{"id":"5d5926f7-7895-42ce-a519-95fff58f2b5a","arxiv_id":"2508.13038","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"Relatively hyperbolic TDLC groups are closed under amalgams and HNN extensions over compact subgroups, and their Bowditch boundaries are determined by the boundaries of the vertex groups.","lead":"This paper builds a framework for relatively hyperbolic totally disconnected locally compact (TDLC) groups and proves combination theorems: amalgams and HNN extensions of such groups over compact subgroups stay relatively hyperbolic. It also shows that the Bowditch boundary of the amalgam is topologically determined by the boundaries of the factors when the group has infinitely many rough ends.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The boundary-rigidity theorems rest on an unproved identification: §5.1–5.3 constructs δ(X^h), but Proposition 5.1 merely asserts, with a skipped topology verification, that it is homeomorphic to the Bowditch boundary.","rationale":"The reader's listed weakest assumption, properness of H_A∪H_B after amalgamation, is not the most dangerous: for a compact edge C, any noncompact H∈H_A fixes a unique vertex in the Bass-Serre tree, and a conjugate landing in B would force H to fix two vertices and hence an edge with compact stabilizer, so properness is automatic and only needs a short argument. The real gap is the boundary identification. The paper itself flags the skipped topology check, and Proposition 5.1 is the bridge from the abstract δ(X^h) to the actual Bowditch boundary; all boundary-homeomorphism claims inherit this risk. I also note that Lemma 2.3's proof is flawed in the bounded-domain case, but that is auxiliary and fixable. The recommendation stays UNCHANGED relative to the reader's CONDITIONAL verdict: the main theorems are plausible and likely repairable, but the boundary compactification is not yet fully supported as written.","tokens_in":25835,"tokens_out":43944,"duration_ms":522696,"concrete_test":"Verify Proposition 5.1 in the simplest nontrivial case: let G=Z*Z with A=B=Z and C=1, where X^h is the tree of augmented lines and ∂relG is the Cantor set. Write the V_U and V_n bases explicitly from §5.3 and prove that they define a compact Hausdorff topology matching the Gromov topology on ∂relG; in particular, show the natural bijection ∂relG→δ(X^h) is a homeomorphism. Then repeat for a TDLC example such as G=A*_C B with A,B hyperbolic TDLC and C a compact open subgroup, checking that each vertex space X^h_v satisfies the hypotheses of [MS15, Section 3]. If the topology axioms fail or the [MS15] hypotheses are not met, Proposition 5.1 must be replaced before Theorems 1.5/1.7 can be accepted.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central new results are the Bowditch-boundary combination theorems. Their proof requires that the ad hoc compactification X^h of Subsection 5.3 is actually the Gromov compactification of the augmented Cayley-Abels graph of G from Definition 4.6. The paper explicitly says 'We skip a verification that the above collections of sets satisfy the axioms for the basis of open neighborhoods,' and Proposition 5.1 is proved only by a 'word on the proof' saying that continuity follows from [MS15, Section 3]. But [MS15] concerns trees of hyperbolic spaces whose vertex spaces are Cayley graphs of hyperbolic groups; here the vertex spaces are augmented Cayley-Abels graphs with horoballs attached to every peripheral coset. It is not automatic that the V_U(ξ) and V_n(η) collections form a neighborhood basis, that δ(X^h) is Hausdorff, or that the natural bijection from ∂relG is open at points of ∂T. If the topology is wrong, Proposition 5.4's dense-amalgam conditions describe a space that need not be homeomorphic to the actual Bowditch boundary, so Theorems 1.5 and 1.7 and their corollaries would not be established. This is a genuine missing proof in the main line of argument.