{"id":"ae766731-bdbd-448e-8985-a79cf645ffcc","arxiv_id":"1908.06343","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":8.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"For finite group actions with the weak tracial Rokhlin property on simple stably finite C*-algebras, the fixed point algebra has radius of comparison at most that of A, and the crossed product at most one over the group size times rc(A).","lead":"The paper proves new bounds on the radius of comparison, a numerical invariant of C*-algebras, when a finite group acts with the weak tracial Rokhlin property. It also constructs an example showing these bounds can be exactly achieved.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The weak tracial Rokhlin results rest on unpublished [14], and the most load-bearing pieces are matrix-amplification permanence ([14, Cor. 4.6]) and crossed-product simplicity ([14, Cor. 3.3]), used in Lemma 3.11 and Theorems 4.1/4.5.","rationale":"The reader's weakest-assumption analysis identifies the same load-bearing dependency: the unpublished preprint [14] supplies the equivalence in Definition 3.2, matrix-amplification permanence, and simplicity of the crossed product. I found no independent internal flaw that would change the verdict. The main theorems are carefully argued conditional on [14], and the example section is largely self-contained aside from Niu's mean-dimension theorem. A minor algebraic typo in Lemma 3.7's final equality, where (a'−3ε/4)_+ should be interpreted after reparameterizing ε or after defining a'=(a−ε/4)_+, is readily repaired and does not affect the argument. The concern is therefore external and verifiable rather than an observed contradiction. The CONDITIONAL verdict remains appropriate.","tokens_in":47566,"tokens_out":30702,"duration_ms":307781,"concrete_test":"Independently verify [14, Prop. 3.10], [14, Cor. 4.6], and [14, Cor. 3.3] directly from Definition 3.2. Specifically, check that for every n≥1, the action id_{M_n}⊗α on M_n(A) satisfies condition (4) of Definition 3.2 for an arbitrary positive x∈M_n(A) of norm 1, not only for x of the form 1_{M_n}⊗x_0. If this fails, re-examine Lemma 3.11 and Theorem 4.1. For Theorem 4.5, check whether fullness of p can be proved directly from the weak tracial Rokhlin property; if not, the equality rc(C*(G,A,α))=|G|^{-1}·rc(A^α) is unsupported in the weak tracial case.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim for weak tracial Rokhlin actions is not proved from first principles: Section 3's injectivity theorem and Section 4's radius bounds use imported framework results. Lemma 3.11(1) and (2) explicitly invoke [14, Cor. 4.6] to get the weak tracial Rokhlin property for id_{M_n}⊗α, and Theorem 4.1 repeats this. Theorem 4.5 uses [14, Cor. 3.3] to conclude that C*(G,A,α) is simple, which makes the projection p full in the corner calculation. Without these, the chain A^α ≅ pCp and rc(pCp)=|G|·rc(C*(G,A,α)) breaks. The rest of the paper's internal argument is plausible; I found only reparametrizable minor issues, not a fatal internal flaw. The risk is that a nontrivial definitional equivalence or a permanence property in [14] is subtly wrong or has hidden hypotheses, for example that condition (4) of Definition 3.2 fails for the amplified action on M_n(A) for arbitrary positive x of norm 1. This is a correctness risk, not an inconsistency.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper studies finite group actions with the weak tracial Rokhlin property on infinite-dimensional stably finite simple unital C*-algebras. The main results are the radius-of-comparison bounds rc(A^α) ≤ rc(A) (Theorem 4.1) and rc(C*(G,A,α)) ≤ (1/|G|) rc(A), together with the stronger identity rc(C*(G,A,α)) = (1/|G|) rc(A^α) (Theorem 4.5), and the ordered-semigroup isomorphism Cu+(A^α)∪{0} ≅ Cu+(A)^α∪{0} (Theorem 5.5), with stable-rank-one W-semigroup variants. The proof strategy is to establish injectivity and surjectivity of the natural map on purely positive Cuntz classes, using a substantial amount of Cuntz-semigroup machinery, and then to relate radii of comparison via traces and corners. Section 6 constructs a Z/2-action with the Rokhlin property on a simple unital AH algebra for which rc(A)=κ>0, rc(A^α)=κ, and rc(C*(Z/2,A,α))=κ/2, so the inequalities are sharp in this example. The arguments are detailed and internally coherent, but the central weak-tracial-Rokhlin results and the example's lower bounds depend on results imported from unpublished preprints, and one step in Lemma 4.4 appears to assert an unjustified commutativity.","tokens_in":47768,"tokens_out":12756,"duration_ms":123429,"significance":"The results, if fully supported, are a substantial contribution: they give the first general