{"id":"2a19d83a-3993-4f99-9d2d-dc5b7febaa68","arxiv_id":"2412.06031","paper_version":4,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":8.0,"correctness_risk":"low","formal_verification":"none","parameter_count":0,"one_line_summary":"The reduced C*-algebras of free groups F_n for n ≥ 2, and of finitely generated acylindrically hyperbolic groups with trivial finite radical and rapid decay, have strict comparison and are selfless.","lead":"This paper proves that the reduced group C*-algebras of the free groups on two or more generators have strict comparison, settling a question open since 1998. It also proves this property for a broad family of non-amenable groups and gives applications to the Elliott classification program.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The quantitative contraction bound in Appendix A is asserted by reference; Theorem B hinges on it, and the proof as written does not establish it.","rationale":"The reader's weakest assumption is exactly the same as mine: the uniform quadratic contraction bound asserted in Appendix A. I agree with the conditional verdict. I also considered two other potential weaknesses. First, in Theorem 3.5 the step 'as lim inf f(n)^{1/n}=1, we may pick m such that P(f(2mR))^{1/(2m)} < 1 + eps/3' is formally not justified for an arbitrary f with only liminf 1, because the good subsequence need not lie in the arithmetic progression 2mR; however the concrete functions f produced in Proposition 3.2 and Theorem 3.3 have actual limit 1, so this is a fixable wording issue. Second, the dependence on Robert's preprint [59, Theorem 2.6] is external, but the paper's contribution is the construction of the required embedding and the external result is explicitly cited. The contraction bound is different: it is internal to the paper's most general theorem and is merely asserted. A positive re-derivation would strengthen the paper; a failure would invalidate Theorem B. Since Theorem A and the free-product examples do not rely on Appendix A, REJECT is not warranted. The appropriate outcome is to keep the conditional verdict, requiring the authors to supply the missing verification of the quadratic contraction bound in Appendix A.","tokens_in":18263,"tokens_out":19644,"duration_ms":193069,"concrete_test":"Independently re-derive the contraction claim in Theorem A.4 from Lemma A.1 and the proof of [10, Corollary 3.4]. Concretely: fix an acylindrical action and a loxodromic g with tau(g)=C; for an arbitrary (lambda, 0)-quasi-geodesic q and a quasi-axis p of a conjugate of g, track the best bound on the projection diameter ||pi_p(q)|| as a function of lambda, including the Morse-lemma constants, and check whether the resulting mu, epsilon are truly bounded by a quadratic in lambda. If the bound is not quadratic, recompute C_{lambda,0}, B_{lambda,0}, R, Lambda, and D in Theorem A.4; the injectivity of phi_N in Theorem 3.3 depends on the final degree of P remaining polynomial in |g|_S.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The load-bearing step is in the proof of Theorem A.4 (Appendix A): the family X of quasi-axes of conjugates of a fixed loxodromic g is claimed to be (mu, epsilon)-contracting with respect to all (lambda, 0)-quasi-geodesics, with mu and epsilon bounded by a quadratic Q3(lambda). The paper justifies this by 'Lemma A.1 and the proof of [10, Corollary 3.4]', but no derivation is given. This is not a cosmetic detail: the subsequent estimates C_{lambda,0} = O(|g|_S^3), D = O(|g|_S^5), and the final degree-6 polynomial P all depend on Q3 being quadratic. If the true dependence is only polynomial of larger degree, or worse, exponential in lambda, then Theorem A.4 is not established. Theorem A.4 is the only tool proving injectivity of phi_N on B_{S cup {z}}(N) in Theorem 3.3; without that injectivity, selflessness for acylindrically hyperbolic groups is unsupported, and Theorem B collapses. The free-group case (Theorem A) is not affected because Proposition 3.2 gives a direct argument avoiding Appendix A. The concern is about a missing verification, not a known counterexample, but it is the most load-bearing unproved premise in the paper's central claim.