Error Mitigation via Verified Phase Estimation

The accumulation of noise in quantum computers is the dominant issue stymieing push algorithms beyond their classical counterparts. We do not expect to be able afford overhead required for error correction next decade, so meantime we must rely on low-cost, unscalable mitigation techniques bring computing its full potential. This paper presents a new technique based phase estimation that can also reduce errors expectation value (e.g., variational algorithms). general idea apply while effectively post-selecting system register starting state, which allows us catch and discard knock away from there. refer this as "verified estimation" (VPE) show it adapted function without use control qubits order simplify circuitry near-term implementations. Using VPE, demonstrate values numerical simulations intermediate scale circuits with multiple orders magnitude improvement over unmitigated at rates (even after accounting additional complexity estimation). Our results suggest VPE mitigate against any single might occur; i.e., estimated often O(p^2), where p probability an occurring point circuit. property, combined robustness sampling reveal practical mitigating experiments.