{"id":"6833203e-0784-4e59-adfc-3705be9a1ead","arxiv_id":"2608.01513","paper_version":1,"verdict":"ACCEPT","confidence":"MODERATE","novelty_score":8.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"Every hierarchically hyperbolic group that is not hyperbolic contains a Z^2 subgroup, and every virtually Z^n subgroup lies in an A-invariant uniform quasi-flat whose points are joined by hierarchy paths.","lead":"This paper proves a flat torus theorem and a quasiflat closing theorem for hierarchically hyperbolic groups: any virtually Z^n subgroup sits in an invariant quasi-flat, and every non-hyperbolic such group contains a Z^2 subgroup. It also yields algebraic control over centralisers, normalisers and commensurators of abelian subgroups, with applications to Coxeter groups.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"No significant objection identified","rationale":"The reader's verdict of ACCEPT with moderate confidence seems appropriate. The paper is a substantial research preprint with a detailed, structured proof. The most load-bearing assumptions—hierarchical semisimplicity and the [PS23] classification—are explicitly stated and are satisfied in the primary case of HHGs. I checked the key reductions: the construction of the equivariant median model (Theorem 3.1) is supported by the product-of-trees construction in Section 3; the bounded-orbit reduction (Proposition 4.1) and the invariant subspace analysis (Proposition 5.6) are coherent; and the application of the CAT(0) Flat Torus Theorem in Theorem 6.1 is valid because the median space Q carries a compatible CAT(0) metric with a semisimple proper action. The proof of Theorem 7.14 (flat closing) follows from the highest-abelian-subgroup result and standard rank arguments. I found no internal inconsistency or missing case that would materially affect the central claims. The minor notational issue in Proposition 4.1(6) (where the symbol for 'orthogonal' appears as '&') is cosmetic and does not affect the argument.","tokens_in":72224,"tokens_out":41410,"duration_ms":423451,"concrete_test":"Independently verify the precise statement of [PS23, Thm. 5.1] used in Lemma 5.2: for a finitely generated virtually abelian group A acting properly and hierarchically semisimply on an HHS (X,S), there exist pairwise-orthogonal domains U_1,...,U_n such that (a) A preserves the set {U_i}, and (b) for every W∈S, if π_W(A·x0) is unbounded for some x0, then W⊑U_i for some i. Confirm that this is exactly what [PS23] proves, and that the construction of the boundary points p± from the lineal actions on C_{U_i} yields supp(p±)={U_1,...,U_n}.","verdict_should_be":"UNCHANGED","load_bearing_attack":"After a close reading of the proof of Theorem 6.1 and its supporting machinery (Sections 3–6), I could not identify a clear internal gap or unjustified step that would threaten the central claim. The main theorem is carefully reduced, via Lemma 5.2 and Proposition 5.6, to the equivariant median model (Theorem 3.1) and the CAT(0) Flat Torus Theorem. The proof is long and intricate, but the stated assumptions—especially hierarchical semisimplicity (Definition 5.1)—are used precisely where needed, and the dependence on external results ([PS23, Thm. 5.1], [DHS20, Thm. 3.1], [Dur23], [Bow16]) is explicit. The only mildly under-justified point is Lemma 5.2(4), which asserts an 'if and only if' characterization of unbounded projections; the 'only if' direction is the contrapositive of the cited [PS23, Thm. 5.1] and the 'if' direction follows from the construction of the boundary points p±, so this is not a real gap. The proof of the quasiflat closing theorem (Theorem 7.14) is also coherent, relying on Corollary 7.8 and standard properties of HHS rank. Overall, the central claim appears well-supported, given the explicitly stated and well-motivated assumptions.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper proves a coarse flat torus theorem and a quasiflat closing theorem for hierarchically hyperbolic spaces and groups. For a virtually abelian group A acting properly and hierarchically semisimply by HHS automorphisms on an HHS, Theorem 6.1 constructs an A-invariant EHQC subspace quasi-isometric to Euclidean space, together with a detailed product structure for a coarse minset. For HHGs, Theorem 7.14 concludes that an HHG of rank ν contains Z^ν, and Theorem 7.13 gives the special case that a non-hyperbolic HHG contains Z^2. The paper also derives applications to ascending chain conditions, centralisers, normalisers, commensurators, virtually solvable subgroups, and HHG structures on Coxeter groups. The proof strategy is to pass to an equivariant median model (Theorem 3.1), apply the CAT(0) flat torus theorem, and combine this with bounded-orbit results for injective