Quantum Circuit Learning on NISQ Hardware

Current quantum computers are small and error-prone systems for which the term noisy intermediate-scale quantum (NISQ) has become established. Since large scale, fault-tolerant quantum computers are not expected to be available in the near future, the task of finding NISQ suitable algorithms has received a lot of attention in recent years. The most prominent candidates in this context are variational quantum algorithms. Due to their hybrid quantum-classical architecture they require fewer qubits and quantum gates so that they can cope with the limitations of NISQ computers. An important class of variational quantum algorithms is the quantum circuit learning (QCL) framework. Consisting of a data encoding and a trainable, parametrized layer, these schemes implement a quantum model function that can be fitted to the problem at hand. For instance, in combination with the parameter shift rule to compute derivatives, they can be used to solve differential equations. QCL and related algorithms have been widely studied in the literature. However, numerical experiments are usually limited to simulators and results from real quantum computers are scarce. In this paper we close this gap by executing QCL circuits on a superconducting IBM quantum processor in conjunction with an analysis of the hardware errors. We show that exemplary QCL circuits with up to three qubits are executable on the IBM quantum computer. For this purpose, multiple functions are learned and an exemplary differential equation is solved on the quantum computer. Moreover, we present how the QCL framework can be used to learn different quantum model functions in parallel, which can be applied to solve coupled differential equations in an efficient way.