{"id":"9fb0f072-3e51-4dba-b576-7b34bfdb3a4f","arxiv_id":"2412.15112","paper_version":4,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"Hochschild and cyclic homology of Steinberg algebras of ample groupoids decompose in terms of groupoid homology, and for Exel-Pardo algebras these invariants and K-theory are computed by explicit cones and exact sequences.","lead":"This paper computes Hochschild, cyclic, and K-theoretic invariants of Steinberg algebras associated to ample groupoids, with detailed formulas for Exel-Pardo groupoids from self-similar graph actions. It also constructs a Dennis trace map from K-theory to groupoid homology and shows it is often an isomorphism onto the relevant homology summand.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Pseudo-freeness is used in two inequivalent senses: the introduction's edge-level definition is weaker than the path-level definition in Lemma 6.5.7, so Theorem 1.5(iii) may assert more than the proof establishes.","rationale":"The reader identified the pseudo-free hypothesis as the weakest assumption, and I agree that it is load-bearing; however, the sharper problem is that the paper uses the term in two different senses. The introduction defines pseudo-freeness on edges, while Lemma 6.5.7 and Theorem 6.5.13 require the stronger path-wise version, namely that no nontrivial group element strongly fixes a finite path. I constructed a valid row-finite Exel-Pardo tuple with trivial action on vertices in which the edge-level condition holds but path-wise pseudo-freeness fails: b fixes the loops e and f, φ(b,e)=φ(b,f)=a≠1, yet b strongly fixes the path ee. This shows the stated hypothesis of Theorem 1.5(iii) is not sufficient for the proof's main restriction lemma. The impact is direct: Lemma 6.5.7(b) underlies the identification of twisted groupoid homology with cone(I−τ), which in turn gives the exact sequence and the Dennis-trace diagram in Theorem 1.5(iii), Corollary 6.5.15, Corollary 6.5.16, and Lemma 6.7.1. The issue is not a disagreement with established consensus; it is an internal inconsistency in the manuscript's own definitions. The theorem is likely repairable by adopting the path-wise definition throughout, and the unconditional claims in parts (i) and (ii) appear unaffected. For these reasons the verdict should remain CONDITIONAL, but the condition should explicitly require aligning the pseudo-freeness hypothesis with the proof's path-wise version or supplying a proof of equivalence.","tokens_in":57902,"tokens_out":18523,"duration_ms":165961,"concrete_test":"Check the displayed one-vertex/two-loop example: E has loops e,f; G=(Z/2)^2 with a(e)=f,a(f)=e and b(e)=e,b(f)=f; set φ(a,e)=φ(a,f)=1, φ(b,e)=φ(b,f)=a, extended by (6.2.1). Verify the cocycle identities for all g,h and both loops. Then compute b(ee)=ee and φ(b,ee)=φ(a,e)=1, while b≠1 and e is b-fixed with φ(b,e)=a≠1. If these computations are correct, edge-wise pseudo-freeness holds but path-wise pseudo-freeness fails, so Theorem 1.5(iii)'s stated hypothesis does not match Lemma 6.5.7(b)'s hypothesis. As a second check, recompute Lemma 6.5.7(b)(ii) for the n=0 element ξ=a0 b a0* in this example: res(ξ) is supported at units [b,θ] with θ∈Z_ee, so res(ξ)≠0 although g0=b≠1, showing the statement fails under the edge-level reading.","verdict_should_be":"CONDITIONAL","load_bearing_attack":"Theorem 1.5(iii) and Theorem 6.5.13 rest on Lemma 6.5.7(b), where pseudo-freeness is defined path-wise: no nontrivial g strongly fixes a path (g(γ)=γ and φ(g,γ)=1). But the introduction defines pseudo-freeness edge-wise: g(e)=e with g≠1 implies φ(g,e)≠1. These notions are not equivalent in the row-finite, trivial-vertex-action setting. Concrete divergence: take E with one vertex and two loops e,f; G=(Z/2)^2 with a swapping e,f and b fixing both loops; set φ(a,e)=φ(a,f)=1 and φ(b,e)=φ(b,f)=a. The cocycle identities (6.2.1)-(6.2.2) hold, and the edge-level condition holds (b fixes e,f but φ(b,e)=φ(b,f)=a≠1). Yet b(ee)=ee and φ(b,ee)=φ(φ(b,e),e)=φ(a,e)=1, so b strongly fixes the path ee. Hence the edge-level hypothesis stated for Theorem 1.5 does not imply the path-level hypothesis used to prove Lemma 6.5.7(b) and Theorem 6.5.13. In particular, the key restriction/decomposition argument (nonzero res(ξ) iff g0...gn=1, and finiteness of minimal strongly fixed paths from [16, Theorem 12.2]) is only justified under the stronger path-wise definition. The theorem can be repaired by using the path-wise definition throughout, but as written the main hypothesis is ambiguous and the proof does not cover all tuples satisfying it.