{"id":"72ae10a7-07df-4b7c-a115-80e5a3227612","arxiv_id":"1908.08826","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":8.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"If H is an almost normal subgroup of G, both are of type VFP, and vcd(G)=vcd(H)+1, then G is the fundamental group of a finite graph of groups with vertex and edge groups commensurable to H.","lead":"This paper proves a higher-dimensional version of Stallings' theorem: a group containing a codimension-one almost normal subgroup must split into pieces commensurable to that subgroup. A generalist should care because the result gives a new geometric splitting criterion with applications to one-relator, duality, and 3-manifold groups.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The proof of Theorem 1.2 rests on Theorem 4.5, whose key Lemma 4.6 assumes uniform fibre transfer maps and controlled chain homotopies without a complete construction; this under-proved step is the load-bearing risk.","rationale":"I read the paper as aiming to prove Theorem 1.2 by a Stallings-style argument: a coarse Kunneth theorem gives cohomological one-dimensionality of the quotient, end detection gives a splitting, and relative accessibility terminates the process. The least secure premise is Theorem 4.5. Its proof is not machine-checked, and the key Lemma 4.6 is an inductive construction that assumes a uniform family of transfer maps between fibre complexes and then asserts controlled chain homotopies. The reader's verdict already flags Lemma 4.6 and Theorem 5.2; I concur but put the weight specifically on the unproved transfer-map existence inside Lemma 4.6, because Theorem 4.5 is applied directly in Lemma 5.4 to identify H^1_coarse(G/H) with the obstruction to vcd(G) = vcd(H)+1. Theorem 5.2 is imported from the author's preprint and would also need checking, but even if it is correct the Kunneth step must stand. I found no counterexample and no internal contradiction in the main theorem; the concern is a rigorous gap, not a known falsehood. Therefore I keep the reader's CONDITIONAL verdict.","tokens_in":23542,"tokens_out":15369,"duration_ms":154287,"concrete_test":"Attempt to construct the metric complex E_* of Lemma 4.6 for the trivial coarse bundle X = Z x Z over B = Z with fibre Z, using standard uniformly acyclic metric complexes on Z, and verify that the boundary map satisfies d^2 = 0 and that the transfer maps f^{b,b'}_# satisfy the control inequalities (3)-(4) with displacement independent of b-b'. If this simplest nontrivial case cannot be made explicit, the inductive lemma is under-proved; if it works, the concern narrows to the generality of the induction.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central argument depends on the coarse Kunneth theorem, Theorem 4.5, and its proof via Lemma 4.6 is the least secure part of the paper. In the first paragraph of the proof of Lemma 4.6, the author assumes, without construction or citation, that for every pair b,b' in B there is a chain map f^{b,b'}_#: D^b_* -> D^{b'}_* with finite displacement over a closest-point projection, with displacement depending only on d_B(b,b'). The subsequent inductive definition of the boundary map on E_* and the chain homotopies f^b_#, g^b_#, h^b_# all rest on this transfer family and on the control inequalities (3)-(4). If those transfer maps cannot be chosen with uniform control, the metric complex E_* is not well defined and the isomorphism H^*_coarse(G;F) = H^*_coarse(H;F) tensor H^*_coarse(G/H;F) used in Lemma 5.4 collapses. I do not claim Lemma 4.6 is false; it is under-proved at exactly the point where the main theorem's correctness risk is concentrated. The imported Theorem 5.2 is a further dependency, but it is secondary because the Kunneth step is what converts the codimension-one hypothesis into the splitting conclusion.