{"id":"38fe5df1-b6e0-4f80-8470-13079a2e8470","arxiv_id":"2502.05125","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"A discrete quantum group has the approximation property exactly when every element of its C*- or von Neumann crossed products has a Fejér-type series representation.","lead":"This paper proves that a discrete quantum group has the approximation property exactly when its crossed products admit a Fejér-type series expansion, generalizing a known theorem for classical groups. The result gives a structural way to recognize and use the approximation property in quantum settings, with applications to operator bimodules and crossed products.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"AP-to-Fejér step for arbitrary von Neumann algebras rests on an imported stable-convergence replacement that the paper does not justify for quantum groups.","rationale":"The reader identified the same technical step as the weakest assumption, and I agree that it is the most load-bearing point in the paper. However, I do not think the cited Haagerup–Kraus result is inherently limited to classical groups: Proposition 1.7 in [21] is a general von Neumann algebra statement about normal completely bounded maps, and the maps Θ^ℓор(\\f_ι) are normal c.b. maps on L^∞(\\widehat{G}ор). Thus the step is very likely valid, but the manuscript should say so explicitly rather than leaving the reader to infer it. The central Theorems 3.2 and 3.7 otherwise appear coherent: Lemma 3.1 supplies the finite Fejér sums, the stable point-weak* convergence is converted to crossed-product convergence via the standard \\widetilde{W} identity, and the C*-version uses Mazur convex combinations to obtain norm convergence. I found no independent mathematical error in the main equivalence that would force a change of verdict. The flagged Ditkin assumption in Proposition 4.8 appears genuinely unused in the proof, but that is a peripheral blemish, not a threat to the central claim. Since the reader's CONDITIONAL verdict is based on a real but likely addressable gap in justification, I recommend leaving the verdict unchanged pending the proposed verification.","tokens_in":28554,"tokens_out":35112,"duration_ms":326082,"concrete_test":"Check the exact statement and proof of [21, Proposition 1.7]. If it is stated for an arbitrary von Neumann algebra (or if its proof transfers verbatim to M = L^∞(\\widehat{G}ор)), then insert an explicit sentence in §3 identifying Θ^ℓор(\\f_ι) as a normal c.b. map on L^∞(\\widehat{G}ор) and applying Proposition 1.7 verbatim; this closes the gap. If the proposition is group-specific, then verify directly that AP stable point-weak* convergence on L^∞(\\widehat{G}ор) ⊗ B(H) for all H implies point-weak* convergence on G ⋉ N for every von Neumann algebra N, for instance by working in a faithful normal representation N ⊂ B(H_N) and reducing normal functionals to separable subspaces.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The load-bearing step is in the proof of Theorem 3.2 (page 10): after defining the AP via stable point-weak* convergence of Θ^ℓ(\\f_ι) on B(H) for Hilbert spaces H, the proof replaces B(H) by an arbitrary von Neumann algebra N, citing [21, Proposition 1.7]. This replacement is what turns AP into the Fejér representation (18) for every action on every von Neumann algebra N. Without it, the forward direction of Theorem 3.2 is only established for N = B(H), and the von Neumann-algebraic Fejér theorem is not proven in the stated generality. The paper neither states the proposition nor explains why its hypotheses are satisfied by the maps Θ^ℓор(\\f_ι) on L^∞(\\widehat{G}ор). The concern is not that the statement is false: Haagerup–Kraus Proposition 1.7 is a general result about normal completely bounded maps on von Neumann algebras and should apply to M = L^∞(\\widehat{G}ор). But as written, the proof contains an unexplained importation from the classical group literature, and this is exactly the condition that must hold for the central equivalence to be valid for arbitrary N.