{"id":"3abed20b-0f76-40fb-acdc-7f2652e93867","arxiv_id":"1908.08315","paper_version":1,"verdict":"ACCEPT","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"low","formal_verification":"none","parameter_count":0,"one_line_summary":"For every subshift, the Matsumoto and Carlsen-Matsumoto C*-algebras are realized as groupoid C*-algebras from the inverse hull of the language semigroup, and this universal groupoid is amenable.","lead":"The paper builds a unified inverse-semigroup framework for the two best-known C*-algebras attached to a subshift, showing both arise as groupoid C*-algebras from one universal object. It also proves the universal groupoid is amenable and gives a condition under which the two algebras agree.","discovery_kind":"unification","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Corollary 10.4 hinges on an asserted but unproved equivalence between Definition 6.4 and the amended Carlsen-Matsumoto condition (*); Theorem 10.3 itself stands.","rationale":"The reader's weakest assumption was the imported full invariance result [22:13.4] used in Proposition 8.3. That is a legitimate concern, but it is a cited theorem from the authors' companion paper, and the direct check for subshift semigroups via Proposition 3.2 and 3.5 is straightforward; the risk is low. A more load-bearing gap is the equivalence between Definition 6.4 and the amended Carlsen-Matsumoto condition (*), which is explicitly left unproved in Section 6 while also correcting the published statement of that condition. Corollary 10.4, part of the reader's strongest claim, depends exactly on this equivalence. If the equivalence is wrong or incomplete, the corollary does not follow from Theorem 10.3, even though Theorem 10.3 and the main groupoid models remain intact. The appropriate verdict is therefore CONDITIONAL: accept the main theorem, but require the authors to supply the missing equivalence proof or to restate Corollary 10.4 solely in terms of Definition 6.4. The concern is specific, testable, and confined to the corollary rather than the central construction; no issue was found with the internal logic of Theorem 10.3, the support computations, or the amenability arguments.","tokens_in":40567,"tokens_out":45135,"duration_ms":420658,"concrete_test":"State the amended condition (*) of [11:Section 3] explicitly, with the infinite-range requirement, and prove both implications with Definition 6.4 for the language semigroup S_X. Specifically, show: (a) if the amended CM condition holds, then for every finite Λ,Γ ⊆ S~_X with F^θ_Λ,Γ infinite there is an infinite word ω ∈ X such that tω ∈ X for all t ∈ Λ and rω ∉ X for all r ∈ Γ; (b) conversely, the semigroup formulation implies the amended CM condition. Pay attention to elements equal to 1, to words of length zero, and to the requirement that the chosen sequence has infinite range. If either implication fails for a particular subshift, Corollary 10.4 should be reworded to use Definition 6.4 directly.","verdict_should_be":"CONDITIONAL","load_bearing_attack":"The paper's Corollary 10.4 claims that if S_X satisfies the Carlsen-Matsumoto condition (*), then M_X is naturally isomorphic to O_X. The proof invokes Theorem 6.7, whose condition (*) is Definition 6.4, formulated in terms of the semigroup S_X and the sets F^θ_Λ,Γ. In Section 6 the authors state that this is equivalent to the condition (*) in Carlsen-Matsumoto [11:Section 3], but only after amending that statement by requiring the sequence {μ_i} to have infinite range, and they leave the verification to the reader. This is not a routine translation: it involves comparing a condition on all finite pairs (Λ,Γ) of subsets of S~_X with a condition on infinite words and finite collections of words in the original subshift, including edge cases with the unit 1 and with words that are not admissible. Because the original [11] statement is said to be incorrect, the precise corrected formulation is not available in the cited literature. If the equivalence fails in either direction, Corollary 10.4 would not follow from Theorem 10.3, and the paper's advertised connection to the classical Carlsen-Matsumoto condition would be unsupported. Theorem 10.3 itself, and the groupoid models for M_X and O_X, do not depend on this equivalence.","agreement_with_reader":"disagree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper develops a unified inverse-semigroup framework for the C*-algebras associated with a one-sided subshift X over a finite alphabet. The authors study the language semigroup S_X = L_X ∪ {0} (concatenation when admissible, zero otherwise) and its inverse hull H(S_X). They introduce the notion of essentially tight representations and characters (Definition 2.2), and compare four invariant subspaces of the character space: E^max ⊆ E^∞ ⊆ E^tight and