{"id":"9b47f529-0d02-4bdf-b307-c21f0fba09cc","arxiv_id":"2608.11193","paper_version":1,"verdict":"ACCEPT","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"Derived block spaces for the Drinfel'd double of S3 in characteristic 3 carry new SL(2,Z)-representations in cohomological degrees 1 and 2 mod 4 that are not generated from degree zero under the Yoneda product.","lead":"This paper computes the action of mapping class groups on derived block spaces for Drinfel'd doubles of finite groups in positive characteristic. It shows that, for the symmetric group on three letters over characteristic 3, the derived spaces contain genuinely new representations that cannot be obtained from the ordinary block spaces.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"No significant objection identified.","rationale":"The reader's verdict of ACCEPT with moderate confidence is reasonable. The reader's identified weakest assumption, the Theorem 4.28 dictionary between the Lyubashenko action and the representation variety action, is also the point I examined most closely. The proof of Theorem 4.28 is explicit for the generators, and the final non-isomorphism is certified by the central element s^2 = -I, whose action is robust under the conventional sign choices that could plausibly vary. The cohomology computations in Section 7 are internally consistent: the characteristic-3 decompositions in Paragraph 6.5, the fixed-point calculations via Proposition A.5, and the Yoneda module structure arguments in Theorem 7.13 all cohere. No circular step or missing proof was found in the portions needed for the headline claim. A direct computational re-derivation of the SL(2,Z)-module structure using the explicit dictionary would still be worthwhile as an independent check, but it is a verification step, not a demonstrated flaw.","tokens_in":67297,"tokens_out":40279,"duration_ms":353659,"concrete_test":"Independently compute Z^1 and Z^2 for G = S_3 over K of characteristic 3 by applying Corollary 5.4 to the explicit commuting-pair basis of M_1, using the standard R-matrix isomorphism Theta of Corollary 4.37 and the generator action formulas assembled in Theorem 4.28; verify that s^2 acts as -1 on the m congruent to 1 or 2 mod 4 summands and as +1 on Z^0, and that the H(S_3, K)-submodule generated by Z^0 has no degree-1 or degree-2 components. This isolates any sign or basepoint error in the geometric dictionary from the cohomological reductions of Section 5.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The most delicate point is the isomorphism Theta-tilde of Theorem 4.28, on which every example depends: it identifies the Lyubashenko action on W^e with the Gamma_g(x)-action on the representation variety M_g, and a sign or basepoint error there could change the SL(2,Z)-module K_chi2[P^1_3] into the untwisted K[P^1_3], destroying the claimed non-isomorphism. However, the proof checks the generators t'_i, s'_i, and n''_i explicitly, the transport from y to x is described in Paragraph 4.9, and the distinguishing test used in the paper, the action of the central element s^2 = -I, is insensitive to the plausible pre/postcomposition sign choices. I found no internal inconsistency in Sections 4 through 7, and the module-theoretic conclusion of Theorem 7.13 follows from Theorem 7.7 together with the vanishing of the degree-1 and degree-2 components of the Yoneda algebra H(S_3, K) in characteristic 3. I therefore do not regard the basepoint dictionary as a live objection.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"This paper is the second part of the authors' program on mapping class group actions on derived block spaces of non-semisimple modular tensor categories. It specializes to the category of representations of the Drinfel'd double D(G) of a finite group G over an algebraically closed field, in the non-semisimple case where the characteristic divides |G|. The main technical steps are: (i) an explicit comparison (Theorem 4.28) between the Lyubashenko action on W^e = (L^{⊗g})^e and the action on the linearized representation variety M_g = K[Hom(π_1(Σ_g),G)], together with a companion statement for the standard R-matrix (Corollary 4.37); (ii) a reduction of the derived block spaces Z^m(Σ_g) to group cohomology groups H^m(G,M_g) (Corollary 5.4); and (iii) a decomposition of H^m(G,M_g) according to conjugacy classes of cyclic subgroups, reducing the computation to cohomology of centralizers and eigenspaces for Dirichlet characters. The paper then computes the full SL(2,Z)-module structure of Z^*(Σ_1) for G=S_3 in characteristics 2 and 3. In characteristic 3 it finds that for m≡1 or 2 mod 4 the representation is K_{χ2}[P^1_3], which is not isomorphic to any summand of the degree-zero block space because s^2 acts by -1, and Theorem 7.13 shows that this summand is not contained in the submodule generated by Z^0(Σ_1) under the Yoneda product.","tokens_in":67528,"tokens_out":13274,"duration_ms":115402,"significance":"If correct, the paper achieves its stated goal: derived block spaces can carry genuinely new mapping class group representations that are invisible in the ordinary block spaces and are