hulls.","tokens_in":72525,"tokens_out":20617,"duration_ms":204229,"significance":"If the main results hold, this is a substantial contribution. The quasiflat closing theorem gives a positive answer in the HHS setting to a phenomenon that fails for general CAT(0) spaces, and the coarse flat torus theorem answers questions raised in recent work. The applications are broad: an ascending chain condition for virtually abelian subgroups, hierarchical quasiconvexity of highest abelian subgroups, control of centralisers and commensurators, and obstructions to HHG structures on Coxeter groups via hyperoctahedral point groups. The paper is careful about constants: the appendix explicitly tracks how constants depend on HHS parameters, and the main proofs are structured so that the use of external results such as [PS23, Thm. 5.1], [DHS20, Thm. 3.1], and [Dur23] is transparent. The equivariant median model and the reduction to the CAT(0) flat torus theorem are natural and well-motivated.","major_comments":[{"comment":"Lemma 6.3 is false as stated. For n=1, the map f:R→R given by f(x)=|x|+1 is (1,1)-coarsely Lipschitz, and for every r≥0 every preimage f^{-1}(N_r(x)) has diameter at most 2r, but f is not coarsely surjective. The proof's Borsuk-Ulam step cannot establish surjectivity: it shows only that a non-surjective proper map identifies an antipodal pair, and in this example f(1)=f(-1)=2, which is not a contradiction. This lemma is used in the proof of Theorem 6.1.(5) and again in Corollary 7.3 to derive coarse surjectivity of certain maps between A-invariant quasi-flats. In the Theorem 6.1.(5) application the maps are quasi-isometric embeddings E^n→E^n, for which coarse surjectivity is true and should be proved directly. In Corollary 7.3 the map π_n additionally preserves A-orbit distances, so a similar repair is possible. As written, the proof of the main theorem relies on a false statement.","section":"Section 6, Lemma 6.3"}],"minor_comments":[{"comment":"The phrase 'The later is a consequence' should read 'The latter is a consequence'.","section":"Abstract"},{"comment":"The heading contains the typo 'Throguhout'; it should be 'Throughout'.","section":"Appendix A heading"},{"comment":"In the paragraph defining M_A as a retract, 'coarse lipshcitz retract' should be 'coarse Lipschitz retract'.","section":"Proof of Theorem 6.1"}],"recommendation":"major_revision","confidential_remarks":"The paper is well-structured and the main theorems are likely correct, but Lemma 6.3 is false and is used in load-bearing arguments. I expect the authors can repair the proof by proving the needed coarse-surjectivity statement for the specific quasi-isometric embeddings and orbit-preserving maps that actually arise, rather than for general coarse Lipschitz maps. Because the issue is localized but substantive, major revision is appropriate."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"This is a substantial paper, and the main results are real. The quasiflat closing theorem for HHGs (rank nu implies Z^nu subgroup) and the coarse flat torus theorem answer open questions and give a clean geometric obstruction to HHG structures. The authors introduce existential hierarchical quasiconvexity (EHQC), which is the right weakening of hierarchical quasiconvexity for this problem: it lets them build invariant quasiflats that are not median subalgebras, which is necessary by their own Figure 1 example. The equivariant median model in Theorem 3.1 is a genuine technical advance, not a routine application of Durham's theorem; the equivariance and the product-of-quasilines structure in Proposition 3.2 are new.\n\nThe paper is well organized. The main proof is long but the structure is clear: reduce via Proposition 5.6 to a product situation, apply the median model, then use the CAT(0) flat torus theorem. The appendix explicitly tracks constants, and Remark 1.2 and Section 5 flag the hierarchical semisimplicity assumption and why it is necessary. The applications are credible: ascending chain condition, hierarchical quasiconvexity of highest abelian subgroups, normaliser/centraliser control, and the Coxeter group obstructions. The new proof that virtually solvable subgroups are virtually abelian, avoiding Gromov's polynomial growth theorem, is a nice bonus.\n\nSoft spots are mostly about verification burden. The proof leans heavily on external results: [PS23, Thm 5.1], [Dur23, Prop 4.3 and 4.14], and [DHS20, Thm 3.1]. I did not machine-check anything, and neither did the stress-test. That said, the dependencies are explicit, and the stress-test found no internal gap. The weakest assumption—hierarchical semisimplicity—is clearly stated and satisfied for HHGs by [DHS20], so it is not a hidden restriction. The non-uniform constants in Remark 3.3 are a minor annoyance but do not affect the main theorem. One small caveat: Proposition 3.2 and parts of Section 6 are dense enough that a referee will need real time; this is not a paper to skim.