{"id":"f5993b6b-d124-4a14-9ba7-0bfbd1c5c65d","arxiv_id":"2606.09654","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"Every selfless C*-algebra is either stable rank one (tracial case) or purely infinite simple (nontracial case); non-faithful selfless states are shown to land in the purely infinite class.","lead":"This math paper closes the last gap in the dichotomy for selfless C*-probability spaces: a non-faithful selfless state forces the algebra to be purely infinite and simple. That completes the classification of selfless C*-algebras into stably finite (stable rank one) or purely infinite, and implies every selfless C*-algebra is pure.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"No significant objection identified to the central Theorem 1; Section 4's pure-state examples remain a non-central overclaim.","rationale":"The reader's weakest_assumption targeted Lemma 2.3(ii), but that excision step appears correct. The key mechanism is that centered elements from factors in G annihilate all vectors whose first letter is not in G; since the right a_F restricts to first-letter-F vectors, every word from G maps them to first-letter-G vectors, which the left a_F annihilates. Thus a_F d_0 a_F = 0 is valid. The I_j ⊆ I step also holds, using Lemma 2.3(i) to identify the Cuntz class of any normalized positive kernel element with α. The rest of Theorem 6 is a coherent contradiction argument, and Theorem 1 follows from existential-embedding preservation of the relevant continuous existential formula. I therefore find no central flaw. The Section 4 overclaim about pure states on C*(Z) is real, but it concerns the applications and not the dichotomy itself; the reader's CONDITIONAL verdict is still reasonable, so I recommend UNCHANGED rather than ACCEPT/REJECT.","tokens_in":7430,"tokens_out":48403,"duration_ms":484045,"concrete_test":"Verify Lemma 2.3(ii) numerically in a free product of B(ℓ^2) with vector state: choose a finite G, a centered d in the G-algebra, and positive kernel a_F from disjoint factors; check on Fock basis words that a_F d a_F = ω(d)a_F^2. Equivalently, independently re-derive the containment I_j ⊆ I by showing every normalized positive kernel element has Cuntz class α.","verdict_should_be":"UNCHANGED","load_bearing_attack":"After re-checking the excision step that the reader flagged, I find no load-bearing flaw in Theorem 1. In Lemma 2.3(ii), the key identity a_F d_0 a_F = 0 is justified: once the right a_F has selected vectors whose first letter lies in F (disjoint from G), every reduced word from G acts on such a vector by producing a first letter in G (no cancellation is possible because F-letters are not inverses of G-letters), so d_0 maps this subspace into first-letter-G words, which the left a_F annihilates. Similarly a_i a_j = 0 holds as a consequence of centered elements annihilating words with a different first letter. The containment I_j ⊆ I follows because any nonzero positive kernel element, after normalization, is Cuntz equivalent to a_j by Lemma 2.3(i) via a third factor. The contradiction argument in Theorem 6 and the model-theoretic transfer (existential embeddings preserve the continuous ∃-formula for pure infiniteness) are sound. The main remaining problem is §4: several isomorphisms use pure states on C*(Z) whose GNS representations are not faithful, so Corollary 5 cannot be invoked; this affects the applications, not the dichotomy.","agreement_with_reader":"disagree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper claims to complete the selfless dichotomy for C*-probability spaces by proving Theorem 1: if (A, φ) is selfless and φ is nonfaithful, then A is purely infinite and simple. The proof passes through Theorem 6, which asserts that certain infinite reduced free products of nonfaithful GNS-faithful states are purely infinite and simple whenever a local δ_j condition holds. The announced consequences are that every selfless C*-algebra is pure, that several permanence properties can be strengthened, and that new Kirchberg-algebra isomorphisms follow from reduced free products. The main technical content is in Section 2: Lemma 2.3 is supposed to establish Cuntz comparison and the existence of a minimal essential purely infinite ideal in the free product, and Theorem 1 then transfers pure infiniteness back to a factor using existential embeddings.","tokens_in":7678,"tokens_out":14318,"duration_ms":133667,"significance":"If Theorem 1 were correct, it would settle the last open case of Robert's dichotomy and imply that