not generated from degree zero by Yoneda products. The proof is unusually explicit. The formulas for S, T, N, A, B, C, G, and U are given on the basis u(a,b); the capping argument in Paragraph 4.9 is checked; the reduction to group cohomology in Section 5 is proved; the decomposition in Section 6 is careful about the finite-group actions; and the final computations in Section 7 use the periodic resolution and the Yoneda/cup product compatibilities from the appendix. The paper is honest about the set-theoretic universe assumption and about its reliance on the authors' Part I. The most delicate point, the isomorphism Θ-tilde of Theorem 4.28, is handled by explicit checks on the generators t'_i, s'_i, and n''_i; the sign and basepoint conventions are discussed, and the distinguishing invariant s^2 = -I is insensitive to the plausible sign ambiguities.","major_comments":[],"minor_comments":[{"comment":"The authors state that they do not know a reference for this naturality statement. The proof given is correct, but since the lemma is used to define the vector-space structure on Yext and thereby the Yoneda module structure, a reference to the standard naturality of the Yoneda product would be helpful.","section":"Paragraph 5.1, Lemma 5.1"},{"comment":"The set-theoretic axiom of a universe is stated once in the introduction, but the reader is not told where the smallness assumption is used, e.g., in forming the big groups Yext^m. A footnote at the first occurrence of 'big group' would improve the exposition.","section":"Introduction and Paragraph 5.1"},{"comment":"The isomorphism Θ-tilde is the hinge of the paper. The proof is explicit on the generators, but a short remark that the isomorphism is independent of the chosen diffeomorphism φ and that the non-generation conclusion is unchanged under the alternative pre/postcomposition convention would eliminate a possible source of doubt.","section":"Paragraph 4.10, Theorem 4.28"},{"comment":"The paper relies heavily on notation and results from Part I, in particular the definition of Z^m(Σ_g), the capping homomorphism, and the generators t_i, s_i, n_l. This dependence is clearly declared, but a brief 'Notation and prerequisites' section or a more detailed recap in the introduction would make the paper more readable for a journal audience.","section":"Throughout"},{"comment":"The sentence 'we can subtract from this element some elements from the submodule so that the resulting element has a scalar λ∈K as its first component' would benefit from one more sentence explaining why this reduction is possible, for instance by using the explicit S-basis and the form of the two generators of the kernel.","section":"Paragraph 7.5, proof of Proposition 7.11"},{"comment":"There are small stylistic issues, such as the use of the Kronecker symbol δ_{h,[b^{-1},a]} in Proposition 4.3 and the repeated use of 'in total' at the ends of proofs. These are harmless and can be smoothed in the final version.","section":"Various"}],"recommendation":"minor_revision","confidential_remarks":"The paper is a direct continuation of the authors' Part I and builds substantially on their earlier work [LMSS1] and [SZ]. The citation pattern is heavy, but it is appropriate here because the genuinely new content is the explicit dictionary, the cohomological reduction, and the non-generation example. If the journal requires the paper to be readable without Part I, the lack of a self-contained summary of Part I would be a concern; otherwise I see no issue. The set-theoretic universe assumption is nonstandard for some readers, but it is explicitly flagged and does not affect the algebraic content of the computations."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The paper answers the question it left open in Part I: the mapping class group action on derived block spaces is not redundant, and the nontriviality is genuine. The main new content is explicit: the Lyubashenko action for Drinfel'd doubles is identified with the representation variety action (Theorem 4.28), the cohomology reduces cleanly to group cohomology of the finite group (Corollary 5.4), and the S3 examples in characteristic 3 give SL(2,Z)-representations in higher cohomological degree that are neither isomorphic to the degree-zero ones nor generated by them under the Yoneda product (Theorems 7.7, 7.13). That is a real result, and the proof is thorough: the S, T, N, A, B, C formulas are checked in detail, the capping argument is handled, and the reduction to group cohomology is set up carefully. I did not find a load-bearing error. The central claim holds up.\n\nThe soft spots are minor. The set-theoretic universe convention is stated honestly and is standard; the Yoneda-product treatment in Section 5 is a bit heavy but does the job. Lemma 5.1 is flagged as without reference, which is an honest caveat rather than a flaw. The self-citation is heavy, but the cited Part I and earlier work genuinely set up the framework, and the new claim is not an input of that framework. The most delicate point, the basepoint/sign dictionary in Theorem 4.28, is addressed explicitly in Paragraphs 4.9–4.12, and the distinguishing test used later (the action of s^2 = -I) is insensitive to the plausible sign ambiguities. I do not regard that as a live objection.