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper develops a framework for relative hyperbolicity of totally disconnected locally compact (TDLC) groups. It introduces three definitions (TDRH-I, TDRH-II, TDRH-III), proves their equivalence for compactly generated groups, and uses them to prove combination theorems: if A and B are relatively hyperbolic TDLC groups and the edge group C is compact, then the amalgamated free product A*_C B and the HNN extension A*_C are relatively hyperbolic. The main new results are the construction of the Bowditch boundary for such amalgams and HNN extensions via a tree of augmented Cayley-Abels graphs, and the claim that, when the rough ends are infinite, the topology of the Bowditch boundary is determined by the Bowditch boundaries of the vertex groups (Theorems 1.5 and 1.7). The paper also proves connectedness of the Bowditch boundary for one rough end (Theorem 1.8) and a splitting theorem for hyperbolic TDLC groups with totally disconnected Gromov boundary (Proposition 1.9).","tokens_in":26129,"tokens_out":8140,"duration_ms":92425,"significance":"If the main claims hold, the paper provides a substantial extension of classical combination theorems and boundary-rigidity results from discrete relatively hyperbolic groups to TDLC groups, and it connects the existing notions of relative hyperbolicity in this setting. The use of Świątkowski's dense amalgam technique is appropriate, and the paper is clearly organized, with explicit credit to the external results on which it builds ([BF92], [KM08], [Sis12], [S16], [MS15]). However, several load-bearing steps are asserted rather than proved, most importantly the construction of the topology on the compactification X^h in Section 5.3. For this reason the boundary rigidity theorems should be regarded as conditional on a complete proof.","major_comments":[{"comment":"The proof of properness of the embedding i: X_H → X_G contains an invalid counting argument. The sentence \"for every N there exists h_N with d_{X_H}(U_H,h_N U_H) ≤ M but d_{X_G}(U,h_N U) ≤ N\" does not imply that there are infinitely many distinct points in the ball of radius M in X_H: the elements h_N could be the same or could repeat. To derive the required contradiction, one must instead use local finiteness of X_G and the fact that, if d_{X_G} is bounded on infinitely many distinct cosets of U_H, then a finite ball in X_G contains infinitely many vertices. This lemma is used in Section 4 to justify the quasi-isometry invariance of the augmented and coned-off Cayley-Abels graphs, so the argument needs to be corrected even if the statement is true.","section":"§2.3, Lemma 2.3"},{"comment":"The central topological identification is not proved. After defining the sets V_U(ξ) and V_n(η), the text says \"We skip a verification that the above collections of sets satisfy the axioms for the basis of open neighborhoods,\" and Proposition 5.1 is then justified only by a \"word on the proof\" referring to [MS15, Section 3]. The cited argument applies to trees of hyperbolic spaces whose vertex spaces are Cayley graphs of hyperbolic groups, whereas here the vertex spaces are augmented Cayley-Abels graphs with horoballs attached along peripheral cosets. It is not automatic from [MS15] that the V_U and V_n collections form a neighborhood basis, that δ(X^h) is Hausdorff, or that the natural bijection from ∂_{rel}G to δ(X^h) is continuous and open. Without this verification, the homeomorphism type of the Bowditch boundary has not been established, and Proposition 5.4 and Theorems 1.5 and 1.7 are not justified.","section":"§5.3, Proposition 5.1"},{"comment":"The proof of Lemma 3.1 asserts that the constructed graph X is a Cayley-Abels graph of G with respect to H_A ∪ H_B, but it does not verify that (G, H_A ∪ H_B) is a proper pair in the sense of Definition 2.7. This requires showing that no two distinct non-compact subgroups from H_A and H_B become conjugate in the amalgam or HNN extension. Since the definition of relative hyperbolicity and the entire boundary construction depend on properness, this missing verification is load-bearing. A proof using the Bass-Serre tree and the compactness of C is plausible, but it needs to be supplied.","section":"§3, Theorems 1.1 and 1.2, Lemma 3.1"},{"comment":"The proof of condition (2) (nullness of the family Y) uses the compactness of δ(X^h) to pass to a finite subcover of an arbitrary open cover. Compactness of δ(X^h) is exactly part of the conclusion of Proposition 5.1, which is not proved. Thus the argument is circular unless Proposition 5.1 is established independently. This is not a mere technicality, because the dense-amalgam theorem [Ś16, Theorem 0.2] is applied to a space whose compact metrizability is assumed as a hypothesis.","section":"§5.4, Proposition 5.4"}],"minor_comments":[{"comment":"The statement says that A and A' are hyperbolic relative to H_A and H_B, respectively; the second collection should presumably be H'_A, not H_B.","section":"§1, Theorem 1.7"},{"comment":"The