{"id":"9ec047c8-3c91-44c8-8665-b2c45a96afb9","arxiv_id":"1909.00039","paper_version":4,"verdict":"ACCEPT","confidence":"HIGH","novelty_score":7.0,"correctness_risk":"low","formal_verification":"none","parameter_count":0,"one_line_summary":"For f(z)=z^2-1, the arboreal Galois group always lies in a new group M∞ and equals M∞ exactly when [K(√-x0,√(1+x0),ζ8):K]=16, which is true for infinitely many x0 over Q.","lead":"This paper computes the full Galois symmetry group of the iterated preimage tree of f(z)=z^2-1 for a large family of starting points and fields. It defines an explicit group, the arithmetic basilica group, and gives a simple degree-16 condition that guarantees the Galois group equals it.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Theorem 5.1's exact sequence (28) appears to identify the geometric group Gal(K'_∞/K') with a subgroup of G∞; this identification is not literally a subgroup inclusion, and if Pink's theorem does not supply the needed compatibility, the Main Theorem's bridge to M∞ is unproven.","rationale":"The paper's internal algebra is largely solid: the labeling construction in Lemma 1.4, the identification of P(σ,x) with the cyclotomic action in Theorem 2.2, the proof that M∞ is a group in Theorem 3.2, the order computations in Theorems 4.5 and 4.6, and the Frattini argument in Theorem 5.5 are coherent when read carefully. The reader's weakest-assumption diagnosis points to Pink's external theorems, and I agree that this is the correct locus. However, the concern is more specific than a generic reliance on cited results: as written, the exact sequence (28) in Theorem 5.1 is not literally a sequence of subgroups of G∞, since Gal(K'_∞/K') is the geometric Galois group over \\bar{k}(t), not a subgroup of Gal(K∞/K). The proof needs a supplied compatible identification between the kernel of the cyclotomic character and B∞, and between that kernel and the kernel of P under the embedding G∞→M∞. This is likely fixable from Pink's theorems, and the computational evidence for small n is reassuring, but the paper as written leaves the verification implicit. Therefore the appropriate verdict is conditional acceptance: the central claim is credible and well-motivated, but the Pink bridge should be checked and the notation in (28) corrected before the proof is fully rigorous.","tokens_in":21880,"tokens_out":38847,"duration_ms":354396,"concrete_test":"Independently re-derive the exact sequence in Theorem 5.1 from Pink's Theorems 2.8.2 and 2.8.4: write the kernel of the cyclotomic character as Gal(K∞/K(ζ_2^∞)) (or equivalently as B∞), and verify that the embedding G∞→M∞ restricts to an isomorphism from that kernel onto ker P = B∞, with the second square of the diagram in Theorem 5.1 commuting. If this derivation succeeds, the notation in (28) is harmless; if it cannot be carried out, Theorem 5.1 is not established. A computational cross-check is to compute |G5| for K=Q(t) via the Galois group of f^5(z)-t over Q(t) and compare with |M5| = 2^25 from Theorem 4.6.","verdict_should_be":"CONDITIONAL","load_bearing_attack":"The Main Theorem's equivalence (1)⇔(5) and the Frattini upgrade in Theorem 5.5 rest on Theorem 5.1, whose proof uses Pink's Theorems 2.8.2 and 2.8.4. In that proof the paper defines K' = \\bar{k}(t) and K'_∞ = \\bigcup_n K'(f^{-n}(t)), then writes the short exact sequence (28) as 0 → Gal(K'_∞/K') → G∞ → Z_2^× → 0. But G∞ = Gal(K∞/K) with K = k(t), and K∞ contains k(t, preimages), not \\bar{k}; hence K'_∞ is not a subfield of K∞ and Gal(K'_∞/K') is not naturally a subgroup of G∞. What Pink's Theorem 2.8.4 can supply is an extension 0 → B∞ → G∞ → Z_2^× → 0, where B∞ is the closed basilica group; the identification of B∞ with Gal(K'_∞/K') is an abstract isomorphism from Pink 2.8.2, not an inclusion. The five-lemma diagram then requires a compatible map from the kernel of the cyclotomic character to the kernel of P, and the paper only asserts that the embedding from Corollary 3.3 restricts to an isomorphism. If this compatibility is not justified, Theorem 5.1 fails, and with it the Frattini computation in Theorem 5.3 and the equivalence (a)⇔(c) of the Main Theorem. The internal construction of M∞ in Sections 2–4 is coherent; the load-bearing point is this external bridge.