comparison-theoretic bounds for fixed-point algebras and crossed products by finite group actions with the weak tracial Rokhlin property, and they identify the purely positive part of the Cuntz semigroup of the fixed-point algebra with the fixed-point subsemigroup. The paper also contains useful new tools, notably the corner radius-of-comparison estimates in Theorem 2.18 and the quasitrace computation in Lemma 4.4, and it constructs an explicit example with positive radius of comparison and sharp bounds. The authors are honest about limitations and open problems, and the main internal derivation is plausible. The main caveats are external: several load-bearing statements are taken from unpublished preprints, and one local proof step in Lemma 4.4 needs repair.","major_comments":[{"comment":"The weak tracial Rokhlin half of the paper is not self-contained. The equivalence in Definition 3.2 is imported from [14, Proposition 3.10]; the permanence of the weak tracial Rokhlin property under matrix amplification, used in Lemma 3.11, Theorem 4.1, and Lemma 5.4, is [14, Corollary 4.6]; and simplicity of C*(G,A,α), used in Theorem 4.5 and Corollary 5.8, is [14, Corollary 3.3]. Since [14] is an unpublished arXiv preprint, the main claims of Sections 3–5 are conditional on an external source whose correctness the referee cannot verify from the manuscript. Please provide complete proofs of these specific facts, or state and prove precise versions of them, or replace [14] with published references.","section":"§3, Definition 3.2; §4, Theorems 4.1 and 4.5; §5, Theorem 5.5"},{"comment":"The proof of Lemma 4.4 asserts that τ(∑_{h∈G} f_h p f_h) = ∑_{h∈G} τ(f_h p f_h) 'since the elements f_h p f_h, for h ∈ G, commute with each other.' This commutativity is not justified by the stated hypotheses: f_h p f_h = (1/|G|)∑_g f_h α_g(f_h) u_g, and the group-unit terms u_g prevent the operators f_h p f_h from commuting without additional structure. This step is load-bearing because it is used to identify ∑_h τ(f_h p f_h) with τ((1/|G|)f^2), which in turn supplies the key estimate τ(p)=1/|G| used in Theorem 4.5. The gap is likely reparable using the already proved approximation ‖f_h p f_h − (1/|G|)f_h^2‖ and the fact that the elements f_h^2 commute as elements of A, but the proof as written is incomplete.","section":"§4, Lemma 4.4, equation (4.9)"},{"comment":"The exact lower bounds in Theorems 6.15 and 6.21, namely rc(A)=κ and rc(C*(Z/2,A,α))=κ/2, depend on Lemma 6.9, which is quoted from [22, Lemma 1.9], an unpublished preprint listed as 'in preparation.' The nonembedding statement L×k not embedding in a trivial bundle of rank less than 2k is used essentially in Corollaries 6.13 and 6.20 to force rank inequalities. This is a central feature of the paper, not a peripheral remark, so the authors should provide a proof of Lemma 6.9 or cite a published source.","section":"§6, Lemma 6.9 and Corollaries 6.13 and 6.20"},{"comment":"A large amount of the Cuntz-semigroup machinery is imported from [38], which is also an unpublished arXiv preprint. In particular, Lemma 3.9 relies on [38, Lemma 3.2 and Corollary 3.3], Lemma 3.10 uses [38, Lemma 2.1], Lemma 4.4 uses [38, Corollary 2.5], and Lemma 5.4 and Proposition 5.7 use [38, Theorem 1.16 and Lemma 1.25]. Since these facts are load-bearing for Theorem 5.5 and for the quasitrace arguments, the authors should either include the relevant proofs, state the exact results with sufficient context, or replace [38] with a published reference.","section":"§2–§5, references to [38]"}],"minor_comments":[{"comment":"The notation Cu+(A)^α is used in the abstract before Notation 3.1 is introduced; a one-line definition in the introduction would improve readability.","section":"Abstract and §1"},{"comment":"In the sentence after equation (4.1), 'also has the weak tracial property' should read 'also has the weak tracial Rokhlin property.'","section":"Theorem 4.1"},{"comment":"The phrase 'This result holds when α has the Rokhlin property, without the requirement that 0 be a limit point of sp(b)' is helpful, but the reader must infer that the cited Theorem 4.1(ii) of [16] also requires no simplicity or stable finiteness; please state the hypotheses explicitly.","section":"Lemma 3.7"},{"comment":"The density condition on the points x_m is stated in a compact but difficult way; a short explanatory sentence about the role of the condition would help the reader verify that such a choice is possible.","section":"Section 6, Construction 6.1(5)"},{"comment":"When introducing the UHF algebra D and the action α via Example 2.8 of [37], the authors should explicitly note that the Cuntz-semigroup non-injectivity phenomenon concerns projections and therefore does not contradict Lemma 