State infidelity for error-suppressed (error supp.) and standard schemes are shown in blue and orange, respectively. On the x -axis, a state is reconstructed with either logical or physical tomography. The correction for the initial S Z measurement in Fig. 2a is implemented using either real-time feedforward (FF) or post-selection (PS). For the physical data points, the state from physical tomography is projected onto the logical subspace before computing the infidelity by fitting to ideal projectors. Error bars represent 1σ from bootstrapping. For all tomographic methods, the error-suppressed scheme achieves a lower state infidelity compared with the standard scheme. The unencoded magic state prepared directly on two physical qubits gives an average (avg.) infidelity across 28 qubit pairs as approximately 6.2 x 10 -2 (green dashed line) using 18 repetitions over a 24-h period with 10 5 shots per circuit. Of these, the best-performing pair yields a minimum (min.) infidelity of (2.354 ± 0.271) x 10 -2 (red solid line) found over all repetitions for all qubit pairs. In all cases, the error-suppressed scheme exceeds the fidelity of the best two-qubit unencoded magic state.
