{"id":"7a4e239c-e4c0-4082-9243-abf24b813c22","arxiv_id":"2411.15337","paper_version":3,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"The paper defines Cayley-Abels-Rosendal graphs for Polish groups and uses them to classify when homeomorphism groups of countable Stone spaces are coarsely bounded, locally bounded, and boundedly generated.","lead":"This paper introduces a graph, called a Cayley-Abels-Rosendal graph, that lets certain large topological groups be studied with the tools of geometric group theory. It then uses these graphs to completely classify the coarse geometry of homeomorphism groups of countable Stone spaces.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Lemma 21's strong triangle inequality is false under the printed 'for all i' quantifier; the limit-ordinal proof of Proposition 19 is therefore unsupported as written.","rationale":"The reader's weakest-assumption points to the same place I would: the limit-ordinal proof of non-bounded-generation is the least secure part of Theorem A. My independent reading confirms the issue is real, and slightly sharpens it: it is not merely that the proof of Lemma 21 does not go through; the strong triangle inequality is actually false as printed. The explicit n=3, alpha=omega example shows that the universal quantifier in the definition of h is incompatible with the inequality the later lemmas require. The subsequent proofs of Lemmas 22 and 23 also rely on the existential reading to conclude that a homeomorphism moving one point of high rank from P_1 to P_2 has h(gP) >= beta. Under 'for all i', such a g can have h(gP) = 0 when n >= 3. Thus Proposition 19, as written, does not establish that Homeo(X_{alpha,n}) is not boundedly generated for limit alpha and n > 1. I do not think this warrants rejection: the fix is a one-word quantifier change ('there exists i'), the intended existential reading is consistently used in Lemmas 22-24, and the surrounding CAR-graph framework appears sound. But the proof needs this correction, so the conditional verdict is appropriate. No other concern I checked (successor-ordinal graph connectivity, self-similarity argument, Milnor-Schwarz application) seems as load-bearing; those parts are sketched but plausible and less central to the classification's failure direction.","tokens_in":20307,"tokens_out":12623,"duration_ms":118219,"concrete_test":"Construct the explicit counterexample on X_{omega,3}: fix a base good partition P = P_1 ⊔ P_2 ⊔ P_3, choose distinct rank-1 points a_i in P_i, let R be obtained from P by moving a_1 from P_1 to P_2, and let Q be obtained from R by moving a_2 from R_2 to R_3. Using the printed definition of h with 'for all i', compute h(P,R), h(R,Q), and h(P,Q), and verify that h(P,R) = h(R,Q) = 0 while [P_i △ Q_i]_1 is nonempty for i = 1,2,3, so h(P,Q) >= 1. If the definition is changed to 'there exists i', recompute and confirm the strong triangle holds; this check determines whether Proposition 19 can be repaired by the evident quantifier correction.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The limit-ordinal half of Theorem A (Proposition 19) rests on the height function h(P,Q) = sup{ beta < alpha : [P_i △ Q_i]_beta is nonempty for all i = 1,...,n }. Under this universal reading, the proof of Lemma 21's strong triangle inequality is not merely incomplete but false. The pointwise inclusion [P_i △ Q_i]_beta is contained in ([P_i △ R_i]_beta) union ([R_i △ Q_i]_beta) only yields that if both sets on the right are empty for a fixed i, then the left is empty; it does not control the case where one piece is changed in P versus R and a different piece is changed in R versus Q. Concretely, for n=3 and alpha=omega, take good partitions P, R, Q with rank-1 points a_1 in P_1, a_2 in P_2, a_3 in P_3; let R move a_1 from P_1 to P_2, and let Q move a_2 from R_2 to R_3. Then P and R differ at rank 1 only in pieces 1 and 2, and R and Q differ only in pieces 2 and 3, so h(P,R) = h(R,Q) = 0. But P and Q differ at rank 1 in all three pieces, so h(P,Q) >= 1, violating h(P,Q) <= max{h(P,R), h(R,Q)}. The later lemmas implicitly use the existential reading 'there exists i'; under that reading the inclusion argument does establish the strong triangle, and Lemma 22's properness argument also works. As typeset, however, the chain {Stab(P_beta)} may fail to be proper, so the non-bounded-generation conclusion does not follow from the written proof.