{"id":"a2ee242d-6e41-4881-97a0-8707d37308fa","arxiv_id":"1908.01290","paper_version":6,"verdict":"ACCEPT","confidence":"HIGH","novelty_score":7.0,"correctness_risk":"low","formal_verification":"none","parameter_count":0,"one_line_summary":"A new spectral sequence and an algebraic excision theorem control the cohomology of Dehn fillings of groups with hyperbolically embedded subgroups, with applications to Poincaré duality, simplicial volume, and acylindrically hyperbolic quotients.","lead":"This paper derives a spectral sequence that computes the cohomology of group theoretic Dehn filling quotients, and an algebraic excision theorem that preserves relative cohomology. These tools generalize results on Poincaré duality and simplicial volume, and yield new acylindrically hyperbolic quotients with prescribed cohomological finiteness properties.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"No significant objection identified: the main load-bearing input is the external Cohen-Lyndon theorem [Sun20, Thm 5.1], but it is published, explicitly cited, and applied with the correct deepness and coefficient hypotheses.","rationale":"The reader's ACCEPT verdict is appropriate. The paper proves substantial new results conditional on the companion paper's Cohen-Lyndon theorem. I checked the main structural steps: the Cohen-Lyndon decomposition yields the cohomological decomposition, the spectral sequence, and the excision isomorphism; Theorem A's module hypothesis is correctly stated; and the applications follow from the stated results with no visible circularity. The only caveat is the external dependency, which is standard and explicitly cited. I therefore recommend no change to the verdict.","tokens_in":38060,"tokens_out":39888,"duration_ms":378061,"concrete_test":"Test the Cohen-Lyndon input in the minimal nontrivial case: G=F_2=<a,b>, H=<a>, N=<a^k>. Compute the normal closure <<N>> in F_2 as a free product of conjugates over a transversal of H<<N>>, and verify Proposition 4.1 for q=1 with trivial coefficients by checking H^1(<<N>>;Z) ≅ CoInd_H^G H^1(N;Z). In addition, re-read [Sun20, Theorem 5.1] to confirm that its deepness condition is the same as the finite-set avoidance used here. If the example matches and the hypotheses align, the central dependency is sound.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The single most load-bearing place in the argument is the Cohen-Lyndon theorem, quoted as Theorem 3.22 from [Sun20, Theorem 5.1]. Proposition 4.1, the spectral sequence in Theorem 4.2, the excision isomorphism (Proposition 4.3/Remark 4.4), and the applications in Theorems B–E are built directly on the free-product decomposition xN y = ∗ tN_λ t^{-1}. If that decomposition failed, Theorem A(i) would not follow. I do not find a defect in the way the paper uses this input: the paper's hypotheses give a hyperbolically embedded subgroup, so the relative metric is locally finite, and Remark 3.18 correctly converts 'sufficiently deep' to the finite-set formulation used in [Sun20]; moreover, the coefficient-module caveat (N acts trivially on A) is stated in Theorem A and is exactly what is needed for the q=1 row. The present paper does not reprove Theorem 3.22, but [Sun20] is an external published proof; this is a dependency, not an internal gap. No other load-bearing assumption appears under-supported.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The manuscript develops a cohomological framework for group-theoretic Dehn fillings. The main result, Theorem A, gives, for sufficiently deep normal subgroups N of a hyperbolically embedded subgroup H in G, a spectral sequence computing H^*(Gbar; A) from H^*(Hbar; H^*(N; A)) and H^*(G; A), an excision-type isomorphism between the relative cohomology of (G,H) and (Gbar,Hbar), a direct-sum decomposition of H^n(Gbar; A) for large n, a cohomological dimension bound, and an FP_n criterion. The paper then applies this framework to prove preservation of Poincaré duality under Dehn fillings (Theorem B), simplicial volume bounds (Theorem C), and theorems on common quotients of acylindrically hyperbolic groups with controlled cohomology (Theorems D and E), followed by several striking applications in Section 8. The proofs are built on the Cohen-Lyndon property for sufficiently deep Dehn fillings, quoted as Theorem 3.22 from Sun's companion paper [Sun20].","tokens_in":38280,"tokens_out":22560,"duration_ms":225429,"significance":"If