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper proves that for every n ≥ 2 the reduced group C*-algebra of the free group F_n has strict comparison. The main theorem, Theorem B, asserts that if G is a finitely generated acylindrically hyperbolic group with trivial finite radical and the rapid decay property, then C*_r(G) is selfless in the sense of Robert and hence has strict comparison. The proof introduces a quantitative group-theoretic selflessness property (Definition 3.1), proves it for free products (Proposition 3.2) and for acylindrically hyperbolic groups (Theorem 3.3), and then, using rapid decay, proves that G*Z is existentially C*-residually-G (Theorem 3.5). This yields an embedding of C*_r(G*Z) into an ultrapower of C*_r(G), and Robert's theorem [59] converts this into selflessness and strict comparison. Several applications are drawn, including uniqueness of Jiang-Su embeddings and full Cuntz semigroup computations.","tokens_in":1660,"tokens_out":3799,"duration_ms":195196,"significance":"Assuming the quantitative contraction bound in Appendix A can be supplied, this is a major breakthrough: strict comparison for C*_r(F_n) resolves a problem open since Dykema-Rørdam and listed by Robert and in the recent problem set [62]. The proof is modular and opens a promising route from geometric group theory to C*-regularity. The free-product case has a self-contained proof, and the paper provides a wealth of new examples. The applications to Cuntz semigroup computations and unique Z-embeddings are concrete and well motivated.","major_comments":[{"comment":"The claim that the family X of quasi-axes of conjugates of a fixed loxodromic element is (mu,epsilon)-contracting with respect to all (lambda,0)-quasi-geodesics, with mu and epsilon bounded by a quadratic Q3(lambda), is asserted by reference to Lemma A.1 and the proof of [10, Corollary 3.4] but no derivation is given. This bound is load-bearing: the subsequent estimates C_{lambda,0}=O(|g|_S^3), D=O(|g|_S^5), and the final degree-6 polynomial P all depend on Q3 being quadratic. If the contraction constants grow only polynomially with larger degree, or worse exponentially in lambda, Theorem A.4 is not established, and with it the injectivity of phi_N in Theorem 3.3 and hence Theorem B collapse. The appendix must provide an explicit proof or a precise quotation of the quadratic dependence.","section":"Appendix A, proof of Theorem A.4"},{"comment":"The application of Theorem A.4 is stated for h1,...,hm in B_S(|g_n|^{1/2}_S), with the conclusion h1 g^{k1} ... hm g^{km} != e, but Theorem A.4 requires each h_i not in E(g). In the word arising from the proposed normal form, the endpoint factors h1 and hm may be the identity, and after free reduction some h_i could also become identity. The proof therefore needs a preliminary step that eliminates identity factors and combines adjacent powers of g before applying Theorem A.4. Without this reduction, the asserted injectivity of phi_N on B_{S union {z}}(N) is not justified. This is a fixable gap, but it is central to the proof of Theorem B.","section":"Theorem 3.3"},{"comment":"The final step from the embedding of C*_r(G*Z) into C*_r(G)_U to strict comparison depends entirely on Robert's [59, Theorem 2.6], which is an unpublished preprint. The manuscript should state this dependence explicitly and quote the theorem so that the reader can verify that its hypotheses are met. This is an external dependency rather than an internal inconsistency, but it is load-bearing for the main result.","section":"Section 3.3, Corollary 3.8"}],"minor_comments":[{"comment":"The displayed definition of phi_N has subscripts and superscripts swapped: as printed it writes phi_N(z) = g_{P(D(4CN^2+1))}^{2CN^2}, whereas the surrounding estimates require phi_N(z) = g_{2CN^2}^{P(D(4CN^2+1))}.","section":"Theorem 3.3"},{"comment":"The phrase 'for all n ≥ 1 2 C2D' is garbled; it should probably be 'for all n ≥ max(1, 2 C^2 D)'. The symbol C is also used for the word-length quasi-equivalence constant and for the stable length constant in Theorem A.4; these should be distinguished.","section":"Theorem 3.3"},{"comment":"The family