Fig. 2: Time-resolved benchmarking & tomography on simulated data.
From: Detecting and tracking drift in quantum information processors
a, b Time-resolved RB on simulated data for gates with time-dependent phase errors. a Inset: the simulated phase error over time. Main plot: the true, time-dependent RB error rate (r) versus time (grey line) and a time-resolved estimate obtained by applying our techniques to simulated data (black line). b Instantaneous average-over-circuits (points) and per-circuit (distributions) success probabilities at each circuit length, estimated by applying our spectral analysis techniques to the simulated time-series data, and fits to an exponential (curves), for the three times denoted by the vertical lines in a. Each instantaneous estimate of r, shown in a, is a rescaling of the decay rate of the exponential fit at that time. c, d Time-resolved GST on simulated data, for three gates Gi, Gx, and Gy that are subject to time-dependent coherent errors around the \(\hat{z}\), \(\hat{x}\), and \(\hat{y}\) axes, respectively, by angles θi, θx, and θy. The estimates of these rotation angles (denoted \({\hat{\theta }}_{{\mathsf{i}}}\), \({\hat{\theta }}_{{{x}}}\) and \({\hat{\theta }}_{{{y}}}\)) track the true values closely. The shaded areas are 2σ (~95%) confidence regions.
