{"id":"c5d21c44-a6fd-4c4b-a9fd-706764c467de","arxiv_id":"2607.04274","paper_version":2,"verdict":"ACCEPT","confidence":"HIGH","novelty_score":7.0,"correctness_risk":"low","formal_verification":"none","parameter_count":0,"one_line_summary":"A hyperrigid three-dimensional operator system in a matrix amplification of the reduced free-group C*-algebra produces a Calkin subsystem without the lifting property, countering Smith–Ward and giving a non-exact dual.","lead":"A three-dimensional operator system inside the Calkin algebra has no unital completely positive lift. This settles the Smith–Ward problem in its critical dimension and yields the first non-exact three-dimensional operator system.","discovery_kind":"extension","skeptic_critique":{"model":"grok-4.5","headline":"No significant objection identified","rationale":"The reader correctly isolates the Haagerup–Thorbjørnsen non-invertible Ext class as the sole external load-bearing input and notes that the rest of the argument is algebraic and self-contained. My own line-by-line check of the unique-extension proof (the only novel piece) finds no gap: the endpoint spectral subspaces of D are invariant under the Stinespring isometry, the partial-isometry edges of K propagate the invariance to the remaining two subspaces by norm preservation, and the resulting reducing subspace forces the u.c.p. extension to be a representation. All subsequent steps (Proposition 4.1, Theorem 4.2, Corollary 4.3) are then routine. Because the weakest assumption is a published theorem and the internal reasoning is transparent, the ACCEPT verdict with high confidence stands without modification.","tokens_in":8542,"tokens_out":540,"duration_ms":4132,"concrete_test":"Independently re-derive the unique-extension argument of Theorem 3.2 for the concrete matrices D and K: verify that the successive norm-preservation steps force Vπ(pj)E⊆ρ(pj)L for j=1,2,3,4 and that the resulting reducing subspace for ρ(D) and ρ(K) is reducing for the whole M4(C*_r(F2)). If the inclusions hold and Φ coincides with π, the hyperrigidity claim is confirmed.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim rests on three standard pillars that the manuscript deploys correctly: (i) Haagerup–Thorbjørnsen’s theorem that Ext(C*_r(F2)) is not a group, yielding a non-invertible injective unital representation τ into the Calkin algebra; (ii) the elementary Ext-obstruction of Lemma 2.1 that a u.c.p. lift would force invertibility; and (iii) the unique-extension property of the concrete three-dimensional system S=span{1,D,K} proved in full in Theorem 3.2. The novel step—recovering the four spectral projections of D from the two endpoint subspaces and propagating them along the partial-isometry edges of K—is written line-by-line and appears free of algebraic gaps. No hidden boundedness, continuity, or non-separability assumption is required. Consequently the reduction from four to three dimensions is secure, and the dual non-exactness corollary follows immediately from Kavruk’s theorem.","agreement_with_reader":"agree"},"referee_report":{"model":"grok-4.5","summary":"The paper constructs a three-dimensional hyperrigid operator system S = span{1, D, K} inside M4(C*_r(F2)), with D a diagonal self-adjoint matrix having four distinct eigenvalues and K a self-adjoint matrix whose nonzero entries are the free unitaries u, v and the unit. It proves that C*(S) equals the full matrix algebra and that every unital representation of M4(C*_r(F2)) restricts to a map on S with the unique extension property (Theorem 3.2). Combining this with a non-invertible injective representation τ of C*_r(F2) into the Calkin algebra (from Haagerup–Thorbjørnsen) and the elementary Ext-obstruction of Lemma 2.1, the authors obtain a three-dimensional Calkin subsystem ẊS whose identity map admits no u.c.p. lift (Proposition 4.1). Equivalently, the operator T = T1 + i T2 formed by self-adjoint lifts of τ4(D) and τ4(K) has no compact perturbation realizing the full joint matrix range of q(T) (Theorem 4.2). By Kavruk’s duality the dual of ẊS is therefore a three-dimensional non-exact operator system (Corollary 4.3).","tokens_in":8817,"tokens_out":980,"duration_ms":8907,"significance":"The result settles the generalized Smith–Ward problem in dimension three and supplies the first example of a three-dimensional operator system that fails to be exact. The reduction from Harris’ four-dimensional coding to a hyperrigid three-dimensional system is