of quasi-axes is denoted X, the same symbol as the hyperbolic space X; this makes the statement that X is contracting confusing. A script or fraktur symbol for the family would help.","section":"Appendix A"},{"comment":"The displayed formula 'lambda = tau(g)d(go,o)' should be 'lambda = d(go,o)/tau(g)'. The conclusion lambda = O(|g|_S) is still correct because tau(g) is fixed in Theorem A.4, but the displayed equation is dimensionally wrong.","section":"Appendix A, proof of Theorem A.4"},{"comment":"The function g(n) used for subexponential growth conflicts with the use of g for group elements; one of these should be renamed.","section":"Remark 3.6"},{"comment":"Writing Cu(C*_r(F_2)) = N ⊔ [0,∞] requires a brief explanation that the second component is the strictly positive part of the lower-semicontinuous affine functions on the singleton trace space; otherwise the notation [0,∞] is ambiguous.","section":"Theorem D"}],"recommendation":"major_revision","confidential_remarks":"The paper is likely correct in broad strategy, but the Appendix leaves the main geometric estimate as an assertion. I would encourage the editor to ask for a complete proof of the quadratic contraction bound before publication. The endpoint-identity issue in Theorem 3.3 is fixable but should be addressed. The dependence on Robert's unpublished preprint should also be clearly flagged. These are substantial but local issues, hence major revision rather than rejection."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"David — the key thing to know: this paper proves strict comparison for C*_r(F_n), n≥2, which has been open since Dykema–Rørdam's work in the 90s, and it does it by proving the stronger selflessness property from Robert's paper. The method is new — the quantitative 'selfless group' definition, the reduction from selflessness to existential C*-residualness via rapid decay, and the group-theoretic input for acylindrically hyperbolic groups. The free-group case is clean: selflessness follows from a short free-product argument (Prop 3.2), and the operator-algebraic transfer via Louder–Magee is elegant. The applications (Z-embedding uniqueness, Cuntz semigroup computation Cu(C*_r(F_2)) = N ⊔ [0,∞]) then fall out from known theorems. The paper is well-written and the citations look right.\n\nThe soft spots are real but manageable. The main one is in Appendix A. Theorem A.4 is a quantitative admissible path lemma, and its proof asserts, by reference to the proof of Bestvina–Fujiwara's Corollary 3.4, that the family of quasi-axes is (µ,ε)-contracting with polynomial bounds Q3(λ). No derivation is given. The stress-test is right that this is load-bearing for Theorem B: Theorem 3.3's injectivity claim for acylindrically hyperbolic groups uses Theorem A.4 directly. But I don't think it's a fatal flaw. The quantitative Morse lemma (Lemma A.1) is already quoted with quadratic bounds, and the Bestvina–Fujiwara contracting property is standard; the needed track of λ is a finite verification, not a new idea. Even if the final degree came out higher than 6, the argument would still go through — selflessness only needs subexponential growth, so any polynomial bound suffices. An exponential bound is not on the table from the cited geometry. So this is a missing verification, not a known counterexample, and it does not affect Theorem A at all.\n\nMinor issues: a couple of typos (swapped sub/superscripts in the definition of φ_N, a garbled inequality), and the application of Theorem A.4 technically excludes endpoint elements equal to the identity, which the selflessness argument needs — again, a small fix. The final step relies on Robert's preprint [59]; that's a normal dependency, though worth flagging until it's published.