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper defines a subgroup M∞ of the automorphism group of the infinite binary rooted tree, called the arithmetic basilica group, using a 2-adic quantity P(σ,x) built from the parities of a tree automorphism at certain nodes. For the postcritically finite quadratic polynomial f(z)=z^2−1, the authors prove that for any field K of characteristic not 2 and any root point x0∈K with x0≠0,−1, the arboreal Galois group G∞ embeds equivariantly into M∞. Their Main Theorem then asserts that G∞≅M∞ is equivalent both to the fifth-level condition G5≅M5 and to the explicit degree condition [K(√−x0,√1+x0,ζ8):K]=16. The proof strategy is to establish the isomorphism for the generic function field K=k(t) via two theorems of Pink, compute the Frattini subgroup of M∞ using the special case K=Q(t), and then use a Frattini argument to lift the isomorphism from a finite-level condition to the full pro-2 group. The paper also proves that for K=Q the degree condition holds for infinitely many x0 and formulates two broader conjectures about arboreal Galois groups of PCF maps.","tokens_in":22236,"tokens_out":28777,"duration_ms":254263,"significance":"If the main theorem is correct, it provides one of the first explicit, non-full arboreal Galois groups for a postcritically finite quadratic polynomial over number fields, together with a computationally checkable criterion for the full arithmetic basilica group to occur. The internal machinery in Sections 2–4 is a genuine contribution: the quantity P(σ,x) is defined combinatorially on Aut(T∞) without reference to Galois groups, the proof that M∞ is a subgroup and that P is a homomorphism is detailed, and the order computations for the finite quotients Mn, Bn, and En are explicit and agree with the table. The resulting condition (2c) is a clean, falsifiable prediction for infinitely many rational x0, and the paper appropriately credits Pink’s theorems as the external input connecting the combinatorial group to actual Galois groups. The main weakness is not the internal construction but the terseness and, in one place, an inclusion-direction error in the proof of the Frattini computation, which needs correction before the central equivalence can be regarded as fully proved.","major_comments":[{"comment":"The proof of Theorem 5.1 states that restricting the injection G∞→M∞ yields an isomorphism Gal(K′∞/K′)≅B∞, and that the first square of the diagram commutes because this isomorphism is the restriction of the injection. As written, Gal(K′∞/K′) is not literally a subgroup of G∞: K′∞ = \\bar{k}(t)(f^{-n}(t) for all n) is not a subfield of K∞, and the natural restriction map from Gal(\\bar{k}K∞/\\bar{k}K) to Gal(K∞/K) must be introduced to realize the geometric group as a subgroup of G∞. The paper should spell out this standard identification and verify explicitly that Pink’s Theorems 2.8.2 and 2.8.4 supply the compatibility of the left square of the diagram. Without that compatibility, the five-lemma conclusion that G∞→M∞ is an isomorphism does not follow from the two exact sequences.","section":"Section 5, Theorem 5.1, Eq. (28)"},{"comment":"The sentence \"By the claim of Step 1, then, we have K^H_∞⊆L. Thus, the Frattini subgroup of G∞ is contained in Ψ := Gal(K∞/L)\" has the inclusion reversed. From K^H⊆L it follows that H⊇Ψ for every maximal subgroup H, so the Frattini subgroup Φ contains Ψ rather than being contained in Ψ. The subsequent \"Conversely\" paragraph only proves Φ⊆Ψ, since Ψ is the intersection of the four displayed maximal subgroups. As printed, the proof of Theorem 5.3 establishes at most one inclusion and does not prove equality. The intended argument is clear, but the inclusion direction and the two inclusions needed for equality should be stated correctly.","section":"Theorem 5.3, Step 2"}],"minor_comments":[{"comment":"The stated value Δ2 = −64(1+x0)^2 x0 is