Stochastic search for approximate compilation of unitaries

TL;DR

STOQ is a stochastic search method for approximate quantum unitary compilation that offers a different approach from traditional techniques, with comparable runtime and potential applications in device characterization.

Contribution

This paper introduces STOQ, a novel stochastic search protocol for approximate unitary compilation, expanding the toolkit beyond product-formula methods.

Findings

STOQ produces less accurate but similarly fast compilations compared to product-formula techniques.

It can generate approximate compilations for random unitaries and shallow circuits.

STOQ's diverse paths in state space may benefit quantum device characterization.

Abstract

Compilation of unitaries into a sequence of physical quantum gates is a critical prerequisite for execution of quantum algorithms. This work introduces STOQ, a stochastic search protocol for approximate unitary compilation into a sequence of gates from an arbitrary gate alphabet. We demonstrate STOQ by comparing its performance to existing product-formula compilation techniques for time-evolution unitaries on system sizes up to eight qubits. The compilations generated by STOQ are less accurate than those from product-formula techniques, but they are similar in runtime and traverse significantly different paths in state space. We also use STOQ to generate compilations of randomly-generated unitaries, and we observe its ability to generate approximately-equivalent compilations of unitaries corresponding to shallow random circuits. Finally, we discuss the applicability of STOQ to tasks such as characterization of near-term quantum devices.