achieved by a transparent spectral-propagation argument that recovers the four projections of D from the two endpoint subspaces and the partial-isometry edges of K. All steps rely only on standard tools (Stinespring dilation, Arveson extension, Powers simplicity, Haagerup–Thorbjørnsen non-group Ext) and are written in full detail; the construction is therefore both novel and immediately usable for further work on lifting and exactness questions in low-dimensional operator systems.","major_comments":[],"minor_comments":[{"comment":"Abstract and page 1: the phrase “a counterexamples” should be “counterexamples” (or “a counterexample”).","section":"Abstract"},{"comment":"Page 1, line 3 of the introduction: “into the Calkin algebra gives explicit four-dimensional…” reads more smoothly as “into the Calkin algebra yields explicit four-dimensional…”.","section":"Introduction"},{"comment":"Lemma 3.1: the phrase “Simple computations now imply eij ⊗ 1 ∈ C*(S)” could be expanded by one sentence indicating which products of the already-obtained corners produce the remaining matrix units.","section":"Lemma 3.1"},{"comment":"Theorem 3.2 proof, middle of page 6: the sentence beginning “Using V(π(p4)E) ⊆ ρ(p4)L the left hand side changes to…” is slightly dense; a brief parenthetical reminder that the other summands vanish by the already-established inclusions would improve readability.","section":"Theorem 3.2"},{"comment":"References: the arXiv identifier for Harris [6] is given as arXiv:2508.00113v2; if a published version appears before final production it should be updated.","section":"References"}],"recommendation":"accept","confidential_remarks":"The manuscript is short, self-contained and of clear interest to the operator-algebra community working on lifting, Ext and exactness. No concerns about novelty disclosure or citation pattern; the dependence on Haagerup–Thorbjørnsen and Kavruk is properly acknowledged. Suitable for a short note or research announcement in a strong OA journal."},"author_rebuttal":null,"desk_editor":{"model":"grok-4.5","letter":"The new thing here is the three-dimensional hyperrigid system S = span{1, D, K} inside M4(C*_r(F2)). Harris already got four-dimensional counterexamples from the same non-invertible Ext class; Scherer isolates the obstruction and replaces the four-generator coding with a concrete pair of self-adjoints whose C*-algebra is the full matrix algebra and whose restriction has the unique-extension property. That drops the dimensional threshold for the generalized Smith–Ward problem to three and, via Kavruk, yields the first three-dimensional non-exact operator system.\n\nWhat works: Lemma 2.1 is the standard Ext-obstruction written cleanly. Theorem 3.2 is the real contribution—the spectral projections of D are recovered from the two endpoint eigenspaces and then propagated along the partial-isometry edges of K by a sequence of norm-preservation arguments. It is elementary Stinespring plus linear algebra, written line-by-line, and I do not see a gap. Once unique extension is in hand, the rest is immediate: a u.c.p. lift of the identity on the Calkin image would invert [τ4], contradicting Haagerup–Thorbjørnsen. The dual non-exactness corollary is free.\n\nSoft spots are minor. The construction is tailored to this particular graph of K; the paper notes that a more general statement is possible but does not pursue it. Everything rests on the existence of a non-invertible injective representation of C*_r(F2), which is a published theorem, not a new assumption. No circularity, no hidden continuity or separability issues.\n\nThis is for people who work on operator systems, matrix ranges, or Ext. It is short, self-contained, and checkable. I would send it to a serious referee without hesitation; the central claim is supported and the novelty is real even if incremental. Worth reading if the dimension-three case matters to you.","headline":"Clean reduction of the Smith–Ward counterexample from four to three dimensions via an explicit hyperrigid triple; the argument holds.","tokens_in":9414,"tokens_out":498,"would_cite":true,"duration_ms":5264,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["46L07"],"pacs":[],"model":"grok-4.5","headline":"A three-dimensional operator system in the Calkin algebra has no unital completely positive lift of the identity, giving the first three-dimensional non-exact operator system.","keywords":["Smith-Ward problem","lifting property","hyperrigidity","operator