\n\nOverall: this deserves serious refereeing. The main theorem is likely correct, the free-group case is solid, and the paper will be an important reference. A referee should ask for the Appendix A contraction bound to be written out or cited precisely.","headline":"Settles the 25-year-old strict comparison problem for C*_r(F_n) with a genuinely new method; the one contested step is a fixable technical lemma in the appendix.","tokens_in":19068,"tokens_out":5350,"would_cite":true,"duration_ms":48082,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["46L05","20F65","22D25","46L35"],"pacs":[],"model":"deepseek-v4-flash","headline":"This paper proves strict comparison for the reduced group C*-algebras of free groups with two or more generators, settling an open problem from the late 1990s, and extends the result to a broad class of acylindrically hyperbolic groups…","keywords":["strict comparison","reduced group C*-algebra","free groups","acylindrically hyperbolic groups","rapid decay property","selfless C*-algebra","Cuntz semigroup","Jiang–Su algebra"],"falsifier":"A direct falsification would be a group satisfying the hypotheses of Theorem B whose reduced C*-algebra lacks strict comparison: two positive elements $a,b$ with $d_\\tau(a) < d_\\tau(b)$ but $a$ not Cuntz-subequivalent to $b$. A sharper test attacks the proof's geometric lemma: in a concrete acylindrically hyperbolic group, compute the bounded-projection constants of the axis family and search for a word $h_1 g^{n_1} \\cdots h_k g^{n_k} = e$ with all exponents below the paper's degree-6 bound; finding such a word would invalidate Theorem A.4.","tokens_in":18061,"feed_emoji":"🎯","tokens_out":14603,"duration_ms":123951,"temperature":0.7,"pith_summary":"This paper proves that the reduced group $C^*$-algebras of the non-abelian free groups, $C^*_r(\\mathbb{F}_n)$ for $n \\geq 2$, have strict comparison, a property that has been open since the late 1990s. Strict comparison means the unique trace completely decides when one positive element is Cuntz-subequivalent to another, making the trace a complete invariant for the natural order on the algebra. The authors obtain this by proving a stronger structural property called selflessness, and they extend it to every finitely generated acylindrically hyperbolic group with trivial finite radical and the rapid decay property. The method is group-theoretic: it builds retraction homomorphisms from $G * \\mathbb{Z}$ to $G$ with subexponential distortion, then uses rapid decay to turn them into an embedding of $C^*_r(G * \\mathbb{Z})$ into an ultrapower of $C^*_r(G)$. From that embedding, strict comparison, uniqueness of Jiang–Su embeddings, and the full Cuntz semigroup computation $\\mathrm{Cu}(C^*_r(\\mathbb{F}_2)) = \\mathbb{N} \\sqcup [0,\\infty]$ all follow.","feed_headline":"Free group C*-algebras have strict comparison at last","feed_subtitle":"The result extends to hyperbolic and acylindrically hyperbolic groups and yields full Cuntz semigroup computations.","key_machinery":"The load-bearing mechanism is a group-level strengthening the paper calls selflessness (Definition 3.1), modeled on the C*-algebraic selflessness of [59] and on a quantitative word-collapse lemma of [46]. A finitely generated group $(G,X)$ is selfless if, for every radius $n$, there is an epimorphism $\\varphi_n: G * \\langle a \\rangle \\to G$ that fixes $G$, is injective on the ball of radius $n$ in the generating set $X \\cup \\{a\\}$, and expands the ball by at most a subexponential factor. When $G$ also has the rapid decay property — the operator norm of a finitely supported group-ring element is polynomially bounded by its $\\ell^2$-norm and its support radius — selflessness implies that $G * \\mathbb{Z}$ is existentially $C^*$-residually-$G$ (Theorem 3.5), which then produces the injective ultrapower embedding. For acylindrically hyperbolic $G$ with trivial finite radical, the maps $\\varphi_n$ are built from a hyperbolically embedded copy of $\\mathbb{F}_2$, with the conjugate elements $b^n a b^{-n}$ playing the role of a free generator, and the non-collapse of words is certified by the quantitative admissible path lemma (Theorem A.4), a polynomial-constant version of the admissible path lemma of [72] proved in Appendix A.","core_discovery":"The paper's central claim has two layers. Theorem A states that for every $n \\geq 2$, the reduced group $C^*$-algebra of the free group $\\mathbb{F}_n$ is selfless in the sense of [59], hence has strict comparison. Theorem B states the same for every finitely generated acylindrically hyperbolic group $G$ with