{"id":"26a3a1ad-7df3-44ab-86f0-bf531ce11b4e","arxiv_id":"2506.20161","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"Every nonelementary hyperbolic group has a weakly malnormal, quasiconvex subgroup of the form F_n × A with A finite, extending Kapovich's theorem.","lead":"Every nonelementary hyperbolic group contains a weakly malnormal, quasiconvex subgroup that is a free group of rank at least two times a finite group. This extends Ilya Kapovich's malnormal subgroup theorem from torsion-free hyperbolic groups to all hyperbolic groups, and yields a non-quasiconvex embedding theorem.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Lemma 4.16's omitted Case 2 is the load-bearing gap: if the claimed interior property fails for any w and ray, Proposition 4.17 and hence Theorem 5.2 collapse, but the proof defers that case.","rationale":"After reading the proof carefully, the dependency chain is: Theorem 5.2 Step 2 uses Proposition 4.17; Proposition 5.1 Step 2 also uses Proposition 4.17; Proposition 4.17 relies on Lemma 4.16. The explicit gap in Lemma 4.16 is therefore the single most load-bearing unproved step in the paper. The issue is not an internal inconsistency with established results; the construction appears standard and likely repairable, but the manuscript itself does not supply the omitted argument, and the main theorem is unsupported if the omitted case is false or requires extra hypotheses. The separate normality assertion in Proposition 5.3 is also genuinely unsupported by Proposition 5.1, but the final theorem's Step 2 does not obviously use normality of F in Comm_G(F), and replacing F by its normal core in Comm_G(F) would repair that statement, so I rank it secondary. The reader's weakest_assumption identifies the same primary concern, and the recommended verdict remains conditional acceptance pending a written proof of the omitted Case 2 and a correction to Proposition 5.3.","tokens_in":17014,"tokens_out":20148,"duration_ms":217477,"concrete_test":"Work in F2 = <x1,x2> and set w = x1 x2 (Case 2). For the ray α with α(n0+1) = x1, define g_n = x1^n x2 x1^{-n} x2^{-1} ∈ [F2,F2]. Check that w g_n is reduced and cyclically reduced and that [1, w g_n] contains [1, w x1^n], so (w g_n)^∞ → α(∞) as n → ∞. Repeat this construction for the remaining Case-2 rays, including w = x1 x2^{-1} and rays turning to x2. If every Case-2 ray admits such a sequence, write out the uniform construction and Lemma 4.16 is fillable; if any ray does not, Lemma 4.16 is false and Proposition 4.17 must be re-examined.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central construction ultimately rests on Proposition 4.17, which supplies a free rank-2 subgroup avoiding a finite list of infinite-index subgroups. Proposition 4.17 needs two elements g1, g2 in A = {wg : g ∈ [G,G], wg reduced and cyclically reduced} whose endpoints avoid the limit sets of all conjugates of the E_i, and this uses Lemma 4.16: the closure of {g^∞ : g ∈ A} has nonempty interior. The proof of Lemma 4.16 handles only Case 1, where w begins and ends with powers of the same basis letter; Case 2, where w begins and ends with powers of different basis letters, is dismissed as 'a simple exercise for the reader'. Every geodesic ray in V_w through w must be approximated by endpoints (w g_n)^∞ with g_n ∈ [G,G] and w g_n cyclically reduced. If one Case-2 configuration cannot be handled, the interior conclusion may fail, and Proposition 4.17 need not produce the ping-pong pair; Proposition 5.1 Step 2 and Theorem 5.2 Step 2 both depend on it. The case is plausibly true — for example, in F2 with w = x1 x2 and a ray continuing by x1, g_n = x1^n x2 x1^{-n} x2^{-1} works — but the written proof gives no argument and no reference, so the dependency is not secured. Separately, Proposition 5.3 asserts 'Clearly' that F = F ∩ F3 is normal in Comm_G(F); this does not follow from Proposition 5.1 as stated. That second gap is real but probably repairable, for instance by replacing F with its normal core in Comm_G(F), and it is less central to Theorem 5.2 than Lemma 4.16.