Fig. 4
From: Experimental quantum verification in the presence of temporally correlated noise
Demonstration of GST sensitivity to correlated error models. a, b Sensitivity of GST sequences to \(\hat \sigma _x\), \(\hat \sigma _z\) errors using the length of the sequence-dependent walk vector \({\vec{\boldsymbol V}}_{2D}\). GST sequence walks are shown as red crosses on a background colour scale illustrating the distribution over 106 RB walks and their average (yellow line). Here gates are defined as constituent Clifford operations of length τπ/2. c Flow diagram for the numerical analysis of the diamond norm estimation under correlated errors concurrent with gates G. d, e Results of the analysis for the standard gate set G I , G x , G y with the calculated diamond distance shown as solid lines (dashed lines) without (with) gauge optimisation on all graphs, and GST estimation depicted as symbols. Both overrotation errors on the G x , G y gates d and concurrent detuning errors e are studied. For overrotation errors the ideal rotation angle, \(\theta \to \left( {1 + \epsilon } \right)\theta\). f, g Analysis is repeated by extending the gate set to include −G x , −G y . In panels (d) and (f) which employ only overrotation errors, the calculated diamond distance for G I vanishes and we do not show the noise floor for visual clarity. h Experimental investigation of concurrent detuning \(\hat \sigma _z\) errors via a deliberately engineered detuning Δ. Markers indicate GST estimates from experimental data and solid lines represent analytical calculations performed within the pyGSTi toolkit
