Robust characterization of average gate set noise: Filtered randomized benchmarking and cross-talk tomography

Description

As quantum processors scale, robust and scalable tools to characterize gate-level noise become indispensable.
Randomized benchmarking (RB) has emerged as a SPAM-robust, bottom-up approach: it yields a single, meaningful number — the average gate fidelity — to characterize a gate set.
At the same time, RB can in principle be extended into a full tomographic framework (RB tomography), but practical implementations have long been hindered by the need for sequence inversion, explicit gate interleaving, and an infeasible amount of shot.
In this talk, I present two applications that explore both ends of this spectrum. First, I show how filtered randomized benchmarking (filtered RB) — a variant of RB where the sequence inversion is performed in classical post-processing — can be adapted to benchmark analog quantum simulators with bosonic and fermionic dynamics. This yields robust fidelity benchmarks for systems governed by particle-number-conserving dynamics.
Second, I demonstrate the practical application of RB tomography on digital platforms: using only one random sequence experiment, we reconstruct marginals of the average noise channel.
This allows us to learn classically correlated quantum noise, including cross-talk, from experimental data.
These protocols are lightweight, scalable, and compatible with current hardware.
I will present both numerical simulations and experimental results, including what may be the first direct observation of average gate-set noise channels reconstructed from real superconducting quantum device data.

Period	23 Sept 2025
Event title	Assessing Performance of Quantum Computers (APQC) 2025
Event type	Conference
Location	Estes Park (Colorado), United States, ColoradoShow on map
Degree of Recognition	International