\n\nThe appendix contains useful material, though some of it (the acyclic assembly lemma, tensor products of resolutions) is standard and could be trimmed or referenced. The cup-product discussion is careful but the paper does not need it for the main result.\n\nWho gets value from this: people working on non-semisimple TQFTs, finite-group Dijkgraaf-Witten theory in positive characteristic, and modular tensor categories. It is a serious paper that deserves a real referee; it is not a desk reject. I would send it to review, and I would expect the referee to engage with the S3 computations and the representation-variety dictionary in detail. My own verdict is positive: accept, with minor revisions.","headline":"A careful, explicit continuation that delivers the first nonabelian example where derived block space representations genuinely differ from ordinary ones; the S3 characteristic-3 computation is the real payoff.","tokens_in":68034,"tokens_out":615,"would_cite":true,"duration_ms":8734,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["16E40","18M15","20J06","57K20"],"pacs":[],"model":"deepseek-v4-flash","headline":"For the Drinfel'd double of S3 in characteristic 3, the torus's derived block spaces carry an SL(2,Z)-representation in degrees 1 and 2 mod 4 that is absent from ordinary block spaces and cannot be generated from them by Yoneda products.","keywords":["Hochschild cohomology","derived block spaces","mapping class groups","Drinfel'd double","modular tensor categories","group cohomology","representation variety","Yoneda product"],"falsifier":"Recompute the action of $s^2$ on $Z^1(\\Sigma_1)$ for the Drinfel'd double of $S_3$ in characteristic 3, using the standard R-matrix directly. The paper's Theorem 7.7 predicts that this element acts as $-\\mathrm{id}$, so that its trace equals minus the dimension; finding the identity instead would show the summand $K_{\\chi_2}[\\mathbb{P}^1_3]$ is absent and the non-isomorphism claim fails.","tokens_in":67122,"feed_emoji":"📐","tokens_out":11637,"duration_ms":99391,"temperature":0.7,"pith_summary":"This paper establishes that passing from ordinary block spaces to derived block spaces genuinely changes the mapping class group representations attached to a surface. It proves this by explicit computation for the Drinfel'd double of the symmetric group S3 over fields of characteristic 2 and 3, on a torus. In characteristic 3, the cohomological degrees congruent to 1 or 2 modulo 4 carry an SL(2,Z)-representation that does not occur in degree zero, and Theorem 7.13 shows it is not contained in the submodule generated by the degree-zero block spaces under the Yoneda product. The route to this result is an identification of the categorical action with the action on a linearized representation variety, followed by a reduction of the relevant Ext groups to ordinary group cohomology.","feed_headline":"Higher-degree block spaces carry new SL(2,Z) representations","feed_subtitle":"For S3 in characteristic 3, torus block spaces in degrees 1 and 2 mod 4 cannot be built from degree zero.","key_machinery":"The load-bearing object is the isomorphism $\\tilde{\\Theta}$ of Theorem 4.28 between the linearized representation variety $M_g=K[\\mathrm{Hom}(\\pi_1(\\Sigma_g,x),G)]$ and the degree-zero homogeneous component $W^e$ of $L^{\\otimes g}$, sending a homomorphism $f$ to $u(a_1,b_1^{-1})\\otimes\\cdots\\otimes u(a_g,b_g^{-1})$ with $a_i=f([\\alpha_i])$, $b_i=f([\\beta_i])$. It identifies the coend-based action built from the alternative R-matrix with the geometric action of $\\Gamma_g(x)$ on fundamental-group homomorphisms, and Corollary 5.4 converts derived block spaces into group cohomology $H^m(G,M_g)$. The subsequent decomposition of $M_g$ according to subgroups, centralizers, normalizers and Dirichlet characters, together with the periodic projective resolution for cyclic groups, turns this into explicit SL(2,$\\mathbb{Z}$)-computations for $S_3$.","core_discovery":"The central claim is that the mapping class group representations on derived block spaces are, in general, genuinely new: for the Drinfel'd double of $G=S_3$ over a field of characteristic 3 and a torus (genus $g=1$), the representation on $Z^m(\\Sigma_1)$ is isomorphic to $K_{\\chi_2}[\\mathbb{P}^1_3]$—a four-dimensional, sign-twisted SL(2,$\\mathbb{Z}$)-representation on functions on the projective line over $\\mathbb{F}_3$—when $m\\equiv 1$ or $2 \\pmod 4$, while degree zero is $K\\oplus K[\\mathbb{P}^1_2]\\oplus K[\\mathbb{P}^1_3]$. Theorem 7.13 then shows this summand is not contained in the submodule generated by $Z^0(\\Sigma_1)$ under the Yoneda product. In other words, the derived block spaces contain mapping class group information that the ordinary block spaces, together with products with the cohomology ring of the point, do not see.","pith_inferences":["The 4-periodic pattern in characteristic 3 likely reflects the 2-periodic resolution of the cyclic subgroup $A_3$ tensored with a sign twist from the normalizer quotient $S_3/A_3$; one