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper introduces Cayley–Abels–Rosendal (CAR) graphs, countable connected graphs on which a Polish group acts continuously, vertex-transitively, with finitely many edge orbits and coarsely bounded vertex stabilizers. It proves a Milnor–Schwarz-type lemma showing that groups admitting such graphs are boundedly generated and quasi-isometric to them, and then applies this framework to homeomorphism groups of countable Stone spaces X_{α,n}. The main result, Theorem A, asserts that Homeo(X_{α,n}) is always locally bounded, is coarsely bounded exactly when n=1, and is boundedly generated but not coarsely bounded exactly when n>1 and α is a successor ordinal; for successor ordinals the authors construct explicit CAR graphs, and for limit ordinals they attempt to show non-bounded-generation via an exhaustive chain of open subgroups.","tokens_in":20641,"tokens_out":28691,"duration_ms":271949,"significance":"If the technical issue below is repaired, the paper is a valuable contribution. It gives a clean topological analogue of finite generation for Polish groups, provides explicit CAR graphs for the homeomorphism groups of countable Stone spaces in the successor-ordinal case, recovers and extends results of Mann–Rafi, and supplies new non-bounded-generation results in the limit-ordinal case. The exposition is largely self-contained, including a proof of the Mazurkiewicz–Sierpiński classification of countable Stone spaces, and the main structural arguments (Milnor–Schwarz, the double-coset graph construction, and the subgroup-chain criterion) are coherent. The most serious obstacle is a quantifier error in Section 4.4 that currently invalidates the limit-ordinal half of Theorem A; because the fix is local and the surrounding argument is otherwise sound, the central claims appear defensible after revision.","major_comments":[{"comment":"The quantifier in the definition of h is incompatible with the proofs. The text defines h(P,Q) as the supremum of β<α such that [P_i△Q_i]_β is nonempty for all i=1,...,n, but the proofs of Lemma 21's strong triangle inequality and of Lemmas 22–24 all use the existential reading ('there exists i'). Under the printed universal quantifier, Lemma 21 is false. For example, in X_{ω,3}, let R be obtained from P by moving a rank-1 point from P_1 to P_2, and let Q be obtained from R by moving a rank-1 point from R_2 to R_3; then h(P,R)=0 and h(R,Q)=0, because the rank-1 symmetric difference is not nonempty in all three pieces, while h(P,Q)≥1 because P△Q is nonempty at rank 1 in all three pieces. This violates h(P,Q)≤max{h(P,R),h(R,Q)}. Consequently, Lemma 22's properness argument fails as written: a homeomorphism that moves a single high-rank point from P_1 to P_2 has h(P,gP)=0 under the universal reading, so it lies in every Stab(P_β), and the chain of open subgroups may fail to be proper and may fail to exhaust Homeo(X_{α,n}). Since Proposition 19 and the limit-ordinal half of Theorem A rest entirely on this chain, the proof is unsupported as printed. The fix is local and clear: replace 'for all i' with 'there exists i' in the definition of h. Under that reading, the pointwise inclusion argument proves the strong triangle inequality, h(P,gP) is at least the rank of any point moved by g, and Lemmas 22–24 go through.","section":"Section 4.4, definition of h(P,Q) and Lemma 21"}],"minor_comments":[{"comment":"The restatement of Theorem A at the start of Section 4 omits the hypothesis α>0 that is present in the introduction. For α=0, Homeo(X_{0,n}) is a finite symmetric group and hence coarsely bounded even when n>1, so the clause 'coarsely bounded if and only if n=1' needs the α>0 caveat.","section":"Section 4, Theorem A statement"},{"comment":"The induction hypothesis is stated only for characteristic pairs (α,1), but the pieces A_n and B_n in the back-and-forth construction can have finitely many maximal points. The proof should either state the induction hypothesis for all finite n or explicitly reduce each piece to the rank-one case before applying induction.","section":"Section 4.1, proof of Theorem 12"},{"comment":"There is a typo near the end of the sequential-continuity argument: 'f (xi) → xi' should read 'f (xi) → f (x)'.","section":"Section 4.1, proof of Theorem 12"},{"comment":"The statement that ω^α·n+1 with the order topology has characteristic pair (α,n) is