\n\nBottom line: this deserves a serious referee. I would send it out. It is likely to become a standard reference for HHG rank and abelian subgroups.","headline":"Major new theorems on quasiflats in HHGs, with a long but carefully structured proof and no load-bearing flaw found.","tokens_in":72965,"tokens_out":1949,"would_cite":true,"duration_ms":20940,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["20F65","20F67","20F55"],"pacs":[],"model":"deepseek-v4-flash","headline":"A quasiflat closing theorem holds for hierarchically hyperbolic groups: rank $\\nu$ forces a $\\mathbb{Z}^\\nu$ subgroup.","keywords":["hierarchically hyperbolic groups","quasiflat closing","coarse flat torus theorem","existentially hierarchically quasiconvex subspaces","virtually abelian subgroups","hierarchical quasiconvexity","rank of HHGs","Coxeter groups"],"falsifier":"Exhibit a hierarchically hyperbolic group of rank 2 that contains no $\\mathbb{Z}^2$ subgroup; Theorem 7.14 predicts none exists. Equivalently, construct a virtually $\\mathbb{Z}^2$ group acting properly and hierarchically semisimply on a hierarchically hyperbolic space with unbounded orbits but no $A$-invariant EHQC subspace quasi-isometric to $E^2$.","tokens_in":72134,"feed_emoji":"📐","tokens_out":7152,"duration_ms":69499,"temperature":0.7,"pith_summary":"This paper establishes a quasiflat closing theorem and a coarse flat torus theorem for hierarchically hyperbolic groups (HHGs). It proves that a non-hyperbolic HHG always contains a $\\mathbb{Z}^2$ subgroup, and more generally that an HHG of rank $\\nu$ contains $\\mathbb{Z}^\\nu$. For every virtually abelian subgroup $A$ acting properly and hierarchically semisimply on a hierarchically hyperbolic space, it constructs an $A$-invariant quasiflat $F$, quasi-isometric to Euclidean $n$-space, such that any two points of $F$ are joined by a uniform-quality hierarchy path staying close to $F$. This coarse flat torus theorem is the engine behind a cluster of structural consequences: an ascending chain condition for virtually abelian subgroups, hierarchical quasiconvexity of highest abelian subgroups, control over normalisers, centralisers and commensurators, and a new proof that virtually solvable subgroups of HHGs are virtually abelian. A careful reader should care because the paper converts coarse geometric information into finitary algebraic information for a very broad class of groups.","feed_headline":"Rank-2 HHGs always contain a Z^2 subgroup","feed_subtitle":"A coarse flat torus theorem places every virtually abelian subgroup on an equivariant Euclidean quasiflat.","key_machinery":"The central object is the coarse minset: a canonical $A$-invariant EHQC subspace assembled from all the $A$-invariant quasiflats produced by the flat torus argument. To build it, the paper constructs an equivariant median model $Q$ for the hull of a pair of boundary points fixed by $A$: one takes the quasi-linear domains whose projections are unbounded on $A$-orbits, forms their associated trees or lines $T_U$, and then passes to the subspace of consistent tuples, obtaining a complete median space on which $A$ acts by isometries. A standard theorem supplies a compatible CAT(0) metric on this median space, and the classical CAT(0) flat torus theorem extracts the Euclidean factor. The other lo","core_discovery":"The paper's central claim is Theorem 6.1: if $A$ is a virtually $\\mathbb{Z}^n$ group acting properly and hierarchically semisimply by HHS automorphisms on a hierarchically hyperbolic space $(X,\\mathcal{S})$, then there is an $A$-equivariant quasi-isometry from an $A$-invariant 'existentially hierarchically quasiconvex' (EHQC) subspace $F\\subseteq X$ to Euclidean space $E^n$, with $A$ acting properly and cocompactly on $E^n$. The quasiflat $F$ is EHQC in the weak sense that any two of its points are joined by a uniform hierarchy path lying close to $F$. From this the authors derive the coarse flat torus theorem for HHGs and the quasiflat closing theorem: an HHG of rank $\\nu$ contains $\\mathbb","pith_inferences":["If the quasiflat closing theorem is correct, then the presence of $\\mathbb{Z}^2$ is not only sufficient but necessary for non-hyperbolicity in every HHG, giving a cheap algebraic obstruction for candidate HHG structures before any cubulation or hierarchical geometry is constructed.","The EHQC notion is strictly weaker than median convexity, and the paper's flats cannot in general be median subalgebras, as the cube-tiling example illustrates; any attempt to strengthen the conclusion to a median-convex flat will need extra hypotheses.","The explicit