You know how modern CPUs have debug registers — tiny hardware hooks that let you set breakpoints and watchpoints without recompiling or slowing down the main program? Imagine trying to do that on a machine where every bit is actually spread across dozens of physical bits for error protection, and touching any one of them wrong collapses the whole computation. That's the problem SynDebug solves for fault-tolerant quantum computers. The committed claim: SynDebug is the first systems-level debugging framework that operates directly within encoded fault-tolerant quantum computing (FTQC) infrastructure. Prior quantum debugging tools work at the logical circuit level — they assume you're looking at clean qubits, ignoring the massive QEC (quantum error correction) machinery that keeps those qubits alive. SynDebug's assertion primitives live inside that machinery, which is where the bugs actually hide. The architecture is assertion-based: lightweight checks injected into the QEC pipeline that verify properties of the encoded quantum state (state preparation correctness, entanglement validity) without requiring additional logical qubits — no extra "surface area" in the surface code lattice. This is the key design choice. Adding logical qubits for debugging would blow up the physical resource budget; SynDebug avoids it entirely by piggybacking on existing syndrome extraction rounds. The framework is evaluated across multiple QEC code families and code distances, on workloads including state preparation, entanglement validation, and Shor-style factoring programs. The headline numbers are striking: 1.8% average additional circuit depth, 0% in most cases, and no measurable increase in logical error rate. That last point matters enormously. Any debugging tool that degrades error correction defeats its own purpose — you'd be injecting bugs while hunting them. SynDebug threads this needle by operating within the temporal gaps already present in syndrome extraction cycles. On the ladder: there is no direct prior art to beat here. Existing quantum debuggers (Huang et al.'s assertion-based approaches, various circuit-level tools) don't target FTQC execution at all. The comparison isn't "we're faster than X" — it's "X doesn't exist at this layer." The relevant comparison is against running without debugging at all, and SynDebug's overhead relative to that null baseline is negligible. Integrity is simulation-only, which is honest and appropriate — nobody has a fault-tolerant quantum computer large enough to run Shor's algorithm yet. The authors evaluate across multiple code families (surface codes and others), multiple code distances, and multiple workload types, which guards against cherry-picking. The ATC 2026 acceptance provides peer review signal. Still, every result here lives inside a simulator; hardware noise profiles could change the overhead story. The successor question is obvious: run this on real hardware. The authors didn't because real FTQC hardware at the scale needed (hundreds to thousands of physical qubits per logical qubit, running actual programs) doesn't exist yet. This is compute constraint, not evasion. The honest next experiment is deploying SynDebug on the first generation of real FTQC devices — IBM's or Google's planned machines in the 2026-2028 window — and checking whether the 1.8% overhead holds when real noise, real decoder latency, and real timing constraints enter the picture.
