Strategies for variational quantum compiling of a zero-phase beamsplitter on the Xanadu X8 processor

TL;DR

This paper develops and tests a cost function for variationally compiling a zero-phase beamsplitter on the Xanadu X8 processor, demonstrating feasibility despite hardware noise and providing methods for future large-scale CV quantum compiling.

Contribution

It introduces a faithful cost function tailored for the X8's constraints and shows its practical computation, advancing variational quantum compiling for continuous-variable systems.

Findings

Cost function exhibits optimum parameter resilience despite noise

Feasibility of variational compiling demonstrated on near-term hardware

Methods applicable to larger, more complex CV quantum problems

Abstract

In the context of variational compiling of a continuous-variable (CV) unitary operation, the architecture and parameter space of the Xanadu X8 processor constrain both the set of feasible compiling problems and the allowed cost functions. In this paper, we motivate a faithful cost function for variational compiling of a two-mode, continuous-variable beamsplitter gate with real matrix elements (i.e., a "zero-phase" beamsplitter) that complies with the constraints of the X8 processor. This cost function is then computed on the X8. Despite the noise in the processor, we find that the cost function exhibits optimum parameter resilience and, therefore, that this variational compiling problem is feasible on the X8. The intent of the paper is partly to report a proof-of-principle cost function calculation on near-term CV hardware, and partly to present methods that may be relevant for CV variational compiling problems in more complicated, large scale settings.