Generalized Markovian Noise as a Resource for Suppressing Markovian Errors in Superconducting Qubits: Part II (Experiment)

In superconducting qubits, non-Markovian noise environments, i.e. environments with non-negligible memory effect and temporal correlation, are often considered harmful to quantum computational tasks. Recent theoretical work has predicted that non-Markovian environments can also be used to reduce error rates. We experimentally test how non-Markovian noise can be used as a resource to suppress Markovian dephasing noise in a transmon qubit. Theory predicts that by injecting classical non-Markovian noise, the qubit dephasing time can be extended. The best dephasing noise suppression can be achieved by choosing an optimal memory kernel function of the non-Markovian noise, as predicted by our quantum trajectory simulations and the master equation description (see Part I of this talk). We discuss the implications of our results on correlations in non-Markovian noise and how they provide an extra degree of freedom in controlling and engineering qubit dissipation process. We further discuss how, comparing with other deterministic control sequences, our noise suppression protocol, being stochastic in nature, is potentially more flexible in design and more robust to control errors.