trivial finite radical and the rapid decay property. The proof reduces selflessness to an embedding statement: if $G * \\mathbb{Z}$ is existentially $C^*$-residually-$G$ — meaning finite pieces of the group ring of the free product can be approximately pushed into $G$ without increasing the reduced operator norm — then $C^*_r(G * \\mathbb{Z})$ embeds injectively into an ultrapower of $C^*_r(G)$ extending the diagonal inclusion, and a theorem of [59] converts that embedding into strict comparison. The free group case is handled by explicit retractions from $G * \\mathbb{Z}$ to $G$; the acylindrically hyperbolic case uses a hyperbolically embedded copy of $\\mathbb{F}_2$ inside $G$ and a quantitative admissible path lemma to control word collapse.","pith_inferences":["Remark 3.6 in the paper notes that full injectivity of the retraction maps can be relaxed to a subexponential fiber bound; testing whether this relaxation suffices for strict comparison in groups without rapid decay, such as higher-rank lattices, is a direct next question.","The reduction to an embedding of $C^*_r(G * \\mathbb{Z})$ into an ultrapower gives a common criterion that could be checked for other families of groups, for example those with proximal boundary actions mentioned in the paper's footnotes, potentially separating group-theoretic rapid decay from C*-algebraic selflessness.","Because the constants in the quantitative admissible path lemma are tracked as a degree-6 polynomial in the word length of the loxodromic element, the method suggests explicit quantitative questions about how large a ball must be inspected to witness Cuntz subequivalence in terms of the gap between traces."],"forward_implications":["For every $n \\geq 2$, $C^*_r(\\mathbb{F}_n)$ has strict comparison, settling an open question that had been recorded in the literature since the late 1990s.","For every group covered by Theorem B, the Jiang–Su algebra embeds unitally into $C^*_r(G)$, and the embedding is unique up to approximate unitary equivalence (Theorem C).","The Cuntz semigroup of $C^*_r(\\mathbb{F}_2)$ is fully computed as $\\mathbb{N} \\sqcup [0,\\infty]$ (Theorem D), and analogous computations hold for every exact acylindrically hyperbolic group with trivial finite radical and rapid decay.","Stable rank one for the reduced $C^*$-algebras of these groups is recovered by a new route.","Selflessness, and hence strict comparison, is not confined to rapid decay groups: the paper shows $C^*_r(\\mathbb{F}_2 * \\mathrm{SL}_3(\\mathbb{Z}))$ is selfless even though the group fails rapid decay."],"supporting_citations":[{"why":"Supplies the theorem that an embedding of the reduced free product into an ultrapower extending the diagonal implies selflessness and strict comparison; the final step of the argument.","marker":"[59]"},{"why":"Introduces the existential residual method and quantitative word-collapse lemma that the paper adapts to build and control retraction homomorphisms.","marker":"[46]"},{"why":"Provides the admissible path lemma whose quantitative, constant-tracked version (Theorem A.4) rules out collapse of words in powers of a loxodromic element.","marker":"[72]"},{"why":"Gives the hyperbolically embedded copy of F_2 in acylindrically hyperbolic groups with trivial finite radical and the Lipschitz quasi-retract length bounds used to build the maps.","marker":"[22]"},{"why":"Proves the rapid decay property for free groups, the norm control needed to pass from ball-injective homomorphisms to operator-norm inequalities.","marker":"[35]"},{"why":"Supplies rapid decay for hyperbolic groups, extending the argument from free groups to the wider acylindrically hyperbolic class.","marker":"[41]"},{"why":"Provides the bounded projection lemma for acylindrical actions that is quantitatively refined in Lemma A.3 to bound how far distinct quasi-axes can see each other.","marker":"[24]"},{"why":"Supplies the contracting property of quasi-axes (cited in the proof of Theorem A.4) that the quantitative admissible path lemma relies on.","marker":"[10]"}],"fun_headline_variants":["Strict comparison at last for free group C*-algebras","Free group C*-algebras get strict