{"id":"56ba7c22-e428-4270-ba3c-8a3c08aa7fcd","arxiv_id":"2511.02760","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"For acyclic graphs and graph algebras with finitely many ideals, Z-stability is equivalent to a new graph condition ('distinct detours'); with Condition (K) this condition also characterizes purity.","lead":"This paper introduces a graph condition called \"distinct detours\" and shows it exactly captures when the associated operator algebra absorbs the Jiang–Su algebra, in acyclic graphs and graphs with finitely many ideals. The result gives a combinatorial test for Z-stability and connects it to purity, a step toward the generalized Toms–Winter conjecture.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The asserted density of corners of finite in-trees in p_v C*(E)p_v is false for acyclic distinct-detour graphs; the proof of Theorem A case (a) rests on an invalid step.","rationale":"The reader identified the same load-bearing step but treated it as a gap repairable by a density argument. My analysis shows the step is actually false: there are graphs satisfying the hypotheses of Theorem A case (a) for which the corners of finite entrance-complete in-trees are not dense in p_v C*(E)p_v. The bi-infinite ladder graph provides a concrete counterexample to the asserted density. This is not an internal inconsistency in the statement of Theorem A, but it invalidates the proof of the sufficient direction. The central claim may still be true by a different argument, but as written the paper's main theorem is unsupported. Hence I recommend REJECT rather than CONDITIONAL. I credit the reader for locating the exact assertion, but the severity is greater than 'likely repairs it.'","tokens_in":16213,"tokens_out":33480,"duration_ms":337866,"concrete_test":"In the bi-infinite ladder graph E defined in the attack (vertices a_n,b_n for n∈Z; edges a_n→a_{n-1}, a_n→b_{n-1}, b_n→a_{n-1}, b_n→b_{n-1}), set v=a_0, μ=a_2→a_1→a_0, ν=a_2→b_1→a_0, and u=s_μ s_ν^*. Let S be the union of p_v L_C(T)p_v over all finite entrance-complete in-trees T⊂E containing v. Compute (or prove lower bound for) dist(u, S). If dist(u, S) ≥ 1, the density assertion in the proof of Theorem A case (a) is false. An analytic check: for any x∈p_v L_C(T)p_v, if T omits μ, then (s_μ s_μ^*) x = 0, while (s_μ s_μ^*) u = u, so ||u-x|| ≥ 1; similarly if T omits ν. Since a tree cannot contain both μ and ν, every T omits at least one, forcing the distance to be at least 1.","verdict_should_be":"REJECT","load_bearing_attack":"In the proof of Theorem A, case (a), the sentence 'For each n, we may find an entrance-complete finite in-tree (F_n,v) such that the distance between F_n and p_v L_C(F_n)p_v is less than 1/2^n' asserts that the union of corners p_v L_C(T)p_v over all finite entrance-complete in-trees T⊂E is dense in p_v C*(E)p_v. This is not merely unproved; it is false under the stated hypotheses. Consider the row-finite acyclic graph with vertices a_n,b_n (n∈Z) and edges a_n→a_{n-1}, a_n→b_{n-1}, b_n→a_{n-1}, b_n→b_{n-1}. It has no sources and satisfies distinct detours: for any infinite simple path, an alternative route via the other chain gives a distinct detour. Let v=a_0, and let μ=a_2→a_1→a_0 and ν=a_2→b_1→a_0 be two paths from a_2 to v. The element u=s_μ s_ν^* lies in p_v C*(E)p_v. Any entrance-complete finite in-tree containing both μ and ν would force a_2 to have two outgoing edges, contradicting the tree property; any tree containing at most one of these paths has a corner that cannot contain u. Moreover, because u is a partial isometry between the orthogonal projections s_μ s_μ^* and s_ν s_ν^*, the distance from u to any such corner is at least 1. Thus the asserted approximability fails, and the density argument cannot repair this step. The proof of the sufficient direction (distinct detours ⇒ Z-stability) collapses.