David Schmid, John H. Selby, Matthew F. Pusey, Robert W. Spekkens (2020), Quantum 8, 1283 (2024) — ARXIV-2005.07161
Key takeaways
- In any diagram-preserving generalized-noncontextual ontological model of a tomographically local operational theory, the model must be an exact (non-overcomplete) frame representation of the associated GPT, so the number of ontic states equals the GPT dimension (for example exactly d² for a qudit).
Seeded from the abstract and a skim, before the reading in NOTE-017. What follows is what the work says about itself.
The authors take a theory to be classically explainable when it admits a generalized-noncontextual ontological model. They extend ontological models and generalized noncontextuality from prepare-measure scenarios to arbitrary compositional scenarios, using process theories. Under reasonable assumptions such models have a rigid structure: a frame representation that is not overcomplete. It follows that the number of ontic states is bounded by the dimension of the associated generalized probabilistic theory, which helps prove noncontextuality no-go theorems and certify contextuality experimentally. They also prove that three notions of classical explainability coincide in compositional scenarios: a noncontextual model of the operational theory, a positive quasiprobability representation of its GPT, and an ontological model of its GPT.
Standing in the record
Filed by the reading-time triage of 2026-09-25: 660 seconds of active reading over 2 sessions in the papers-feed tracker. It was filed Deferred, unread. NOTE-017 is the close reading of 2026-09-25, and it placed the work: Active — It is the formal backbone of the “classicality = generalized noncontextuality = positive quasiprobability” position that LIT-007 and LIT-019 appeal to. The main theorem is fully proved and short. Read §3–4 and App. B. §5.2 illustrates why tomographic locality is necessary, and its numbers should be checked (see corrections).
When seeded, before the reading in NOTE-017, the entry gave a priority for a deeper reading: medium — technically central to the Spekkens cluster, and the result is stated crisply enough that the skim captures it; only worth a deep read if the owner wants the proof machinery (process theories, App. B).
What the seeding asked a deeper reading to check (NOTE-017 is that reading):
- Strongest claim: uniqueness and rigidity of noncontextual models. A deeper read should scrutinize diagram preservation, the load-bearing assumption (§5.1, App. B), and the proof of Theorem 4.1 (App. B).
- It formalizes the classicality criterion items 14 and 17 appeal to. The §5.2 example ties back to item 14’s odd-prime stabilizer/epistricted equivalence.
- ML link: none direct. The “positive quasiprobability representation” criterion is a nonnegativity-of-a-linear-representation condition, but any ML analogy is speculative.
Access when seeded: arXiv abs page and full PDF (41 pp.) retrieved; read outline, §1.1 Results, §1.2 Assumptions heading, §5.2 (necessity of tomographic locality). Proofs in Appendices B-E not read. DOI and venue confirmed via Crossref.