IBM Quantum devices are accessible via the cloud and allow users free access for running quantum computation and quantum dynamics experiments. I will present a detailed experimental characterization of the noise model of idle qubits on the 127-qubit devices, accounting for qubit spontaneous emission, finite temperature, dephasing, frequency drifts, charge-parity (quasiparticle) oscillations, crosstalk (“ZZ” coupling) and initialization and readout errors. Using this model, we study the decay dynamics of stabilizers characterizing graph states, and their protection using tailored dynamical-decoupling sequences suppressing both single-qubit and interaction terms [1]. We then employ this noise model to analyze how different approximations of the noise capture the performance of the five-qubit error correction code in recovering a decaying quantum memory logical qubit. We find that the commonly used single-qubit Pauli channel completely fails, and a two-qubit Pauli approximation is sensitive to the details of the noise, state, and decoder, accurate in many cases only at short timescales. We calculate the code pseudo-threshold emerging within this model and demonstrate how knowledge of the qubit parameters and connectivity can be used to design better decoders [2].

[1] L. Shirizly, G. Misguich, H. Landa, Dissipative Dynamics of Graph-State Stabilizers with Superconducting Qubits, Phys. Rev. Lett. 132, 010601 (2024).

[2] Z. Schwartzman-Nowik, L. Shirizly, H. Landa, Modeling error correction with Lindblad dynamics and approximate channels, arXiv:2402.16727.