UC Berkeley

Quantum computers have the potential to revolutionize the computational power accessible to humanity. But because the scale and quality of current devices is limited, maximizing their usefulness requires that researchers make efficient use of the resources that they provide.

Motivated by this, we introduce in this work several techniques for efficient verification and operation of near-term quantum devices. We first develop and demonstrate a stochastic quantum compilation technique (STOQ) which produces approximate compilations of unitaries in terms of any arbitrary native instruction set. We then construct a novel verification protocol for quantum devices called randomized analog verification (RAV), which uses STOQ to generate verification sequences that ideally leave the system near a measurement basis state, which allows for efficient estimation of the success rate.

We first apply RAV to the context of quantum computers with continuously-parameterized native gate sets. We demonstrate both numerically and experimentally that RAV provides an efficiency advantage over cross-entropy benchmarking (XEB) when estimating an error rate experimentally. We then adapt the RAV protocol for verification of analog quantum simulators, and we demonstrate its sensitivity to various types of error both numerically and experimentally.

In addition, we discuss practical techniques for efficient operation of quantum computers. We develop and implement a surrogate-based optimization (SBO) technique for variational quantum algorithms, and we demonstrate that the technique has significant performance advantages over the most commonly-used optimization methods. Then, applying a more experimental lens, we describe work toward realistic simulation of various aspects of a trapped-ion experiment, such as experimental control sequences and calibration runs.

Taken together, the techniques introduced in this work are a collection of incremental steps toward the broader goal of increasing efficiency in near-term quantum computers and quantum simulators.