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper proves Theorem 1.2: if H is an almost normal subgroup of G, both are of type VFP, and vcd(G)=vcd(H)+1, then G is the fundamental group of a finite graph of groups in which every vertex and edge group is commensurable to H. The proof passes through a coarse Künneth theorem (Theorem 4.5) for coarse bundles: under coarse uniform acyclicity of fibre and base, the coarse cohomology of the total space is computed as a Künneth-type extension. This is applied to the quotient space G/H, whose coarse one-dimensionality forces more than one end and hence a splitting over a subgroup commensurable to H; a relative accessibility theorem then yields the finite graph-of-groups decomposition. The paper also contains applications to one-relator groups, virtual duality groups, and 3-manifold groups, as well as a Gorenstein cohomological dimension refinement. The main structural risk is concentrated in the proof of Lemma 4.6, which constructs the metric chain complex E_• used in the Künneth theorem.","tokens_in":23797,"tokens_out":9259,"duration_ms":88860,"significance":"If the main theorem and its supporting coarse Künneth theorem are correct, this is a substantial advance: it unifies and extends earlier splitting theorems of Bieri, Kropholler, and Walker under a single hypothesis about cohomological codimension one, and it introduces a new coarse-bundle Künneth technique that is likely to be useful beyond this paper. The applications to one-relator groups and virtual duality groups are significant and non-obvious. The paper is honest about its dependencies: it relies on the author's prior work [Mar18] and the preprint [Mar19] for coarse cohomology, quotient-space geometry, and relative accessibility, rather than restating the main theorem. No fitted parameters or target-restating normalizations appear. The central claim is falsifiable and cleanly stated. However, the significance is conditional on a complete proof of Lemma 4.6, which is currently under-specified at a load-bearing point.","major_comments":[{"comment":"The existence of the transfer chain maps f^{b,b'}_# is assumed without proof. For each pair b,b' in B, the manuscript says: 'let f^{b,b'}_# : D^b_* -> D^{b'}_* be a chain map with finite displacement over f^{b,b'}, where the displacement depends only on d_B(b,b')'. No construction or citation is provided for these maps, and it is not obvious that arbitrary uniformly acyclic metric complexes over the fibres admit such uniformly controlled chain maps over closest-point projections, nor that the composites f^{b,b'}_# f^{b',b}_# can be taken to be chain homotopic to the identity. The boundary map on E_*, the chain maps g^b_#, f^b_#, and the chain homotopies h^b_# and hbar^b_# all rest on this transfer family and on the control inequalities (3)-(4). Since Theorem 4.5, Lemma 5.4, and hence Theorem 1.2 depend on the construction of E_*, this is a load-bearing gap that needs to be filled with a full proof or a precise reference to a statement in [KK05] or [Mar18] that supplies exactly this uniform transfer property.","section":"§4, paragraph before Lemma 4.6 and Lemma 4.6"},{"comment":"Lemma 4.8 as stated appears to contain a typographical error and its proof is incomplete. The statement says 'Then σ is the boundary of a chain...' but the intended conclusion should be that the cycle τ is a boundary; the base case and the inductive step strongly suggest that τ rather than σ is the object to be filled. More importantly, in the inductive step the proof asserts, for each ρ, the existence of ω_ρ with ∂ω_ρ equal to a specified cycle, citing acyclicity of the fibre complexes, but it does not verify that the assembled chain ω satisfies the required control bounds in (3) and (4), nor does it show that the constants R_j and μ_j can be chosen uniformly and independently of b. The subsequent construction of f^b_#, h^b_#, and hbar^b_# is only described as done 'in a similar way', and the assertion that f^{b,b'}_# f^{b',b}_# is chain homotopic to the identity is not justified. The proper chain homotopy between E_* and B_* ⊗ D^b_* is central to the Künneth theorem, so these omissions need to be addressed.","section":"§4, Lemma 4.8"},{"comment":"The proof of Theorem 5.5, and hence of Theorem 1.2, relies essentially on Theorem 5.2, which is quoted as [Mar19, Theorem 3.24] from an unpublished preprint. The relative accessibility result is nontrivial and is not proved in this paper. Since the graph-of-groups decomposition is the main conclusion of Theorem 1.2, this dependency should be made explicit in the introduction and either Theorem 5.2 should be proved in an appendix or the referee should be able to verify it in the companion preprint. As it stands, the correctness of the main theorem is contingent on an external unpublished result, which is a legitimate concern for a journal submission.","section":"§5, Theorem 5.2"}],"minor_comments":[{"comment":"There is a typo in 