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper claims a generalization of Ilya Kapovich's theorem: in every nonelementary hyperbolic group G, every nonelementary quasiconvex subgroup H contains a free subgroup H_1 of rank at least 2 such that Comm_G(H_1) is of the form H_1 × A with A finite, and H_1A is weakly malnormal and quasiconvex in G. The proof proceeds by combining tools from hyperbolic geometry (coned-off graphs, limit sets, ping-pong) with an algebraic analysis of virtually free groups, and then derives an application to non-quasiconvex embeddings via the Bestvina–Feighn combination theorem. The strategy is coherent and follows the broad outline of Kapovich's work, but several load-bearing steps are underproved as written.","tokens_in":17477,"tokens_out":21972,"duration_ms":224930,"significance":"If the main theorem is correct, it is a natural and useful extension of Kapovich's malnormal-subgroup theorem to hyperbolic groups with torsion, and it feeds into a clean non-quasiconvexity embedding result. The paper is largely self-contained against standard prior results (Kapovich, GMRS, Bestvina–Feighn, Stallings, Baumslag–Taylor) and the overall approach is credible. The main gaps are technical rather than architectural: the authors leave a key boundary-approximation case as an exercise, and one normality assertion is made without proof. I regard the result as likely true, but the manuscript as submitted does not yet fully secure the central claims.","major_comments":[{"comment":"The proof of Lemma 4.16 explicitly leaves Case 2, where w starts and ends with powers of two different basis elements, as 'a simple exercise for the reader.' This case is load-bearing: Proposition 4.17, Proposition 5.1, Proposition 5.3, and Theorem 5.2 all depend on the conclusion that the closure of the set A has nonempty interior in ∂G. The manuscript must supply a complete argument for Case 2, including the verification that the constructed elements g_n lie in [G,G] and that w g_n is reduced and cyclically reduced. The existing Case 1 discussion also does not address possible cancellation between a terminal power of w and an initial power of w_k when the signs of the exponents are opposite.","section":"4.2 (Lemma 4.16)"},{"comment":"After showing that A\\∪_{gE_i∈F_D}Λ(gE_i) is infinite, the proof asserts 'Hence, we may choose g_1,g_2 from A\\∪... satisfying the desired properties.' Infinite cardinality alone does not guarantee two elements with disjoint pairs of fixed points. An explicit argument is needed: for instance, one can use the nonempty interior supplied by Lemma 4.16 to find two boundary points whose full geodesic axes avoid the relevant limit sets and then realize them by elements of A. As written, the existence of an independent pair is not established.","section":"4.2 (Proposition 4.17)"},{"comment":"The assertion 'Clearly then F = F ∩ F_3 satisfies the properties of the proposition' is not justified. Proposition 5.1, as stated, does not conclude that its subgroup H_1 is normal in Comm_G(H_1), and the preceding argument in Proposition 5.3 does not establish normality of F_3 in Comm_G(F_3). Since normality is part of Proposition 5.3(2), the authors must supply a proof (for example, by replacing F_3 with its normal core in Comm_G(F_3) after checking that the intersection-triviality properties are preserved) or reformulate Proposition 5.3.","section":"5 (Proposition 5.3)"},{"comment":"The final paragraph of Step 2 reads as a sketch rather than a complete proof. The claims that H_1A is weakly malnormal in G_3, that Comm_G(H_1)=Comm_{G_3}(H_1), and that H_1A is weakly malnormal in G are asserted without detailed verification of the cases (elements of G_3\\H_1A, elements of G\\G_3, and mixed elements). Since Theorem 5.2 is the main result, this step should be written out in full.","section":"5 (Theorem 5.2, Step 2)"}],"minor_comments":[{"comment":"The conjugation notation E^h is used without stating whether it means hEh^{-1} or h^{-1}Eh; since Proposition 5.3's intersection-triviality property depends on this convention, it should