E^max ⊆ E^ess. They prove that H(S_X) is strongly 0-E-unitary with universal group the free group and that the associated partial action is semi-saturated and orthogonal, so the universal groupoid is a Deaconu-Renault groupoid and hence amenable (Theorem 9.6, Corollaries 9.7-9.8). The main results are the groupoid models M_X ≅ C*(G^ess_X) and O_X ≅ C*(G^max_X) (Theorem 10.3), with Corollary 10.4 asserting that M_X ≅ O_X under a condition (*) (Definition 6.4) identified with the Carlsen-Matsumoto condition. A worked counterexample (Section 4) shows that the boundary-finiteness hypothesis of Theorem 2.10 cannot be removed.","tokens_in":40795,"tokens_out":32674,"duration_ms":279301,"significance":"If correct, the paper gives a systematic construction of both the Matsumoto and Carlsen-Matsumoto algebras from the single object H(S_X), identifying the Matsumoto algebra with the reduction to the essentially tight spectrum and the Carlsen-Matsumoto algebra with the reduction to the closure of the maximal-string spectrum. The essentially tight spectrum is a natural new notion in the tight-representation theory of Exel and Paterson, and Theorem 9.6 is a generally useful result on semi-saturated orthogonal partial actions of free groups, with amenability consequences for all reductions of the universal groupoid. The support computations (7.14, 8.9) and the Deaconu-Renault realization (9.6) are proven in detail, and the counterexamples in Sections 3 and 4 are concrete and checkable. The principal caveats are the unproved equivalence between Definition 6.4 and the amended Carlsen-Matsumoto condition (*), which is the advertised basis of Corollary 10.4, and the sketched proof of the O_X half of Theorem 10.3.","major_comments":[{"comment":"The paper asserts, immediately after Prop. 6.5, that Definition 6.4 of condition (*) is equivalent to the condition (*) of Carlsen and Matsumoto [11: Section 3], but only after amending the [11] statement with an additional requirement that the sequence {μ_i} have infinite range, and it leaves the verification to the reader. This equivalence is load-bearing: Corollary 10.4 advertises the natural isomorphism M_X ≅ O_X under 'Carlsen and Matsumoto's condition (*) (see (6.4))', and since the paper states that the [11] formulation is incorrect as written, the reader has no published statement to fall back on. The translation is not a routine restatement: it must relate a condition over all finite pairs (Λ,Γ) of subsets of S̃_X to a condition on sequences of words in the subshift, with edge cases involving the unit 1 and inadmissible concatenations. The authors should either prove the equivalence of (6.4) with the amended [11] condition in full, or restate Corollary 10.4 as a theorem about Definition 6.4 alone and mark the comparison with [11] as a conjecture. Theorem 10.3 and the groupoid models themselves do not depend on this equivalence.","section":"§6.4–6.5, Cor. 10.4"},{"comment":"The proof of part (ii) of Theorem 10.3 — the isomorphism O_X ≅ C*(G^max_X) — is the second half of the paper's headline theorem, yet the final paragraph delegates the verification to the reader. The stated key point is that for non-idempotent α ∈ H(S_X) one has d(α) ≠ 1, so ρ(α) = π(α) ⊗ λ_{d(α)} has no non-zero diagonal coefficients; what is not written out is the analogue of the part (i) injectivity argument: verification of the commuting diagram for the conditional expectations P and Q, identification of the kernel of Φ on C0(E^max(S_X)) using the support computation (8.9), and the use of amenability (9.8) to get faithfulness of P. Since this is one of the two central claims of the paper, the details should be supplied rather than left as an exercise.","section":"Thm 10.3(ii)"}],"minor_comments":[{"comment":"In the proof of Proposition 6.7, Z is defined as Z := X \\ ⋂_{j=1}^m Y_j, but the immediately following display (1 = φ(X) = ⋁_{j=1}^m φ(Y_j) = 0) and the later step asserting W ∩ F_{Δ_j} = ∅ for every j both require Z = X \\ ⋃_{j=1}^m Y_j. Please correct this typo, which currently makes the argument look inconsistent with the definition of essential tightness in (5.2).","section":"§6.7, proof of Prop. 6.7"},{"comment":"Proposition 8.3 constructs the Carlsen-Matsumoto representation π on ℓ²(X) using the full invariance of the maximal-string set S^∞_X under the action of H(S_X) on the string space, imported as [22:13.4] from the companion preprint; the proof of Theorem 10.3(ii) inherits this dependence. The authors should state the publication status of [22] and confirm that the numbering refers to the final version, so that this background theorem (and the other imports, e.g. [22:7.13, 7.21, 10.19]) can be checked in the published source.","section":"Prop. 8.3, [22:13.4]"},{"comment":"In the proof of Theorem 10.3(i), the text says that