'We fix a a PID R'; it should be 'a PID R'.","section":"§2, first paragraph"},{"comment":"In the proof of Lemma 4.8, 'there is some some ω_ρ' contains a duplicated 'some'. The statement of the lemma should also be corrected from 'σ is the boundary' to the intended 'τ is the boundary'.","section":"§4, Lemma 4.8"},{"comment":"The word 'Gorentstein' is misspelled; it should be 'Gorenstein'.","section":"§3, before Proposition 3.6"},{"comment":"The sentence 'It follows from Proposition 2.6 that G is a duality group of dimension n over the field R_i' cites the wrong proposition. Proposition 2.6 is a criterion for coarse uniform 0-acyclicity; the intended statement appears to require the universal coefficient theorem for coarse cohomology (Proposition 2.19) together with the duality-group hypotheses. The reference should be corrected.","section":"§7, proof of Theorem 7.2(1)"},{"comment":"The statement of Theorem 5.2 reads 'Let G be an almost finitely presented containing...'; the word 'group' is missing. It should be 'Let G be an almost finitely presented group containing...'.","section":"§5, statement of Theorem 5.2"}],"recommendation":"major_revision","confidential_remarks":"This is a promising and potentially important paper, but it should not be accepted until the proof of Lemma 4.6 is made complete. The under-specified transfer maps and the terse inductive construction are exactly where the correctness of the Künneth theorem, and therefore of Theorem 1.2, is decided. The dependence on the unpublished [Mar19] for the relative accessibility theorem should also be resolved, either by including a proof or by waiting until that preprint is available in refereed form. I do not see a circularity or a fundamental false step; the issues are about missing details in a central technical lemma, which is the appropriate subject of a major revision rather than outright rejection."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Main take: this paper deserves a serious referee. Theorem 1.2, if correct, is a genuine advance: it replaces Bieri's normal-subgroup result and the Kropholler/Walker special cases with one statement for all VFP almost normal subgroups, and it gives real applications to one-relator groups, duality groups, and 3-manifold groups. The reader's conditional verdict is fair, and the stress-test note hits the right spot.\n\nWhat is good: the proof architecture is coherent. Proposition 3.3 reduces to field coefficients, Lemma 5.4 turns codimension one into nontrivial first coarse cohomology of the quotient, and Theorem 5.5 converts that into a graph-of-groups splitting. The reliance on [Mar18] and [Mar19] is not circular: those are separate prior results. The coarse Kunneth theorem is a useful tool in its own right.\n\nWhere it gets soft: Lemma 4.6. The proof assumes, without construction or citation, a family of transfer chain maps over closest-point projections, with displacement depending only on the distance in the base. Everything after that, the boundary operator on E_*, the chain homotopies, and the proper chain homotopy equivalence to the tensor product, rests on those maps. The stress-test note is accurate: if uniform control of the transfer maps fails, E_* is not well-defined and the Kunneth isomorphism in Lemma 5.4 collapses. I do not think the lemma is false; the argument has the shape of a standard uniform acyclicity construction. But it is under-proved at exactly the load-bearing joint. A referee should ask for the full construction or a precise reference.\n\nSecondary issue: Theorem 5.2, the relative accessibility result, is imported from an unpublished preprint. That is a smaller concern, but it should be stated as a dependency or proved in an appendix.\n\nAudience: geometric group theorists working on splittings, commensurated subgroups, or coarse cohomology. I would bring it to reading group. If the Kunneth step gets fixed, I would cite it.