Fig. 4: Error map of Hamiltonian implementation across the Sycamore processor.
From: Robustly learning the Hamiltonian dynamics of a superconducting quantum processor
Over the grid of 27 qubits, we randomly choose subsets of connected qubits and couplers of size N = 5. On each subset we implement two Hamiltonians with b = 0, 0.5 and run the identification algorithm. Two instances are shown in (a). For each subset, we compute the deviation of the identified Hamiltonian and initial map from their respective target and assign it to each qubit or coupler involved. Due to overlap of subsets, each qubit or coupler has been involved in at least five different choices of subsets. b, c Show the median deviation for the Hamiltonian and initial map implementations, respectively. d Shows the mean of the sign flips in the identified (diagonal ± 1) final map for each qubit.
