Quickly extracting fidelity decay rates in random circuit benchmarking experiments
Randomized benchmarking (RB) with structured circuits struggles to accurately estimate fidelity at short circuit depths because transient modes are not fully mixed, pushing signals below noise levels. The authors introduce a one-parameter family of estimators, including a 'trace filter function,' that suppresses these unwanted transient modes. This allows for interpretable fidelity decay estimates at short depths, matching standard RB's theoretical guarantees while increasing variance.