inaccurate for α=0, where ω^0·n+1 = n+1 has n+1 points. The statement should be restricted to α>0 or adjusted to the ordinal n in the finite case.","section":"Section 4.1, final sentence of Theorem 12"},{"comment":"The map Φ:gV↦g·x is only well-defined up to the bounded error coming from V⊂U; this should be stated explicitly, since otherwise the reader may wonder why the choice of coset representatives does not matter for the Lipschitz estimate.","section":"Section 4.2, Corollary 14"},{"comment":"The sentence 'Repeating this process for each ordered pair (i,j) yields the desired path' should justify that clearing P_i∩Q_j for one ordered pair is not undone by later shifts. This is true, because later operations only add Q_i-points to P_i, but the proof as written leaves that verification to the reader.","section":"Section 4.3, Lemma 17"}],"recommendation":"major_revision","confidential_remarks":"The paper is well within the scope of a geometric group theory or topological groups journal. The main issue is the quantifier error in Section 4.4, which currently makes the limit-ordinal half of Theorem A unproven; however, the intended fix is clear and the rest of the paper is largely sound. I would be happy to see a revised version."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Colleague,\n\nYou should know about this paper if you care about coarse geometry of Polish groups. It introduces Cayley–Abels–Rosendal (CAR) graphs—countable connected graphs on which a Polish group acts with coarsely bounded vertex stabilizers and finitely many edge orbits—and proves that any boundedly generated Polish group with an open coarsely bounded subgroup admits one (Proposition 8). That is a clean, genuinely useful generalization of Cayley–Abels graphs, and the paper does a good job explaining why 'coarsely bounded generating set' is the right replacement for 'finite generating set.'\n\nThe concrete payoff is a full classification for homeomorphism groups of countable Stone spaces. The successor-ordinal case (n>1, α = β+1) is handled by a graph of good partitions connected by maximal shifts; the graph is shown to be connected, of infinite diameter, and of the form Γ(Stab(P), F), so it really is a CAR graph. The n=1 case is coarsely bounded via self-similarity, and local boundedness for all n follows from a neat short-exact-sequence argument. I found this part convincing.\n\nThe soft spot is the limit-ordinal half (Section 4.4). The height function h(P,Q) is printed with the quantifier 'for all i = 1,...,n' in the set over which the sup is taken. Under that reading, the strong triangle inequality in Lemma 21 is false; the stress-test counterexample with three partitions and n=3 is correct. The later lemmas (22–24) implicitly use the existential reading 'there exists i', and under that reading the inclusion argument does give the strong triangle and the proper exhaustion of Homeo(X_{α,n}) by open subgroups. So the non-bounded-generation conclusion is very likely true, but the printed proof does not establish it. This is a load-bearing typo, not a foundational gap—still, the authors need to fix the quantifier before the result is fully reliable.\n\nMinor things: Proposition 13's diameter-at-most-three proof is sketched more than I would like, and the proof of the Mazurkiewicz–Sierpiński classification (Theorem 12) is compressed. These are surfaces, not holes.\n\nCitation pattern is fine: [BL24] is used for the concrete model of Homeo(X) as automorphism group of a graph, and the Mann–Rafi connection is flagged honestly.\n\nWho is this for? Geometric group theorists who want to see Rosendal's theory made concrete, and anyone working on non-Archimedean Polish groups. I would send it to a serious referee—the typo is exactly what refereeing is for—and I would cite the CAR graph construction in my own work. Bring it to a reading group if you want a good discussion of Milnor–Schwarz in this setting.","headline":"Worth reading: a genuinely useful new CAR graph framework and a mostly solid classification, with a fixable quantifier bug in the limit-ordinal proof.","tokens_in":21201,"tokens_out":4676,"would_cite":true,"duration_ms":39302,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["20F65","22A05","57S05"],"pacs":[],"model":"deepseek-v4-flash","headline":"This paper proves a complete coarse-geometric classification of homeomorphism groups of countable Stone spaces, and