The resource cost of magic in a code block
A new theoretical result bounds the non-Clifford resource of a post-selected logical measurement by the resource required to produce it. The setting is a single logical qubit in one code block under an adaptive protocol that measures, feeds forward, and accepts. The proposed witness checks each accepted outcome against the free set of magic resource theory, rather than an averaged ensemble.
AI analysis — not reported by the source
What this could mean
- 0–2 yearsPlausible
This outcome-resolved witness could be incorporated into resource estimators for early fault-tolerant processors, allowing compilation tools to reject or re-route logical measurements that would carry more magic than the protocol can afford.
Because the bound operates per accepted outcome rather than on averages, it matches how adaptive circuits actually branch. The main gap is turning the bound into an efficiently computable test for specific codes and noise models, which is a named engineering precondition but not a fundamental blocker.
This is a brief. The day’s lead story carries the full analysis.