H(S_X) is 0-E-unitary by (8.7), but Proposition 8.7 proves the stronger property 'strongly 0-E-unitary'. The argument uses the idempotent-pure property d^{-1}(1) = E(S_X) that comes with the stronger statement; please make the implication explicit.","section":"Thm 10.3(i)"},{"comment":"In Section 8, the symbol ρ is used for two different maps: the representation of H(S_X) on the string space in the proof of Proposition 8.3, and the representation of H(S_X) into O_X in Proposition 8.8. Renaming one of them would prevent confusion, since both appear within a few pages.","section":"§8"}],"recommendation":"major_revision","confidential_remarks":"The main results are announced in the authors' earlier conference paper [21], but the present manuscript is substantially more detailed, so I see no novelty-disclosure problem. The paper depends heavily on the companion preprint [22], which was arXiv-only at the time of submission; I would ask the editor to confirm that it has appeared in final, citable form before acceptance. The gap described in major comment 1 is genuine but localized, and in my view fixable within the scope of a revision."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Two things you should know. First, this is a genuine advance: Exel and Steinberg show that both Matsumoto's algebra M_X and the Carlsen-Matsumoto algebra O_X arise as C*-algebras of reductions of one universal groupoid built from the inverse hull of the language semigroup. The new essentially tight spectrum is the right spectrum for M_X, and the amenability theorem (via Deaconu-Renault) is a nice stand-alone result. Second, the one real gap is in Corollary 10.4: the equivalence between their condition (6.4) and Carlsen-Matsumoto's condition (*) is asserted without proof, and they warn that the original statement in [11] is wrong. That is not a load-bearing flaw for Theorem 10.3, but it does mean the advertised 'condition (*) implies isomorphism' result is not yet established.\n\nThe paper does several things well. The support calculations in Props 7.14 and 8.9 are detailed and credible. The proof that the universal groupoid is a Deaconu-Renault groupoid for semi-saturated orthogonal actions (Theorem 9.6) is clean and has independent value. The example in Section 4 shows the essential tightness hypothesis is genuinely needed. There is also an interesting discussion of why the naive Toeplitz representation fails.\n\nThe main weakness, beyond the Cor 10.4 gap, is the heavy reliance on the companion paper [22]. Normal forms, string machinery, and invariance results are imported without much explanation. For the target audience this is probably acceptable—[22] is prior work, not a circular dependency—but it makes the paper hard to read alone.\n\nMy overall read: the central claim, Theorem 10.3, is likely correct and is a substantial contribution. The condition (*) equivalence should be filled in; as written, the proof of Cor 10.4 is incomplete. This is fixable and does not undermine the rest.\n\nRecommendation: send it to a serious referee. It deserves the referee time even with the known gap; a good referee will push for a complete proof of the equivalence or a clear reference. I would cite it for the groupoid models.","headline":"Inverse-semigroup unification of subshift algebras with solid groupoid models and an amenability theorem, but Corollary 10.4 rests on an unproved equivalence.","tokens_in":41356,"tokens_out":2932,"would_cite":true,"duration_ms":29217,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["37B10","46L05","46L55","20M18","22A22"],"pacs":[],"model":"deepseek-v4-flash","headline":"For any one-sided subshift, the Matsumoto and Carlsen-Matsumoto C*-algebras are reductions of the universal groupoid of the inverse hull of the language semigroup.","keywords":["subshift","language semigroup","inverse hull","Matsumoto algebra","Carlsen-Matsumoto algebra","etale groupoid","tight spectrum","partial crossed product"],"falsifier":"Take the concrete subshift of Section 4 (alphabet {0,1,2,3,4} with forbidden words 10+4[0,2,3,4], 20+4[0,1,3,4], and 30+4), where F{1,2} has infinite boundary and E(S_X) is not essentially tight. Compute the K-theory or the ideal structure of C*(G^ess_X) and C*(G^max_X) (equivalently, of M_X and O_X) for this subshift; if they differ, the two groupoid reductions are genuinely non-isomorphic, and if they agree despite E^max not being dense in E^ess, then condition (*) is not necessary for M_X ≅ O_X.","tokens_in":40323,"feed_emoji":"🔗","tokens_out":5314,"duration_ms":133172,"temperature":0.7,"pith_summary":"The paper claims that the two C*-algebras most often attached to a one-sided subshift, Matsumoto's algebra M_X and the Carlsen-Matsumoto algebra O_X, are not separate ad hoc constructions: both are groupoid C*-algebras obtained