{"id":"7baed399-a322-4b2e-a5cf-37ce92377c00","arxiv_id":"2607.13776","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":8.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"The Sierpiński carpet group is the first infinite finitely generated amenable group with Property FW; more generally, FW for contracting branch groups is equivalent to being just-infinite and having a no-local-cut-point limit space.","lead":"The paper classifies all commensurating actions of contracting self-similar branch groups and proves that the Sierpiński carpet group—an infinite, finitely generated, amenable group—has Property FW, a strong fixed-point property for cube complexes. This is the first known example of its kind, settling a question of Cornulier.","discovery_kind":"unification","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The Section 6 application hinges on asserted finite computations (Props 6.2 and 6.4) that are not exhibited; if any fails, the Sierpiński carpet group is not a contracting branch just-infinite group and the FW conclusion does not follow.","rationale":"The reader's weakest assumption identifies exactly the place where the main theorem's hypotheses are least secure. The proof chain for Theorem 6.5 is otherwise coherent: the geometric description of the limit space via a quotient of C×G is argued with tile arguments and is plausible; the amenability is imported from the published [MBNZ25]; and the abstract-versus-content consistency is good. The finite checks in Props 6.2 and 6.4, however, are not exhibited, leaving the possibility that the Sierpiński carpet group G might not be a contracting branch just-infinite group. If that happened, Corollary C would be inapplicable, and the FW conclusion (the paper's headline application) would not follow. This is not a circularity or an appeal to consensus; it is an internal verification gap that can be closed by computation. The same concern applies to the generation of Alt(X) from four Alt(5)'s, which is a concrete finite fact. The suggested test directly checks these assertions and, if passed, would raise confidence substantially. Therefore the appropriate verdict remains CONDITIONAL, and no adjustment to the reader's assessment is needed.","tokens_in":38142,"tokens_out":16878,"duration_ms":155032,"concrete_test":"Run an independent computer verification, e.g., in GAP: (1) Construct the group defined by the wreath recursion on the free product of four involutions with relators (ab)^6, (bc)^6, (cd)^6, (da)^6, and compute the nucleus by iterating sections until stabilization; verify it contains exactly 41 elements and is contained in <a,b>∪<b,c>∪<c,d>∪<d,a>. (2) Verify the faithfulness condition: check that every non-trivial nucleus element acts nontrivially on some vertex of the tree. (3) Verify that the commutator subgroup of <(ab)^2,(da)^2> is the rooted group Alt({1,2,4,6,7}), and similarly for the other three pairs, and confirm these groups generate Alt(8). (4) Check RiSt(1)=G^X by verifying that for each generator g, the element (g,1,...,1) lies in the subgroup generated by the first-level sections. If all checks pass, the conditional concern is resolved.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central application (Theorem 6.5) depends on Corollary C, which requires G to be a finitely generated contracting self-replicating branch group, and on the additional fact that the limit G-space has no local cut points. The latter is argued geometrically with a plausible tiling construction, but the former rests on two unshown finite checks. Proposition 6.2 asserts that the wreath recursion defines a contracting self-similar group with a specific 41-element nucleus, saying 'It is checked directly...' without presenting the check. Proposition 6.4 asserts branchness and just-infiniteness: it claims Alt(X) < G and RiSt(1) = G^X, using further unshown claims that four specific Alt(5) subgroups are generated by commutators of elements like (ab)^2 and (da)^2, and that 'the same arguments' produce the other three Alt(5)'s. If any of these checks fail—for instance, if the nucleus contains an element that becomes trivial in the faithful quotient, or if the four Alt(5)'s do not generate Alt(X), or if RiSt(1) is a proper subgroup of G^X—then G is not a contracting branch just-infinite group, Corollary C cannot be applied, and the FW conclusion is unsupported. The paper's own AI-referee statement mentions 'minor inaccuracies' without listing them, heightening the need for explicit verification of these computational assertions.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper develops a complete theory of commensurating actions for finitely generated contracting self-similar branch groups acting on rooted trees. Theorem 4.8 (Theorem A) shows that in a finitely generated branch group with finite germ groups, every multi-ended Schreier graph is a finite cover of a graph of germs, and that Property FW is equivalent to just-infiniteness together with one-endedness of all graphs of germs. Theorem 5.15 (Theorem B) identifies the supremum of the number of ends of graphs of germs with the supremum of local degrees of points in the limit G-space for contracting self-replicating groups. Corollary C combines these into a criterion for Property FW. The paper then applies this criterion to the Sierpiński carpet group, proving that it is amenable (using the authors' earlier work [MBNZ25]) and has Property FW, which would be the first infinite finitely generated amenable group with FW and would answer a question