arXiv · 2611.21906
Probing the real-time decoding accuracy bounds in trapped ion architectures
Abstract
Fault tolerant quantum computers will rely on real-time decoders that process syndromes faster than the QEC cycle duration. The millisecond cycle times offered by trapped ion platforms relax low-latency decoding demands, but simulations and experiments have not yet exploited this time budget to reach maximum logical error suppression. We show how a hybrid sliding window decoding scheme composed of a lower-accuracy and a higher-accuracy decoder can yield substantially improved logical error rates, while preserving low latency. All windows are handled by the lower-accuracy decoder, with some windows being infrequently switched to the higher-accuracy decoder. We extract soft-output information from the lower-accuracy decoder efficiently for any QLDPC code, which allows us to identify decoding windows of low confidence. We simulate memory experiments of bivariate bicycle (BB) codes, using trapped-ion informed noise models. Depending on the physical error rate and BB code, we find up to 3-4 orders of magnitude logical error rate reduction compared to the $\texttt{BP-LSD}$ decoder, and up to 2-3 orders of magnitude lower logical error rate compared to the $\texttt{Beamsearch}$ decoder. By leveraging available classical processing time, our decoding protocol significantly boosts decoding accuracy, ultimately reducing the demands on increasing code size.
Explore related subjects
Keep this discovery
Explore connections, maps & timelines
Arianna Meinking, Evangelia Takou, Kenneth R. Brown. 2026-11-16. Probing the real-time decoding accuracy bounds in trapped ion architectures. https://arxiv.science/abs/2611.21906
Citations retain the original authors and DOI and include this site’s record URL. If this discovery service contributed to your research, you can also cite arXiv Science by Indiciae: BibTeX · RIS.