off by a factor of 4; a direct discriminant computation for f2(z)−x0 = z4−2z2−x0 gives Δ2 = −256 x0(1+x0)^2. Since the square root of Δ2 is changed only by an element of K, the fields L and the parity arguments are unaffected, but the displayed formula should be corrected.","section":"Eq. (29) and Theorem 5.5"},{"comment":"In the sentence describing β, the text says \"again satisfies P(α)≡1 (mod 8)\"; this should read P(β)≡1 (mod 8).","section":"Theorem 5.5, Step 3"},{"comment":"Even after the identification of the geometric group with a subgroup of G∞ is made explicit, the map ρ in sequence (28) should be defined as the cyclotomic character on 2-power roots of unity, and the hypothesis [k(ζ8):k]=4 should be used to justify that the image of the cyclotomic character is all of Z2^× rather than merely a subgroup.","section":"Theorem 5.1"}],"recommendation":"major_revision","confidential_remarks":"The paper’s central equivalence rests on theorems from Pink’s preprint [25]. The referee did not independently verify the exact formulation of Pink’s Theorems 2.8.2 and 2.8.4; it would be prudent for the handling editor to confirm that the statements used here match Pink’s hypotheses, particularly the identification of the kernel of the cyclotomic character with the geometric Galois group. The internal Sections 2–4 appear sound and are the most original part of the work."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"This paper computes the arboreal Galois group of f(z)=z^2-1 over number fields, under a simple split condition, and names the resulting group M∞, the arithmetic basilica group. That is a real new result: Pink had the function-field case, and Benedetto et al. had a cubic analogue, but this is the first complete quadratic PCF arboreal group over number fields for infinitely many starting points.\n\nThe construction is the best part. The function P(σ,x) on Aut(T∞) is a tidy combinatorial encoding of the cyclotomic action; Theorem 2.2, which proves P matches the Galois action on 2-power roots of unity, is argued in detail. The Frattini upgrade—degree-16 field condition iff full group at level 5 iff full group at infinity—is elegant, and the order computations for Mn, Bn, and En line up. The paper is also careful about what it does not claim: it states the finite-index conjecture and notes that periodic x0 must be excluded.\n\nThe soft spots are modest. The proof of Theorem 5.1 is compressed at the point where the abstract group M∞ is matched with Pink's function-field results. The exact sequence (28) writes Gal(K'_∞/K') as a subgroup of G∞; literally, K'_∞ is not a subfield of K∞, so the correct statement is that Gal(K'_∞/K') is identified with the kernel of the cyclotomic character via Galois correspondence. Pink's Theorems 2.8.2 and 2.8.4 are what make that identification compatible with the tree action. The authors cite those results rather than reproving them, which is normal, but they could have added a sentence explaining the identifications. The stress-test worry about this point seems to me closer to a notation gap than a real flaw—unless Pink's preprint has hidden hypotheses, which the paper has no way to rule out. The internal field classification in Theorem 5.3 (the √a lemma) is also a bit dense, but the discriminant formula checks out.\n\nWho is this for? Arithmetic dynamicists, anyone working on arboreal Galois representations, and people who like pro-2 group structures. The paper deserves a serious referee: the main theorem is likely correct, the construction is reusable, and the write-up is mostly clear. I would send it to review and ask for a clarifying paragraph around Theorem 5.1's diagram.","headline":"A genuinely new computation of an arboreal Galois group for z^2-1, with a clean Frattini criterion; the only real soft spot is the compressed use of Pink's theorems in Theorem 5.1.","tokens_in":22784,"tokens_out":7788,"would_cite":true,"duration_ms":70758,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["37P05","11R32","14G25"],"pacs":[],"model":"deepseek-v4-flash","headline":"For f(z)=z^2−1, the