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"This paper studies homological invariants of Steinberg algebras of ample groupoids over a commutative ring k, with applications to twisted Exel-Pardo groupoids arising from self-similar group actions on graphs. The main structural results are: groupoid homology embeds into Hochschild homology of the Steinberg algebra and is a direct summand in the Hausdorff case; cyclic, negative cyclic, and periodic cyclic groupoid homology are computed from groupoid homology; and, under a discreteness assumption on isotropy, a Burghelea-type decomposition is obtained in terms of centralizer subgroups. For twisted Exel-Pardo data, assuming row-finiteness and trivial action on vertices, the paper computes each weight component of Hochschild homology as the cone of an explicit map 1 - sigma_m, computes twisted groupoid homology as the cone of 1 - tau under pseudo-freeness, gives homotopy algebraic K-theory exact sequences, and, under strengthenings such as regular supercoherent group rings, identifies K-theory with these cones and obtains a Dennis trace diagram. Consequences include K_0(L) = BF(E) and computations of K_1 in favorable cases.","tokens_in":58252,"tokens_out":9253,"duration_ms":65433,"significance":"If the main theorems hold as stated, this is a substantial and useful contribution. It unifies and extends several existing computations for Leavitt path algebras, Katsura algebras, and group algebras, and it provides explicit, formula-level descriptions of the relevant maps, not merely existence statements. The paper is well structured: Lemma 2.8.5, Proposition 3.8, Theorem 4.2, and Theorem 5.3.4 form a coherent chain, and the later Exel-Pardo computations are anchored by detailed cone descriptions. The reliance on the second author's previous work [12] is heavy but is not circular: those results are quoted with proofs and are logically independent. The paper also explicitly acknowledges the recent work of Miller and Steinberg [28] and the relation to the present results. The main issue, detailed below, concerns an inconsistency in the hypothesis of pseudo-freeness; it is localized and fixable, but it is load-bearing for Theorem 1.5(iii) and Theorem 6.5.13.","major_comments":[{"comment":"The manuscript uses the term 'pseudo-free' in two inequivalent senses, and the stronger sense is load-bearing. In the introduction, a tuple is declared pseudo-free if, for every edge e and every g ≠ 1, the condition g(e) = e implies φ(g,e) ≠ 1. In Section 6.5, however, pseudo-free is defined path-wise: 1 is the only element g such that g(γ) = γ and φ(g,γ) = 1 for some finite path γ. Lemma 6.5.7(b) is exactly the statement that this path-level condition is equivalent to the restriction property 'res(ξ) ≠ 0 implies g_0 ... g_n = 1', and Theorem 6.5.13 relies on that property, as well as on finiteness of minimal strongly fixed paths from [16, Theorem 12.2]. The edge-level condition does not imply the path-level condition. For example, take one vertex with two loops e and f, let G = (Z/2)^2 with a swapping e and f and b fixing both loops, and set φ(a,e) = φ(a,f) = 1, φ(b,e) = φ(b,f) = a. The cocycle identities (6.2.1)-(6.2.2) hold and the edge-level pseudo-freeness condition holds, but b strongly fixes the path ee: b(ee) = ee and φ(b,ee) = φ(φ(b,e),e) = φ(a,e) = 1. Thus Lemma 6.5.7(b) is false under the edge-level definition, and the restriction/decomposition argument in Theorem 6.5.13 is not justified for all tuples satisfying the hypothesis as stated in Theorem 1.5. The repair is straightforward: replace the introduction's definition with the path-level definition, whose edge-level version then follows as an immediate consequence; alternatively, state explicitly in Theorem 1.5 and throughout that the stronger path-level definition is intended. As written, the proof covers a strictly smaller class of tuples than the theorem asserts.","section":"§1; §6.5"}],"minor_comments":[{"comment":"The displayed exact sequence (1.6) appears to contain a stray '0' before HH_{n+1}(L/ℓ) and the arrow decorations are garbled; it should be reset to a clean exact sequence of Hochschild homology groups.","section":"Introduction"},{"comment":"The statement that the elements ξ with g_0 ... g_n = 1 generate H(G,k/ℓ) 'as an abelian group' should presumably be 'as an ℓ-module', since the complexes are ℓ-modules; this would avoid ambiguity about coefficients in k ⊗_ℓ^{n+1}.","section":"§6.5"},{"comment":"The