it constructs explicit Cayley-Abels-Rosendal graphs that compute the quasi-isometry type in every boundedly generated case.","keywords":["Cayley-Abels-Rosendal graph","coarse boundedness","Polish groups","homeomorphism groups","countable Stone spaces","Cantor-Bendixson rank","bounded generation","quasi-isometry"],"falsifier":"Compute the height function for a concrete triple of good partitions at a limit ordinal $\\alpha$ with $n\\ge 3$, where at some rank $\\beta$ the partition $P$ differs from $R$ only in one piece, $R$ differs from $Q$ only in a different piece, and $P$ differs from $Q$ in every piece. Under the definition as printed, this gives $h(P,Q)\\ge\\beta$ while $h(P,R)$ and $h(R,Q)$ are below $\\beta$, refuting the strong triangle inequality on which the exhausting-chain proof depends; the existence or nonexistence of such triples is a direct check using the clopen sets that define the partitions.","tokens_in":20088,"feed_emoji":"🪨","tokens_out":15505,"duration_ms":136715,"temperature":0.7,"pith_summary":"This paper introduces Cayley-Abels-Rosendal graphs: countable connected graphs on which a Polish group (a separable, completely metrizable topological group) acts continuously, vertex transitively, with coarsely bounded vertex stabilizers and finitely many edge orbits. Coarse boundedness is the topological replacement for finiteness: a subset is coarsely bounded when every continuous isometric action of the group keeps its orbit bounded. These graphs generalize Cayley graphs and Cayley-Abels graphs, and the Milnor-Schwarz lemma makes any group admitting one boundedly generated and quasi-isometric to the graph. The main theorem classifies the homeomorphism groups of countable Stone spaces, the compact totally disconnected spaces that are classified by a countable ordinal $\\alpha$ and an integer $n$. It proves that $\\operatorname{Homeo}(X_{\\alpha,n})$ is always locally bounded, coarsely bounded exactly when $n=1$, and boundedly generated but not coarsely bounded exactly when $n>1$ and $\\alpha$ is a successor ordinal, with an explicit graph $\\Gamma(\\alpha,n)$ computing the quasi-isometry type in the boundedly generated cases.","feed_headline":"Homeomorphism groups of Stone spaces split into three coarse types","feed_subtitle":"Cayley-Abels-Rosendal graphs give each boundedly generated group an explicit quasi-isometry type.","key_machinery":"The central object is the Cayley-Abels-Rosendal graph: a connected countable graph on which a Polish group acts continuously, vertex transitively, with coarsely bounded vertex stabilizers and finitely many edge orbits. The Milnor-Schwarz lemma is the engine: it turns such an action into bounded generation and a quasi-isometry between the group, with a word metric from a coarsely bounded generating set, and the graph. For the successor-ordinal case the load-bearing construction is the graph $\\Gamma(\\alpha,n)$ whose vertices are good partitions, meaning $n$ clopen pieces each containing exactly one maximal point, and whose edges are maximal shifts. For the limit-ordinal case the load-bearing object is a height function $h(P,Q)$ on pairs of good partitions, whose sublevel sets define the chain of proper open subgroups used to rule out bounded generation.","core_discovery":"The paper's central claim is that the coarse geometry of $\\operatorname{Homeo}(X_{\\alpha,n})$ is fully determined by the pair $(\\alpha,n)$. If $n=1$, the homeomorphism group is coarsely bounded. If $n>1$ and $\\alpha=\\beta+1$ is a successor ordinal, the group is boundedly generated but not coarsely bounded, and the paper constructs a Cayley-Abels-Rosendal graph whose vertices are the good partitions of $X_{\\alpha,n}$ and whose edges are maximal shifts, thereby identifying the group's quasi-isometry type with that graph. If $n>1$ and $\\alpha$ is a limit ordinal, the group is not boundedly generated; the witness is a countable chain of proper open subgroups that exhausts the group. The graph construction is new in every boundedly generated case.","pith_inferences":["Correcting the height function to the existential reading used in the later lemmas would repair the limit-ordinal argument; the strong triangle inequality is the single step to recheck.","The successor-ordinal graph $\\Gamma(\\alpha,n)$ carries a natural Hamming-like distance, the number of points of rank $\\beta$ that must be shifted between two good