Bayesian vs Frequentist Statistical Model for Randomized Benchmarking

004_demo.py

# curve fit
popt, pcov = curve_fit(f = gsp, xdata = x,
                       ydata = np.mean(counts, axis = 0) / shots,
                       sigma = np.std(counts, axis = 0) / shots / np.sqrt(num_samples),
                       absolute_sigma = True,
                       bounds = ([scale - .1, .9, .9, (1. - scale) - .1],
                                 [scale + .1, 1., 1., (1. - scale) + .1]))

# statistical error on pararameters
perr = np.sqrt(np.diag(pcov))

# estimated error of the interleaved gate EPC
epc_est = scale*(1. - popt[2])

# statistical error on EPC
epc_est_err = scale*perr[2]