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper proves an equivalence, for a discrete quantum group G, between the approximation property (AP) and the existence of Fejér-type representations for elements of arbitrary C*-algebraic or von Neumann algebraic crossed products by G (Theorems 3.2 and 3.7). The proof is built on Lemma 3.1, which expresses the action of the multiplier Θ̂ℓop(f) on crossed-product elements as an explicit finite sum involving the conditional expectation E and matrix coefficients of the dual compact quantum group. The authors then apply this Fejér theorem to obtain structural results for invariant L∞(Ĝ)-bimodules in B(ℓ2(G)) and C(Ĝ)-bimodules in K(ℓ2(G)), to characterize jointly invariant subspaces in terms of closed left ideals of ℓ1(G), and to prove a slice-map property for actions of discrete quantum groups with the AP.","tokens_in":28750,"tokens_out":26145,"duration_ms":267748,"significance":"If the main equivalence holds, it is a natural and substantial quantum-group analogue of the classical Fejér theorem for crossed products, and it identifies the AP precisely as the summability condition that recovers every crossed-product element from its Fourier coefficients. The paper is well organised, gives detailed proofs of the central statements, and contains several genuinely useful applications: the bimodule characterizations in Theorems 4.4, 4.5, 4.25 and 4.26, the harmonic-operator corollaries, and the Fubini crossed-product slice-map property. The main definitions are standard, and the formulas are explicit enough to be checked. The principal weakness is a load-bearing imported step in the proof of Theorem 3.2, which is discussed below; it is repairable but should be addressed before publication.","major_comments":[{"comment":"The forward direction uses the assertion, attributed to [21, Proposition 1.7], that stable point-weak* convergence of Θ̂ℓop(fι) on B(H) for Hilbert spaces H implies weak* convergence of Θ̂ℓop(fι)⊗idN on L∞(Ĝop)⊗N for every von Neumann algebra N. This assertion is the only step that upgrades the AP from its Hilbert-space formulation to arbitrary crossed products, and it is exactly what yields the Fejér representation (18) for arbitrary N. The manuscript neither states the proposition nor verifies its hypotheses (normality and complete boundedness of the maps, and a uniform cb-bound for the net). Please add this verification, or replace the appeal by the direct argument: represent N faithfully on a Hilbert space K, apply the stable convergence with H = L2(G)⊗K, and restrict to the subalgebra L∞(Ĝop)⊗(B(L2(G))⊗N).","section":"§3, proof of Theorem 3.2 (p. 10)"}],"minor_comments":[{"comment":"The stated hypothesis that Ĝ has Ditkin's property is never used in the proof: (ii) follows from Theorem 4.4 and equation (27), and (iii) follows from (i) and Theorem 4.4. Moreover, Corollary 4.9 applies (ii) without the Ditkin assumption. Please either prove the stronger statements or explain the role of Ditkin.","section":"§4.1, Proposition 4.8(ii)-(iii)"},{"comment":"The definition of the AP says 'for any (separable) Hilbert space H' with the parentheses; this ambiguity matters for the passage to arbitrary von Neumann algebras. Please specify whether stable convergence is required for all Hilbert spaces or only separable ones, and ensure consistency with the definition used in [17].","section":"§3, AP definition"},{"comment":"In the displayed computation near equation (23), the sums written as 'β∑' and 'γ∑' should presumably be 'nβ∑' and 'nγ∑'; the notation is otherwise confusing.","section":"Lemma 3.3, displayed computation"},{"comment":"The sentence 'Repeating the proof of Lemma 3.1' is not literally accurate, because Lemma 3.1 is stated for von Neumann actions, whereas Theorem 3.7 concerns C*-actions on reduced crossed products. Please add a sentence explaining that the C*-case follows by applying the lemma to the double-dual action α~ on A** and restricting to Gα⋉rA.","section":"§3, proof of Theorem 3.7"}],"recommendation":"major_revision","confidential_remarks":"The central claim appears defensible and the gap in Theorem 3.2 is local and easily repairable. I do not see grounds for rejection. The Ditkin assumption issue in Proposition 4.8 is a minor internal inconsistency, not a threat to the main results."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Quick take: the main AP/Fejér equivalence for discrete quantum groups is real, and the applications are substantial. The paper deserves a serious referee; the issues I found are minor and fixable.\n\nWhat's new: The equivalence between AP and Fejér representability in crossed products—Theorem 3.2 and 3.7—is genuinely new for quantum groups and doesn't reduce to the classical Crann–Neufang result. The proof via Peter–Weyl coefficients in Lemma 3.1 is clean and detailed. The bimodule applications, especially Corollary 4.7, extend known results and give something new even for discrete groups. The slice map property for AP actions is a natural and useful extension.