Quantum computers are theoretically capable of tackling problems well outside the remit of existing classical computers. As the first generation of quantum computers are now reaching the point where they can answer such otherwise impossible questions, it is necessary to consider how the answers of these early devices can be confirmed correct. Certain problems, such as factoring, can be checked but not solved efficiently. The full power of quantum computing includes a wide range of problems for which this is not possible, instead demanding new techniques to test quantum computers.

Previous methods rely on extrapolating from the performance of individual components, which can miss effects on the non-local correlations that are present in quantum algorithms, or on techniques which can verify an output but require trust of certain quantum resources on top of a significant overhead in additional quantum resources. A new test has now been proposed by Samuele, Theodoros, and Animesh which can be used to make sure the quantum computer is working correctly without using excessive additional resources while still testing the entire quantum computer. Published in the New Journal of Physics (DOI:10.1088/1367-2630/ab4fd6), this protocol uses circuits which have the same form as the desired circuit but are formulated to give known outcomes. Based on the accuracy of these circuits they are able to place a statistical bound on how close the distribution the quantum computer gives is to the correct distribution.

A press release has been issued by the University of Warwick and the paper has been reported on Phys.org and ZDNet.