the results hold, this is a substantial contribution to the cohomological study of Dehn fillings. Theorem A provides a genuine spectral sequence tool for Dehn fillings, and the excision isomorphism is an algebraic analog of topological excision that is used to obtain new results on Poincaré duality pairs, simplicial volume, and cohomological finiteness of acylindrically hyperbolic quotients. The paper is careful and detailed: the spectral sequence arguments in Section 4 are explicit, and Example 4.9 shows that the dimension threshold in Theorem A(iii) is sharp. The main external input, the Cohen-Lyndon property, is cited from a published companion paper and is applied with the correct deepness and coefficient hypotheses; I see no circularity or hidden parameter-fitting. The presentation is generally clear, and the applications are well motivated.","major_comments":[],"minor_comments":[{"comment":"The parenthetical in the statement of Proposition 4.1 is ambiguous: it reads \"for any G-module A and q > 1 (also for any G-module A and q > 0)\", but the proof establishes q = 1 only under the additional hypothesis that <N> = <langle>N<rangle> acts trivially on A. This should be reformulated, for example as \"for q > 1, and also for q > 0 when A is trivial as a <N>-module\".","section":"Section 4, Proposition 4.1"},{"comment":"The displayed inequality in the proof of Lemma 7.6, \"2n * 50D <= 3nD\", does not match the counting: there are 2l components, each of p_ell-length at least 50D, bounding a 3l-gon. The printed inequality appears to contain a typographical error and should be something like \"2l * 50D <= 3lD\"; the contradiction is unaffected.","section":"Section 7.2, Lemma 7.6"},{"comment":"In the proof of statement (v), the phrase \"in which case (iii) is a void statement\" should refer to statement (v): when the original group G has torsion, cd(G) = infinity, so the desired inequality cd(Gbar) <= max{cd(G), cd(C)} is vacuous. Statement (iii) concerns torsion-freeness and is not the statement being justified at that point.","section":"Section 7.1, proof of Theorem D"},{"comment":"In the proof of Theorem 5.2, the vanishing H^1(G; ZG) = 0 is used without explanation. It follows from the fact that (G,H) is a PD(n)-pair with n >= 3, which forces G to be a duality group of dimension n-1; adding a sentence with this justification would improve clarity.","section":"Section 5, Theorem 5.2"},{"comment":"In the remark after Corollary 8.5, \"the quotients {G_k}_{k>=1}\" should be \"{G_k}_{k>=2}\", consistent with the statement of the corollary.","section":"Section 8.3, Corollary 8.5"}],"recommendation":"minor_revision","confidential_remarks":"This is a strong paper and the central arguments are sound. The dependency on the Cohen-Lyndon theorem from [Sun20] is explicit and published, and the paper uses it under the correct hypotheses. The only issues I found are local presentation and typographical problems, which can be fixed without changing the mathematics."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"I agree with the reader's verdict: this is a strong paper that deserves a serious referee. The genuinely new content is the spectral sequence in Theorem A(i) and the relative cohomology excision in Theorem A(ii), both built on the Cohen-Lyndon property from the companion paper [Sun20]. The authors are upfront about that dependency, and it is the right division of labor: [Sun20] is published, the statement they quote is explicit, and they correctly convert 'sufficiently deep' to the finite-set formulation (Remark 3.18) so the hypotheses line up. The coefficient-module caveat in Theorem A, that N acts trivially on A, is stated and is exactly what the spectral sequence needs for the q=1 row.\n\nWhat the paper does well beyond Theorem A is the range of applications. Theorem B preserves Poincare duality under Dehn fillings, and Theorem C gets simplicial volume bounds, generalizing Fujiwara-Manning. Theorems D and E add cohomological control to earlier SQ-universality and common quotient results, and Section 8 shows the machinery has teeth: quotients with property (T) and prescribed FP_n gaps, and a negative answer to an actual Betti number question. Example 4.9 gives a clean sharpness example for the dimension bound in Theorem A(iii). The proof structure is clear: Proposition 4.1 and the spectral sequence comparison in Theorem 4.2 do the heavy lifting, and the later sections are honest applications.