\n\nSoft spots:\n- The step in Theorem 3.2 replacing B(H) by an arbitrary von Neumann algebra N cites Haagerup–Kraus Prop 1.7 without stating it. The stress-test worried this is not justified for quantum groups. I don't think the concern lands: HK Prop 1.7 is a general statement about normal completely bounded maps on von Neumann algebras, not specific to groups. Once AP gives stable point-weak* convergence on L∞(Ĝop), the stable convergence passes to any N by that proposition. The paper should state the proposition and verify the maps are normal c.b., but that's a presentation issue, not a gap.\n- Proposition 4.8 states a Ditkin assumption for parts (ii) and (iii), but the proof never uses it. Either the assumption is extraneous or the proof is missing a step. This is worth asking about, but it's peripheral to the main theorem.\n- The pre-annihilator step in Prop 4.8(iii) (identifying (I1∩I2)^⊥ with I1^⊥+I2^⊥) is terse and needs a bit more justification, though it looks like a standard fact if the spaces are closed.\n\nOverall: the central equivalence is well argued and the applications are meaningful. The paper is honest: no sign of circularity, no fitted parameters. It's a serious piece of work.\n\nBottom line: send it to a capable referee. I'd engage with it—cite it if I work in this area. For a reading group, it's a good choice for people interested in quantum groups and approximation properties.","headline":"A genuinely new AP/Fejér equivalence for discrete quantum groups with solid proofs and useful applications; the flagged concerns are mostly presentation issues, not real gaps.","tokens_in":29298,"tokens_out":4430,"would_cite":true,"duration_ms":40925,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["46L67","46L55","46L07","46L89","43A55"],"pacs":[],"model":"deepseek-v4-flash","headline":"A discrete quantum group has the approximation property exactly when its crossed products admit Fejér summability.","keywords":["discrete quantum groups","crossed products","Fejér representation","approximation property","completely bounded multipliers","bimodules","Fubini crossed product","slice map property"],"falsifier":"Find one discrete quantum group $\\mathbb G$ without the approximation property and one action $\\alpha$ on a von Neumann algebra $N$ for which the net in (18) fails to converge weak* to some $T\\in\\mathbb{G}\\ltimes N$; equivalently, exhibit a Hilbert space $H$ and a von Neumann algebra $N$ such that the stable-convergence replacement used on page 10 (citing [21, Proposition 1.7]) breaks for a non-classical quantum group, because then the 'only if' direction of Theorem 3.2 collapses.","tokens_in":28328,"feed_emoji":"🧮","tokens_out":10198,"duration_ms":90473,"temperature":0.7,"pith_summary":"This paper establishes that a discrete quantum group has the approximation property exactly when every element of any of its crossed products—von Neumann or C*-algebraic—can be recovered as the weak* limit (norm limit in the C*-case) of an explicit Cesàro-style Fejér average built from the dual group. The average is a finite sum over irreducible co-representations of the dual, with coefficients obtained by applying the canonical conditional expectation to the element twisted by the corresponding matrix coefficients. The equivalence is proved as Theorems 3.2 and 3.7, and it converts a stability condition about completely bounded multipliers into a concrete summability statement for Fourier series in crossed products. The paper then uses this representation to characterize the invariant $L^\\infty(\\hat{\\mathbb G})$-bimodules of $\\mathcal B(\\ell^2(\\mathbb G))$ and $C(\\hat{\\mathbb G})$-bimodules of $\\mathcal K(\\ell^2(\\mathbb G))$, and to show that every action of such a quantum group has the slice map property. If the paper is right, the approximation property is exactly the summability condition that makes quantum crossed products behave like commutative ones.","feed_headline":"Quantum group AP equals Fejér summability in crossed products","feed_subtitle":"Cesàro-type averages from the dual group recover every crossed-product element exactly when AP holds.","key_machinery":"The load-bearing object is the operator-valued weight $E=E_{\\hat\\alpha}$ induced by the dual action $\\hat\\alpha$ of $\\hat{\\mathbb G}^{\\mathrm{op}}$ on the crossed product, together with the completely bounded slice maps $\\Theta^{\\ell op}(f)$ implemented by the fundamental unitary $\\widetilde W$ of $\\hat{\\mathbb G}^{\\mathrm{op}}$. Lemma 3.1 shows