partitions; proving this distance is coarsely equivalent to graph distance would turn the quasi-isometry type into an explicit metric formula.","A testable next step is whether the limit-ordinal obstruction can be formulated purely in terms of the cofinality of the chain of open subgroups, which would make non-bounded-generation checkable without constructing a height function."],"forward_implications":["Every group $\\operatorname{Homeo}(X_{\\alpha,n})$ with $n>1$ and $\\alpha$ a successor ordinal is quasi-isometric to the explicit graph $\\Gamma(\\alpha,n)$, giving a concrete model of its large-scale geometry.","The limit-ordinal groups with $n>1$ admit no Cayley-Abels-Rosendal graph and no coarsely bounded generating set, so no word metric of the relevant kind exists for them.","Because bounded generation and coarse boundedness are preserved by continuous quotients, the trichotomy transfers to big mapping class groups whose end space is $X_{\\alpha,n}$.","For countable discrete groups, coarsely bounded subsets are exactly finite sets, so the new construction reduces to ordinary Cayley graphs and recovers the classical finite-generation picture."],"supporting_citations":[{"why":"It supplies the definition of coarsely bounded subsets, the criterion used to verify them, and the Milnor-Schwarz-type lemmas this paper expands.","marker":"[Ros22]"},{"why":"It gives the Cayley-Abels graph viewpoint for locally compact groups that the paper generalizes to Polish groups.","marker":"[Led22]"},{"why":"It provides the self-similar Stone-space coarse boundedness result and the big mapping class group comparison used in parts of Theorem A.","marker":"[MR23]"},{"why":"It classifies countable Stone spaces up to homeomorphism by the pair $(\\alpha,n)$, fixing the objects studied in Theorem A.","marker":"[MS20]"},{"why":"It supplies the Cantor-Bendixson derivative background and the stabilization theorem that defines rank $\\alpha$.","marker":"[Kec95]"},{"why":"It shows $\\operatorname{Homeo}(X)$ is the automorphism group of a countable graph and describes the clopen-partition basis of the compact-open topology.","marker":"[BL24]"},{"why":"It supplies the Pettis lemma that converts a non-meagre coarsely bounded generating set into a coarsely bounded identity neighborhood.","marker":"[Pet50]"}],"fun_headline_variants":["Coarse trichotomy for Stone homeomorphism groups","Three coarse types emerge for Stone homeo groups","Ordinal growth dictates Stone homeo group coarse type","Cayley-Abels-Rosendal graphs reveal three coarse classes","Stone Space homeo groups: bounded, boundedly generated, or neither"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The limit-ordinal half of the classification rests on the height function $h(P,Q)$ and its strong triangle inequality; as printed, $h$ is defined by a 'for all pieces' condition, under which the inequality's proof is invalid, so the chain of open subgroups built from $h$ may fail to exhaust the group.","fun_headline_variants_meta":{"raw":{"variants":["Coarse trichotomy for Stone homeomorphism groups","Three coarse types emerge for Stone homeo groups","Ordinal growth dictates Stone homeo group coarse type","Cayley-Abels-Rosendal graphs reveal three coarse classes","Stone Space homeo groups: bounded, boundedly generated, or neither"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000358,"raw_usage":{"total_tokens":1940,"prompt_tokens":947,"completion_tokens":993,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":563,"completion_tokens_details":{"reasoning_tokens":920}},"tokens_in":563,"tokens_out":993,"duration_ms":9946,"temperature":1.0,"reasoning_tokens":920,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-12T14:26:29.089048+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Compute the height function for a concrete triple of good partitions at a limit ordinal $\\alpha$ with $n\\ge 3$, where at some rank $\\beta$ the partition $P$ differs from $R$ only in one piece, $R$ differs from $Q$ only in a different piece, and $P$ differs from $Q$ in every piece. Under the definition as printed, this gives $h(P,Q)\\ge\\beta$ while $h(P,R)$ and $h(R,Q)$ are below $\\beta$, refuting the strong triangle inequality on which the exhausting-chain proof depends; the existence or nonexistence of such triples is a direct check using the clopen sets that define the partitions.","supporting_citations":[],"review_version":1}