{"id":"0191fa90-e877-44ea-942a-f48aa4a2c667","arxiv_id":"1908.10676","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":2,"one_line_summary":"Nonlocality without entanglement occurs in polygon-based generalized probabilistic theories, and the pentagon model exhibits a strictly larger gap between local and global discrimination success than quantum theory does.","lead":"The paper shows that 'nonlocality without entanglement', a quantum effect in which product states are easier to decode together than separately, also appears in polygon-based toy theories of physics. It argues that quantum theory is weaker in this effect than a pentagon model and links that weakness to the continuity of quantum state space.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Theorem 2's strict inequality rests entirely on the unproved claim that the optimal local success probability for the pentagon octet is 7/8; perfect indistinguishability (Theorem 1) does not imply this optimality, and Appendix B.2 asserts it without proof.","rationale":"The reader and I converge on the same load-bearing assumption: the exact value Δ[Pentagon]=1/8 is asserted, not proved. I checked the surrounding text: Theorem 1 establishes perfect global discriminability and only sketches local indistinguishability; Appendix B.2 provides a flow chart and then asserts optimality. Nothing in the proof excludes a first-round non-extremal measurement or a non-perfect strategy with success probability between 7/8 and 1. This is exactly the type of gap that can hide a counterexample in convex optimization. The quantum side is only an upper bound on Δ, so the pentagon side must be certified. I also noted the reader's Lemma 1 objection: the stated proof has a branch error, but that lemma concerns asymmetric local discrimination, not the main comparative theorem; it should be repaired but does not change the central verdict. The Discussion's claim that continuity of the state space causes limited NWE is asserted, not derived; I treat it as a speculation layered on Theorem 2, not as a separate load-bearing result. The verdict should remain CONDITIONAL: the explicit constructions and the quantum optimization are strong independent support, but the quantitative claim is incomplete pending the pentagon optimality check.","tokens_in":15135,"tokens_out":8531,"duration_ms":99185,"concrete_test":"Implement an exact finite-dimensional search for the optimal one-way LOCC protocol for the octet in Eq. (B5). Since Ply(5) state and effect spaces are polytopes, model the first round by a variable effect f∈E(5), parametrized as a convex combination of 0, u, e_i, and ¯e_i, and for each outcome branch solve the optimal two-party discrimination problem for the induced weighted ensemble over the remaining two subsystems using linear programming over the product-effect polytopes of Ply(5)⊗2_min. Maximize the total success probability over f; vertex enumeration makes this exact. If the maximum is 7/8, the missing optimality claim is verified. If the maximum exceeds 1 − (4−√10)/8 ≈ 0.8953, the strict inequality in Theorem 2 fails; if it lies between, the exact gap Δ[Pentagon]=1/8 is wrong even if the theorem survives.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The weakest point is in Appendix B.2, Eqs. (B6)-(B7). The paper needs an exact upper bound PL[Pentagon] = 7/8, or at least PL[Pentagon] < 1 − (4−√10)/8 ≈ 0.8953, to justify Δ[Pentagon] > Δ[Quantum]. What is actually shown is only that no protocol using one of the five extremal measurements M_i={e_i,¯e_i} in the first round discriminates the octet perfectly (Theorem 1 and its 'it is not hard to see' passage). Perfect indistinguishability does not imply that the best success probability is the 7/8 achieved by their particular flow chart; a non-perfect protocol could in principle do better. Moreover, the first party's effect is not restricted to the extremal dichotomy: since E(5) is the convex hull of {0,u,e_i,¯e_i}, the first measurement can be any pair {f,u−f} with f a convex combination. Such a coarse-grained effect changes the posterior weights on the eight states and could reduce the probability of the ambiguous two-state case. Appendix B.2 merely states 'It follows from the proof of Proposition-2, any other strategy is no good' without supplying the required optimization over E(5) and over later rounds. Until that optimization is done, the numerical comparison in Theorem 2 is not established.