that for $f\\in\\mathrm{Pol}(\\hat{\\mathbb G})\\cdot\\hat\\phi$, the slice $(\\Theta^{\\ell op}(f)\\otimes\\mathrm{id})(T)$ can be rewritten as the displayed finite Fejér sum; the coefficients $d_\\beta/\\lambda^\\beta_i\\,\\langle f,\\hat u^\\beta_{ji}\\rangle$ come exactly from the Peter–Weyl orthogonality relations, so they match the Fourier coefficients of $f$ with respect to the dual co-representations. The approximation property supplies a net $f_\\iota\\to 1$ in the stable point-weak* topology of $M^\\ell_{cb}(L^1(\\hat{\\mathbb G}))$, and the proof reduces to whether this stable convergence survives tensoring with an arbitrary von Neumann algebra $N$.","core_discovery":"The paper's central claim is that a discrete quantum group $\\mathbb G$ has the approximation property if and only if, for every action $\\alpha$ on a von Neumann algebra $N$ and every $T$ in the crossed product $\\mathbb G\\ltimes N$, $$T=\\mathrm{w}^*-\\lim_\\iota\\sum_{\\$\\beta$\\in F_\\iota}\\sum_{i,j,k=1}^{n_\\$\\beta$}\\frac{d_\\$\\beta$}{\\$\\lambda$^\\beta_i}\\langle f_\\iota,\\hat u^\\beta_{ji}\\rangle E\\bigl(T((\\hat u^\\beta_{ki})^*\\otimes 1)\\bigr)(\\hat u^\\beta_{kj}\\otimes 1),$$ where $(f_\\iota)$ is a net in $\\mathrm{Pol}(\\hat{\\mathbb G})\\cdot\\hat\\phi$, the $\\hat u^\\beta_{ij}$ are matrix coefficients of irreducible co-representations of $\\hat{\\mathbb G}$, $d_\\beta$ is the quantum dimension, $\\lambda^\\beta_i$ are the weights from the Peter–Weyl orthogonality relations, and $E$ is the operator-valued weight induced by the dual action (a conditional expectation in the C*-case). The same statement holds in norm topology for C*-crossed products. Thus the AP is not merely a technical property of the dual Banach algebra of completely bounded multipliers; it is precisely the summability condition under which the Cesàro–Fourier series of every crossed-product element converges back to that element.","pith_inferences":["If the unproved replacement of $\\mathcal B(H)$ by an arbitrary von Neumann algebra fails for a non-classical quantum group, the 'if' direction of Theorem 3.2 survives but the 'only if' may not; this hinge can be probed directly by testing the displayed net with $\\mathbb G$ acting on $N=L^\\infty(\\mathbb G)$ by the co-multiplication, where the crossed product is concrete.","The closed-form coefficients suggest a truncation scheme: on finite-dimensional truncations of the dual (e.g., root-of-unity deformations), the Fejér average becomes a finite matrix expression, giving a direct numerical route to approximate crossed-product elements.","The bimodule characterization is a natural stepping stone to a Galois correspondence for intermediate subalgebras of discrete quantum group crossed products, which the authors explicitly postpone; in the classical setting the same representation supplies exactly that correspondence.","If the slice map property holds for all invariant subspaces, it should imply rigidity of crossed products under equivariant embeddings; conversely, known classical examples where slice maps fail should correspond to discrete quantum groups without the AP."],"forward_implications":["For every discrete quantum group with the AP, each element of any von Neumann crossed product is a weak* limit of the displayed Fejér averages, and each element of any C*-crossed product is a norm limit of them (Theorems 3.2 and 3.7).","The AP is necessary for the representation: if a discrete quantum group lacks the AP, some crossed product element cannot be recovered by these Cesàro averages.","For an AP discrete quantum group, every weak*-closed $L^\\infty(\\hat{\\mathbb G})$-bimodule in $\\mathcal B(\\ell^2(\\mathbb G))$ that is invariant under completely bounded left multipliers is of the form $\\mathrm{Bim}(J^\\perp)=\\mathrm{Ran}(J)^\\perp$ for a closed left ideal $J$ of $\\ell^1(\\mathbb G)$, with the norm-closed analogue for $C(\\hat{\\mathbb G})$-bimodules in $\\mathcal K(\\ell^2(\\mathbb G))$.","Every action of an AP discrete quantum group on a von Neumann algebra or C*-algebra has the slice map property: the Fubini crossed product of any invariant subspace coincides with its ordinary crossed-product subspace.","Compact groups, which always have the AP, obtain the new bimodule characterization in $\\mathcal K(L^2(G))$ as a special case."],"supporting_citations":[{"why":"Supplies the classical AP definition and the stable-convergence replacement of B(H) by a von Neumann algebra N, the step the quantum proof relies on.","marker":"[21]"},{"why":"The