arboreal Galois group G∞ always lies in an explicitly defined arithmetic basilica group M∞, and it equals M∞ exactly when a one-shot degree-16 field condition holds.","keywords":["arboreal Galois group","postcritically finite polynomial","basilica group","arithmetic basilica group","2-adic cyclotomic character","iterated monodromy group","preimage tree"],"falsifier":"Take K=Q and x0=1. Since [Q(√−1,√2,ζ8):Q]=4, the theorem predicts G∞ is not isomorphic to M∞ and in particular |G5|<$2^{25}$; computing the degree of the splitting field of $f^{5}$(z)−1 over Q would directly test this, and if that degree equals $2^{25}$, the main equivalence is false.","tokens_in":21685,"feed_emoji":"🌳","tokens_out":11652,"duration_ms":106754,"temperature":0.7,"pith_summary":"The paper establishes a complete Galois description for the iterated preimage tree of the quadratic map f(z)=$z^{2}$−1. It defines a group M∞, the arithmetic basilica group, which acts on the infinite binary tree and is an extension of the basilica group by Z_2^×. The main theorem shows that over any field K of characteristic different from 2, with root x0 not 0 or −1, the arboreal Galois group G∞ embeds into M∞, and equality holds exactly when a depth-five check or an equivalent degree-16 field condition holds. Because that condition holds for infinitely many rational x0, the full arithmetic basilica group is genuinely realized over Q. This matters because it gives a postcritically finite map whose arboreal Galois group is understood at every level, with the obstruction to fullness reduced to a single easy field extension.","feed_headline":"One degree-16 test gives the whole Galois group for z^2 - 1","feed_subtitle":"A degree-16 field extension over K decides whether the full arithmetic basilica group is the Galois group at all depths.","key_machinery":"The central object is a function P(σ,x), taking values in the 2-adic units Z_2^×, attached to an automorphism σ of the binary tree and a node x. It is defined as (−1) to the parity of σ at x plus a weighted alternating sum of parities at selected higher nodes, and it computes the exact power to which σ sends primitive 2-power roots of unity: Theorem 2.2 proves σ(ζ)=$ζ^{{P(σ,y)}}$ for every Galois element and every node y. The arithmetic basilica group M∞ is then the subgroup of all tree automorphisms for which P(σ,x) is independent of x; the closed basilica group B∞, generated by the two recursively defined automorphisms α and β, is the kernel of P. The exact sequence and a Frattini-subgroup computation at level five carry the proof of the main equivalence.","core_discovery":"The central claim is the Main Theorem. Let K be a field of characteristic different from 2 and let x0∈K be nonzero and not −1. For f(z)=$z^{2}$−1, let G∞ be the Galois group of the union of all fields generated by iterated preimages of x0. Then G∞ is equivariantly isomorphic to a subgroup of M∞, and the following are equivalent: G∞≅M∞; the depth-five quotient G5 is isomorphic to M5; and the extension K(√−x0,√1+x0,ζ8) has degree 16 over K. The paper also shows M∞ sits in the exact sequence 0→B∞→M∞→Z_2^×→0 with B∞ the closed basilica group, and that for K=Q infinitely many x0 achieve the full group; the stated examples include x0 and −1−x0 in {5,6,10,11,12,13,14,19,...}.","pith_inferences":["By the same mechanism, the depth-5 threshold is probably not special to z^2−1: the Frattini index 16 arises from the first two discriminants and ζ8, so for other PCF quadratics one should expect a finite depth, governed by the critical orbit and ramification in K∞, at which equality of finite-level groups forces equality at all levels.","The function P(σ,x) can be read as a 2-adic cyclotomic character computed from weighted parities; the same construction should produce an arithmetic iterated monodromy group for any PCF quadratic, giving a uniform family of model groups into which every specialization embeds.","Because the strong equality condition has degree 16, almost all rational x0 fail it; the paper's Conjecture 1 would then mean failure is only a finite-index