map τ is introduced in the theorem as a matrix of chain homomorphisms with entries τ_{v,w}, while in the introduction (1.4) the same symbol denotes a single chain map; please align the notation or add an explicit remark that they are the same map under the matrix convention.","section":"§6.5"},{"comment":"The notation I(n)_v is used before it is defined; define it at the start of the proposition or immediately after (6.3.4).","section":"§6.3"}],"recommendation":"major_revision","confidential_remarks":"The pseudo-free inconsistency is the only obstruction I see to the central claims as written. It is a fixable issue, and the rest of the paper is careful and substantial. The heavy use of [12] is acceptable because the cited results are not assumed ad hoc; they are stated with proofs in the prior paper. I would encourage the editor to ask for a consistency pass on the definition of pseudo-free before accepting."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Colleague—\n\nThe paper is a real contribution. It gives a Burghelea-type decomposition for Hochschild and cyclic homology of Steinberg algebras of ample groupoids, with the isotropy summands emerging from the cyclic nerve, and then pushes the computation through for twisted Exel-Pardo algebras, producing explicit cone formulas and a Dennis-trace bridge to groupoid homology. The main theorems (1.1, 5.3.4, 6.4.12, 6.5.13) are new in this generality, and the chain maps σ_m, τ, Φ^t are written out explicitly. The organization is modular and the core chain—Lemma 2.8.5, Proposition 3.8, Theorem 4.2, Theorem 5.3.4—is coherent. I did not find a fatal error in the central arguments.\n\nThe biggest soft spot is the pseudo-freeness hypothesis. The introduction defines it edge-wise: g(e)=e with g≠1 implies φ(g,e)≠1. Section 6.5 redefines it path-wise: the only element strongly fixing a path is 1. These are not equivalent under the standing hypotheses (row-finite, trivial vertex action). Concretely, take one vertex with two loops e,f; G=(Z/2)^2 with a swapping the loops and b fixing both; set φ(a,e)=φ(a,f)=1, φ(b,e)=φ(b,f)=a. The cocycle identities hold, the edge-wise condition holds, yet b strongly fixes the path ee. So Theorem 1.5(iii) and Theorem 6.5.13, whose proof uses the path-wise condition, are not established for all tuples satisfying the introduction's definition. This is repairable by using the path-wise definition throughout, but as written the main hypothesis is ambiguous and the theorem's scope is overstated. The same ambiguity propagates to Theorem 1.9.\n\nThe paper also leans on cited black boxes: [12] (one of the authors) for the isomorphism C(G,E,φ^c)≅A_k(G^u), for kk-equivalences, and for K-regularity, and [16, Theorem 12.2] for finiteness of minimal strongly fixed paths. The citation pattern is not abusive—those results are stated with proofs and are independent—but a referee should check the key invocations, especially [12, Prop 6.2.3 and Thm 6.3.1] and the step in Lemma 6.5.7(c) that uses [16, Thm 12.2]. Steps 2 and 3 of Theorem 6.4.12 are compressed; the source-elimination argument is plausible but not fully detailed. There are also display corruptions (e.g. (1.6)) that make machine-level checking of some formulas impossible; this is a presentation issue, not a mathematical one.\n\nBottom line: this is a serious paper for specialists in K-theory and groupoid homology. It deserves a rigorous referee, but the pseudo-free hypothesis must be straightened and the theorems restated to match the proof. I would send it to review and ask for that clarification, rather than reject.","headline":"Substantial, mostly sound Burghelea-type computations for Steinberg algebras and Exel-Pardo tuples, but the pseudo-free hypothesis is ambiguous and the theorem statement overclaims the edge-level version.","tokens_in":58834,"tokens_out":3653,"would_cite":true,"duration_ms":31631,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["16E40","19D55","22A22"],"pacs":[],"model":"deepseek-v4-flash","headline":"The paper proves that Hochschild and cyclic homology of Steinberg algebras are governed by groupoid homology, and that the Dennis trace maps K-theory of Exel-Pardo algebras into twisted groupoid homology.","keywords":["Steinberg algebras","ample groupoids","Hochschild homology","cyclic homology","Dennis trace","Exel-Pardo algebras","groupoid homology","Bowen-Franks group"],"falsifier":"Take the non-pseudo-free example with one vertex, one loop, G=Z/2 acting trivially, and trivial φ and c, with k=l=Z. The