Speaker
Description
Error mitigation is a key tool for improving the performance of near-term quantum devices, but its efficacy depends strongly on the structure of the underlying noise [1]. In realistic systems, errors can exhibit temporal and spatial correlations, which determine which suppression strategies are effective. Virtual distillation (VD) is a promising approach, where exponential error suppression in the estimation of observable expectation values is achieved by increasing the number of state copies [2,3,4]. Like many mitigation techniques, however, it relies on the assumption that noise acts independently across copies.
Here, we combine error mitigation with a direct probe of this assumption. We demonstrate VD for the first time on a trapped-ion platform and introduce a scheme based on randomised benchmarking using parallel (“k-copy”) sequences that isolates correlated error components with minimal experimental overhead. Applying this method before and after VD allows us to study the effect of correlated errors on its performance. Together, these results provide a scalable route to implementing, diagnosing, and validating error mitigation in the presence of structured noise.
[1] Z. Cai, et al., Rev. Mod. Phys. 95, 045005 (2023).
[2] B. Koczor, Phys. Rev. X 11, 031057 (2021).
[3] W. J. Huggins, et al., Phys. Rev. X 11, 041036 (2021).
[4] T. Araki, J. F. Goodwin, and B. Koczor, Phys. Rev. A 112, 042619 (2025).
| Academic level | PhD student |
|---|