system","exactness","Calkin algebra","Ext","matrix range"],"falsifier":"Exhibit a unital completely positive map from the concrete three-dimensional Calkin system span{1, q(D), q(K)} into B(H) that composes with the quotient map to give the identity, or prove that every injective representation of C*_r(F2) into the Calkin algebra has invertible Ext class.","tokens_in":9419,"feed_emoji":"⊗","tokens_out":717,"duration_ms":5441,"temperature":0.7,"pith_summary":"The paper constructs a three-dimensional operator system inside the Calkin algebra whose identity map admits no unital completely positive lift. The system is the image of a carefully chosen hyperrigid span{1, D, K} living in a matrix algebra over the reduced free-group C*-algebra on two generators; D is diagonal with four distinct eigenvalues and K is a self-adjoint matrix whose nonzero entries are the free generators and the unit. Because the system is hyperrigid, any lift of the identity would force a lift of a non-invertible Ext class, which is impossible. Equivalently, the single operator D + iK supplies a counter-example to the classical Smith–Ward problem: no compact perturbation can match its full joint matrix range. A duality theorem of Kavruk then shows that the dual of this system fails to be exact, producing the first known three-dimensional non-exact operator system. The result sharpens earlier four- and five-dimensional counter-examples and settles the dimension of the Smith–Ward obstruction at three.","feed_headline":"Three-dimensional operator system blocks Smith–Ward lifts","feed_subtitle":"A hyperrigid span in free-group matrices yields the first non-exact three-dimensional operator system.","key_machinery":"The three-dimensional hyperrigid operator system S = span{1, D, K} inside M4(C*_r(F2)), where D is diagonal with distinct eigenvalues and the nonzero entries of K are free unitaries and the unit. Hyperrigidity (unique extension property for every unital representation) converts any supposed lift of the image of S into a lift of a non-invertible Ext class, which is forbidden.","core_discovery":"There exist self-adjoint operators D and K on a separable Hilbert space such that the three-dimensional Calkin operator system span{1, q(D), q(K)} has no unital completely positive lift of the identity; equivalently the operator T = D + iK admits no compact perturbation L with W(T + L) = W(q(T)). The dual of this system is therefore the first three-dimensional operator system that is not exact.","pith_inferences":[],"forward_implications":[],"fun_headline_variants":["3D Calkin system span{1,q(D),q(K)} blocks unital CP lifts","Hyperrigid free-group span yields first non-exact 3D operator system","Self-adjoint D,K give Smith–Ward counterexample in three dimensions","Three-dimensional operator system without Smith–Ward property","Dual of Calkin span{1,q(D),q(K)} is first non-exact 3D system"],"cache_read_input_tokens":128,"weakest_assumption_plain":"The existence of an injective representation of the reduced free-group C*-algebra into the Calkin algebra whose Ext class is not invertible.","fun_headline_variants_meta":{"raw":{"variants":["3D Calkin system span{1,q(D),q(K)} blocks unital CP lifts","Hyperrigid free-group span yields first non-exact 3D operator system","Self-adjoint D,K give Smith–Ward counterexample in three dimensions","Three-dimensional operator system without Smith–Ward property","Dual of Calkin span{1,q(D),q(K)} is first non-exact 3D system"]},"model":"grok-4.5","effort":"low","cost_usd":0.005704,"raw_usage":{"total_tokens":1460,"prompt_tokens":764,"num_sources_used":0,"completion_tokens":118,"cost_in_usd_ticks":57040000,"prompt_tokens_details":{"text_tokens":764,"audio_tokens":0,"image_tokens":0,"cached_tokens":0},"completion_tokens_details":{"audio_tokens":0,"reasoning_tokens":578,"accepted_prediction_tokens":0,"rejected_prediction_tokens":0}},"tokens_in":764,"tokens_out":118,"duration_ms":5936,"temperature":1.0,"reasoning_tokens":578,"cache_read_input_tokens":0,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-07-14T16:22:20.248663+00:00","model_set":{"reader":"grok-4.5"},"falsifier":"Exhibit a unital completely positive map from the concrete three-dimensional Calkin system span{1, q(D), q(K)} into B(H) that composes with the quotient map to give the identity, or prove that every injective representation of C*_r(F2) into the Calkin algebra has invertible Ext class.","supporting_citations":[],"review_version":2}