3.11; the paragraph does this, but the connection could be made earlier.","section":"Example 4.7"}],"recommendation":"major_revision","confidential_remarks":"The paper's central claims depend heavily on [14], [22], and [38], all unpublished preprints; [14] is coauthored by the second author and [22] by the third author with Hirshberg. This creates a nontrivial referee burden and a correctness risk. I would ask the editor to require that the authors either provide complete proofs of the specific imported results used here or replace them with published references before the paper is accepted. The apparent commutativity gap in Lemma 4.4 should also be resolved explicitly; if the intended approximation argument works, the proof needs to be rewritten. The topic is suitable for the journal and the main ideas appear sound, but the manuscript is not yet self-contained enough for publication as it stands."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Here's my take. The paper proves genuinely new inequalities: if a finite group acts on an infinite-dimensional stably finite simple unital C*-algebra with the weak tracial Rokhlin property, then rc(A^α) ≤ rc(A) and rc(C*(G,A,α)) ≤ (1/|G|)rc(A), and the purely positive part of Cu(A^α) plus zero is isomorphic to the α-invariant purely positive part of Cu(A). The Rokhlin-property version of the Cuntz semigroup statement was already in Gardella-Santiago, and the authors say so; the weak tracial case, the radius bounds, and the sharp Z/2 example are new. I also found the corner radius estimate (Theorem 2.18) and Lemma 4.4 genuinely useful and carefully proved.\n\nThe main body is very well done. The proofs are detailed, the hypotheses are stated cleanly, and the hardest part—the Section 6 example—looks like real work: a Rokhlin action on a simple AH algebra with rc(A) > 0, equality rc(A^α) = rc(A), and rc(crossed product) = (1/2)rc(A). The counterexample in Example 4.7 to Osaka-Teruya's Proposition 6.2 and Corollary 6.3 is a useful service to the literature.\n\nThe soft spot is the one the authors themselves leave visible: Sections 3–5 lean on two unpublished papers from the same circle. [14] supplies the equivalence in Definition 3.2, matrix-amplification permanence ([14, Cor. 4.6]), and simplicity of the crossed product ([14, Cor. 3.3]); the amplification and simplicity facts are load-bearing in Lemma 3.11, Theorem 4.1, and Theorem 4.5. The sharpness example uses [22, Lemma 1.9], the nontrivial vector-bundle fact, as Lemma 6.9. None of this is a contradiction or a circular argument within the paper itself; it is an external-correctness risk. A referee should check [14] in particular, and the authors should either state the needed results or wait until those papers are public and vetted. This is a real caveat, but it is proportionate: the present text is coherent, the missing dependencies are clearly identified, and the authors do not hide them.\n\nWho is this for? Specialists in Cuntz semigroups, radius of comparison, and crossed products by finite group actions. Someone working on strict comparison or Elliott invariants would get a solid new tool, not just a technical exercise.\n\nRecommendation: this deserves a serious referee. I would not desk-reject it; I would send it out and specifically ask the referee to verify the unpublished dependencies, especially [14] and [22, Lemma 1.9].","headline":"Solid, careful paper with genuinely new radius-of-comparison and Cuntz semigroup results under weak tracial Rokhlin actions, plus a hard sharp example; the main caveat is heavy reliance on unpublished companion papers that a referee must check.","tokens_in":48373,"tokens_out":3613,"would_cite":true,"duration_ms":37677,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["46L55","19K14","46L80"],"pacs":[],"model":"deepseek-v4-flash","headline":"This paper proves that for a finite group action with the weak tracial Rokhlin property on a simple stably finite C*-algebra, the radius of comparison of the crossed product is at most one over the group order times that of the original…","keywords":["C*-algebras","Cuntz semigroup","radius of comparison","crossed products","finite group actions","Rokhlin property","tracial Rokhlin property","AH algebras"],"falsifier":"Find a finite group G, an infinite-dimensional stably finite simple unital C*-algebra A, and a weak tracial Rokhlin action α for which rc(A^α) > rc(A), or for which the map Cu+(A^α) ∪ {0} → Cu+(A)^α ∪ {0} is not surjective. A more targeted check: compute the radius of comparison of the example in Section 6 using Niu's mean dimension formula and the Bott bundle obstruction; if the values differ from κ and κ/2, the construction