\n\nRecommendation: send to peer review. Do not desk reject; it needs a referee who can spend time on Section 4 and verify Lemma 4.6.","headline":"A strong and significant generalization of Stallings' theorem whose main risk is concentrated in the under-proved transfer maps of Lemma 4.6; worth a serious referee.","tokens_in":24352,"tokens_out":3780,"would_cite":true,"duration_ms":39118,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["20J06","20F65","20E06","20J05"],"pacs":[],"model":"deepseek-v4-flash","headline":"A codimension-one almost normal subgroup forces a group to split over a commensurable subgroup.","keywords":["almost normal subgroups","commensurated subgroups","virtual cohomological dimension","graph of groups","coarse cohomology","coarse bundles","relative ends","duality groups"],"falsifier":"Run the paper's Section 4 construction on a concrete coarse bundle, for example G→G/H for a known pair of groups, and check whether $H^*_{coarse}(G)$ equals the Künneth product of $H^*_{coarse}(H)$ and $H^*_{coarse}(G/H)$; any failure of that equality would refute Theorem 4.5, the engine of the main theorem.","tokens_in":23308,"feed_emoji":"🧩","tokens_out":12578,"duration_ms":129983,"temperature":0.7,"pith_summary":"This paper proves a higher-dimensional analogue of the classical theorem that groups of cohomological dimension one are free. The main theorem says: if H is an almost normal subgroup of G, meaning every conjugate of H is commensurable to H, and both groups are of type VFP (each has a finite-index subgroup with a finitely dominated classifying space), then the virtual cohomological dimension of G being exactly one more than that of H forces G to be the fundamental group of a finite graph of groups in which every vertex and edge group is commensurable to H. In less technical terms, a subgroup that is 'nearly normal' and sits inside a group with exactly one additional dimension must be part of the group's splitting structure. The result unifies earlier special cases and yields new structural conclusions for one-relator groups and virtual duality groups.","feed_headline":"Codimension-one almost normal subgroups force a group to split","feed_subtitle":"The theorem covers all finite-type groups, unifying earlier special cases and adding one-relator and duality-group consequences.","key_machinery":"The load-bearing object is the quotient space G/H, equipped with the metric d(gH,kH)=d_Haus(gH,kH); this space is well-defined up to quasi-isometry and carries a coarse fibre bundle structure p:G→G/H with fibre H. The key theorem is a Künneth formula for such coarse bundles: if fibre and base are coarsely uniformly acyclic, the coarse cohomology of the total space is the tensor product of the coarse cohomologies of fibre and base, up to Tor corrections. This formula converts the hypothesis on virtual cohomological dimension into the statement that the base G/H has coarse cohomology only in degrees 0 and 1, with nonzero degree-1 cohomology, meaning the base has more than one end. Ends then feed into a relative splitting theorem and a relative accessibility theorem that together terminate the induction.","core_discovery":"The central claim is Theorem 1.2: under the stated hypotheses, G decomposes as a finite graph of groups whose vertex and edge groups are all commensurable to H. The proof puts a metric on the space of cosets G/H so that G becomes a coarse fibre bundle over G/H, and uses a Künneth theorem for such bundles to show that the numerical hypothesis vcd(G)=vcd(H)+1 makes the base G/H cohomologically one-dimensional and hence multi-ended. The author then follows the classical ends-to-splitting pattern: multi-endedness yields a splitting of G over a subgroup commensurable to H, and a relative accessibility theorem ensures the splitting process terminates, producing the desired graph of groups. The same machinery proves a criterion relating finite index to equality of virtual cohomological dimension, and drives applications to one-relator groups and virtual duality groups.","pith_inferences":["An implicit consequence is that virtual cohomological dimension behaves like a rank function on almost normal VFP subgroups: equal dimension forces commensurability up to finite index. This suggests a commensurability-invariant dimension theory for such subgroups, with potential use in studying lattices and groups acting on products of trees.","The coarse Künneth theorem is likely a general tool, not just a group-theoretic one: any coarse bundle with uniformly acyclic fibre and base should satisfy the same product formula, giving a way to compute large-scale cohomology of other metric spaces and to detect multi-endedness geometrically.","Since the proof only needs a dimension function with monotonicity and top-cohomology detection, analogues of the main theorem should hold for Gorenstein cohomological dimension and for