notation \"H ∈ H\" is overloaded, since H is used both for a subgroup and for the collection of subgroups. Writing H ∈ H_A ∪ H_B explicitly would improve readability.","section":"§3.1, Lemma 3.1"},{"comment":"The sentence \"Since G is compact, there exists a subsequence {a_{m_k}} that converges to η ∈ ∂_{rel}G\" is not correct as written; G is not compact in general. The intended statement is that the compactification of X^h is compact, so a bounded sequence of vertices has a subsequence converging to a boundary point.","section":"§6, proof of Theorem 1.8"},{"comment":"The set δStab(X) should be δStab(X^h) for consistency with the amalgamated case.","section":"§5.2"},{"comment":"In condition (2), the symbol U is used both for an arbitrary finite open cover and for a neighborhood basis of a boundary point; this makes the argument hard to follow.","section":"§5.4, proof of Proposition 5.4"}],"recommendation":"major_revision","confidential_remarks":"The main risk of the paper is the unproved topological identification in §5.3. If the authors can provide a complete proof that the neighborhood bases defined there give a compact metrizable Hausdorff topology and that the natural bijection from ∂_{rel}G to δ(X^h) is a homeomorphism, the main theorems would be established. The Section 3 combination theorems are more straightforward and likely correct, but they also need a proper-pair verification."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Here's my take. The paper does two genuinely useful things in the first half. It proves combination theorems for relatively hyperbolic TDLC groups with compact edge groups, and it shows that three candidate definitions of relative hyperbolicity — Arora–Pedroza's, the Bowditch/Groves–Manning style, and Farb-style — are equivalent for compactly generated groups. The equivalence is a real contribution, and the combination theorems look right. The proof strategy via trees of augmented Cayley-Abels graphs is natural, and the examples (SL(2,Zp)×Z, Aut(Td), p-adic semidirect products) help the reader see the scope.\n\nThe soft spots are real, and they are in the main line of the boundary results. The proof of Lemma 2.3 has a bad counting step: the argument finds, for each N, an element h_N with bounded distance to the base point, but nothing prevents these h_N from representing the same vertex of the locally finite ball. The lemma is probably true, but the written proof does not establish it.\n\nThe second issue is that Theorems 1.1 and 1.2 never prove that (G, H_A ∪ H_B) is a proper pair. The construction of the relative Cayley-Abels graph and the whole relative hyperbolicity framework require this. It may follow from a Bass–Serre argument — a non-compact subgroup fixing two distinct vertices would be conjugate into an edge stabilizer, hence compact — but the paper does not spell it out.\n\nThe big problem, though, is Proposition 5.1 and the topology on X^h. This is where the boundary rigidity theorems rest. The paper explicitly skips the verification that the V_U(ξ) and V_n(η) collections form a neighborhood basis, and then the proof of the homeomorphism is a 'word on the proof' citing [MS15]. That citation is not a substitute. [MS15] is about trees of hyperbolic spaces whose vertex spaces are Cayley graphs of hyperbolic groups. Here the vertex spaces are augmented Cayley-Abels graphs with horoballs on every peripheral coset. It is not automatic that the bijection from the Gromov boundary to δ(X^h) is continuous, that δ(X^h) is Hausdorff, or that the proposed basis is actually a basis. The stress-test note lands exactly here. Without fixing this, Theorems 1.5 and 1.7 and their corollaries are not established as written.\n\nWho should read this? People working on large-scale geometry of TDLC groups — the first half gives them working tools and a clean dictionary between definitions. The boundary part needs a real revision, not just a referee's suggestion.