\n\nSoft spots: the biggest one is the dependence on [Sun20, Theorem 5.1]. If that theorem had a gap, this paper would inherit it. That is a dependency, not a defect, and the stress-test confirms the usage is correct. The paper is long and dense, especially Sections 7.2 and 7.3, but the technical work there is aimed at a real purpose. I did not find any place where the argument hides a step or overclaims. The citation pattern looks fair: the relevant prior work by DGO, Hull, Wang, and Fujiwara-Manning is credited, and the companion paper is clearly flagged.\n\nWho is this for? Geometric group theorists and people who work on cohomological finiteness properties. A reader comfortable with spectral sequences and relatively hyperbolic groups will get a lot out of it. I would send it to peer review without hesitation, and I would not be surprised if it lands in a good journal. If I were a referee I would check the [Sun20] dependency and the proof of Theorem 5.2, but I see no reason to expect a problem there.","headline":"Strong, careful paper with a genuinely new spectral sequence for Dehn filling cohomology; the main load-bearing input is an external published theorem, and that dependency is handled cleanly.","tokens_in":715,"tokens_out":1870,"would_cite":true,"duration_ms":44918,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["20F67","20F10","20E06"],"pacs":[],"model":"deepseek-v4-flash","headline":"The paper proves that for sufficiently deep Dehn fillings of a hyperbolically embedded subgroup, the relative cohomology of the group pair is isomorphic to that of the filled pair, and a spectral sequence computes the absolute cohomology…","keywords":["group cohomology","Dehn filling","hyperbolically embedded subgroup","Cohen-Lyndon property","relative cohomology","Poincaré duality pair","simplicial volume","acylindrically hyperbolic group"],"falsifier":"One could look for a hyperbolically embedded subgroup $H$ in a group $G$ and a sufficiently deep normal subgroup $N$ of $H$ such that the normal closure of $N$ in $G$ is not a free product of conjugates of $N$; if such an example exists, the main spectral sequence and its consequences would collapse.","tokens_in":1573,"feed_emoji":"🧪","tokens_out":7664,"duration_ms":137667,"temperature":0.7,"pith_summary":"The paper proves that for a group $G$ with a hyperbolically embedded subgroup $H$, when $N$ is a sufficiently deep normal subgroup of $H$, the cohomology of the Dehn filling quotient $G/\\langle\\langle N\\rangle\\rangle$ is controlled by a spectral sequence built from the cohomology of $H$ and $N$. It also establishes an algebraic form of excision: the relative cohomology of $(G,H)$ is isomorphic to that of the filled pair $(\\overline{G}, \\overline{H})$. This yields explicit bounds on cohomological dimension, preservation of finiteness properties $FP_n$, and applications to Poincaré duality groups and simplicial volume.","feed_headline":"Dehn fillings preserve relative cohomology","feed_subtitle":"A spectral sequence computes the cohomology of the filled group, with implications for Poincaré duality and simplicial volume.","key_machinery":"The key object is the Cohen–Lyndon triple: a group triple $(G,\\{H_\\lambda\\},\\{N_\\lambda\\})$ for which the normal closure of the $N_\\lambda$'s is a free product of conjugates $t N_\\lambda t^{-1}$ over a transversal. This structural decomposition, quoted from the companion paper, permits an identification of $H^q(\\langle\\langle N\\rangle\\rangle;A)$ with a co-induced product, making the spectral sequence of Theorem 4.2 computable. That spectral sequence, together with its morphism to the Lyndon–Hochschild–Serre spectral sequence, is the engine that yields the cohomological decomposition and the excision isomorphism.","core_discovery":"The central discovery is that the Cohen–Lyndon structure of the normal closure $\\langle\\langle N\\rangle\\rangle$ allows one to identify the cohomology of $\\langle\\langle N\\rangle\\rangle$ with a direct sum of co-induced modules, leading to a morphism of Lyndon–Hochschild–Serre spectral sequences that becomes an isomorphism on the $E^2$ page for $q>0$. Consequently, for all sufficiently deep $N$, the relative cohomology $H^n(G,H;A)$ is naturally isomorphic to $H^n(\\overline{G},\\overline{H};A)$ for every $G$-module $A$. This algebraic excision underlies the paper's applications to Poincaré duality pairs and to