimization steps and speedup achieved during the optimization process.Seed: The random seed used in the optimization for reproducibility.Additional Metadata: Other relevant details like molecule basis set, tapering, and the experimental timestamp.Data StructureThe main table in the database, runs, includes the following columns:id: Unique identifier for each run.ts: Timestamp of the experiment.molecule: The molecule used in the experiment (e.g., 'H2', 'LiH', 'BeH2').basis: The basis set used for the calculations (e.g., 'STO-3G').mapper: The qubit mapping strategy used (e.g., 'parity').tapering: Whether tapering was used in the optimization.reps: Number of repetitions for the experiment.optimizer: The optimizer used in the VQE algorithm (e.g., 'COBYLA', 'L-BFGS-B').seed: The random seed used for the experiment.n_qubits: The number of qubits used in the simulation.bond_length: The bond length for the molecule.energy_init: The initial energy before optimization.energy_final: The final energy after optimization.energy_total: The total energy calculated (sum of energies or other relevant aggregation).steps: Number of optimization steps taken.de_init: The initial derivative of energy.de_final: The final derivative of energy.speedup: The speedup factor compared to a baseline.note: Any additional remarks related to the run.final_params: The final optimized parameters for the quantum circuit.best_energy: The best energy obtained during the optimization.best_params: The parameters associated with the best energy found.vqe_energy: The energy solution obtained by the VQE method, specifically for each molecule. This is the key result of the VQE optimization process and is essential for validating the effectiveness of the quantum circuit parameters.Purpose and InnovationThis database is designed to record both the optimal quantum circuit parameters and the VQE-solved energy for each molecule (H2, LiH, BeH2). The data can then be leveraged to initialize parameters for future quantum optimizations and to train machine learning models, particularly Graph Neural Networks (GNNs). These models can predict optimal quantum circuit parameters more efficiently, reducing the computational cost and improving the precision of quantum simulations.