Stochastic gradient line Bayesian optimization for efficient noise-robust optimization of parameterized quantum circuits

Optimizing parameterized quantum circuits is a key routine in using near-term devices. However, the existing algorithms for such optimization require an excessive number of quantum-measurement shots estimating expectation values observables and repeating many iterations, whose cost has been critical obstacle practical use. We develop efficient alternative algorithm, stochastic gradient line Bayesian (SGLBO), to address this problem. SGLBO reduces measurement-shot by appropriate direction updating circuit parameters based on descent (SGD) further utilizing (BO) estimate optimal step size each iteration SGD. In addition, we formulate adaptive strategy introduce technique suffix averaging reduce effect statistical hardware noise. Our numerical simulation demonstrates that augmented with these techniques can drastically cost, improve accuracy, make noise-robust.