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper argues that nonlocality without entanglement (NWE) is not peculiar to quantum theory but appears in generalized probabilistic theories (GPTs). It constructs an eight-state product ensemble in the tripartite minimal composition of the pentagon model Ply(5)^3, exhibits a separable product-effect measurement that discriminates it perfectly, and claims that no local protocol can do so. It then proposes a quantitative measure of NWE strength, Δ = 1 − P_L, the gap between global and optimal local success probability, with elementary systems matched by signaling dimension. The central quantitative result is Theorem 2, which states Δ[QT] ≤ (1/8)(4−√10) < 1/8 = Δ[Pentagon], and a biased-prior version is claimed to show the same ordering for all prior weights. The authors attribute the limited NWE of quantum theory to the continuity of the quantum state space.","tokens_in":15300,"tokens_out":3681,"duration_ms":41436,"significance":"If the proofs were complete, the paper would make a valuable conceptual contribution: it would show that a phenomenon previously considered quantum is generic in GPTs, and it would provide a quantitative criterion for comparing NWE strength across theories. The construction of a product-effect measurement for the pentagon octet is explicit and checkable, and the quantum one-parameter optimization giving P_err = (1/8)(4−√10) is analytic and correct. The framework using signaling dimension to put different elementary systems on an equal footing is natural and potentially useful. However, the paper's main quantitative claim rests on an unproved optimality assertion for the pentagon model, and several auxiliary proofs are incomplete or inconsistent at the level of internal references. These issues are load-bearing rather than cosmetic.","major_comments":[{"comment":"The equality Δ[Pentagon] = 1/8 is asserted, not proved. The flow-chart protocol achieves P_succ = 7/8, but optimality over all local protocols is justified only by the sentence 'It follows from the proof of Proposition-2, any other strategy is no good', and Proposition-2 is never stated in the manuscript. Since E(5) is the convex hull of {0, u, e_i, ¯e_i}, a first-round measurement may use any effect f ∈ E(5), not only the extremal dichotomies {e_i, ¯e_i}, and a rigorous upper bound must rule out all such strategies and all later-round choices. Without a proof that P_L[Pentagon] ≤ 7/8 (or at least P_L[Pentagon] < 1 − (4−√10)/8 ≈ 0.8953), the strict inequality in Theorem 2 is not established.","section":"Appendix B.2, Eqs. (B6)–(B7)"},{"comment":"The local indistinguishability part of Theorem 1 is not proved. The sentence 'It is not hard to see that whichever measurement Alice starts with no perfect discrimination is possible' substitutes for an argument that must cover arbitrary effects in E(5), not merely the five extremal measurements M_i. Since the octet is perfectly discriminated by a global separable measurement, the entire NWE phenomenon in the pentagon model depends on this claim. Please provide a complete proof or a precise reference to one.","section":"Theorem 1 (main text)"},{"comment":"The described local protocol for the four states in Ply(4)^2 appears to be incorrect as written. When Alice measures M_0 = {e_0, e_2}, the outcome e_0 on Alice's subsystem also has nonzero probability for the state ω_1⊗ω_0, so it does not uniquely select {ω_0⊗ω_0, ω_0⊗ω_3} as the proof claims. Consequently Bob's subsequent measurement cannot perfectly resolve all cases in the way described. Please verify the transition probabilities p(e_0|ω_i) for the squit model and either correct the protocol or adjust the state set and proof.","section":"Lemma 1 (main text)"},{"comment":"The appendix refers repeatedly to 'Proposition-2' and 'Proposition-3', but no such propositions are stated in the paper. This makes the claimed local indistinguishability of the hexagon and heptagon constructions, and the optimality claim for the pentagon protocol, impossible to verify. The numbering of results must be made consistent, and the referenced proofs must either be included or given explicit locations.","section":"Appendix B.1–B.2"}],"minor_comments":[{"comment":"There is a typo in the last sentence: 'perfetly' should be 'perfectly'.","section":"Lemma 1 proof"},{"comment":"In the biased-prior paragraph, 'p4 = p4 = p' should presumably read 'p4 = p5 = p'; please correct