of Cornulier. It also proves that a contracting self-replicating regular branch group does not have Property PW, and applies this to the Grigorchuk group.","tokens_in":38424,"tokens_out":17841,"duration_ms":175957,"significance":"If correct, the main results are substantial: they give a complete characterization of Property FW for a major class of groups acting on rooted trees, connect it to a geometric no-local-cut-point condition on limit spaces, and provide a long-sought example of an infinite finitely generated amenable group with Property FW. The structural theorems are proved in detail and the paper gives a coherent, self-contained framework. The main weakness is that the flagship application in Section 6 rests on several finite computations that are asserted rather than exhibited. The theorems themselves appear well supported; the computational gap is localized but load-bearing for the headline example.","major_comments":[{"comment":"The proposition asserts that the wreath recursion defines a contracting self-similar group whose nucleus is the 41-element set N = <a,b> ∪ <b,c> ∪ <c,d> ∪ <d,a>. The displayed computations only verify that the relators of H are respected by the recursion. The sentence 'It is checked directly, that for every triple h1,h2,h3 ... (⟨h1,h2⟩·h3)|x ⊂ N' is exactly the step that establishes contraction, but no verification is shown. Likewise, 'It is also easy to check that every element of N is a section of an element of N' is asserted. These checks are finite but essential: Theorem 6.5 applies Corollary C, which requires G to be contracting. Please supply a full verification, e.g. a table of sections of elements of N and a recursive argument showing that all sufficiently deep sections of arbitrary words lie in N.","section":"§6, Proposition 6.2"},{"comment":"The proof that G is branch and just-infinite depends on several unshown algebraic identities. In particular, the claim that L = <(ab)^2, (da)^2> has commutator subgroup equal to the rooted group Alt({1,2,4,6,7}) is not immediate: the two generators have nontrivial self-similar sections at positions 3 and 8, so one needs to compute the commutator structure, not only the projected permutation group. The 'same arguments' for the other three Alt(5) subgroups, and the 'Similar arguments' producing (b,1,...,1), (c,1,...,1), (d,1,...,1), are likewise not shown. These identities imply RiSt(1) = G^X and hence branchness and just-infiniteness; without them Corollary C cannot be applied. Please provide the explicit computations or a verifiable computer-assisted check.","section":"§6, Proposition 6.4"},{"comment":"The geometric part of the proof that X_G has no local cut points is asserted rather than demonstrated. The text says 'It follows that every point of X has a neighborhood homeomorphic to the Sierpiński carpet', but at points on gluing edges and at corners with dihedral isotropy one needs a precise local model; the figure is illustrative but does not replace a formal verification. Since 'no local cut point' is one of the two hypotheses of Corollary C, this step is load-bearing. Additionally, the identification of the quotient space X with the limit G-space X_G via the G-equivariant homeomorphism X⊗B ≅ X and the claimed contraction factor 1/3 are described only informally; these need to be checked against Theorem 5.9.","section":"§6, Theorem 6.5 proof"}],"minor_comments":[{"comment":"The sentence following the proposition states that all non-trivial elements of the nucleus of H remain non-trivial in G, so the nucleus of G also has 41 elements. This is not shown explicitly and should be justified.","section":"§6, Proposition 6.2"},{"comment":"The statement says that an AI-generated referee report helped find 'some typos and minor inaccuracies (none affecting the core validity of proof)', but the inaccuracies are not listed. For transparency, please list them.","section":"Statement on AI use"},{"comment":"The arrow for virtual homomorphisms appears as a corrupted LaTeX token ('/axisshort/axisshort/arrowaxisrightZ') in several places. This should be typeset correctly.","section":"§4.2 and Corollary 4.14"},{"comment":"Figure 3 is referenced but the manuscript text does not display the actual local neighborhood picture. If the figure is included, its labels should be explained in the caption.","section":"§6, Figure 3"}],"recommendation":"major_revision","confidential_remarks":"The structural results (Theorems 4.8, 5.15, Corollary C, Corollary D) appear sound and well argued. The main barrier is verifiability of Section 6: the Sierpiński carpet application depends on finite but nontrivial computations that are only asserted, and the geometric no-local-cut-point argument is also terse. These gaps are likely fixable, but they are not merely cosmetic. I would not reject on present evidence, but I would require the authors to supply the missing verifications before publication."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Short version: this is a strong paper that answers Cornulier's question — first infinite finitely generated amenable group with Property FW — and it does so with a genuinely useful structural theory, not just a one-off example. Theorems A and B (with Corollaries C and D) give a complete description of commensurating actions for contracting self-similar branch groups, and the duality between ends of germ graphs and local cut points in the limit G-space is a nice idea that should be reusable. Theorem 3.1 on direct products is independently interesting. The writing is careful, and the main proofs are supplied in full.