every selfless C*-algebra is pure. The strategy of using infinitely many nonfaithful factors as 'tails' to excise arbitrary positive elements is attractive, and the model-theoretic transfer from an infinite free product to a factor is a plausible use of the existing literature. A correct proof of the dichotomy would be a substantial contribution. However, the central free-product lemma contains a false Hilbert-space calculation, and this invalidates the main theorem as written. The paper also overstates the applicability of the free-product results to pure states in Section 4.","major_comments":[{"comment":"The proof asserts 'As the free product representation is faithful, a_i a_j = 0 whenever i ≠ j.' This is false. Faithfulness of the reduced free-product representation does not imply that elements from different factors annihilate each other. Concretely, take nonzero a_i ∈ (A_i)_+ and a_j ∈ (A_j)_+ with φ_i(a_i)=φ_j(a_j)=0, and choose k ∉ {i,j} and a nonzero vector h_k in the centered Hilbert space of the k-th factor. Then a_j h_k = a_j ⊗ h_k and a_i(a_j ⊗ h_k) = a_i ⊗ a_j ⊗ h_k, a nonzero reduced word in H. Hence a_i a_j ≠ 0. The same erroneous annihilation claim is used in the computation a_F d_0 a_F = ω(d) a_F^2 in (ii): if d_0 contains a centered word from A_j, then a_i d_0 a_i is generally nonzero, not zero. Since (ii) is the source of the comparison mα ≤ [b], Lemma 2.3(ii) and all results depending on it (Theorem 6, and hence Theorem 1) are unsupported.","section":null},{"comment":"Theorem 4.1 invokes Theorem 6 and Corollary 5 for pure states φ_j, j ≥ 2. But Theorem 6 and Corollary 5 require each state to induce a faithful GNS representation. A pure state on a non-simple C*-algebra generally has a nonzero GNS kernel; for example, a pure state on C*(Z) ≅ C(S^1) has a one-dimensional, non-faithful GNS representation. Since A_j is only assumed separable, nuclear, and UCT (not simple), the hypotheses are not met. The proof of δ_j = 1 also uses ∥x∥ = ∥π_φ(x)∥, which is exactly faithfulness of π_φ and can fail for pure states on non-simple algebras. Consequently, the Kirchberg-algebra and isomorphism claims in Section 4 do not follow from the preceding theorems as written.","section":null},{"comment":"Even setting aside the false excision claim, the step from 'mα ≤ [b] for every m' to '2[b] ≤ [b]' is asserted rather than proved: 'some standard Cuntz semigroup analysis grants 2[b] ≤ [b].' This is a load-bearing implication in the proof that I is purely infinite and simple. A precise proof or a stated theorem reference is needed; without it, Theorem 6(ii) is incomplete even if the earlier comparison were repaired.","section":null}],"minor_comments":[{"comment":"In Definition 2.1, the symbol ψ is used both for a state and for a quantifier-free formula; this is confusing and should be changed (e.g., use σ(¯x,¯y)).","section":null},{"comment":"The abstract says the paper completes the 'stably finite/purely infinite dichotomy,' while Theorem 2 states the tracial case as stable rank one. The terminology is not always consistent; please unify.","section":null},{"comment":"The approximation of c by d in a finite free-product subalgebra is plausible, but the phrase 'sufficiently large so that there is self-adjoint d ... with ∥c−d∥<t/3' should justify that c is in the closure of the union of finite free products; since C is the reduced free product, this is true but should be stated.","section":null}],"recommendation":"reject","confidential_remarks":"The main theorem is appealing and the model-theoretic transfer idea is sound in isolation, but the central Hilbert-space calculation in Lemma 2.3 is plainly false and invalidates Theorem 6 and Theorem 1. This is not a presentation issue that can be resolved by local revision; it is the mechanism of the whole proof. Section 4 has an additional independent applicability gap regarding pure states. I cannot recommend publication in the current form."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Let me give you the short version. This paper closes a real gap in the selfless dichotomy: it shows that a selfless C*-probability space with nonfaithful state is purely infinite simple. That was explicitly open in Robert's work, and it completes the dichotomy so that every selfless algebra is either stably finite with stable rank one or purely infinite simple. The central mechanism, Theorem 6, is a genuinely new tail-excision criterion for infinite reduced free products. It uses nonfaithfulness positively: positive elements in the kernels of infinitely many factors let you excise any positive element. The proof of Theorem 6 checks out on the free-product Hilbert space; I rechecked the key identity a_F d_0 a_F = 0 and the containment argument, and they hold. The model-theoretic transfer from the infinite free product to the original algebra via existential embeddings is standard and fine.