Error mitigation with Clifford quantum-circuit data

Achieving near-term quantum advantage will require accurate estimation of observables despite significant hardware noise. For this purpose, we propose a novel, scalable error-mitigation method that applies to gate-based computers. The generates training data {Xinoisy,Xiexact} via circuits composed largely Clifford gates, which can be efficiently simulated classically, where xmlns:mml="http://www.w3.org/1998/Math/MathML">Xinoisy and class="MJX-TeXAtom-ORD">exact are noisy noiseless respectively. Fitting linear ansatz then allows for the prediction noise-free arbitrary circuits. We analyze performance our versus number qubits, circuit depth, non-Clifford gates. obtain an order-of-magnitude error reduction ground-state energy problem on 16 qubits in IBMQ computer 64-qubit simulator.