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper introduces a graph-theoretic condition called \"distinct detours\" and claims it characterizes Z-stability of graph C*-algebras under additional hypotheses. Theorem A asserts that for a countable row-finite graph that is either acyclic or has C*(E) with finitely many ideals, the following are equivalent: distinct detours, Z-stability of C*(E), and absence of elementary subquotients. Theorem B gives a finite-graph characterization, and Theorem C asserts that Condition (K) plus distinct detours is equivalent to C*(E) being pure. The paper also proposes a conjecture extending these equivalences to all row-finite graphs. The proofs rely on approximations by finite in-trees, composition series for ideals, and several recent results on pure C*-algebras.","tokens_in":16644,"tokens_out":34075,"duration_ms":324205,"significance":"If correct, the paper would provide a useful combinatorial criterion for Z-stability and pureness in a broad class of graph algebras, with direct bearing on the Generalized Toms–Winter Conjecture. The proposed condition of distinct detours is natural and the overall program is well motivated. However, the manuscript contains a simple counterexample to Theorem A case (b) and to Theorem C, and the proof of Theorem A case (a) rests on a false density assertion. These are not mere presentation issues; they invalidate the main theorems as stated. The paper cannot be accepted in its present form.","major_comments":[{"comment":"The statement of Theorem A case (b) is false. Let E be the graph with one vertex v and one loop e. Then E is row-finite and C*(E) is isomorphic to C(T), which has only two ideals, so case (b) applies. Since E is finite and has no sources, the author's own observation before Theorem B implies that E trivially has distinct detours. Moreover, C(T) has no elementary subquotients: C(T) is simple and not isomorphic to K(H). But C(T) is not Z-stable. Thus (i) and (iii) hold while (ii) fails. The proof's assertion that C*(E) having finitely many ideals implies Condition (K) is incorrect: C(T) is a simple graph algebra with finitely many ideals and fails Condition (K).","section":"Theorem A, case (b); Section 3"},{"comment":"The same one-vertex, one-loop graph disproves the equivalence (ii) => (i) in Theorem C. It satisfies (ii): C(T) has no elementary subquotients. It satisfies distinct detours vacuously, since there is no simple path in E^{≤∞}. But it fails Condition (K), since the vertex on the cycle has only one return path. The proof states that failure of Condition (K) \"produces a subquotient stably isomorphic to C(T). This subsequently yields an elementary subquotient.\" The final step is wrong: C(T) (and also C(T)⊗K) is simple and non-elementary, so it does not yield an elementary subquotient. Theorem C is therefore false as stated.","section":"Theorem C; Section 4"},{"comment":"The sentence \"For each n, we may find an entrance-complete finite in-tree (F_n,v) such that the distance between F_n and p_v L_C(F_n)p_v is less than 1/2^n\" is load-bearing and is not justified. In fact the asserted density is false. Consider the row-finite acyclic graph with vertices a_n,b_n (n∈Z) and edges a_n→a_{n-1}, a_n→b_{n-1}, b_n→a_{n-1}, b_n→b_{n-1}. This graph has distinct detours. Let v=a_0, μ=a_2→a_1→a_0, ν=a_2→b_1→a_0, and u=s_μ s_ν^*. Any entrance-complete finite in-tree containing both μ and ν would give a_2 two outgoing edges, contradicting the tree property; any tree missing one of these paths cannot contain u in its corner, and the element u is not approximable by such a corner. Thus the approximation step, and with it the proof of (i)⇒(ii) for acyclic graphs, collapses.","section":"Proof of Theorem A, case (a); density assertion after defining F_n"},{"comment":"Two further problems affect the proof of Theorem A case (a). First, the construction in Lemma 2.7 does not preserve the in-tree property: adding all edges r^{-1}(F_n^0) can give a single vertex two outgoing edges. In the graph described above, if F_0 contains a_1→a_0, then F_1 contains both a_2→a_1 and a_2→b_1, so F_1 is not a tree. Second, the proof asserts that (iii)⇒(i) \"was proven in Lemma 2.4,\" but Lemma 2.4 proves only that Z-stability implies distinct detours. It does not show that absence of elementary subquotients implies distinct detours. An additional argument would be needed to upgrade the ideal I constructed in Lemma 2.4 to an elementary subquotient in the acyclic case.","section":"Lemma 2.7; Theorem A, case (a), (iii)=> (i)"}],"minor_comments":[{"comment":"The notation F_n is used both for a finite subset of p_v C*(E)p_v and for an entrance-complete finite in-tree. This overloading is confusing and should be fixed.","section":"Proof of Theorem A, case (a)"},{"comment":"The sentence \"If v has a source\" appears to be a typo; it should presumably read \"If v is a source\" or \"If E has a source v.\"","section":"Proof of Theorem B"}],"recommendation":"reject","confidential_remarks":"The counterexample to Theorem A case (b) is elementary and also disproves Theorem C. Because the central theorems are false as stated, no amount of local revision can make the paper acceptable without substantial reformulation of the main claims."