be fixed explicitly.","section":"Throughout"},{"comment":"In the definition of w_k, the word is written as w_k=x_{p_1}^{r_1}...x_{p_i}^{r_i} with p_1=1, but the condition that p_1=1 depends on the subcase; this notation is confusing and should be clarified.","section":"4.2 (Lemma 4.16)"},{"comment":"The proof uses the fact that a cyclically reduced word's geodesic axis passes through 1, but this is only implicit; stating it explicitly would improve readability.","section":"4.2 (Proposition 4.17)"}],"recommendation":"major_revision","confidential_remarks":"The two principal gaps (Lemma 4.16, Case 2; Proposition 5.3 normality) are exactly the kind of omissions that a careful revision can repair. The third and fourth major comments concern presentation of arguments that are likely fillable but are currently too compressed for a journal proof. I would not recommend rejection, because the strategy is sound and the missing pieces appear local. However, the authors should be asked to provide complete proofs of these points before publication."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Dear colleague,\n\nThe headline: this is a genuinely new extension of Ilya Kapovich's malnormal-subgroup theorem, dropping the torsion-free hypothesis, and the overall strategy is credible. But the proof as written has two underproved claims — the missing case in Lemma 4.16 and the normality assertion in Proposition 5.3 — so the result is not yet fully secure.\n\nWhat's new: Theorem 5.2 gives, in any nonelementary hyperbolic group G, a quasiconvex subgroup H1 ≅ F2 such that Comm_G(H1) = H1 × A for some finite A, and H1A is weakly malnormal in G. This is absent from the literature; Kapovich's version only covers torsion-free groups. The proof adapts Kapovich's ping-pong approach to the coned-off graph of a hyperbolic group, uses a Baire-category argument to avoid finitely many limit sets, and then uses virtually free groups and Stallings' theorem to get the commensurator structure. The paper is careful with definitions and includes useful lemmas (Lemma 4.6 and Corollary 4.8, for instance). The non-quasiconvex embedding application (Theorem 5.5) is a nice payoff. The citation pattern looks appropriate; the self-citations are background lemmas, not something on which the main result circularly depends.\n\nThe soft spots are real but not obviously fatal, and they match the reader's concern. Lemma 4.16 is load-bearing: it asserts that endpoints of cyclically reduced words wg (g in the commutator) have closure with nonempty interior. The proof handles one case and leaves the other, where w starts and ends with powers of different basis elements, as \"a simple exercise.\" That omitted case is exactly what Proposition 4.17 needs to produce a ping-pong pair avoiding a finite list of subgroups, and Theorem 5.2 depends on it. The case is plausibly true — a quick example in F2 suggests a construction exists — but the written proof doesn't secure it. A referee should ask the authors to write out that case.\n\nThe second gap: Proposition 5.3 ends with \"Clearly then F satisfies the properties,\" where F is defined as an intersection of two finite-index subgroups inside Comm_G(F3). The claimed normality of F in Comm_G(F) does not follow from the preceding lines. This is probably repairable by passing to the normal core, but the proof as written skips the step. Smaller than the first gap, but still requires fixing.\n\nIf both gaps are filled, this is a solid contribution: significant theorem, standard techniques well deployed, readable exposition. I would send it to a serious referee; it deserves a careful look, but it is not ready in its current form. I would not cite it in my own work yet.","headline":"A genuinely new result extending Kapovich's theorem, but with two underproved claims — the missing case in Lemma 4.16 and the normality assertion in Proposition 5.3 — that need fixing before the proof is convincing.","tokens_in":17915,"tokens_out":4710,"would_cite":false,"duration_ms":43245,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["20F65","20F67"],"pacs":[],"model":"deepseek-v4-flash","headline":"Every