could test $A_4$ or $S_4$ in characteristic 2 or 3 for analogous new representations.","Because the new representation lives in the summand attached to the cyclic subgroup $A_3$, the same mechanism should appear for any finite group with a cyclic subgroup whose normalizer quotient acts by a nontrivial character on the subgroup's cohomology; a systematic check across finite groups would delimit the phenomenon.","The gluing map in Proposition 7.14 is exactly the kind of datum a derived modular functor would have to assemble consistently; constructing such a functor at the level of categories, rather than only representations, remains an open direction suggested by these computations."],"forward_implications":["The two R-matrices of the Drinfel'd double give isomorphic mapping class group representations (Theorem 4.36), so the new higher-degree representations are intrinsic rather than an artifact of a braiding choice.","For abelian $G$, $Z^\\infty(\\Sigma_g)$ is free as a right module over $Z^\\infty(\\Sigma_0)$, so no genuinely new representations appear; nonabelian finite groups are necessary for the phenomenon.","For $S_3$ in characteristic 2 all higher-degree representations already occur in degree zero, whereas in characteristic 3 new ones appear in a 4-periodic pattern; the characteristic matters sharply.","The new summand can still be generated in degree zero if one labels the torus boundary by the nontrivial module $X=I(e,K_\\varepsilon)$ and multiplies with $\\mathrm{Ext}(K,X)$ (Proposition 7.14), pointing toward a genuine gluing calculus rather than a defect of the Yoneda product.","By Corollary 5.4, derived block spaces for finite-group doubles are group cohomology groups with coefficients in the character variety, which suggests a derived version of Dijkgraaf-Witten theory in finite characteristic."],"supporting_citations":[{"why":"first part of this work; defines derived block spaces and proves the mapping class group action descends to cohomology, the setting this paper computes","marker":"[LMSS2]"},{"why":"provides the ribbon-normalized integral and alternative R-matrix conventions used for the Drinfel'd double computations","marker":"[SZ]"},{"why":"the coend-based construction of mapping class group actions whose non-semisimple extension is being computed","marker":"[Ly]"},{"why":"defines the Drinfel'd double of a finite group and its standard R-matrix and integral","marker":"[Ka]"},{"why":"supplies the description of Drinfel'd double modules via centralizers and the reduction of Ext over the double to group cohomology","marker":"[W1]"},{"why":"standard group cohomology facts, including the Eckmann-Shapiro lemma and finite-generation inputs","marker":"[E]"},{"why":"gives the periodic projective resolution for cyclic groups and the cohomology ring computations used in characteristic 2 and 3","marker":"[Be]"},{"why":"supplies the Dehn-Lickorish generators and the action of mapping classes on the fundamental group used in the representation-variety dictionary","marker":"[FM]"},{"why":"provides the orbit classification of $\\mathrm{Sp}(2g,\\mathbb{Z}_n)$ on $\\mathbb{Z}_n^{2g}$ used to decompose the genus-one representation spaces","marker":"[FF]"}],"fun_headline_variants":["Derived blocks reveal new SL(2,Z) representations","For S3 in char 3, torus degrees 1 and 2 mod 4","Derived block spaces go beyond ordinary blocks","New mapping class group reps on derived block spaces","SL(2,Z) on derived blocks is not from degree zero"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The whole example chain relies on the dictionary that identifies the algebraically defined mapping class group action with the geometric action on homomorphisms from the fundamental group of the surface to S3; that dictionary involves a chosen basepoint, a diffeomorphism moving the capping point to the basepoint, and the convention on whether the group acts by precomposition or postcomposition.","fun_headline_variants_meta":{"raw":{"variants":["Derived blocks reveal new SL(2,Z) representations","For S3 in char 3, torus degrees 1 and 2 mod 4","Derived block spaces go beyond ordinary blocks","New mapping class group reps on derived block spaces","SL(2,Z) on derived blocks is not from degree zero"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000783,"raw_usage":{"total_tokens":3429,"prompt_tokens":887,"completion_tokens":2542,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":503,"completion_tokens_details":{"reasoning_tokens":2466}},"tokens_in":503,"tokens_out":2542,"duration_ms":16713,"temperature":1.0,"reasoning_tokens":2466,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-12T04:17:48.963639+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Recompute the action of $s^2$ on $Z^1(\\Sigma_1)$ for the Drinfel'd double of $S_3$ in characteristic 3, using the standard R-matrix directly. The paper's Theorem 7.7 predicts that this element acts as $-\\mathrm{id}$, so that its trace equals minus the dimension; finding the identity instead would show the summand $K_{\\chi_2}[\\mathbb{P}^1_3]$ is absent and the non-isomorphism claim fails.","supporting_citations":[],"review_version":1}