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Greg, quick take on 2511.02760. The paper introduces a promising new condition — \"distinct detours\" — and Lemma 2.4, showing it is necessary for Z-stability, is clean and correct. The finite-graph Theorem B is also plausibly assembled from earlier work. But the sufficiency proof of Theorem A, case (a), has a load-bearing step that is not merely unproved; it is false.\n\nThe problem is the assertion that for every finite subset of p_v C*(E)p_v one can find an entrance-complete finite in-tree F approximating it within 1/2^n. Consider the acyclic row-finite graph with vertices a_n,b_n (n∈Z) and edges a_n→a_{n-1}, a_n→b_{n-1}, b_n→a_{n-1}, b_n→b_{n-1}. It has distinct detours. Let v=a_0, and take the two paths μ=a_2→a_1→a_0 and ν=a_2→b_1→a_0. Then u=s_μ s_ν^* belongs to p_v C*(E)p_v. Any entrance-complete finite in-tree that contained both μ and ν would force a_2 to have two outgoing edges, impossible in a tree. If the tree contains at most one of these paths, its corner cannot contain u; indeed, since uu^* and u^*u are orthogonal projections, the distance from u to any such corner is at least 1. So the asserted density fails.\n\nThat kills the proof of distinct detours ⇒ Z-stability for acyclic graphs. Since Theorem A case (a) is also used in Theorem B and Theorem C, the paper's main results are not supported as they stand. The statements are likely true — the AF case is known to experts — but this graph-theoretic proof does not work. The author would need a substantially different argument.\n\nWhat the paper does well: the distinct detours notion is a real contribution, and the necessity lemma is solid. The writing is clear and the citations are honest. The black-box use of deep results in Theorem C is a minor concern compared with the central gap.\n\nA serious referee could check whether a different approximation scheme can rescue the proof, but as it stands the paper is a promising draft, not a finished proof.","headline":"The distinct detours condition is a genuinely nice idea and the necessity proof works, but the sufficiency proof of Theorem A rests on a false approximation assertion; the main theorems are not established as written.","tokens_in":17076,"tokens_out":29297,"would_cite":false,"duration_ms":245245,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["46L05"],"pacs":[],"model":"deepseek-v4-flash","headline":"A new graph condition, 'distinct detours,' is shown to be equivalent to Z-stability of the associated C*-algebra when the graph is acyclic or has finitely many ideals, and to pureness in general for row-finite graphs.","keywords":["distinct detours","graph C*-algebra","Z-stability","Jiang–Su algebra","pureness","Condition (K)","row-finite graph","Toms–Winter conjecture"],"falsifier":"Compute whether C*(E) is Z-stable for the acyclic graph with two sources u_1, u_2 each having an edge to v_0, and edges v_n→v_{n-1} for n≥1 (an infinite ray). This graph has distinct detours. If its C*-algebra has an elementary subquotient or fails to be Z-stable, Theorem A(a) is false; if not, the density step survives this test.","tokens_in":16124,"feed_emoji":"🕸️","tokens_out":6026,"duration_ms":63749,"temperature":0.7,"pith_summary":"The paper introduces a combinatorial condition on directed graphs called 'distinct detours': every simple infinite path (or path from a source) must have a detour that uses an edge not on the path. It proves that for a countable row-finite graph, this condition is equivalent to the graph C*-algebra tensoring with the Jiang–Su algebra Z without changing, provided the graph is acyclic or the algebra has finitely many ideals. Together with Condition (K), distinct detours is shown equivalent to the algebra being pure in the Cuntz-semigroup sense, which verifies the generalized Toms–Winter conjecture for these cases. If the proposed conjecture holds, Z-stability of graph algebras would be a purely graph-theoretic property.","feed_headline":"Distinct detours characterize Z-stable graph algebras","feed_subtitle":"A purely combinatorial path property tells when a graph's operator algebra absorbs the Jiang–Su algebra.","key_machinery":"The central object is the 'distinct detours' condition, a divisibility-type property on vertex projections. The proof machinery combines two tools: (1) a corner-wise reduction showing C*(E) is D-stable iff each corner p_v C*(E)p_v