nonelementary hyperbolic group contains a weakly malnormal quasiconvex subgroup of the form $H_1 \\times A$, where $H_1$ is free of rank at least 2 and $A$ is finite.","keywords":["hyperbolic groups","quasiconvex subgroups","weakly malnormal subgroups","virtually free subgroups","commensurator","coned-off spaces","ping-pong lemma","non-quasiconvex embedding"],"falsifier":"In the free group $F_2 = \\langle x_1, x_2 \\rangle$, take $w = x_1 x_2$, which starts and ends with powers of different generators, and check whether the closure of $\\{(wg)^\\infty : g \\in [F_2,F_2],\\, wg \\text{ cyclically reduced}\\}$ has nonempty interior in $\\partial F_2$; if not, Lemma 4.16's omitted Case 2 fails and the ping-pong construction of Proposition 4.17 would need a different proof.","tokens_in":16813,"feed_emoji":"","tokens_out":8920,"duration_ms":82545,"temperature":0.7,"pith_summary":"This paper proves that every nonelementary hyperbolic group $G$ contains a weakly malnormal, quasiconvex subgroup that is virtually free: a subgroup $H_1 \\times A$ with $H_1$ free of rank at least two, $A$ finite, and $\\mathrm{Comm}_G(H_1) = H_1 \\times A$. Weakly malnormal means that any conjugate of the subgroup by an element outside it intersects it in a finite subgroup, so distinct conjugates are almost disjoint. The theorem removes the torsion-free assumption that was needed in the earlier malnormal-subgroup theorem for hyperbolic groups, and answers positively the paper's stated question for quasiconvex subgroups. A direct corollary is that every nonelementary hyperbolic group embeds injectively into another hyperbolic group as a non-quasiconvex subgroup.","feed_headline":"Weakly malnormal free subgroups found in all hyperbolic groups","feed_subtitle":"The result covers groups with torsion, giving subgroups whose distinct conjugates meet only finitely.","key_machinery":"The carrying mechanism is the coned-off (electrified) Cayley graph: one adds a vertex for each coset of a finite family of quasiconvex subgroups and joins it to every point of that coset, producing a hyperbolic graph whose boundary is a homeomorphic copy of the 'undistorted' part of $\\partial G$. Proposition 4.13 is the ping-pong criterion: two loxodromic isometries with disjoint fixed point sets, and with no power conjugate into any coned-off subgroup, generate a free group whose intersection with every conjugate of every coned-off subgroup is trivial. The algebraic half uses the fact that finite subgroups of $\\mathrm{Aut}(F_n)$ act faithfully on the abelianization, so a word $w = x_1 x_2^N$ can be chosen whose abelianized image is not an eigenvector of any nontrivial finite-automorphism image; this forces the finite part of the commensurator to centralize the free subgroup, giving $H_1 \\times A$.","core_discovery":"The central claim is Theorem 5.2. If $G$ is a nonelementary hyperbolic group and $H$ is a nonelementary quasiconvex subgroup of $G$, then there is a free subgroup $H_1 \\cong F_2$ of $H$ such that $\\mathrm{Comm}_G(H_1) = H_1 \\times A$ for some finite subgroup $A$ of $G$, and the product $H_1 A \\cong H_1 \\times A$ is weakly malnormal and quasiconvex in $G$. This is the first written treatment for hyperbolic groups that may have torsion; the torsion-free case was known. The authors' construction first produces a quasiconvex free subgroup whose commensurator is a free-by-finite extension with finite normal part, then passes to a rank-two subgroup inside it so that the finite part centralizes it and the whole product is weakly malnormal.","pith_inferences":["If the construction were run inside a torsion-free finite-index subgroup of $G$, one might hope for full malnormality rather than weak malnormality; the paper does not attempt this, and torsion in $G$ is the stated obstacle.","The proof of Lemma 4.16 divides into two cases and leaves the second as an exercise; the main theorem therefore currently rests on that omitted verification.","The normality assertion for $F$ in $\\mathrm{Comm}_G(F)$ inside Proposition 