{"id":"89a3c6b4-2bdb-41ea-a84f-787647b5607b","arxiv_id":"2502.10286","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"For each 0<c<1, the operator system generated by four concrete operators on L2(S3)⊕C is not hyperrigid, while every irreducible representation of its C*-algebra restricts to a boundary representation.","lead":"This paper constructs a finite-dimensional operator system that violates Arveson's hyperrigidity conjecture, even though all its irreducible representations have the unique extension property. It settles the finite-dimensional case of a conjecture that was already known to fail in infinite dimensions.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Arveson's boundary theorem is applied to T1,c, which is not in the operator system Sc; the proof that the identity representation is a boundary representation is incomplete as written.","rationale":"The paper's central construction is plausible and the main line of argument is coherent, but the proof of Theorem 2.5 contains a concrete gap at the application of Arveson's boundary theorem. The operator T1,c used in the displayed norm inequality is not in the operator system Sc, whereas the boundary theorem requires the test operators to be in Sc. This is not merely cosmetic: it is the step that establishes the identity representation as a boundary representation, which is half of the theorem's conclusion. The gap is readily fixable by using T_{1,c/2} or \\tilde{T}_{1,c} instead, and the required strict inequality does hold, so the concern does not overturn the construction. The reader's verdict was CONDITIONAL, and this finding supports keeping that verdict: the paper should be accepted only after the boundary-theorem application is corrected and the norm computation is supplied. I do not see a more serious, irreparable flaw: the classification of irreducibles, the use of Davidson-Kennedy, and the hyperrigidity-failure argument all appear sound. The reader's weakest assumption pointed to the Davidson-Kennedy/classification steps; my concern is adjacent but more specific, hence 'partial' agreement.","tokens_in":5039,"tokens_out":42201,"duration_ms":408867,"concrete_test":"First verify T1,c ∉ Sc: set T1,c = α1 + β\\tilde{T}_{1,c} + γ\\tilde{T}_{1,c}^* + Σ_{i=2}^4 δ_i T_i and compare block entries. The (2,2) entry forces α=0; (1,2) forces β=1; (2,1) forces γ=1; then (1,1) requires t1 + Σ_{i=2}^4 δ_i t_i = 0 on S3, which contradicts linear independence of 1,t1,t2,t3,t4. Second, recompute the Arveson-boundary-theorem norm with X1 = T_{1,c/2} (or X1 = \\tilde{T}_{1,c}): find a unit vector (g,a) with g concentrated near t1=1 and a chosen so the cross term is positive, and show ||X1^*X1 + Σ_{i=2}^4 T_i^*T_i|| > 1 for every 0<c<1. If this strict inequality holds, the boundary-representation proof can be repaired; if it fails, the identity representation would not be boundary and the counterexample would be in jeopardy.","verdict_should_be":"UNCHANGED","load_bearing_attack":"In Theorem 2.5, the proof that the identity representation of C*(Sc) is a boundary representation invokes Arveson's boundary theorem with the estimate 1 = ||Σ_{i=1}^4 T_i^*T_i|| < ||T_{1,c}^*T_{1,c} + Σ_{i=2}^4 T_i^*T_i||. However, Sc is generated by \\tilde{T}_{1,c}, T2, T3, T4, not by T1,c. The operator T1,c = [[Mt1, cP*],[cP,0]] is not in Sc: every element of Sc lies in span{1, \\tilde{T}_{1,c}, \\tilde{T}_{1,c}^*, T2, T3, T4}, and a linear-independence comparison of the (1,1), (1,2), (2,1), and (2,2) block entries shows T1,c cannot be expressed in this span. Arveson's boundary theorem requires the test operators to belong to the operator system itself. Thus Theorem 2.5's proof does not, as written, supply a valid tuple from Sc. The gap is repairable: X1 = T_{1,c/2} = (\\tilde{T}_{1,c}+\\tilde{T}_{1,c}^*)/2 lies in Sc, and a direct norm estimate gives ||X1^2 + Σ_{i=2}^4 T_i^2|| > 1 while every point evaluation e_z gives the value |z1|^2+...+|z4|^2 = 1. With this replacement the boundary-representation argument goes through, but the current text uses an operator outside Sc and omits the needed verification.