tight groupoid has Z/2 isotropy at every infinite path. Computing the twisted groupoid homology H(G,Z/Z) directly and comparing it with the homology of cone(I−τ), where τ is the identity map on the single vertex, gives a difference in degree 1: the cone acquires an H0 summand that genuine groupoid homology does not have, so any proof of Theorem 6.5.13 must break exactly at the step using finitely many minimal strongly fixed paths.","tokens_in":57680,"feed_emoji":"🧮","tokens_out":10057,"duration_ms":69573,"temperature":0.7,"pith_summary":"This paper establishes a tight computational bridge between homological invariants of an ample groupoid's Steinberg algebra and the groupoid's own homology. For ample Hausdorff groupoids, the groupoid homology complex embeds as a direct summand of the Hochschild complex, and for principal groupoids, or Hausdorff groupoids with discrete isotropy outside the unit space, the paper gives explicit decompositions of the cyclic nerve. In the Exel-Pardo setting, the authors compute Hochschild homology weight by weight as the cone of an explicit endomorphism built from the graph and the twisting cocycle. Under a pseudo-freeness condition, they then identify twisted groupoid homology with the cone of a related map and show that the Dennis trace fits into a commutative diagram relating K-theory to twisted groupoid homology. The payoff is concrete: for many self-similar graph actions, K0 of the Exel-Pardo algebra is the Bowen-Franks group, and the Dennis trace becomes an isomorphism after scalar extension.","feed_headline":"Hochschild homology of Steinberg algebras reduces to groupoid homology","feed_subtitle":"K-theory of Exel-Pardo algebras maps into twisted groupoid homology, and K0 of many examples is the Bowen-Franks group.","key_machinery":"The central machinery is the weight decomposition of the Z-graded Exel-Pardo algebra L=L(G,E,φ^c), whose m-th Hochschild component is represented by the cone of I−σ_m for an explicit chain map σ_m built from the reduced incidence matrix, the self-similar cocycle φ, and the twisting cocycle c. A parallel map τ acts on the twisted groupoid homology complex H(G,k/l), and a spectrum-level map Φ^t acts on homotopy K-theory; the paper proves these three cones match. The groupoid-homology side rests on the semicyclic module H(G)=C_c(G^(•)) and an embedding whose splitness for Hausdorff groupoids gives the Dennis-trace map D_*=res∘D_* from K-theory into groupoid homology.","core_discovery":"For any ample groupoid G, H_*(G) is a direct summand of HH_*(A_k(G)); when G is principal, or Hausdorff with GIso\\G(0) discrete, cyclic homology decomposes into groupoid homology and isotropy contributions, recovering Burghelea's theorem in the group case. For a row-finite Exel-Pardo tuple with trivial vertex action and a flat ring extension l⊂k, each m-th weight component of HH(L(G,E,φ^c)/l) is naturally quasi-isomorphic to cone(I−σ_m), where σ_m is an explicit chain map on Hochschild complexes of k[G]. Under pseudo-freeness, the twisted groupoid homology complex H(G,k/l) is quasi-isomorphic to cone(I−τ), and if k[G] is regular supercoherent the Dennis trace yields the commutative diagram (1.8) with exact rows, linking K_*(L) to twisted groupoid homology. Consequently K0(L)=BF(E) and D0 induces an isomorphism K0(L)⊗_Z k ≅ H0(G^ω,k/l).","pith_inferences":["The cone presentations likely extend beyond pseudo-freeness if one replaces H(G,k/l) by a complex indexed by strongly fixed paths, in the spirit of Burghelea's isotropy decomposition; this would restore the diagram (1.8) for non-Hausdorff Exel-Pardo groupoids.","The commutative diagram (1.8) is a template: any excisive, homotopy-invariant, matricially stable functor satisfying the hypotheses of Theorem 6.6.4 should satisfy the same exact triangle, so the paper's machinery applies beyond K-theory.","The non-pseudo-free one-loop example suggests that the obstruction to the groupoid-homology exact sequence is concentrated in isotropy over periodic infinite paths, and that a finite-path truncation argument might weaken the pseudo-free hypothesis without losing K0 computations.","The isomorphism K0(L)⊗k≅H0(G^ω,k/l) can be read as a torsion-free shadow of the Bass trace conjecture for groupoids; testing it on groupoids with torsion would clarify how isotropy must enter the trace map."],"forward_implications":["For every ample Hausdorff groupoid, the Dennis trace factors through groupoid homology, giving a canonical comparison from K-theory of Steinberg algebras to a