fails.","tokens_in":47323,"feed_emoji":"🧮","tokens_out":7160,"duration_ms":58540,"temperature":0.7,"pith_summary":"This paper proves that for a finite group acting on an infinite-dimensional stably finite simple unital C*-algebra with the weak tracial Rokhlin property, the radius of comparison of the crossed product is at most one over the group order times the radius of comparison of the original algebra, and the fixed point algebra has radius no larger than the original. It also establishes that the inclusion of the fixed point algebra induces an isomorphism between the purely positive part of the Cuntz semigroup of the fixed point algebra and the fixed points of the purely positive part of the original semigroup. The paper constructs an explicit example where G = Z/2Z, the action has the Rokhlin property, the algebra is a simple unital AH algebra with stable rank one, and the bounds are equalities: rc(A^α) = rc(A) and rc(C*(G,A,α)) = (1/2)rc(A). A sympathetic reader should care because the Cuntz semigroup is usually too complicated to compute, and this result shows a tractable rigidity for a broad class of actions.","feed_headline":"Finite group actions divide comparison radius by the group size","feed_subtitle":"The radius of comparison drops by the group order, and a constructed Z/2 example shows the drop is exact.","key_machinery":"The weak tracial Rokhlin property (existence of approximately equivariant orthogonal positive contractions f_g whose sum f satisfies 1-f ≼ x and ‖fxf‖ close to 1) is the engine. The paper's Lemma 3.5 averages a Cuntz subequivalence in A over the group to produce subequivalence in the fixed point algebra, and the projection p = (1/card(G))Σ u_g identifies A^α with a corner pC*(G,A,α)p. The corner estimate Theorem 2.18 then converts the constant quasitrace value τ(p)=1/card(G) (Lemma 4.4) into the factor 1/card(G) for the crossed product. For the example, a diagonal AH-system built from two copies of a system over products of spheres is merged by point evaluations, and the Bott line bundle obstruction (via Lemma 6.9) gives the lower bounds on rc.","core_discovery":"The central claim is that for an action with the weak tracial Rokhlin property, the map Cu+(A^α) ∪ {0} → Cu+(A)^α ∪ {0} induced by the inclusion is an isomorphism of ordered semigroups, and as a consequence rc(A^α) ≤ rc(A) and rc(C*(G,A,α)) ≤ (1/card(G))rc(A). The paper proves the reverse-style equalities in an example: an action of Z/2Z with the Rokhlin property on a simple unital AH algebra A with rc(A)>0, such that rc(A^α)=rc(A) and rc(C*(Z/2Z,A,α))=(1/2)rc(A).","pith_inferences":["The merging construction of two copies of a diagonal AH system, with point evaluations connecting them, seems generalizable: presumably any finite group G can be made to act with the Rokhlin property on a simple AH algebra whose radius of comparison is divided exactly by |G| in the crossed product, and possibly any prescribed value can be realized.","If the unpublished framework results of the second author's preprint are correct, the weak tracial Rokhlin property should force the same Cuntz semigroup rigidity for the crossed product even when the action is not Rokhlin; the paper's Example 6.22 shows pointwise outerness alone is far from sufficient.","A natural testable extension: for actions where every tracial state is invariant (Question 7.2), one might expect rc(A^α) = rc(A) to force equality in the crossed product bound, since the trace space is unchanged; the paper leaves this open.","The corner estimate Theorem 2.18 relating rc of a full corner to rc of the algebra is likely of independent use in other crossed product problems, since it converts the projection p = (1/|G|)Σ u_g into a sharp factor of |G|."],"forward_implications":["If the weak tracial Rokhlin property holds, the crossed product's radius of comparison is at most a fraction 1/|G| of the original algebra's, so taking a crossed product by a finite group makes the algebra more comparable in a precise trace sense.","The fixed point algebra's radius of comparison never exceeds that of the ambient algebra, so passing to fixed points cannot increase the obstruction to strict comparison.","The purely positive part of the Cuntz semigroup of the fixed point algebra is isomorphic to the fixed-point sub-semigroup of the original, giving a complete invariant for this part when the action is weak tracial Rokhlin.","The constructed example with G = Z/2Z shows both inequalities can be equalities: rc(A^α) = rc(A) and rc(C*(G,A,α)) = (1/2) rc(A), with the action even having the Rokhlin property.","In the stable rank one case, the isomorphisms pass to the W-semigroup, so W+(A^α) ∪ {0} ≅ W+(A)^α ∪ {0}, making the invariant computable via traces."],"supporting_citations":[{"why":"Provides the weak tracial