Speaker
Description
Temporal fluctuations in the energy relaxation time ($T_1$) of superconducting qubits can occur on fast and irregular timescales. A quantitative understanding of these nonstationary relaxation dynamics is important for improving qubit stability and advancing fault-tolerant quantum processors. Here, we employ a new method for continuous monitoring of qubit relaxation based on an input–output hidden Markov model (IOHMM) applied to single-shot readout data. By explicitly modeling both the underlying state dynamics and the measurement process, the method enables robust extraction of the average decay and excitation rates. Simulations show that relaxation times can be extracted on timescales significantly shorter than conventional $T_1$ measurements, allowing for monitoring of relaxation dynamics two orders of magnitude faster than conventional approaches. Preliminary experimental results demonstrate the method's feasibility and its sensitivity to rapid fluctuations in qubit performance. As we extract the information directly from single-shot data, we do not require any additional classical logic beyond active reset, making the method easily integrable into existing setups.