the notation.","section":"Appendix B.3"},{"comment":"The flow chart is not described algorithmically in the text, which makes the claimed success probability 7/8 hard to reproduce. A short verbal description of the decision tree would improve verifiability.","section":"Figure 3"},{"comment":"The theorem states an upper bound Δ[QT] ≤ (1/8)(4−√10), which is an upper bound on the gap and hence a lower bound on the optimal success probability; this directional language is correct, but it would help to state explicitly that no optimality of the quantum protocol is claimed.","section":"Theorem 2 statement"}],"recommendation":"major_revision","confidential_remarks":"The central comparison in Theorem 2 depends on an unproved optimality claim for the pentagon model, and the manuscript contains several references to nonexistent propositions. These issues are fixable in principle by adding a rigorous optimization proof, but they are substantial enough that the paper is not ready for acceptance in its current form. I also note that the construction relies on the authors' own Ref. [21] for the premise that GPTs admit indistinguishable pure states; this is a legitimate premise, but the reliance should be made more explicit and the relevant result stated."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The real news here is that the pentagon, hexagon, and heptagon models admit explicit tripartite product-state ensembles that are perfectly discriminable by a separable product-effect measurement and not by LOCC. I checked the pentagon octet: the eight effects sum to the unit effect and give p(E_i|φ_j)=δ_ij. That is a legitimate new construction, not a repackaging of Bennett et al. The quantum side is also handled honestly: the one-parameter optimization giving P_err=(1/8)(4−√10) at θ=tan⁻¹(1/3) is correct, and the biased-prior results are analytic. The framework of comparing theories by signaling dimension is sensible, and the paper earns its claim that NWE is not unique to quantum theory.\n\nThe soft spots are real but localized. First, the proof of Lemma 1 is wrong as written: with Alice measuring M0={e0,e2}, outcome e0 also fires on ω1⊗ω0, so the claimed two-state branch is false. The lemma may still be true, but the proof needs fixing. Second, and load-bearing, the comparison theorem's strict inequality Δ[Pentagon]=1/8 > Δ[Quantum] requires that no local protocol in the pentagon model beats 7/8. The paper only rules out perfect discrimination and asserts optimality of the 7/8 flow chart in Appendix B.2 without proof. Since the pentagon effect space includes coarse-grained effects (convex combinations of extremal effects), that assertion is not trivial and needs a real optimization argument. Until that is supplied, Theorem 2 is conditional. The Discussion's claim that limited NWE is due to continuity of the state space is a plausible conjecture, not a derivation, and should be flagged as such.\n\nNone of this is circular or fitted: there are essentially no free parameters beyond the variational angle and biased-prior weight, both handled exactly. The citation pattern is fine; reliance on the authors' own Ref [21] for the GPT premise is acceptable because the constructions here are self-contained.\n\nWho is this for? Foundational GPT people and anyone working on state discrimination under restricted measurements. The pentagon construction is worth knowing even if the comparison theorem needs repair. With the optimality gap closed it would be a solid paper; as it stands, it deserves a serious referee but not acceptance as-is.\n\nRecommendation: send to peer review, ask for a rigorous proof of the pentagon optimality claim and a corrected Lemma 1. My verdict: conditional accept.","headline":"A genuinely new GPT construction of nonlocality-without-entanglement with a clean quantum comparison, but Theorem 2 rests on an unproved optimality claim in the pentagon model.","tokens_in":16026,"tokens_out":637,"would_cite":true,"duration_ms":8461,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"Nonlocality without entanglement is generic across generalized probabilistic theories, and quantum theory's version is strictly weaker than the pentagon model's.","keywords":["generalized probabilistic theories","nonlocality without entanglement","state discrimination","local operations and classical communication","polygonal models","signaling dimension","product states","quantum state space