Description
In this talk, I will give an overview of the latest published and unpublished results of my group on non-Markovian noise in superconducting qubits. Non-Markovian noise arises from unwanted interactions of our qubits with their environment. During a quantum circuit, these interactions causes errors at each step of the circuit. These errors can be correlated, resulting in non-Markovian (or correlated) noise. To fully study this correlated noise, we need to perform "multi-time process tomography", which is tomography of the executed circuit. The tomography protocol, just like in state or channel tomography, consists of a series of operations at each time step of the circuit. We executed such tomography on in-house processors at the University of Queensland and IBM's cloud processors and reported their correlated noise. We also developed a faster protocol using ancillary qubits and an AI model that does not require full tomographic data to estimate the correlated noise. Finally, we use this method to optimise the application of two X gates in the presence of correlated noise. In light of these results, I will discuss the use cases of these tools to improve a quantum processor's performance.
| I am the presenting author | Yes |
|---|