\n\nMy recommendation: send it to a serious referee, but expect heavy revision. The core ideas are good and likely salvageable, but the paper as it stands should not be accepted with the topology of Section 5 in its current state.","headline":"Solid combination theorems and a useful equivalence of definitions, but the boundary-rigidity main theorem is not yet proven: Section 5 skips a load-bearing topology verification.","tokens_in":26633,"tokens_out":4106,"would_cite":true,"duration_ms":47877,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["20F65","20F67","22D05"],"pacs":[],"model":"deepseek-v4-flash","headline":"Combination theorems hold for relatively hyperbolic TDLC groups.","keywords":["TDLC groups","relative hyperbolicity","Bowditch boundary","combination theorems","amalgamated free products","HNN extensions","Cayley-Abels graphs","rough ends"],"falsifier":"Construct an explicit amalgam $G=A\\ast_C B$, with $C$ compact, where $A$ is relatively hyperbolic with parabolic family $H_A$ and $B$ is relatively hyperbolic with parabolic family $H_B$, such that some $h\\in H_A$ is conjugate in $G$ to a distinct $k\\in H_B$; if such an example exists, the pair $(G,H_A\\cup H_B)$ is not proper, so the relative Cayley-Abels graph is undefined and Theorem 1.1 cannot hold as stated.","tokens_in":25640,"feed_emoji":"🔗","tokens_out":9869,"duration_ms":85912,"temperature":0.7,"pith_summary":"This paper establishes combination theorems for relatively hyperbolic totally disconnected locally compact (TDLC) groups. It proves that when two relatively hyperbolic TDLC groups are glued along a compact subgroup, either as an amalgamated free product or as an HNN extension, the resulting group is again relatively hyperbolic, and it constructs its Bowditch boundary from the boundaries of the pieces. A consequence is boundary rigidity: if the rough ends of the amalgam are infinite, the topology of the Bowditch boundary of the amalgam is uniquely determined by the topologies of the Bowditch boundaries of the two factors. The paper also shows that a relatively hyperbolic TDLC group with one rough end has connected Bowditch boundary, and that a hyperbolic TDLC group with totally disconnected Gromov boundary splits as a finite graph of compact groups.","feed_headline":"Combination theorems hold for relatively hyperbolic TDLC groups","feed_subtitle":"Compact edge groups let amalgams and HNN extensions inherit relative hyperbolicity and a determined Bowditch boundary.","key_machinery":"The argument runs through the relative Cayley-Abels graph: a locally finite graph on which the group acts with compact edge stabilizers and vertex stabilizers that are either compact or parabolic. To build the Bowditch boundary, the paper replaces each parabolic coset's Cayley-Abels graph by a combinatorial horoball, forming the augmented Cayley-Abels graph whose Gromov boundary is the Bowditch boundary. For an amalgam or HNN extension, the graph is assembled as a tree of such augmented graphs, one per vertex of the tree of the splitting, and hyperbolicity follows from a classical combination theorem for trees of hyperbolic spaces. The boundary-rigidity statements are then obtained by recognizing the Bowditch boundary as a dense amalgam of the factors' boundaries—a compact metric space assembled from embedded copies of a finite list of compact metric spaces—which records which connected components come from copies of each vertex group.","core_discovery":"The central discovery is that relative hyperbolicity, originally developed for discrete groups, can be carried over to TDLC groups in three equivalent definitions, and that the standard combination and boundary-rigidity results hold in this setting. Theorems 1.1 and 1.2 state that the amalgamated free product and HNN extension of relatively hyperbolic TDLC groups along a compact edge group are relatively hyperbolic TDLC groups. Theorems 1.5 and 1.7 state that, when the rough ends of the resulting group are infinite, the homeomorphism type of its Bowditch boundary is determined by the homeomorphism types of the Bowditch boundaries of the vertex groups. Theorem 1.8 states that one rough end forces connectedness of the Bowditch boundary, and Proposition 1.9 states that a hyperbolic TDLC group with totally disconnected Gromov boundary is a finite graph of compact groups.","pith_inferences":["The dense-amalgam description suggests that the same boundary rigidity should persist for finite graphs of relatively hyperbolic TDLC groups with compact edge groups, by induction on the graph, although the paper only sketches this in Theorem 5.7.","If the properness assumption can fail in a natural construction—for instance, a parabolic subgroup of one factor becoming conjugate to one of the other inside the amalgam—then the relative hyperbolicity statement as written would need a notion that tolerates non-proper collections, or the theorem would be false in that case.","Connectedness of the Bowditch boundary for one rough end, combined with