Multi-time quantum process tomography on a superconducting qubit

Current quantum technologies are at the cusp of becoming useful, but still face formidable obstacles such as noise. Noise severely limits the ability to scale quantum devices to the point that they would offer an advantage over classical devices. To understand the sources of noise it is necessary to fully characterise the quantum processes occurring across many time steps; only this would reveal any time-correlated noise called non-Markovian. Previous efforts have attempted such a characterisation but obtained only a limited reconstruction of such multi-time processes. In this work, we fully characterise a multi-time quantum process on superconducting hardware using in-house and cloud-based quantum processors. We achieve this by employing sequential measure-and-prepare operations combined with post-processing. Employing a recently developed formalism for multi-time processes, we detect general multi-time correlated noise. We also detect quantum correlated noise which demonstrates that part of the noise originates from quantum sources, such as physically nearby qubits on the chip.