\n\nWhat the paper does well beyond the main theorem: it removes faithful assumptions from several permanence properties, sharpens the Choda-Dykema conjecture statement to the selfless case, and the exposition is refreshingly direct. The author also labels the gap honestly.\n\nSoft spots. Section 4 is where I get uncomfortable. Several isomorphism examples use pure states on C*(Z), and pure states need not induce faithful GNS representations. The theorems being invoked, especially Corollary 5, require GNS-faithful states. So the examples as stated are not consequences of the paper's own results. This is an overclaim in the applications, not in the core dichotomy. It should be fixed by either proving the examples under correct hypotheses or stating them as conditional. There's also a compressed step in Lemma 2.3(iv) where 'standard Cuntz semigroup analysis' covers the fact that 2[b] <= [b]; the referee should ask for that to be spelled out. The reliance on [17, 2.6(vi)] and [6, 3.13] for existential embedding and axiomatization is normal, but worth flagging for the referee to verify that the nonfaithful case is covered.\n\nThe central argument holds up. I think this deserves a serious referee. The main theorem is important, the proof is credible, and the flaws are localized to a clearly identifiable section. I'd send it out.\n\nFor a reading group, it's maybe: the core proof is worth working through, but the applications section might mislead. I'd cite it once the Section 4 issues are resolved, or with a caveat.","headline":"Closes the nonfaithful selfless gap with a credible new excision argument; the Section 4 isomorphism examples overreach but the core result is solid.","tokens_in":8178,"tokens_out":2045,"would_cite":true,"duration_ms":19304,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["46L05","46L35","46L54"],"pacs":[],"model":"deepseek-v4-flash","headline":"Nonfaithful selfless C*-algebras are purely infinite, simple.","keywords":["selfless C*-algebras","C*-probability spaces","purely infinite simple","reduced free products","Cuntz semigroup","stable rank one","pure C*-algebras","Kirchberg algebras"],"falsifier":"A concrete counterexample to the excision step: exhibit factors (A_j, φ_j) with nonfaithful GNS-faithful states, a finite F disjoint from the support of an approximant d, and a nonzero word w with a_F d_0 a_F w ≠ 0, or otherwise compute a tail product a_F c a_F that does not reduce to ω(c) a_F^2. Alternatively, find a selfless (A, φ) with φ nonfaithful whose infinite self-free product is not purely infinite simple—this would directly refute Theorem 1's route.","tokens_in":7272,"feed_emoji":"♾️","tokens_out":4200,"duration_ms":40091,"temperature":0.7,"pith_summary":"This paper closes the last gap in the dichotomy for selfless C*-probability spaces: when the distinguished state is nonfaithful, the underlying C*-algebra must be purely infinite and simple. The proof works by analyzing infinite reduced free products of nonfaithful yet GNS-faithful states, showing that such products are often purely infinite and simple when a computable sum of factor-level constants exceeds one. With this, every selfless C*-algebra is either stably finite with stable rank one (tracial case) or purely infinite and simple (nontracial case). A direct consequence is that every selfless C*-algebra is pure, and the result yields new permanence properties, progress on a conjecture of Choda and Dykema, and new Kirchberg algebras satisfying the UCT.","feed_headline":"Nonfaithful selfless C*-algebras are purely infinite, simple","feed_subtitle":"Completes the dichotomy: every selfless C*-algebra is either stably finite or purely infinite, and all are pure.","key_machinery":"The 'tail excision' mechanism on the free-product Hilbert space. Each factor supplies a positive element a_j in the kernel of its state that annihilates the vacuum vector and all centered words whose first letter comes from a different factor; a finite tail a_F built from such elements cuts out arbitrary positive elements, making them comparable to multiples of a_j. Lemma 2.3 turns this into Cuntz-order inequalities that yield a minimal essential ideal and prove it is