5.3 is labelled 'Clearly' but does not follow from the stated Proposition 5.1; a complete proof would need to establish or replace it.","The coned-off ping-pong criterion is not specific to free groups and may provide a template for similar existence results in other classes of groups with quasiconvex subgroups, though the paper does not pursue this."],"forward_implications":["Every nonelementary hyperbolic group, with or without torsion, contains a weakly malnormal quasiconvex subgroup that is virtually free.","The commensurator of the constructed subgroup is as small as possible: it is the product of the free subgroup and a finite subgroup.","The non-quasiconvex embedding theorem follows: any nonelementary hyperbolic group appears as a non-quasiconvex subgroup of some hyperbolic group.","By Remark 5.4 the rank of the free factor can be increased arbitrarily without losing weak malnormality."],"supporting_citations":[{"why":"supplies the torsion-free malnormal-subgroup theorem that this paper extends and the overall ping-pong strategy.","marker":"[Kap99]"},{"why":"proves the coned-off Cayley graph remains hyperbolic when cosets of finitely many quasiconvex subgroups are coned off.","marker":"[KR14, Proposition 2.6]"},{"why":"gives the dichotomy for geodesic rays in the coned-off graph, used to identify loxodromic isometries.","marker":"[AM24, Corollary 6.4]"},{"why":"gives the homeomorphism from the undistorted boundary to the coned-off boundary, used to find independent loxodromic isometries.","marker":"[DT17, Theorem 3.2]"},{"why":"shows finite subgroups of automorphisms of free groups act faithfully on the abelianization, used to choose a word with a non-eigenvector image.","marker":"[BT68, Proposition 1]"},{"why":"identifies torsion-free virtually free groups as free, allowing the commensurator splitting.","marker":"[Sta68, Theorem 0.2]"},{"why":"provides finiteness of the commensurator quotient and of intersection subgroups up to conjugacy, used in the descending-subgroup construction.","marker":"[GMRS98]"},{"why":"Bestvina–Feighn combination theorem is invoked to build the non-quasiconvex embedding in Theorem B.","marker":"[BF96, Theorem 1.2]"}],"fun_headline_variants":["All hyperbolic groups have weakly malnormal free subgroups","Every hyperbolic group admits weakly malnormal F2 subgroups","Weakly malnormal F2 subgroups in every hyperbolic group","Torsion groups included: weakly malnormal free subgroups exist","Hyperbolic groups with torsion get weakly malnormal free subgroups"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is Lemma 4.16, whose proof is completed only for words that start and end with the same generator; the second case is left as a 'simple exercise', and Proposition 4.17, Proposition 5.1, and Theorem 5.2 all depend on it.","fun_headline_variants_meta":{"raw":{"variants":["All hyperbolic groups have weakly malnormal free subgroups","Every hyperbolic group admits weakly malnormal F2 subgroups","Weakly malnormal F2 subgroups in every hyperbolic group","Torsion groups included: weakly malnormal free subgroups exist","Hyperbolic groups with torsion get weakly malnormal free subgroups"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.001832,"raw_usage":{"total_tokens":7108,"prompt_tokens":754,"completion_tokens":6354,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":370,"completion_tokens_details":{"reasoning_tokens":6275}},"tokens_in":370,"tokens_out":6354,"duration_ms":38843,"temperature":1.0,"reasoning_tokens":6275,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-15T18:21:46.021141+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"In the free group $F_2 = \\langle x_1, x_2 \\rangle$, take $w = x_1 x_2$, which starts and ends with powers of different generators, and check whether the closure of $\\{(wg)^\\infty : g \\in [F_2,F_2],\\, wg \\text{ cyclically reduced}\\}$ has nonempty interior in $\\partial F_2$; if not, Lemma 4.16's omitted Case 2 fails and the ping-pong construction of Proposition 4.17 would need a different proof.","supporting_citations":[],"review_version":2}