by reducing one universal groupoid, the groupoid of germs of the inverse hull H(S_X) of the language semigroup S_X. The Matsumoto algebra comes from the reduction to the essentially tight spectrum, while the Carlsen-Matsumoto algebra comes from the reduction to the closure of the maximal-string spectrum. The paper also proves these groupoids are amenable, realized as Deaconu-Renault groupoids, and that when the Carlsen-Matsumoto condition (*) holds the two spectra coincide, giving a natural isomorphism M_X ≅ O_X. A sympathetic reader would care because this unifies previously separate constructions and explains their difference as a choice of invariant subspace, not a difference in foundations.","feed_headline":"One inverse semigroup builds both subshift C*-algebras","feed_subtitle":"Matsumoto and Carlsen-Matsumoto algebras are groupoid algebras of the same universal groupoid, differing only in which spectrum is used.","key_machinery":"The load-bearing object is the inverse hull H(S_X) of the semigroup S_X = L_X ∪ {0}, where multiplication is concatenation when the result is an admissible word and zero otherwise. Its idempotent semilattice E(S_X) consists of constructible sets of finite words, and the paper studies several closed invariant subspaces of the character space of E(S_X): the essentially tight characters E^ess, the maximal-string characters E^max, the ultra-characters, and the tight characters. The essential tightness condition is defined modulo finite sets, and the maximal strings correspond bijectively to infinite words of the subshift. These spectra are the supports of natural representations of H(S_X), and Theorem 10.3 follows by applying the standard groupoid-model machine for inverse semigroups to those representations.","core_discovery":"The central result is Theorem 10.3: for a subshift X on a finite alphabet, the Matsumoto algebra M_X is isomorphic to C*(G^ess_X), the C*-algebra of the reduction of the universal groupoid of H(S_X) to the essentially tight spectrum E^ess(S_X), and the Carlsen-Matsumoto algebra O_X is isomorphic to C*(G^max_X), the reduction to the closure E^max(S_X) of the maximal-string characters. Corollary 10.4 then states that if S_X satisfies Carlsen and Matsumoto's condition (*), the two spectra agree enough that M_X is naturally isomorphic to O_X. The paper further shows that the universal groupoid of H(S_X) is an amenable Hausdorff etale groupoid, isomorphic to a Deaconu-Renault groupoid for a local homeomorphism built from the shift letters, and that both M_X and O_X are partial crossed products of the free group on the alphabet by a commutative C*-algebra.","pith_inferences":["Editorial inference: the same inverse-hull framework should apply to higher-rank shift spaces, since the paper explicitly leaves this open; one would need a language semigroup with a suitable length function and a version of the maximal-string invariance theorem.","Editorial inference: the dichotomy E^max versus E^ess suggests a natural testable refinement: for subshifts where condition (*) fails, compare K-theory or gauge-invariant ideals of M_X and O_X to detect whether the two groupoid models are genuinely different invariants.","Editorial inference: because the paper shows condition (*) is equivalent to density of E^max in E^ess, any subshift with finite constructible sets (like the examples in Sections 3 and 4) gives a concrete place to look for a Matsumoto algebra that is not isomorphic to the Carlsen-Matsumoto algebra.","Editorial inference: if the support computations of Propositions 7.14 and 8.9 extend to other 0-left-cancellative semigroups with locally finite length functions, the same groupoid-model argument may produce a unified picture for semigroup C*-algebras beyond the subshift setting."],"forward_implications":["Matsumoto's algebra M_X and the Carlsen-Matsumoto algebra O_X are both C*-algebras of amenable Hausdorff etale groupoids, so their full and reduced C*-algebras coincide and both admit a faithful conditional expectation onto the unit-space algebra.","When the subshift satisfies condition (*), M_X and O_X are naturally isomorphic, so the condition (*) singles out exactly the case where the essentially tight and maximal-string spectra support the same algebra.","Both algebras can be realized as partial crossed products of the free group on the alphabet by a commutative C*-algebra, which gives a uniform structural description and access to partial-action techniques.","The universal groupoid of H(S_X), and every reduction to an invariant subspace, is amenable because it is isomorphic to a Deaconu-Renault groupoid for a local homeomorphism arising from the shift letters.","The tight spectrum of H(S_X) produces a C*-algebra that has not been previously studied, leaving a new invariant of the subshift