Аннотация:Reliable estimation of expectation values on noisy quantum processors is a key prerequisite for practically useful quantum computations. We propose a hybrid tensor-network workflow for mitigating errors in Pauli-string observables that adaptively combines classical contraction with hardware execution. Whenever the circuit is classically contractable under a prescribed tensor-network budget, the observable is evaluated exactly without running on a quantum device. Otherwise, only the non-contractable residual fragment is executed, and the resulting measurement data are post-processed with scalable tensor-network error mitigation to reduce noise-induced bias.
