Application-Aware Benchmarking on NISQ Hardware using Expectation Value Fidelities

Abstract

We present a low-cost protocol for benchmarking applications on generic quantum hardware in the circuit model. Using families of Clifford circuits which mimic the application circuit structure, we are able to predict how measured expectation value fidelities scale with circuit depth. We consider the specific example of simulating a kicked-Ising model on superconducting hardware, showing our benchmark to be more accurate than predictions which use the gate error data obtained through randomized benchmarking. We also demonstrate how our work can be used to benchmark the performance and limitations of quantum error mitigation techniques. Our method is targeted at applications which have a natural decomposition in terms of Pauli rotations, but can be applied to any input circuit with this decomposition.