accessibility of hyperbolic TDLC groups, suggests an end-theoretic characterization of splittings over compact subgroups, with totally disconnected Gromov boundary as a testable symptom.","Because the proof uses dense amalgams rather than the earlier free-product technique, the same boundary decomposition should describe how geodesic rays from the different vertex spaces accumulate on the boundary of the amalgam, which could connect to limit-set theory for such groups."],"forward_implications":["Any amalgam or HNN extension of relatively hyperbolic TDLC groups along a compact edge group is itself relatively hyperbolic, and the same holds for finite graphs of such groups with compact edge groups.","When the rough ends of the glued group are infinite, its Bowditch boundary is homeomorphic to the dense amalgam of the factors' Bowditch boundaries, so its homeomorphism type depends only on the factors' boundary types.","Distinct parabolic subgroups of a relatively hyperbolic TDLC group intersect compactly, and conjugates of distinct parabolics intersect compactly.","A relatively hyperbolic TDLC group with exactly one rough end has a connected Bowditch boundary.","A hyperbolic TDLC group whose Gromov boundary is totally disconnected splits as a finite graph of groups with compact vertex groups."],"supporting_citations":[{"why":"Supplies the relative Cayley-Abels graph and the existing notion of relative hyperbolicity for TDLC groups that this paper proves equivalent to its two new definitions.","marker":"[AMP23]"},{"why":"Provides Cayley-Abels graphs, their quasiisometry invariance, and the notion of rough ends used throughout.","marker":"[KM08]"},{"why":"Gives the combination theorem for trees of hyperbolic spaces that establishes hyperbolicity of the assembled graph.","marker":"[BF92]"},{"why":"Supplies the discrete relative hyperbolicity and Bowditch boundary notions being generalized to TDLC groups.","marker":"[Bow12]"},{"why":"Provides the combinatorial horoball construction used to form the augmented Cayley-Abels graph.","marker":"[GM08]"},{"why":"Supplies the coned-off space and bounded penetration property that form one of the three equivalent definitions.","marker":"[Far98]"},{"why":"Provides the dense amalgam construction and the regularity criteria that yield the boundary-rigidity conclusions.","marker":"[´S16]"},{"why":"Supplies the tree-of-spaces boundary description and geodesic ray analysis used to topologize the amalgam boundary.","marker":"[MS15]"},{"why":"The discrete boundary-rigidity result that Theorems 1.5 and 1.7 extend to the TDLC setting.","marker":"[Tom25]"}],"fun_headline_variants":["TDLC groups inherit relative hyperbolicity via compact amalgams","Bowditch boundary determined for relatively hyperbolic TDLC groups","One rough end forces connected Bowditch boundary in TDLC groups","Hyperbolic TDLC groups with totally disconnected boundary split"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The construction needs the collection of parabolic subgroups from both factors to remain a proper pair inside the glued group—no two distinct non-compact parabolic subgroups from different factors may become conjugate—and the paper states this rather than proving it.","fun_headline_variants_meta":{"raw":{"variants":["TDLC groups inherit relative hyperbolicity via compact amalgams","Bowditch boundary determined for relatively hyperbolic TDLC groups","One rough end forces connected Bowditch boundary in TDLC groups","Hyperbolic TDLC groups with totally disconnected boundary split"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000649,"raw_usage":{"total_tokens":2979,"prompt_tokens":948,"completion_tokens":2031,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":564,"completion_tokens_details":{"reasoning_tokens":1974}},"tokens_in":564,"tokens_out":2031,"duration_ms":14651,"temperature":1.0,"reasoning_tokens":1974,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-15T17:17:53.303377+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Construct an explicit amalgam $G=A\\ast_C B$, with $C$ compact, where $A$ is relatively hyperbolic with parabolic family $H_A$ and $B$ is relatively hyperbolic with parabolic family $H_B$, such that some $h\\in H_A$ is conjugate in $G$ to a distinct $k\\in H_B$; if such an example exists, the pair $(G,H_A\\cup H_B)$ is not proper, so the relative Cayley-Abels graph is undefined and Theorem 1.1 cannot hold as stated.","supporting_citations":[],"review_version":2}