\n\nThe soft spot is exactly where the reader puts it: Section 6. The application to the Sierpiński carpet group depends on Propositions 6.2 and 6.4, and those rest on 'it is checked directly' and 'the same arguments show' for load-bearing finite facts: the 41-element nucleus, non-triviality of nucleus elements in the faithful quotient, generation of Alt(X) by four specified Alt(5)'s, and RiSt(1) = G^X. None of these computations are exhibited. They are finite and checkable, but without them Corollary C cannot be applied and the FW conclusion does not follow. I would not call this a flaw in the central argument — the structural theorems stand on their own — but it is a real gap in the flagship application.\n\nThe paper's own AI statement says an AI referee found 'minor inaccuracies' without listing them. In ordinary circumstances that would be a small thing, but combined with the unshown computations, it strengthens the case for asking the authors to make Section 6 fully verifiable.\n\nI disagree mildly with the stress-test's framing that the nucleus-element-in-faithful-quotient worry is unaddressed: the text explicitly notes this check, and it is part of what needs verifying. The geometric no-local-cut-point argument for the limit space is plausible and clearly explained. So the issue is not the architecture of the proof; it is the omission of finite verifications.\n\nRecommendation: send it to a serious referee. The result is important enough, and the structural part appears sound. In review, the main demand should be explicit Section 6 computations — in the text or as a code-certified appendix — before the application is accepted as fully verified.","headline":"First amenable FW group, built on a genuinely useful structural theory, but the Section 6 application rests on unshown finite computations that should be supplied before the flagship example is taken as fully verified.","tokens_in":38989,"tokens_out":2264,"would_cite":true,"duration_ms":29850,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["20E08","20F65"],"pacs":[],"model":"deepseek-v4-flash","headline":"The Sierpiński carpet group is the first infinite finitely generated amenable group with Property FW.","keywords":["commensurating actions","Property FW","Property PW","self-similar groups","branch groups","Sierpiński carpet group","graphs of germs","limit space"],"falsifier":"A concrete observation: if one could exhibit a non-trivial commensurated subset $A$ of a transitive $G$-set for the Sierpiński carpet group $G$ that is not transfixed, or a proper cardinal-definite function on $G$, the main theorem would be false. More directly, a computer search could verify the asserted contraction check: for every triple of pairwise distinct generators $h_1,h_2,h_3$, the sections $(\\langle h_1,h_2 \\rangle \\cdot h_3)|_x$ must lie in the 41-element set $N$, and one must check that $\\mathrm{RiSt}(1)=G^X$ and that $\\mathrm{RiSt}(v)'$ has finite index in $\\mathrm{RiSt}(v)$ for every vertex $v$. If any of these checks fails, $G$ is not a contracting branch j","tokens_in":37965,"feed_emoji":"🧩","tokens_out":4795,"duration_ms":72625,"temperature":0.7,"texified_at":"2026-08-05T21:21:34.502319+00:00","pith_summary":"This paper claims a complete description of commensurating actions for contracting self-similar branch groups: every multi-ended Schreier graph either comes from a virtually abelian quotient or is a finite cover of a graph of germs at a boundary point, and the number of ends is governed by the local cut-point degree of the limit space. The punchline is the Sierpiński carpet group: an explicit group acting on an 8-ary rooted tree, which is shown to be amenable, just-infinite, branch, and to have a limit space tiled by copies of the Sierpiński carpet, hence no local cut points. Therefore it has Property FW, giving the first infinite finitely generated amenable group with Property FW and answering a long-standing question in the literature. The same machinery shows that a contracting self-replicating regular branch group can never have Property PW; in particular the Grigorchuk group does not have Property PW.","texify_model":"deepseek-v4-flash","texify_usage":{"total_tokens":5310,"prompt_tokens":905,"completion_tokens":4405,"prompt_tokens_details":{"cached_tokens":0},"prompt_cache_hit_tokens":0,"prompt_cache_miss_tokens":905,"completion_tokens_details":{"reasoning_tokens":3520}},"feed_headline":"First amenable group with Property FW found","feed_subtitle":"A concrete group built from the Sierpiński carpet fixes every action on a CAT(0) cube complex, answering an open question.","key_machinery":"The load-bearing objects are the graph of germs $\\tilde{\\Gamma}_\\xi$ at a boundary point $\\xi$, defined as the Schreier graph of the coset space $G/G^0_\\xi$ where $G^0_\\xi$ is the germ stabilizer (elements fixing a neighborhood of $\\xi$ pointwise), and the limit $G$-space $X_G$, a locally compact space with proper co-compact right $G$-action that uniformizes the limit space $J_G$. Theorem 5.15 establishes the equality of three numbers: the supremum of ends of graphs of germs, the supremum of $|\\pi_0(\\tilde{L}_\\xi \\setminus \\{\\zeta\\})|$ over germ-leaves, and the supremum of local degrees of points of $X_G$. The proof transfers paths between the graphs $\\Xi_n$ and the adjacency graphs of tiles of $X_G$, using the contraction property through the nucleus and Lemma 5.6.","core_discovery":"The central discovery is that for a finitely generated