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The manuscript constructs, for each 0 < c < 1, a finite-dimensional operator system Sc generated by four operators acting on L2(S3) ⊕ C, and claims that Sc is not hyperrigid while every irreducible representation of C*(Sc) restricts to a boundary representation of Sc. The argument follows the strategy of Bilich and Dor-On but replaces the infinite-dimensional construction with a rank-one perturbation controlled by c, so that the ambient algebra C*(Sc) becomes an extension of the compact operators by C(S3). The main technical steps are estimates on the joint numerical range of the generators, the identification of the irreducible representations of C*(Sc), and the verification of the unique extension property via Arveson's boundary theorem and Davidson-Kennedy's maximal dilation theorem.","tokens_in":5399,"tokens_out":15330,"duration_ms":147656,"significance":"If the proof is completed, this gives the first finite-dimensional counterexample to Arveson's hyperrigidity conjecture, a notable result in operator system theory. The construction is explicit and the use of a four-dimensional Euclidean ball together with a rank-one perturbation is elegant. The paper correctly relies on deep tools and gives a detailed proof of the crucial numerical range estimates. The main gap, concerning the application of Arveson's boundary theorem to an operator outside the operator system, is localized and has a clear repair, so the central claim is defensible.","major_comments":[{"comment":"The proof invokes Arveson's boundary theorem with the inequality 1 = ||Σ_{i=1}^4 T_i^*T_i|| < ||T_{1,c}^*T_{1,c} + Σ_{i=2}^4 T_i^*T_i|| to conclude that the identity representation of C*(Sc) is a boundary representation. The boundary theorem, however, requires the test operators to belong to the operator system Sc. The tuple (T_{1,c}, T2, T3, T4) does not satisfy this requirement: Sc is generated by \\tilde{T}_{1,c}, T2, T3, T4, and T_{1,c} differs from \\tilde{T}_{1,c} by the operator [[0,0],[cP,0]], which is not in the span of 1, \\tilde{T}_{1,c}, \\tilde{T}_{1,c}^*, T_i, T_i^* (a block-entry comparison shows that no such linear combination can produce the (2,1) block). This is a load-bearing gap, but it is repairable: replace T_{1,c} by X1 = (\\tilde{T}_{1,c}+\\tilde{T}_{1,c}^*)/2 = T_{1,c/2}, which lies in Sc, and verify the norm estimate ||X1^2+Σ_{i=2}^4 T_i^2|| > 1 while every point evaluation gives norm 1; with that tuple the boundary theorem applies. The authors should either make this replacement or explicitly exhibit another tuple from Sc that satisfies the boundary-theorem hypothesis.","section":"Theorem 2.5, proof of boundary representation for the identity"},{"comment":"The sentence 'It follows from Equation (2) that the restrictions of the maps ez to Sc are extreme points of S(Sc)' omits the argument that explains why the bound in Equation (2) forces the ez to be extreme. The needed step is the strict convexity of the Euclidean unit ball in C^4: if ez = λφ + (1-λ)ψ with 0<λ<1 and φ,ψ∈S(Sc), then the tuples (φ(\\tilde{T}_{1,c}),φ(T2),φ(T3),φ(T4)) and (ψ(\\tilde{T}_{1,c}),ψ(T2),ψ(T3),ψ(T4)) lie in the closed unit ball by Equation (2), and their convex combination is the boundary point z; strict convexity forces both tuples to equal z, hence φ=ψ=ez on the generators, and therefore on Sc. Since this extremity is then used together with Equation (3) to prove maximality of the point evaluations, the missing justification should be supplied explicitly.","section":"Theorem 2.5, proof that point evaluations are maximal"}],"minor_comments":[{"comment":"'Arvson' should be 'Arveson'.","section":"Theorem 2.5"},{"comment":"In the definition of Φ, 'δte' appears to be a typo for 'δt4'.","section":"Lemma 2.2"},{"comment":"The assertion that the inverse of Φ is positive is not proved and is not used elsewhere in the paper; either provide a proof or delete the assertion.","section":"Lemma 2.2"},{"comment":"The notation (T1,c, T2, T3, T4) is used to describe the quotient C*(Sc)/K ≅ C(S3), but T1,c is not one of the original generators of Sc; please state explicitly that T1,c ∈ C*(Sc) follows from Lemma 2.4 before using it in the quotient description.","section":"After Lemma 2.4"},{"comment":"The equality case in the Cauchy-Schwarz step is treated very tersely; an expanded argument that the multiplication operators M_ti have no eigenvalues on L2(S3) (and on the C summand) would improve readability.","section":"Lemma 2.1"},{"comment":"The name 'Carathédory' should be spelled 'Carathéodory'.","section":"Throughout"}],"recommendation":"major_revision","confidential_remarks":"The manuscript addresses a known open problem and the construction appears sound modulo the repairable gap in the proof of Theorem 2.5 identified in the major comments. I recommend major revision and would be willing to see the revised version. No concerns about novelty or attribution arose; the paper credits [5] and [2] appropriately."