Repeated stabilizer measurements in a superconducting distance-3 surface code

We show the calibration and benchmarking of the weight-2 and weight-4 stabilizer measurements of the distance-3 surface code with a 17-transmon processor. We compare measured performance to a density matrix simulation considering various error models with input parameters extracted from experiment. We then initialize a logical state and assess the defect rates for multiple rounds of X- and Z-type stabilizer measurements. We also compare the defect error rates when running stabilizer measurements individually and simultaneously to evidence the impact of crosstalk and leakage errors.