is D-stable; (2) a construction of approximately central matrix-unit systems for M2⊕M3 inside corners, using nondegenerate inclusions of finite entrance-complete in-trees, so that the Robert–Tikuisis criterion (two full orthogonal elements in the central sequence algebra) applies. Lemmas 2.5–2.8 build these matrix units from paths in the graph.","core_discovery":"The paper's central claim is that 'distinct detours' is the right combinatorial shadow of Z-stability for graph C*-algebras. For a row-finite graph E, C*(E)⊗Z ≅ C*(E) if and only if every simple path that is either infinite or starts at a source has a detour containing an edge not on the path, under the hypotheses of Theorem A (acyclic or finitely many ideals). In the finite-graph case this reduces to Condition (K) plus having no sources. The paper further proves that Condition (K) plus distinct detours is equivalent to C*(E) being pure, and conjectures this is exactly Z-stability.","pith_inferences":["The unproved assertion in the proof of Theorem A(a) — that arbitrary finite subsets of a corner can be approximated by corners of finite entrance-complete in-trees — is likely repairable by a density argument using row-finiteness and the Cuntz–Krieger relations, but it is the spot to check first.","The conjecture, if true, would imply a graph-theoretic algorithm for Z-stability: check Condition (K) and distinct detours, both of which are finite-time verifiable on finite graphs.","The distinct-detours condition may be equivalent to a known divisibility property of the Murray–von Neumann semigroup of the graph algebra, which would connect the combinatorial condition to K-theoretic invariants.","For non-row-finite graphs, Remark 4.2 suggests a modification involving cycles through infinite receivers; testing whether that amended condition is necessary and sufficient would be a natural extension."],"forward_implications":["For acyclic row-finite graphs, C*(E)⊗Z ≅ C*(E) is now equivalent to having no elementary subquotients, giving a graph-theoretic proof of a known result about AF algebras.","For graph algebras with finitely many ideals, Z-stability is characterized entirely by distinct detours, and the generalized Toms–Winter conjecture holds in this class.","For finite graphs, Z-stability, O_infinity-stability, absence of elementary subquotients, and Condition (K)+no sources all coincide.","If Conjecture 4.1 holds, Z-stability of any row-finite graph algebra is decidable by checking two graph-theoretic conditions.","The pureness equivalence (Theorem C) gives a large new class where pureness and Z-stability may coincide, matching the generalized conjecture."],"fun_headline_variants":["Distinct detours decide Z-stability in graph algebras","Graph algebras: distinct detours mean Z-stability","Z-stability for graph algebras? Check for distinct detours","Distinct detours: the key to Z-stable graph algebras","When do graph algebras absorb Z? Distinct detours tell you"],"cache_read_input_tokens":2304,"weakest_assumption_plain":"The proof of Theorem A(a) assumes, without proof, that for each finite subset of the corner p_v C*(E)p_v one can find a finite entrance-complete in-tree whose corner approximates that subset within 1/2^n; if this density statement fails, the constructed approximately central matrix units may not exist.","fun_headline_variants_meta":{"raw":{"variants":["Distinct detours decide Z-stability in graph algebras","Graph algebras: distinct detours mean Z-stability","Z-stability for graph algebras? Check for distinct detours","Distinct detours: the key to Z-stable graph algebras","When do graph algebras absorb Z? Distinct detours tell you"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000679,"raw_usage":{"total_tokens":2876,"prompt_tokens":649,"completion_tokens":2227,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":393,"completion_tokens_details":{"reasoning_tokens":2143}},"tokens_in":393,"tokens_out":2227,"duration_ms":16959,"temperature":1.0,"reasoning_tokens":2143,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-04T00:04:27.438929+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Compute whether C*(E) is Z-stable for the acyclic graph with two sources u_1, u_2 each having an edge to v_0, and edges v_n→v_{n-1} for n≥1 (an infinite ray). This graph has distinct detours. If its C*-algebra has an elementary subquotient or fails to be Z-stable, Theorem A(a) is false; if not, the density step survives this test.","supporting_citations":[],"review_version":1}