quotients of acylindrically hyperbolic groups.","pith_inferences":["If the algebraic excision holds, it suggests that Dehn fillings behave like topological surgery in cohomology, so one might expect analogous computations for other cohomology theories such as bounded cohomology or $L^2$ cohomology.","The spectral sequence might be extendable to compute other invariants under Dehn filling, such as the growth of cohomology with twisted coefficients, which would further clarify the structure of Dehn-filled groups.","The reliance on the Cohen–Lyndon property points to a potential program: proving analogous decomposition theorems for wider classes of fillings could extend these cohomological results to graphs of groups or other relatively hyperbolic settings."],"forward_implications":["If the main theorem holds, then for sufficiently deep Dehn fillings the cohomological dimension of the quotient is bounded by the maximum of the original group's dimension, the peripheral dimension plus one, and the filled peripheral dimension.","The algebraic excision implies that any property detected by relative cohomology is invariant under sufficiently deep Dehn fillings, so in particular whether a pair is a Poincaré duality pair is preserved under appropriate fillings.","Theorem B shows that when the original pair is a Poincaré duality pair and the filled peripheral subgroups are Poincaré duality groups, the filled quotient is a Poincaré duality group.","Theorem C gives that the simplicial volume of the filled quotient is bounded above by the relative simplicial volume, and positivity follows when the group is hyperbolic relative to the filled peripheral subgroups.","Theorems D and E provide acylindrically hyperbolic quotients with prescribed cohomology and finiteness properties, including quotients with Property (T) and specific $FP_n$ thresholds."],"supporting_citations":[{"why":"Supplies the Cohen–Lyndon property for sufficiently deep Dehn fillings, the structural free-product decomposition that the spectral sequence argument depends on.","marker":"[Sun20]"},{"why":"Defines hyperbolically embedded subgroups and provides the basic Dehn filling theorem and related tools used throughout the paper.","marker":"[DGO17]"},{"why":"Establishes that sufficiently deep Dehn fillings of relatively hyperbolic groups preserve hyperbolicity and relative hyperbolicity, used in the simplicial volume result.","marker":"[Osi07]"},{"why":"Provides the framework for relative cohomology of group pairs and the Poincaré duality characterization used in Theorem B.","marker":"[BE78]"},{"why":"Gives the simplicial volume results for hyperbolic manifolds that the paper generalizes to Poincaré duality pairs.","marker":"[FM11]"}],"fun_headline_variants":["Deep Dehn fillings preserve relative cohomology","Cohomology of Dehn fillings via spectral sequence","Excision theorem for cohomology of Dehn fillings","Relative cohomology invariant under deep Dehn fillings","New spectral sequence computes Dehn fillings cohomology"],"cache_read_input_tokens":40960,"weakest_assumption_plain":"The entire argument rests on the Cohen–Lyndon property for sufficiently deep Dehn fillings, a theorem imported from a companion paper; if that structural fact failed, the spectral sequence and all later results would not follow.","fun_headline_variants_meta":{"raw":{"variants":["Deep Dehn fillings preserve relative cohomology","Cohomology of Dehn fillings via spectral sequence","Excision theorem for cohomology of Dehn fillings","Relative cohomology invariant under deep Dehn fillings","New spectral sequence computes Dehn fillings cohomology"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000734,"raw_usage":{"total_tokens":3253,"prompt_tokens":888,"completion_tokens":2365,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":504,"completion_tokens_details":{"reasoning_tokens":2283}},"tokens_in":504,"tokens_out":2365,"duration_ms":18348,"temperature":1.0,"reasoning_tokens":2283,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T15:16:24.134893+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"One could look for a hyperbolically embedded subgroup $H$ in a group $G$ and a sufficiently deep normal subgroup $N$ of $H$ such that the normal closure of $N$ in $G$ is not a free product of conjugates of $N$; if such an example exists, the main spectral sequence and its consequences would collapse.","supporting_citations":[],"review_version":1}