to be explored."],"supporting_citations":[{"why":"Supplies the universal groupoid construction, tight characters, and the groupoid-model theorem that Theorem 10.3 invokes.","marker":"[20]"},{"why":"Provides the inverse hull machinery, normal forms, strings, and full invariance of maximal strings used in Proposition 8.3.","marker":"[22]"},{"why":"Gives the original definition of the Matsumoto algebra that Definition 7.5 adapts.","marker":"[25]"},{"why":"Is the source of Definition 8.1 of the Carlsen-Matsumoto algebra and of the partial-action viewpoint.","marker":"[18]"},{"why":"Introduces condition (*), the comparison of Matsumoto and Carlsen-Matsumoto algebras that Corollary 10.4 extends.","marker":"[11]"},{"why":"Shows that the universal groupoid of a strongly 0-E-unitary inverse semigroup is a partial transformation groupoid, used to realize the free-group action.","marker":"[45]"},{"why":"Establishes that Deaconu-Renault groupoids of local homeomorphisms are amenable, which Corollary 9.8 uses.","marker":"[47]"},{"why":"Gives the Paterson groupoid model for inverse semigroup C*-algebras that Theorem 10.3 relies on.","marker":"[46]"}],"fun_headline_variants":["One inverse hull, two subshift C*-algebras","Subshift semigroup unifies Matsumoto algebras","Universal groupoid realizes both subshift algebras","Same semigroup, two spectra: one algebra","Theorem: subshift hull links C*-algebras"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The Carlsen-Matsumoto representation on ℓ2(X) requires the set of maximal strings to be fully invariant under the inverse hull H(S_X), a fact imported from the companion paper rather than verified here, and if that invariance failed for some subshift the groupoid model for O_X would not be defined.","fun_headline_variants_meta":{"raw":{"variants":["One inverse hull, two subshift C*-algebras","Subshift semigroup unifies Matsumoto algebras","Universal groupoid realizes both subshift algebras","Same semigroup, two spectra: one algebra","Theorem: subshift hull links C*-algebras"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000264,"raw_usage":{"total_tokens":1541,"prompt_tokens":822,"completion_tokens":719,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":438,"completion_tokens_details":{"reasoning_tokens":645}},"tokens_in":438,"tokens_out":719,"duration_ms":13529,"temperature":1.0,"reasoning_tokens":645,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T11:47:32.392137+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Take the concrete subshift of Section 4 (alphabet {0,1,2,3,4} with forbidden words 10+4[0,2,3,4], 20+4[0,1,3,4], and 30+4), where F{1,2} has infinite boundary and E(S_X) is not essentially tight. Compute the K-theory or the ideal structure of C*(G^ess_X) and C*(G^max_X) (equivalently, of M_X and O_X) for this subshift; if they differ, the two groupoid reductions are genuinely non-isomorphic, and if they agree despite E^max not being dense in E^ess, then condition (*) is not necessary for M_X ≅ O_X.","supporting_citations":[{"cited_title":"Inverse semigroups and combinatorial C*-alg ebras","cited_arxiv_id":null,"evidence_quote":"Supplies the universal groupoid construction, tight characters, and the groupoid-model theorem that Theorem 10.3 invokes."},{"cited_title":"Representations of the inver se hull of a 0-left-cancellative semigroup","cited_arxiv_id":null,"evidence_quote":"Provides the inverse hull machinery, normal forms, strings, and full invariance of maximal strings used in Proposition 8.3."},{"cited_title":"On C*-algebras associated with subshif ts","cited_arxiv_id":null,"evidence_quote":"Gives the original definition of the Matsumoto algebra that Definition 7.5 adapts."},{"cited_title":"Partial actions and subshift s","cited_arxiv_id":null,"evidence_quote":"Is the source of Definition 8.1 of the Carlsen-Matsumoto algebra and of the partial-action viewpoint."},{"cited_title":"Some remarks on the C*-a lgebras associated with subshifts","cited_arxiv_id":null,"evidence_quote":"Introduces condition (*), the comparison of Matsumoto and Carlsen-Matsumoto algebras that Corollary 10.4 extends."},{"cited_title":"On inverse semigroup C*-alg ebras and crossed products","cited_arxiv_id":null,"evidence_quote":"Shows that the universal groupoid of a strongly 0-E-unitary inverse semigroup is a partial transformation groupoid, used to realize the free-group action."},{"cited_title":"Cuntz-like algebras","cited_arxiv_id":null,"evidence_quote":"Establishes that Deaconu-Renault groupoids of local homeomorphisms are amenable, which Corollary 9.8 uses."},{"cited_title":"Groupoids, inverse semigroups, and their operator algebras","cited_arxiv_id":null,"evidence_quote":"Gives the Paterson groupoid model for inverse semigroup C*-algebras that Theorem 10.3 relies on."}],"review_version":1}