The Threshold Theorem in Watts: Fault Tolerance as a Question About Objective Probability

Abstract:In 2011 Hagar & Sergioli proposed a new interpretation of objective probability in deterministic physics. On it, the probability of a physical state supervenes on the resources, energy over time, required to realize it from a given state, relative to the resources available (arXiv:1101.3521). The motivation for that paper was an objective alternative to QBism, the view that sees quantum probabilities as subjective degrees of belief. Here I apply this interpretation to a more practical subject matter: fault-tolerant quantum computing (FTQC). Under the resource-bounded measure Hagar & Sergioli proposed, the threshold theorem becomes a claim about classification: it asserts that error-corrected logical states belong to the class of relatively cheaply realizable states, whose probability remains near 1 as the machine grows. The theorem originally derived this claim from a resource inventory that was partial, and left out four resources consumed by error correction: calibration of a drifting device, decoding within the correction cycle, coherence as a finite time budget, and entropy flush through fresh ancillas. These entered the original derivation at zero price. Here I translate the feasibility of FTQC into a measurable quantity, watts per decade of suppressed logical error (a decade, in the engineer's usage, being one factor of ten in the error rate). I then show that the published record already contains its first data points for this translation, and I state the two measurements that would settle the question empirically. The three-decade debate on FTQC, conducted so far as an exchange about noise-model assumptions, turns out to be, under this interpretation, a quantitative dispute about a single object: the resource-bounded objective probability of the target logical states.

arXiv:2607.22276 [quant-ph]

(or arXiv:2607.22276v1 [quant-ph] for this version)

https://doi.org/10.48550/arXiv.2607.22276 Focus to learn more arXiv-issued DOI via DataCite (pending registration)

arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.

Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.

Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.