Cycle Benchmarking: The New Paradigm for Assessing All Relevant Errors and Error Correlations...

on YouTube ↗01h 05m

" summary

Cycle benchmarking reconstructs per-cycle Pauli error probabilities (a Pauli channel reconstruction) and shows concrete surprises: trapped-ion single-qubit gates have one-body X errors ~0.3% but correlated two-qubit errors near 0.1% (not quadratically suppressed), and an IBM CNOT with 10^-2 error induces phase and two-body errors on distant qubits at ~10^-1. A software layer (randomized compiling) turns coherent overrotation errors into stochastic ones, removing the square-root blowup versus diamond-norm bounds with ≲20 randomized compilations and often no extra gate depth; exponential decay fits flag strong non‑Markovianity.

AI-written summary · about

" chapters (15)

1. 0:00Cycle Benchmarking Learning Error Correlation for any Gate Combination of Interest
2. 1:31Complexity of the Error Problem Impact on quantum applications
3. 3:34Navigating the Error Problem
4. 5:44Errors in a channel...which channel? Common Inaive assumption: there exists an error channel
5. 7:42Cycle Benchmarking (CB)
6. 11:23CB Error Diagnostics Flow Chart
7. 23:49Pauli Error Eigenvalues via CB
8. 25:46Error Correlations with CB
9. 30:23Cross-talk Errors with CB
10. 36:57The Problem of Control Errors Norm based metrics |TVD, diamond norm behave similarly
11. 41:09Measuring Coherent Gate Errors
12. 42:46Cross-talk Errors (!) on IBM Q
13. 46:04Randomized Compiling Randomized Compiling: Twirling Noise during Universal Circuits
14. 52:45Benchmarking Pauli Error Rates
15. 58:00EB Error Diagnostics Flow Chart