Blake Johnson

This talk will explore recent advances in dynamic circuits, with relevance to both current quantum systems as well as future fault-tolerant architectures. I will revisit key lessons learned from deploying dynamic circuit capabilities on IBM Quantum since 2022, as previously discussed at AQC 2024. I will particularly address challenges encountered when users apply adaptive techniques to wide circuits, especially where control flow is localized to subsets of qubits. I will show how we have addressed these problems with an upgraded execution engine that significantly improves scalability, enabling dynamic circuits to operate at much larger circuit sizes.

The second part of my talk will focus on real-time decoding of quantum low-density parity-check codes (qLDPC). Real-time decoders are a required element to elevate dynamic execution capability to the logical level, because they participate in the process of logical measurement. Earlier this year, we introduced a new algorithm, Relay-BP [arXiv:2506.01779], that offers improved accuracy over prior two-stage decoding pipelines while also being more amenable to hardware acceleration. We have now realized an FPGA implementation of a Relay-BP decoder, capable of split X/Z processing of syndromes from the [[144, 12, 12]] bivariate bicycle code [arXiv:2510.21600]. This decoder is exceptionally fast, completing a belief propagation iteration in just 24ns, and is expected to process full 12-cycle windows in less than 480ns when physical error rates are less than 10^-3.