continuity"],"falsifier":"Search over all one-way local protocols in $\\mathrm{Ply}(5)^{\\otimes 3}_{\\min}$, allowing each party any mixture of the allowed effects; if any protocol exceeds 7/8 success on the octet, then $\\Delta[\\mathrm{Pentagon}]<1/8$ and the strict inequality of Theorem 2 fails.","tokens_in":14749,"feed_emoji":"⚛️","tokens_out":10970,"duration_ms":99839,"temperature":0.7,"pith_summary":"The paper asks whether 'nonlocality without entanglement'—the fact that some sets of product states can be perfectly identified only by a joint measurement, never by separate local measurements with classical communication—is a special feature of quantum theory. It answers no: polygonal generalized probabilistic theories, which replace the quantum state space by a regular polygon, exhibit the same phenomenon with product states and separable product-effect measurements. It then matches the elementary systems of quantum theory and the pentagon model by their signaling dimension, and proves that the pentagon model's gap between global and local success probabilities is strictly larger than quantum theory's gap. The authors conclude that quantum nonlocality without entanglement is limited, and link that limitation to the continuity of the quantum state space.","feed_headline":"Pentagon model beats quantum theory at hiding classical information","feed_subtitle":"A five-sided state space leaves a strictly larger gap between local and joint discrimination than qubits do.","key_machinery":"The load-bearing object is the polygonal GPT model $\\mathrm{Ply}(n)$, whose single-party state space is a regular $n$-gon and whose effects are convex combinations of the unit effect and pairs of extremal effects; the pentagon case is self-dual and has exactly five extremal two-outcome measurements. Within $\\mathrm{Ply}(5)^{\\otimes 3}_{\\min}$, the eight product states and the separable effects $\\{E_i\\}$ realize perfect global discrimination while every one-way local protocol leaves ambiguity. For fair comparison across theories, the paper uses signaling dimension—the least classical dimension that can simulate the system's input-output behavior—to match the pentagon elementary system with a qubit, and quantifies NWE strength as $\\Delta = 1 - P_L$, where $P_L$ is the optimal local success probability.","core_discovery":"The central claim is that nonlocality without entanglement is generic in operational theories, and its strength is theory-dependent. In the minimal tripartite composition of the pentagon model $\\mathrm{Ply}(5)^{\\otimes 3}_{\\min}$, the eight product states $\\{\\phi_1,\\ldots,\\phi_8\\}$ are perfectly distinguishable by a separable measurement whose effects are products of single-party effects (the effects $E_1,\\ldots,E_8$ satisfy $\\sum_i E_i = u^{\\otimes 3}$ and $p(E_i|\\phi_j) = \\delta_{ij}$), yet no local protocol can discriminate them perfectly. The same construction works in hexagon and heptagon models. After matching elementary systems by signaling dimension, the paper proves that for the analogous three-qubit product ensemble the gap is $\\Delta[\\mathrm{QT}] \\le \\frac{1}{8}(4-\\sqrt{10})$, strictly smaller than $\\Delta[\\mathrm{Pentagon}] = \\frac{1}{8}$. This limited behavior in quantum theory is ascribed to the continuity of the state space, which allows a continuous reversible transformation between any two pure states.","pith_inferences":["A numerical search over all local protocols in the pentagon model, allowing convex mixtures of the five basic measurement pairs, would settle the one unproved optimality assertion and either confirm or reduce the strict inequality in Theorem 2.","If the pentagon value is confirmed, the octet becomes a compact NWE witness: any GPT with signaling dimension 2 whose state space is a polygon of at least five sides should show the same $1/8$ gap, yielding a family of testable models.","Constructing a signaling-dimension-3 GPT with a continuous but non-quantum state space would separate the effect of continuity from the effect of Hilbert-space structure, testing the paper's proposed explanation for quantum theory's limited NWE."],"forward_implications":["Nonlocality without entanglement, including asymmetric local discrimination and separable but locally unimplementable measurements, is not a signature of Hilbert-space structure; it arises generically in generalized probabilistic theories.","For tripartite systems with equal signaling