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Quick take: the paper constructs a finite-dimensional operator system S_c that is claimed to be the first finite-dimensional counterexample to Arveson's hyperrigidity conjecture. The construction is explicit and the overall strategy is sound, but the proof as written has a genuine gap: the application of Arveson's boundary theorem in Theorem 2.5 uses T_{1,c}, which is not in S_c. Fortunately the gap is repairable.\n\nWhat's new: Bilich-Dor-On gave an infinite-dimensional counterexample; this paper closes the finite-dimensional case. The key idea is to take a direct sum of multiplication operators on L^2(S^3) with a one-dimensional summand, and to use the Akemann-Weaver technique to show the point evaluations are maximal. The paper does a good job of laying out the construction and proving the numerical range lemmas (Lemma 2.2 and 2.3) and the compactness lemma (2.4). The split exact sequence argument identifying the irreducible representations is standard and correct.\n\nWhere the soft spots are:\n1. In Theorem 2.5, the writer invokes Arveson's boundary theorem with the inequality ||\\Sigma_{i=1}^4 T_i^*T_i|| < ||T_{1,c}^*T_{1,c} + \\Sigma_{i=2}^4 T_i^*T_i||. The tuple on the right includes T_{1,c}, which is not in S_c. S_c is generated by \\tilde{T}_{1,c}, T_2, T_3, T_4, and a quick linear algebra check shows T_{1,c} is not in the span. Arveson's boundary theorem requires the witnessing tuple to lie in the operator system. This is a real gap in the present text. It is repairable: take X_1 = (\\tilde{T}_{1,c} + \\tilde{T}_{1,c}^*)/2 = T_{1,c/2}, which is in S_c, and verify that ||X_1^2 + \\Sigma_{i=2}^4 T_i^2|| > 1 while every point evaluation gives exactly 1. The author should supply this verification.\n2. Lemma 2.2 claims the inverse of Φ is positive without proof; it isn't used later, so this is minor but should be fixed.\n3. The appeal to Arveson's boundary theorem is terse; a precise statement of the version used would help.\n4. Minor typos ('Arvson's', etc.).\n\nThe central argument appears sound. I did not find circularity or fitted parameters. The paper is a short note, likely of high interest to operator algebraists working on hyperrigidity and the noncommutative Choquet boundary.\n\nRecommendation: Send to peer review. The gap in Theorem 2.5 is fixable and the result is significant enough to merit referee time. If the author patches the boundary-representation argument, this should be accepted.","headline":"Finite-dimensional counterexample to Arveson's conjecture, plausible but with a repairable gap in the boundary-representation proof.","tokens_in":5869,"tokens_out":9021,"would_cite":true,"duration_ms":86349,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["46L07","47A20","46L52"],"pacs":[],"model":"deepseek-v4-flash","headline":"A four-generator operator system gives the first finite-dimensional counterexample to the hyperrigidity conjecture: all irreducible representations have the unique extension property, yet the system is not hyperrigid.","keywords":["operator system","hyperrigidity conjecture","unique extension property","boundary representation","joint numerical range","type I C*-algebra"],"falsifier":"Pick $z=(1,0,0,0)$ and some $0<c<1$, and solve for a unital completely positive map $\\Psi$ on $C^*(S_c)$ satisfying $\\Psi(\\tilde T_{1,c})=1$ and $\\Psi(T_2)=\\Psi(T_3)=\\Psi(T_4)=0$ but $\\Psi(P^*P)\\neq0$; existence of such a map would be a second extension of the point evaluation $e_z$, directly contradicting the theorem.","tokens_in":4856,"feed_emoji":"🧩","tokens_out":16373,"duration_ms":162479,"temperature":0.7,"pith_summary":"This paper constructs, for each $0<c<1$, an operator system $S_c$ generated by four operators on $L^2(S^3)\\oplus\\mathbb{C}$, and proves that $S_c$ is not hyperrigid even though the restriction of every irreducible representation of $C^*(S_c)$ to $S_c$ has the unique extension property. It is finite-dimensional as an operator system, spanned by four operators together with the identity, even though those operators act on the infinite-dimensional space $L^2(S^3)\\oplus\\mathbb{C}$. This is the first finite-dimensional counterexample to the hyperrigidity conjecture, which proposed that the unique-extension property on all irreducible representations should force hyperrigidity. The failure of hyperrigidity is explicit: a