Performance Evaluation of Large-Scale VQE Simulation Incorporating Multiple Techniques

Quantum circuit simulation is crucial for advancing quantum algorithm research. While variational quantum eigensolver (VQE) is one of the promising candidates for practical algorithm on near-term quantum computers, simulating quantum circuits with large number of qubits remains challenging problems. Previously, two approaches have been proposed to perform high-speed state-vector simulation of the variational quantum algorithm.One approach utilizes a combination of MPI parallel and distributed processing, and the other utilizes the reduction technique of Hamiltonian terms.In this study, we used the Fujitsu Quantum Simulator, an HPC cluster system consisting of more than 1024 compute nodes. The node is equipped with the A64FX processor utilized for the supercomputer Fugaku.In addition to using the aforementioned approaches for the VQE algorithm, the parameterization operators on the quantum circuit were further reduced from those considered to be less influential. We conducted the simulation of ground-state energy calculations for a target molecule up to 36 qubits. These combined approaches enabled high-speed calculation while suppressing the deterioration of accuracy. We also investigated the impact of varying both the reduction threshold of parameterized operators and that of Hamiltonian terms cutoff on performance. These results demonstrate a step towards enabling larger-scale VQE simulations and accelerating the development of practical variational quantum algorithms.