coarse minset decomposition $Y\\times C\\times E^n$ suggests that virtually abelian subgroups of HHGs have controlled 'elliptic parts' whose geometry is trivial from the viewpoint of the flat; this could be a route to finer quasi-isometric invariants of HHGs.","The proof's reliance on a compatible CAT(0) metric on a median model indicates that any HHS satisfying the semisimplicity hypothesis inherits a Euclidean factor wherever an unbounded virtually abelian subgroup is present, so the phenomenon likely extends to other coarse median settings."],"forward_implications":["Virtually abelian subgroups of HHGs satisfy the ascending chain condition, and every virtually abelian subgroup is virtually contained in a highest virtually abelian subgroup.","Virtually solvable subgroups of HHGs are virtually abelian, by a proof that avoids Gromov's polynomial growth theorem.","Highest virtually abelian subgroups of HHGs are hierarchically quasiconvex, and a suitable finite-index abelian subgroup has hierarchically quasiconvex centraliser.","The normaliser of a virtually abelian subgroup is EHQC and is coarsely described by the coarse minset; each finitely generated subgroup of its commensurator lies in the normaliser of a finite-index abelian subgroup.","An HHG of rank $\\nu$ contains $\\mathbb{Z}^\\nu$, and Coxeter groups containing 'poison' affine subgroups admit no HHG structure."],"supporting_citations":[{"why":"Classifies virtually abelian actions by HHS automorphisms and provides the invariant pairwise-orthogonal domains $U_i$ whose projections are unbounded on $A$-orbits, without which the proof cannot start.","marker":"[PS23, Thm. 5.1]"},{"why":"Supplies hierarchical semisimplicity of HHG actions, so the main theorem applies to all subgroups of HHGs.","marker":"[DHS20, Thm. 3.1]"},{"why":"The CAT(0) flat torus theorem used to extract the Euclidean factor $E^n$ inside the median model.","marker":"[BH99, Cor. II.7.2]"},{"why":"The distance formula, used throughout to pass between projections to hyperbolic domains and metric estimates in the HHS.","marker":"[BHS19, Thm. 4.5]"},{"why":"Cubical approximation theorem, which underpins the construction of the median model and its uniform constants.","marker":"[BHS21, Thm. 2.1]"},{"why":"Cubical model theorem whose strategy is adapted to obtain an equivariant median model of hulls of boundary points.","marker":"[Dur23]"},{"why":"Coarse injectivity and injective hull, used to find a point with bounded $A$-orbit in the quotient part of the decomposition.","marker":"[HHP23, Theorem A]"},{"why":"Provides the compatible CAT(0) metric on a complete connected median space, turning the median quasiflat into a genuine CAT(0) flat.","marker":"[Bow16, Thm. 1.1]"},{"why":"Controls diameters of projections of orthogonal domains, used in the consistency arguments for the median model.","marker":"[DHS17, Lemma 1.5]"},{"why":"Gives the equivariant cone-off construction with uniform constants, used to reduce to the case where $A$ has bounded orbits on the transverse part.","marker":"[CRHK24, Prop. 19.1]"}],"fun_headline_variants":["Virtually abelian subgroups of HHGs lie on Euclidean quasiflats","HHGs hyperbolic iff no Z^2 subgroups","Coarse flat torus theorem for hierarchically hyperbolic groups","Every virtually abelian subgroup of an HHG has a quasiflat"],"cache_read_input_tokens":2816,"weakest_assumption_plain":"The central premise is that any element which moves points infinitely far in a hyperbolic domain it preserves acts by an honest translation rather than a parabolic motion; for HHGs this is automatic, but without it the coarse flat can fail to exist.","fun_headline_variants_meta":{"raw":{"variants":["Virtually abelian subgroups of HHGs lie on Euclidean quasiflats","HHGs hyperbolic iff no Z^2 subgroups","Coarse flat torus theorem for hierarchically hyperbolic groups","Every virtually abelian subgroup of an HHG has a quasiflat"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000453,"raw_usage":{"total_tokens":2134,"prompt_tokens":780,"completion_tokens":1354,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":524,"completion_tokens_details":{"reasoning_tokens":1294}},"tokens_in":524,"tokens_out":1354,"duration_ms":11293,"temperature":1.0,"reasoning_tokens":1294,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-06T00:03:23.524992+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Exhibit a hierarchically hyperbolic group of rank 2 that contains no $\\mathbb{Z}^2$ subgroup; Theorem 7.14 predicts none exists. Equivalently, construct a virtually $\\mathbb{Z}^2$ group acting properly and hierarchically semisimply on a hierarchically hyperbolic space with unbounded orbits but no $A$-invariant EHQC subspace quasi-isometric to $E^2$.","supporting_citations":[],"review_version":1}