purely infinite and simple.","core_discovery":"The paper proves Theorem 1: if (A, φ) is a selfless C*-probability space and φ is nonfaithful, then A is purely infinite and simple. The engine is Theorem 6, which states that for an infinite family of C*-probability spaces with nonfaithful states inducing faithful GNS representations, the reduced free product contains an essential purely infinite simple ideal, and if the sum of δ_j (the norms of the states restricted to the ideals generated by the kernels) exceeds 1, the entire product is purely infinite and simple. This condition is computable factor-by-factor and requires no unitaries in the kernels. Applying the theorem to an infinite self-free product of (A, φ) and using the existential","pith_inferences":["The tail-excision mechanism suggests a general principle: in infinite free products, nonfaithfulness becomes a resource rather than an obstruction, potentially useful for proving pure infinity in amalgamated or graph products where kernel positives behave similarly.","The δ_j > 1 condition is likely far from sharp; the mechanism may work under weaker divergence conditions, e.g., infinitely many δ_j > 0, since the theorem already covers the case of a divergent sum.","The transfer of pure infinity from an infinite free product to a factor via an existential embedding hints at a broader model-theoretic transfer principle: existential embeddings of nonfaithful states may preserve purely infinite simplicity in other contexts.","The new Kirchberg-algebra isomorphisms indicate that infinite reduced free products with pure states are classified by K-theory alone, and the distinction between state-preserving and non-state-preserving isomorphisms is controlled by purity of the factor states."],"forward_implications":["Every selfless C*-algebra is pure (Corollary 3), removing the faithful-state assumption from earlier results.","Permanence properties improve: reduced free products with a selfless factor and a GNS-faithful factor are selfless even without separability (Prop 3.1), and matrix amplifications and UHF tensor products of selfless algebras are selfless (Prop 3.2).","The Choda–Dykema conjecture holds when simplicity is strengthened to one factor being selfless (Corollary 5).","Infinite reduced free products of separable nuclear UCT algebras with pure states are Kirchberg algebras satisfying the UCT, and their isomorphisms are governed by K-theory via the Kirchberg–Phillips theorem (Theorem 4.1).","New concrete isomorphisms arise, such as O_∞ ≅ (Z, τ)∗(Z, φ) and (O_∞, ω) ≅ ∗_{j=1}^∞ (Z, φ) for pure states φ, ω and trace τ."],"fun_headline_variants":["Nonfaithful selfless algebras: purely infinite, simple","Dichotomy complete: every selfless C*-algebra is one of two","Selfless C*-algebras: stable rank one or purely infinite","Nonfaithful states push selfless algebras to pure infinity","Selfless algebras split cleanly: no middle ground"],"cache_read_input_tokens":2304,"weakest_assumption_plain":"Lemma 2.3(ii) assumes that positive elements in the kernels of the factors annihilate all centered words from other factors in the free-product Hilbert space; if that excision property fails, Theorem 6—and with it Theorem 1—does not follow.","fun_headline_variants_meta":{"raw":{"variants":["Nonfaithful selfless algebras: purely infinite, simple","Dichotomy complete: every selfless C*-algebra is one of two","Selfless C*-algebras: stable rank one or purely infinite","Nonfaithful states push selfless algebras to pure infinity","Selfless algebras split cleanly: no middle ground"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000172,"raw_usage":{"total_tokens":1079,"prompt_tokens":678,"completion_tokens":401,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":422,"completion_tokens_details":{"reasoning_tokens":315}},"tokens_in":422,"tokens_out":401,"duration_ms":4321,"temperature":1.0,"reasoning_tokens":315,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-02T12:10:23.411605+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"A concrete counterexample to the excision step: exhibit factors (A_j, φ_j) with nonfaithful GNS-faithful states, a finite F disjoint from the support of an approximant d, and a nonzero word w with a_F d_0 a_F w ≠ 0, or otherwise compute a tail product a_F c a_F that does not reduce to ω(c) a_F^2. Alternatively, find a selfless (A, φ) with φ nonfaithful whose infinite self-free product is not purely infinite simple—this would directly refute Theorem 1's route.","supporting_citations":[],"review_version":2}