dimension, the pentagon model has strictly stronger NWE than quantum theory: $\\Delta[\\mathrm{Pentagon}] = 1/8$ while $\\Delta[\\mathrm{QT}] \\le (4-\\sqrt{10})/8$.","The quantity $\\Delta$ provides a common scale for comparing NWE strength across different theories, making the phenomenon quantitatively testable.","Because limited NWE in quantum theory is tied to continuity of the state space, limited NWE could serve as a candidate principle in axiomatic derivations of quantum theory.","The ordering $\\Delta[\\mathrm{Quantum}] < \\Delta[\\mathrm{polygon}]$ holds also for biased priors over the eight states, not only for the uniform distribution."],"supporting_citations":[{"why":"supplies the original quantum three-qubit product ensemble and the phenomenon of nonlocality without entanglement; it is the baseline that Theorem 2 must beat.","marker":"[12]"},{"why":"introduces the polygonal models Ply(n) whose state and effect spaces carry the pentagon octet construction in Theorem 1.","marker":"[32]"},{"why":"defines signaling dimension, the criterion used to match the pentagon elementary system with a qubit before comparing NWE strength.","marker":"[37]"},{"why":"establishes the quantum bipartite asymmetry in local discrimination that Lemma 1 reproduces in the squit model.","marker":"[23]"},{"why":"documents the difficulty of characterizing general LOCC protocols, supporting the paper's decision to compare one-way local protocols and state the quantum value as an upper bound.","marker":"[41]"}],"fun_headline_variants":["Pentagon model beats quantum at nonlocality without entanglement","Nonlocality without entanglement is stronger in pentagon model","Quantum not special: pentagon model amplifies nonlocal discrimination","Five-sided state space yields stronger nonlocality without entanglement","Pentagon model widens local vs joint discrimination gap vs quantum"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The comparison rests on the unproved assertion that no local protocol in the pentagon model can do better than 7/8 on the eight states, even though each party is allowed to use any mixture of the five basic measurement pairs.","fun_headline_variants_meta":{"raw":{"variants":["Pentagon model beats quantum at nonlocality without entanglement","Nonlocality without entanglement is stronger in pentagon model","Quantum not special: pentagon model amplifies nonlocal discrimination","Five-sided state space yields stronger nonlocality without entanglement","Pentagon model widens local vs joint discrimination gap vs quantum"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000925,"raw_usage":{"total_tokens":3946,"prompt_tokens":912,"completion_tokens":3034,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":528,"completion_tokens_details":{"reasoning_tokens":2951}},"tokens_in":528,"tokens_out":3034,"duration_ms":24392,"temperature":1.0,"reasoning_tokens":2951,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T10:39:53.630803+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Search over all one-way local protocols in $\\mathrm{Ply}(5)^{\\otimes 3}_{\\min}$, allowing each party any mixture of the allowed effects; if any protocol exceeds 7/8 success on the octet, then $\\Delta[\\mathrm{Pentagon}]<1/8$ and the strict inequality of Theorem 2 fails.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"supplies the original quantum three-qubit product ensemble and the phenomenon of nonlocality without entanglement; it is the baseline that Theorem 2 must beat."},{"cited_title":"The principle of information symmetry constrains the state-space in any physical theory","cited_arxiv_id":"1905.09413","evidence_quote":"introduces the polygonal models Ply(n) whose state and effect spaces carry the pentagon octet construction in Theorem 1."},{"cited_title":"Chiribella, G","cited_arxiv_id":null,"evidence_quote":"defines signaling dimension, the criterion used to match the pentagon elementary system with a qubit before comparing NWE strength."},{"cited_title":"Bennett, David P","cited_arxiv_id":null,"evidence_quote":"establishes the quantum bipartite asymmetry in local discrimination that Lemma 1 reproduces in the squit model."},{"cited_title":"Barnum, J","cited_arxiv_id":null,"evidence_quote":"documents the difficulty of characterizing general LOCC protocols, supporting the paper's decision to compare one-way local protocols and state the quantum value as an upper bound."}],"review_version":1}