group-case Fejér theorem whose structure and applications the paper generalizes to discrete quantum groups.","marker":"[15]"},{"why":"Establishes the AP for quantum groups in the stable point-weak* formulation and the duality G vs G′, used to switch to Θ^{ℓop}.","marker":"[17]"},{"why":"Gives the representation of completely bounded multipliers as Θ^ℓ and Θ^r, the maps appearing in the Fejér slices.","marker":"[24]"},{"why":"Proves the corresponding coefficient-functional result for Kac algebras, which Corollary 3.5 extends to discrete quantum groups.","marker":"[28]"},{"why":"Introduces the bimodules Bim(J⊥) and Ran(J) and the harmonic-operator theorems being extended.","marker":"[5]"},{"why":"Provides the locally compact quantum group crossed-product machinery, including the dual action and operator-valued weight E.","marker":"[34]"},{"why":"Supplies the identification K(ℓ2(G)) with the reduced crossed product c0(G) and the calculus for E on compact operators.","marker":"[23]"}],"fun_headline_variants":["Fejér sums recover all crossed product elements iff AP holds","Quantum group AP: dual to Cesàro-Fejér summability in crossed products","Fejér's theorem for discrete quantum groups: equivalent to AP","AP equals Fejér summability in quantum group crossed products","Fejér representations characterize AP in discrete quantum groups"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The key step assumes that stable point-weak* convergence of the net $\\Theta^{\\ell op}(f_\\iota)$ — known when tensored with $\\mathcal B(H)$ for Hilbert spaces $H$ — remains valid when $\\mathcal B(H)$ is replaced by an arbitrary von Neumann algebra $N$; the paper cites a classical-group result for this replacement and supplies no quantum-group proof.","fun_headline_variants_meta":{"raw":{"variants":["Fejér sums recover all crossed product elements iff AP holds","Quantum group AP: dual to Cesàro-Fejér summability in crossed products","Fejér's theorem for discrete quantum groups: equivalent to AP","AP equals Fejér summability in quantum group crossed products","Fejér representations characterize AP in discrete quantum groups"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000739,"raw_usage":{"total_tokens":3314,"prompt_tokens":974,"completion_tokens":2340,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":590,"completion_tokens_details":{"reasoning_tokens":2250}},"tokens_in":590,"tokens_out":2340,"duration_ms":16296,"temperature":1.0,"reasoning_tokens":2250,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-08T20:10:56.738254+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Find one discrete quantum group $\\mathbb G$ without the approximation property and one action $\\alpha$ on a von Neumann algebra $N$ for which the net in (18) fails to converge weak* to some $T\\in\\mathbb{G}\\ltimes N$; equivalently, exhibit a Hilbert space $H$ and a von Neumann algebra $N$ such that the stable-convergence replacement used on page 10 (citing [21, Proposition 1.7]) breaks for a non-classical quantum group, because then the 'only if' direction of Theorem 3.2 collapses.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the classical AP definition and the stable-convergence replacement of B(H) by a von Neumann algebra N, the step the quantum proof relies on."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"The group-case Fejér theorem whose structure and applications the paper generalizes to discrete quantum groups."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Establishes the AP for quantum groups in the stable point-weak* formulation and the duality G vs G′, used to switch to Θ^{ℓop}."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Gives the representation of completely bounded multipliers as Θ^ℓ and Θ^r, the maps appearing in the Fejér slices."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Proves the corresponding coefficient-functional result for Kac algebras, which Corollary 3.5 extends to discrete quantum groups."},{"cited_title":"249 (2019), no","cited_arxiv_id":null,"evidence_quote":"Introduces the bimodules Bim(J⊥) and Ran(J) and the harmonic-operator theorems being extended."},{"cited_title":"Thesis, 2001","cited_arxiv_id":null,"evidence_quote":"Provides the locally compact quantum group crossed-product machinery, including the dual action and operator-valued weight E."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the identification K(ℓ2(G)) with the reduced crossed product c0(G) and the calculus for E on compact operators."}],"review_version":1}