Multi-time quantum process tomography of a superconducting qubit

Non-Markovian noise poses a formidable challenge to the scalability of quantum devices, being both ubiquitous in current quantum hardware and notoriously difficult to characterise. This challenge arises from the need for a full reconstruction of a multi-time process, a task that has proven elusive in previous efforts. In this work, we achieve the milestone of complete tomographic characterisation of a multi-time quantum process on a superconducting qubit by employing sequential measure-and-prepare operations with an experimentally motivated post-processing technique, utilising both in-house and cloud-based superconducting quantum processors. Employing the process matrix formalism, we reveal intricate landscapes of non-Markovian noise and provide evidence that components of the noise originate from quantum sources. Our findings and techniques have significant implications for advancing error-mitigation strategies and enhancing the scalability of quantum devices.