contracting self-replicating branch group $G$ acting on a rooted tree, the existence of non-trivial commensurating actions—equivalently, of multi-ended Schreier graphs—is completely controlled by the local topology of the limit $G$-space $X_G$. Theorem A shows that any faithful transitive action of a branch group with finite groups of germs that has more than one end must have point stabilizer commensurable with the stabilizer of a boundary point, and its number of ends is bounded above by that of the graph of germs at that point. Theorem B equates the supremum of the numbers of ends of graphs of germs with the supremum of local degrees of $X_G$.","pith_inferences":["The criterion suggests a general strategy for finding more amenable groups with Property FW: take a contracting self-replicating group whose limit space is homeomorphic to a Sierpiński carpet (or at least has no local cut points) and then verify, by whatever means, that the group is branch and just-infinite; the present paper does this for one example, but many more may be constructible from ratio","The equality between numbers of ends of graphs of germs and local degrees of the limit space gives a computational route: one could approximate or compute the local degree of X_G directly from the nucleus of the group, potentially yielding a machine-checkable test for Property FW in other examples.","The non-PW result for regular branch groups indicates that the gap between the Haagerup property and Property PW is particularly wide inside the class of contracting self-similar groups; this may inspire a search for other analytic properties (e.g. rapid decay or weak amenability) that are controlled by similar local-topological invariants of the limit space.","The paper leaves open a conceptual—rather than computational—understanding of when a contracting self-similar group is branch and just-infinite in terms of its limit space; if such a description were found, Corollary C would become a purely topological characterization of Property FW for the entire class."],"forward_implications":["The Sierpiński carpet group is the first infinite finitely generated amenable group with Property FW, resolving an open question about whether amenability is compatible with the strongest cubical fixed-point property.","For any finitely generated contracting self-replicating branch group, Property FW is now characterized: it holds exactly when the group is just-infinite and its limit G-space has no local cut point.","No contracting self-replicating regular branch group has Property PW; in particular, the Grigorchuk group does not admit a proper commensurating action, answering a question in the literature.","For iterated monodromy groups of post-critically finite rational maps, the orbital graphs and graphs of germs are one-ended exactly when the Julia set is the whole sphere or a Sierpiński carpet (Corollary 5.22).","Cardinal-definite functions on such groups have a rigid form: they lie at bounded distance from sums of graph-of-germs functions and absolute values of virtual homomorphisms, and for regular branch groups they are dominated by the section word length, precluding properness."],"fun_headline_variants":["Sierpinski carpet yields first amenable group with Property FW","New group answers Cornulier's Property FW question","Amenable group with Property FW from a self-similar tree","Grigorchuk group fails Property PW, new theorem shows","Property FW for amenable groups: first example from a tree"],"cache_read_input_tokens":2304,"weakest_assumption_plain":"The entire application to the Sierpiński carpet group rests on the asserted direct computations that the group defined by the wreath recursion is contracting with a 41-element nucleus, branch, and just-infinite (Propositions 6.2 and 6.4 state these as 'checked directly' without exhibiting the full verification); if any of these checks fails, the group would not satisfy the hypotheses of Corollary C and the Property FW conclusion would not follow.","fun_headline_variants_meta":{"raw":{"variants":["Sierpinski carpet yields first amenable group with Property FW","New group answers Cornulier's Property FW question","Amenable group with Property FW from a self-similar tree","Grigorchuk group fails Property PW, new theorem shows","Property FW for amenable groups: first example from a tree"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000158,"raw_usage":{"total_tokens":1013,"prompt_tokens":648,"completion_tokens":365,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":392,"completion_tokens_details":{"reasoning_tokens":284}},"tokens_in":392,"tokens_out":365,"duration_ms":4985,"temperature":1.0,"reasoning_tokens":284,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-02T03:44:15.915054+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"A concrete observation: if one could exhibit a non-trivial commensurated subset $A$ of a transitive $G$-set for the Sierpiński carpet group $G$ that is not transfixed, or a proper cardinal-definite function on $G$, the main theorem would be false. More directly, a computer search could verify the asserted contraction check: for every triple of pairwise distinct generators $h_1,h_2,h_3$, the sections $(\\langle h_1,h_2 \\rangle \\cdot h_3)|_x$ must lie in the 41-element set $N$, and one must check that $\\mathrm{RiSt}(1)=G^X$ and that $\\mathrm{RiSt}(v)'$ has finite index in $\\mathrm{RiSt}(v)$ for every vertex $v$. If any of these checks fails, $G$ is not a contracting branch j","supporting_citations":[],"review_version":1}