representation acting on $L^2(S^3)$ by multiplication operators dilates non-trivially to the identity representation, so its restriction to $S_c$ has more than one completely positive extension. The other half of the theorem shows that the identity representation and the point evaluations $e_z$, $z\\in S^3$, are maximal on $S_c$, using a joint-numerical-range bound and the fact that every pure state dilates to a boundary representation.","feed_headline":"Four operators break the hyperrigidity conjecture","feed_subtitle":"The first finite-dimensional counterexample: every irreducible representation has unique extensions, yet hyperrigidity fails.","key_machinery":"The carrying mechanism is the joint numerical-range bound for the tuple $(\\tilde T_{1,c},T_2,T_3,T_4)$. Lemma 2.2 shows that the map $\\Phi:A(\\overline{B}_4)\\to\\tilde S_c$ sending $t_i$ to the corresponding operator is positive for $0<c\\le1/2$; positivity is reduced, via the standard block-matrix positivity criterion, to the inequality $c^2P^*P\\le |\\beta|^{-1}M_f$, and the key constant is $\\int_{S^3}(1+t_1)^{-1}dm=2$. Lemma 2.3 then upgrades the bound to all $0<c<1$ by writing the expectation of $\\tilde T_{1,c}$ as a convex combination of expectations of $T_1$ and $T_{1,1/2}$. A second mechanism is the containment of all compact operators in $C^*(S_c)$, which yields the split exact sequence $0\\to K\\to C^*(S_c)\\to C(S^3)\\to0$; this makes the algebra type I and gives the short list of irreducible representations. Maximality of the point evaluations is then certified by the dilation theorem for pure states together with the numerical-range bound.","core_discovery":"The paper's central claim is Theorem 2.5: for every $0<c<1$, the operator system $S_c=\\operatorname{span}\\{1,\\tilde T_{1,c},T_2,T_3,T_4\\}$, where $T_i=M_{t_i}\\oplus0$ and $\\tilde T_{1,c}$ is the upper-triangular block matrix with $M_{t_1}$ in the upper-left corner and $cP^*$ in the upper-right corner, is not hyperrigid, yet the restrictions to $S_c$ of all irreducible representations of $C^*(S_c)$—namely the identity representation and the evaluation maps $e_z$ for $z\\in S^3$—have the unique extension property. The non-hyperrigidity is exhibited by the $*$-homomorphism sending the four generators to the multiplication operators $M_{t_1},M_{t_2},M_{t_3},M_{t_4}$, whose restriction to $S_c$ dilates non-trivially to the identity representation and therefore fails to have the unique extension property. The unique-extension half is proved by showing that the joint numerical range of the four operators in every non-maximal pure state lies strictly inside the unit ball of $\\mathbb{R}^4$, while point evaluations lie on the sphere.","pith_inferences":["This suggests the phenomenon is not tied to the specific sphere $S^3$: replacing the spherical model by the $n$-sphere and the projection $P$ by integration over $S^n$ may yield finite-dimensional operator systems with $n+1$ generators sharing the same dichotomy, with the integral $\\int_{S^n}(1+t_1)^{-1}$ controlling the admissible range of $c$.","One could test the construction numerically on finite-dimensional subspaces of $L^2(S^3)$ spanned by low-degree spherical harmonics: if the joint-numerical-range bound persists under truncation, the counterexample survives in a purely matrix model accessible to computer verification.","The fact that the identity representation is not a boundary representation of $S_c$ while all irreducible representations are suggests that hyperrigidity may depend on how the operator system sits inside the C*-algebra, not just on the extremal structure of its state space."],"forward_implications":["The hyperrigidity conjecture fails within the class of finite-dimensional operator systems, so the earlier infinite-dimensional counterexample is not an artefact of infinite dimension.","For this $S_c$, the unique extension property on irreducible representations does not imply the unique extension property on all representations, so hyperrigidity requires a genuinely global condition.","The counterexample is a one-parameter family indexed by $0<c<1$, so the phenomenon is stable under perturbation of the coupling constant $c$.","Because $C^*(S_c)$ is type I and has only identity-plus-evaluation irreducible representations, the failure cannot be blamed on exotic representation theory.","Any proposed repair of the conjecture must rule out this construction, for instance by imposing extra structure on the operator system beyond the irreducible