Hardware-efficient quantum annealing with error mitigation via classical shadow
Phys. Rev. A 113, 062430 – Published 10 June, 2026
DOI: https://doi.org/10.1103/stb4-wms8
Abstract
Quantum annealing (QA) is an efficient method for finding the ground-state energy of the problem Hamiltonian. However, in practical implementation, the system suffers from decoherence. Recently, “localized virtual purification” (LVP) was proposed to suppress decoherence in the context of noisy intermediate-scale quantum devices. We consider observables with spatially local support in the lattice. In this case, the LVP scheme requires evaluating the expectation value using the reduced density matrix of the subsystem where the observable acts and its surroundings. In this work, we propose a method to mitigate decoherence errors in QA using LVP. The key idea is to use the so-called classical shadow (CS) method to construct the reduced density matrix. Thanks to the use of the CS method, unlike the previous schemes to mitigate decoherence errors in QA, we do not need either two-qubit gates or midcircuit measurements, which means that our method is hardware efficient.
