Classifying and Characterizing Non‑Markovian Noise in Quantum Processors

Classify the relevant non‑Markovian error mechanisms in superconducting‑qubit quantum processors and develop practical methods to characterize these effects, including procedures that can detect and quantify deviations from exponential decay observed in randomized benchmarking experiments.

Background

The paper notes that most established calibration and benchmarking methods implicitly assume Markovian noise. However, real devices often exhibit temporal correlations and drift that are non‑Markovian, which manifest as deviations from exponential decay in randomized benchmarking experiments. A systematic taxonomy of such non‑Markovian effects and corresponding, scalable characterization procedures is lacking.

Resolving this gap is important both for accurate performance assessment and for informing error-suppression, mitigation, and decoding strategies that depend on realistic noise models in large-scale superconducting quantum processors.

References

While many of these effects show up as a deviation from an exponential decay in an RB experiment \citep{magesan2012characterizing,Wallman_2014,Ceasura2022}, the classification of relevant non-Markovian effects and methods for their characterization remains an important open problem.

— Enabling Technologies for Scalable Superconducting Quantum Computing (2512.15001 - Croot et al., 17 Dec 2025) in Section 'QPU Tune-up and Operation' → Subsection 'QPU Pulse-design, Tune-Up and Calibration' → Bullet list 'Some other outstanding challenges include:'