Quantum computing quantum Monte Carlo with hybrid tensor network toward electronic structure calculations of large-scale molecular and solid systems

Quantum computers are expected to solve the problems for quantum chemistry and materials science with higher accuracy than classical computers. Quantum computing quantum Monte Carlo (QC-QMC) is a method that can be combined with quantum algorithms such as variational quantum eigensolver (VQE) to obtain the ground state with fewer quantum resources and higher accuracy than either VQE or QMC alone. In this study, we propose an algorithm combining QC-QMC with hybrid tensor network (HTN) to extend the applicability of QC-QMC for the system beyond the size of a single quantum device, and we named the algorithm HTN+QMC. For HTN with the structure of a two-layer quantum-quantum tree tensor, the proposed algorithm for an $O(n^2)$-qubit reference wave function (trial wave function) in QMC can be performed by using only a $n$-qubit device excluding ancilla qubits. Full configuration interaction QMC is adopted as an example of QMC, and the proposed algorithm is applied to the Heisenberg chain model, the graphite-based Hubbard model, the hydrogen plane model, and MonoArylBiImidazole (MABI). The results show that the algorithm can achieve energy accuracy several orders of magnitude higher than either VQE or QMC alone. In addition, the energy accuracy of HTN+QMC is as same as QC-QMC when the system is appropriately decomposed. These results pave the way to electronic structure calculation for large systems with high accuracy on current quantum devices.
Comment: 27pages, 19 figures, 5 tables