Rokhlin framework: equivalence of definitions (Prop 3.10), matrix amplification permanence (Cor 4.6), and simplicity of the crossed product (Cor 3.3), used throughout Sections 3–5.","marker":"[14]"},{"why":"Supplies the Rokhlin-property version of the Cuntz semigroup isomorphism (Theorem 4.1(ii)) and the result that Cuntz subequivalence in A implies subequivalence in A^α for Rokhlin actions.","marker":"[16]"},{"why":"Supplies the technical Cuntz comparison lemmas (Lemma 1.8, Lemma 2.1, etc.) used to convert trace inequalities into subequivalence in the weak tracial setting.","marker":"[38]"},{"why":"Introduces the radius of comparison and the r-comparison condition that the paper studies and extends.","marker":"[44]"},{"why":"Niu's mean dimension theorem for AH-algebras with diagonal maps gives the upper bound rc(A) ≤ κ and rc(B) ≤ κ/2 used in the example.","marker":"[26]"},{"why":"Contributes the merging of two direct systems idea and Lemma 1.9 on the non-embedding of L^{×k}, used to construct the action with the Rokhlin property and compute the lower bounds.","marker":"[22]"},{"why":"The standard reference for crossed products by finite groups, Rokhlin property methods, and the radius of comparison (Theorem 12.4.4), including the averaging method and Takai duality.","marker":"[17]"}],"fun_headline_variants":["Rokhlin actions cap crossed-product radius at group-size fraction","Z/2 action halves crossed-product radius: exact bound","Group actions divide comparison radius by group size","Crossed-product comparison radius shrinks with finite group order"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The weak tracial Rokhlin results depend on unpublished results from the preprint [14] coauthored by the second author: the equivalence of the definition, permanence under matrix amplification, and simplicity of the crossed product; if any of these is wrong, the central theorems for the weak tracial Rokhlin case would need re-examination.","fun_headline_variants_meta":{"raw":{"variants":["Rokhlin actions cap crossed-product radius at group-size fraction","Z/2 action halves crossed-product radius: exact bound","Group actions divide comparison radius by group size","Crossed-product comparison radius shrinks with finite group order"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.002258,"raw_usage":{"total_tokens":8716,"prompt_tokens":926,"completion_tokens":7790,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":542,"completion_tokens_details":{"reasoning_tokens":7724}},"tokens_in":542,"tokens_out":7790,"duration_ms":52597,"temperature":1.0,"reasoning_tokens":7724,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T12:48:44.634968+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Find a finite group G, an infinite-dimensional stably finite simple unital C*-algebra A, and a weak tracial Rokhlin action α for which rc(A^α) > rc(A), or for which the map Cu+(A^α) ∪ {0} → Cu+(A)^α ∪ {0} is not surjective. A more targeted check: compute the radius of comparison of the example in Section 6 using Niu's mean dimension formula and the Bott bundle obstruction; if the values differ from κ and κ/2, the construction fails.","supporting_citations":[{"cited_title":"The weak tracial Rokhlin property for finite group actions on simple C*-algebras","cited_arxiv_id":"1711.10818","evidence_quote":"Provides the weak tracial Rokhlin framework: equivalence of definitions (Prop 3.10), matrix amplification permanence (Cor 4.6), and simplicity of the crossed product (Cor 3.3), used throughout Sections 3–5."},{"cited_title":"Gardella and L","cited_arxiv_id":null,"evidence_quote":"Supplies the Rokhlin-property version of the Cuntz semigroup isomorphism (Theorem 4.1(ii)) and the result that Cuntz subequivalence in A implies subequivalence in A^α for Rokhlin actions."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Introduces the radius of comparison and the r-comparison condition that the paper studies and extends."},{"cited_title":"Niu, Mean dimension and AH-algebras with diagonal maps , J","cited_arxiv_id":null,"evidence_quote":"Niu's mean dimension theorem for AH-algebras with diagonal maps gives the upper bound rc(A) ≤ κ and rc(B) ≤ κ/2 used in the example."},{"cited_title":"Hirshberg and N","cited_arxiv_id":null,"evidence_quote":"Contributes the merging of two direct systems idea and Lemma 1.9 on the non-embedding of L^{×k}, used to construct the action with the Rokhlin property and compute the lower bounds."},{"cited_title":"Giordano, D","cited_arxiv_id":null,"evidence_quote":"The standard reference for crossed products by finite groups, Rokhlin property methods, and the radius of comparison (Theorem 12.4.4), including the averaging method and Takai duality."}],"review_version":1}