A fidelity time-utility function for the scheduling of quantum applications

Efficient use of limited quantum resources will be fundamental in improving the performance of quantum technologies. One way to improve efficiency is to allow quantum computers to multitask, which is especially useful in the setting of a network where idle times are unavoidable due to the exchange of classical messages. While software tools are being developed with the capability of interweaving multiple applications on the same quantum processor, it is still an open question how one should schedule the execution of component tasks in order to maximise the total performance. A key aspect of this problem is how to efficiently track and predict the accumulation of time-dependent noise, in order to ensure that decoherence is sufficiently low such that applications may be executed successfully, while also enabling multitasking. In our work we define the fidelity time-utility function (FTUF), which connects the projected performance of a quantum application with the time at which it is executed. We then make use of a technique from randomized benchmarking to propose a FTUF that is application-agnostic and fast to compute, allowing real-time comparison of candidate schedules.