unique-extension property."],"supporting_citations":[{"why":"Supplies the block-matrix positivity criterion used to reduce positivity of the block matrix in Lemma 2.2.","marker":"[1]"},{"why":"Provides the technique for bounding a rank-one projection below a family of positive multiplication operators on the sphere.","marker":"[2]"},{"why":"Gives the structure theory for type I C*-algebras used to identify the irreducible representations as the identity and point evaluations.","marker":"[3]"},{"why":"States the hyperrigidity conjecture that the paper disproves.","marker":"[4]"},{"why":"Provides the prior infinite-dimensional counterexample and the overall strategy, including the compact-operator containment argument adapted in Lemma 2.4.","marker":"[5]"},{"why":"Supplies the dilation theorem used to show every pure state dilates to a boundary representation, which certifies maximality of point evaluations.","marker":"[6]"},{"why":"Establishes that the C*-algebra generated by the example is type I, so the classification of irreducible representations applies.","marker":"[7]"}],"fun_headline_variants":["Hyperrigidity conjecture falls to four operators","Unique extension property not sufficient for hyperrigidity","Counterexample: 4 operators, all irreps extend uniquely","First finite-dimensional counterexample to hyperrigidity","Hyperrigidity fails despite unique extension property"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The proof that the point evaluations are maximal rests on two pillars: the classification of the irreducible representations of $C^*(S_c)$ as only the identity and the point evaluations, and the theorem that every pure state dilates to a boundary representation; if either fails to apply here, the unique-extension claim for $S_c$ would not go through.","fun_headline_variants_meta":{"raw":{"variants":["Hyperrigidity conjecture falls to four operators","Unique extension property not sufficient for hyperrigidity","Counterexample: 4 operators, all irreps extend uniquely","First finite-dimensional counterexample to hyperrigidity","Hyperrigidity fails despite unique extension property"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.001444,"raw_usage":{"total_tokens":5754,"prompt_tokens":817,"completion_tokens":4937,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":433,"completion_tokens_details":{"reasoning_tokens":4865}},"tokens_in":433,"tokens_out":4937,"duration_ms":36901,"temperature":1.0,"reasoning_tokens":4865,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-07T18:44:34.729135+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Pick $z=(1,0,0,0)$ and some $0<c<1$, and solve for a unital completely positive map $\\Psi$ on $C^*(S_c)$ satisfying $\\Psi(\\tilde T_{1,c})=1$ and $\\Psi(T_2)=\\Psi(T_3)=\\Psi(T_4)=0$ but $\\Psi(P^*P)\\neq0$; existence of such a map would be a second extension of the point evaluation $e_z$, directly contradicting the theorem.","supporting_citations":[{"cited_title":"McCarthy","cited_arxiv_id":null,"evidence_quote":"Supplies the block-matrix positivity criterion used to reduce positivity of the block matrix in Lemma 2.2."},{"cited_title":"Minimal upper bounds of c ommuting operators","cited_arxiv_id":null,"evidence_quote":"Provides the technique for bounding a rank-one projection below a family of positive multiplication operators on the sphere."},{"cited_title":"An invitation to C∗-algebras, volume No","cited_arxiv_id":null,"evidence_quote":"Gives the structure theory for type I C*-algebras used to identify the irreducible representations as the identity and point evaluations."},{"cited_title":"The noncommutative Choquet boundary I I: hyperrigidity","cited_arxiv_id":null,"evidence_quote":"States the hyperrigidity conjecture that the paper disproves."},{"cited_title":"Arveson’s hyperrigidity c onjecture is false,","cited_arxiv_id":null,"evidence_quote":"Provides the prior infinite-dimensional counterexample and the overall strategy, including the compact-operator containment argument adapted in Lemma 2.4."},{"cited_title":"Davidson and Matthew Kennedy","cited_arxiv_id":null,"evidence_quote":"Supplies the dilation theorem used to show every pure state dilates to a boundary representation, which certifies maximality of point evaluations."},{"cited_title":"Type I C∗-algebras","cited_arxiv_id":null,"evidence_quote":"Establishes that the C*-algebra generated by the example is type I, so the classification of irreducible representations applies."}],"review_version":1}