Optimisation Techniques for Variational Quantum Algorithms using Noisy Quantum Processing Hardware

Abstract

A method of executing a variational quantum algorithm, VQA, using a hybrid computer system comprising a quantum computer and a classical computer. An cost function is optimised with respect to a set of circuit parameters of a quantum circuit. Cost function value is dependent on output of the circuit when run on the quantum computer. The optimisation comprises parallel searches through the search space of possible sets of values of the circuit parameters. The optimisation is performed iteratively, each iteration includes updating a current parameter set for each of the searches based on the cost function in accordance with an optimisation procedure, the current parameter set of a search corresponding to a location of the search in the search space. Searches that stall are identified based on the gradient of the cost function at the location of the search meeting a stall criterion. A new parameter set is assigned to the stalled search, and the stalled search is continued in subsequent iterations from the different location based on the assigned parameter set. Upon termination of the searches, a set of circuit parameter values corresponding to an optimal cost function value found is identified and provided as output.

Images