computable homology theory.","The Hochschild and cyclic homology of Exel-Pardo algebras is reduced to Hochschild homology of k[G] with explicit bimodule coefficients, so computations depend only on the group algebra and graph combinatorics.","Under pseudo-freeness and regular supercoherence of k[G], the long exact sequences for K-theory and twisted groupoid homology align, so K0(L)=BF(E) for torsionfree Farrell-Jones groups over fields or PIDs.","The G={1} case recovers known Leavitt path algebra results, and the G=Z case yields a concrete matrix description of the K-theoretic boundary map, extending computations for Katsura algebras.","If Conjecture 2 holds, excisive, homotopy-invariant, matricially stable functors would be discretization invariant, forcing K-theory and related invariants of universal and tight groupoids to agree in many settings."],"supporting_citations":[{"why":"Defines the Steinberg algebra A_k(G) and its basic generation properties, giving the object whose homology is studied.","marker":"[33]"},{"why":"Constructs the Exel-Pardo groupoid from a self-similar action and supplies pseudo-freeness, Hausdorffness, and finite-minimal-strongly-fixed-path results used in the cone theorems.","marker":"[16]"},{"why":"Establishes twisted Exel-Pardo algebras as twisted Steinberg algebras, the Cohn algebra comparison, kk-stability, and the K-regularity criterion that lets K replace KH.","marker":"[12]"},{"why":"Provides the Leavitt path algebra Hochschild cone computations that the paper recovers and extends when G is trivial.","marker":"[4]"},{"why":"Supplies the flat resolutions, Shapiro-style lemma, and basic groupoid-homology constructions underlying H(G,W) and the cyclic nerve identifications.","marker":"[27]"},{"why":"Gives the recent untwisted, non-pseudo-free version of the groupoid-homology exact sequence, marking exactly which hypothesis the paper still needs in the twisted case.","marker":"[28]"},{"why":"Burghelea's decomposition of Hochschild and cyclic homology of group rings, which the paper recovers as the group case of its cyclic nerve computation.","marker":"[9]"}],"fun_headline_variants":["Groupoid homology is a direct summand of Steinberg algebra Hochschild homology","Hochschild homology of Steinberg algebras contains groupoid homology","K-theory of Exel-Pardo Steinberg algebras via Dennis trace","Exel-Pardo Steinberg algebras recover Bowen-Franks group"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is pseudo-freeness: no nontrivial group element strongly fixes a path, meaning g(γ)=γ and φ(g,γ)=1 only when g=1; if it fails, the paper's twisted groupoid-homology exact sequence and the commuting Dennis-trace diagram are not established.","fun_headline_variants_meta":{"raw":{"variants":["Groupoid homology is a direct summand of Steinberg algebra Hochschild homology","Hochschild homology of Steinberg algebras contains groupoid homology","K-theory of Exel-Pardo Steinberg algebras via Dennis trace","Exel-Pardo Steinberg algebras recover Bowen-Franks group"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000686,"raw_usage":{"total_tokens":3149,"prompt_tokens":1022,"completion_tokens":2127,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":638,"completion_tokens_details":{"reasoning_tokens":2051}},"tokens_in":638,"tokens_out":2127,"duration_ms":9924,"temperature":1.0,"reasoning_tokens":2051,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-11T11:36:57.536140+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Take the non-pseudo-free example with one vertex, one loop, G=Z/2 acting trivially, and trivial φ and c, with k=l=Z. The tight groupoid has Z/2 isotropy at every infinite path. Computing the twisted groupoid homology H(G,Z/Z) directly and comparing it with the homology of cone(I−τ), where τ is the identity map on the single vertex, gives a difference in degree 1: the cone acquires an H0 summand that genuine groupoid homology does not have, so any proof of Theorem 6.5.13 must break exactly at the step using finitely many minimal strongly fixed paths.","supporting_citations":[{"cited_title":"Exel-Pardo algebras with a twist","cited_arxiv_id":"2309.14325","evidence_quote":"Establishes twisted Exel-Pardo algebras as twisted Steinberg algebras, the Cohn algebra comparison, kk-stability, and the K-regularity criterion that lets K replace KH."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides the Leavitt path algebra Hochschild cone computations that the paper recovers and extends when G is trivial."}],"review_version":1}