the invariant d_F highlighted in Remark 5.6, even when virtual cohomological dimension is infinite."],"forward_implications":["If H is an almost normal subgroup of type VFP inside a VFP group G with vcd(G)=vcd(H)+1, then G must split nontrivially over a subgroup commensurable to H, even in cases where the classical relative-end number is not greater than one.","A finitely generated non-trivial almost normal subgroup of a free group has finite index in that free group (Corollary 1.4).","For a finitely presented group of virtual cohomological dimension two, an infinite finitely presented almost normal subgroup is either of finite index, or virtually free with the ambient group splitting as a graph of groups over subgroups commensurable to it (Corollary 1.5).","In a one-relator group, an infinite-index finitely presented almost normal subgroup is either infinite cyclic with the group a generalized Baumslag–Solitar group, or commensurable to a free normal subgroup whose quotient is Z or Z2*Z2 (Theorem 6.1).","If G is a virtual duality or virtual Poincaré duality group, any VFP almost normal subgroup is again such a duality group; in codimension one it is commensurable to a normal subgroup with quotient Z or Z2*Z2 (Theorems 7.2 and 7.3)."],"supporting_citations":[{"why":"Supplies the relative accessibility theorem that lets the splitting process terminate at a graph of groups over H-commensurable subgroups.","marker":"[Mar19, Theorem 3.24]"},{"why":"Provides the metric-complex techniques from which the paper's coarse Künneth theorem (Theorem 4.5) is built.","marker":"[KK05, §11.5]"},{"why":"Brown's criterion is used to prove that the quotient space G/H is coarsely uniformly acyclic.","marker":"[Bro87]"},{"why":"One component of the splitting criterion that passes from many ends of G/H to a splitting over a subgroup commensurable to H.","marker":"[DR93]"},{"why":"The complementary component of that splitting criterion, used to connect splittings over H-commensurable subgroups with the number of ends.","marker":"[SS00]"},{"why":"Shows vertex groups of the graph of groups inherit type FP∞(R), needed to run the dimension comparison at each vertex.","marker":"[Bie81, Proposition 2.13]"},{"why":"Introduces the number of relative ends of (G,H) that connects ends of the quotient space with splitting theory.","marker":"[KR89]"},{"why":"Classical accessibility result invoked to ensure the iterated splitting procedure terminates.","marker":"[Dun85]"},{"why":"Defines the coarse cohomology groups and proves their quasi-isometry invariance, which the Künneth theorem computes.","marker":"[Mar18]"}],"fun_headline_variants":["Almost normal codim-one subgroups force a graph-of-groups splitting","Codim-one almost normal subgroups imply group decompositions","If vcd(G)=vcd(H)+1, an almost normal H splits G","Almost normal codim-one subgroups yield a finite graph of groups","Groups with codim-one almost normal subgroups must split"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The proof rests on the claim that the large-scale cohomology of the whole group is the product of the large-scale cohomologies of the subgroup and the coset space; if that product formula fails, the main conclusion does not follow.","fun_headline_variants_meta":{"raw":{"variants":["Almost normal codim-one subgroups force a graph-of-groups splitting","Codim-one almost normal subgroups imply group decompositions","If vcd(G)=vcd(H)+1, an almost normal H splits G","Almost normal codim-one subgroups yield a finite graph of groups","Groups with codim-one almost normal subgroups must split"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000205,"raw_usage":{"total_tokens":1307,"prompt_tokens":776,"completion_tokens":531,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":392,"completion_tokens_details":{"reasoning_tokens":443}},"tokens_in":392,"tokens_out":531,"duration_ms":5918,"temperature":1.0,"reasoning_tokens":443,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T11:28:56.513726+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Run the paper's Section 4 construction on a concrete coarse bundle, for example G→G/H for a known pair of groups, and check whether $H^*_{coarse}(G)$ equals the Künneth product of $H^*_{coarse}(H)$ and $H^*_{coarse}(G/H)$; any failure of that equality would refute Theorem 4.5, the engine of the main theorem.","supporting_citations":[],"review_version":1}