comparison","Selflessness problem solved for free group C*-algebras","Cuntz semigroup fully computed for free group C*-algebras","Acylindrically hyperbolic groups yield strict comparison"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The proof for the whole family of acylindrically hyperbolic groups rests on a geometric estimate taken from a known result rather than proved here, asserting that certain infinite lines in the group's action uniformly shadow geodesics with error bounded by a fixed polynomial; if that estimate fails, the non-collapse of words that carries the proof is no longer guaranteed.","fun_headline_variants_meta":{"raw":{"variants":["Strict comparison at last for free group C*-algebras","Free group C*-algebras get strict comparison","Selflessness problem solved for free group C*-algebras","Cuntz semigroup fully computed for free group C*-algebras","Acylindrically hyperbolic groups yield strict comparison"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000992,"raw_usage":{"total_tokens":4202,"prompt_tokens":941,"completion_tokens":3261,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":557,"completion_tokens_details":{"reasoning_tokens":3180}},"tokens_in":557,"tokens_out":3261,"duration_ms":23498,"temperature":1.0,"reasoning_tokens":3180,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-11T20:07:33.313039+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"A direct falsification would be a group satisfying the hypotheses of Theorem B whose reduced C*-algebra lacks strict comparison: two positive elements $a,b$ with $d_\\tau(a) < d_\\tau(b)$ but $a$ not Cuntz-subequivalent to $b$. A sharper test attacks the proof's geometric lemma: in a concrete acylindrically hyperbolic group, compute the bounded-projection constants of the axis family and search for a word $h_1 g^{n_1} \\cdots h_k g^{n_k} = e$ with all exponents below the paper's degree-6 bound; finding such a word would invalidate Theorem A.4.","supporting_citations":[{"cited_title":"Selfless C*-algebras","cited_arxiv_id":"2309.14188","evidence_quote":"Supplies the theorem that an embedding of the reduced free product into an ultrapower extending the diagonal implies selflessness and strict comparison; the final step of the argument."},{"cited_title":"Strongly convergent unitary representations of limit groups.J","cited_arxiv_id":null,"evidence_quote":"Introduces the existential residual method and quantitative word-collapse lemma that the paper adapts to build and control retraction homomorphisms."},{"cited_title":"Growth tightness for groups with contracting elements.Math","cited_arxiv_id":null,"evidence_quote":"Provides the admissible path lemma whose quantitative, constant-tracked version (Theorem A.4) rules out collapse of words in powers of a loxodromic element."},{"cited_title":"Hyperbolically embedded subgroups and rotating families in groups acting on hyperbolic spaces.Mem","cited_arxiv_id":null,"evidence_quote":"Gives the hyperbolically embedded copy of F_2 in acylindrically hyperbolic groups with trivial finite radical and the Lipschitz quasi-retract length bounds used to build the maps."},{"cited_title":"An example of a non nuclear C*-algebra, which has the metric approximation property.Invent","cited_arxiv_id":null,"evidence_quote":"Proves the rapid decay property for free groups, the norm control needed to pass from ball-injective homomorphisms to operator-norm inequalities."},{"cited_title":"Rapidly decreasing functions in reduced C*-algebras of groups","cited_arxiv_id":null,"evidence_quote":"Supplies rapid decay for hyperbolic groups, extending the argument from free groups to the wider acylindrically hyperbolic class."},{"cited_title":"K¨ ahler groups,R-trees, and holomorphic families of Riemann surfaces.Geom","cited_arxiv_id":null,"evidence_quote":"Provides the bounded projection lemma for acylindrical actions that is quantitatively refined in Lemma A.3 to bound how far distinct quasi-axes can see each other."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the contracting property of quasi-axes (cited in the proof of Theorem A.4) that the quantitative admissible path lemma relies on."}],"review_version":1}