defect. A natural quantitative check is to count x0 in growing boxes with [Q(√−x0,√1+x0,ζ8):Q]=16 and compare the density with the thin-set prediction."],"forward_implications":["Whenever [K(√−x0,√1+x0,ζ8):K]=16, the full arithmetic basilica group occurs, and over K=Q this happens for infinitely many x0, including 5, 6, 10, 11, 12, 13, 14, 19, and their counterparts −1−x0.","Equality at depth five forces equality at every depth: G5≅M5 implies G∞≅M∞, so a finite computation certifies the entire infinite Galois group.","For every allowed K and x0, G∞ is confined to M∞, which is an extension of the closed basilica group by Z_2^×; the 2-adic cyclotomic character is the only arithmetic information beyond the basilica action.","The criterion is algorithmic: test the degree of the extension K(√−x0,√1+x0,ζ8), and if it is 16, the Galois group at level n has order 2^{m_n} with the explicit value from Theorem 4.6."],"supporting_citations":[{"why":"Supplies the function-field exact sequence 0→B∞→G∞→Z_2^×→0 and the identification of the kernel with the closed basilica group; Theorem 5.1 rests on it.","marker":"[25]"},{"why":"Defines the basilica group as a self-similar group and provides the properties of B∞ used in Section 4.","marker":"[23]"},{"why":"Gives the basilica group as the iterated monodromy group of z^2−1 over C(t), grounding the comparison with G∞.","marker":"[3]"},{"why":"Provides the iterative discriminant formula used to identify the Frattini subgroup in Theorem 5.3.","marker":"[5]"},{"why":"Supplies the earlier cubic PCF analogue and the E∞ model whose quadratic counterpart this paper constructs.","marker":"[4]"}],"fun_headline_variants":["Degree-16 test pins down full arithmetic basilica group","Quadratic PCF arboreal Galois group: one degree-16 check decides","Arithmetic basilica: degree-16 condition yields full Galois group","Full Galois group for z^2-1 decided by degree-16 field extension"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The argument depends on a cited function-field theorem that identifies the generic Galois group as an extension of the closed basilica group by the 2-adic units; the paper does not prove that theorem, and if it fails the main equivalence is unsupported.","fun_headline_variants_meta":{"raw":{"variants":["Degree-16 test pins down full arithmetic basilica group","Quadratic PCF arboreal Galois group: one degree-16 check decides","Arithmetic basilica: degree-16 condition yields full Galois group","Full Galois group for z^2-1 decided by degree-16 field extension"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000613,"raw_usage":{"total_tokens":2823,"prompt_tokens":889,"completion_tokens":1934,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":505,"completion_tokens_details":{"reasoning_tokens":1853}},"tokens_in":505,"tokens_out":1934,"duration_ms":12531,"temperature":1.0,"reasoning_tokens":1853,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T10:05:41.199251+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Take K=Q and x0=1. Since [Q(√−1,√2,ζ8):Q]=4, the theorem predicts G∞ is not isomorphic to M∞ and in particular |G5|<$2^{25}$; computing the degree of the splitting field of $f^{5}$(z)−1 over Q would directly test this, and if that degree equals $2^{25}$, the main equivalence is false.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the function-field exact sequence 0→B∞→G∞→Z_2^×→0 and the identification of the kernel with the closed basilica group; Theorem 5.1 rests on it."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Defines the basilica group as a self-similar group and provides the properties of B∞ used in Section 4."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Gives the basilica group as the iterated monodromy group of z^2−1 over C(t), grounding the comparison with G∞."},{"cited_title":"Benedetto and Jamie Juul, Odoni’s conjecture for num ber fields, Bull","cited_arxiv_id":null,"evidence_quote":"Provides the iterative discriminant formula used to identify the Frattini subgroup in Theorem 5.3."},{"cited_title":"Benedetto, Xander Faber, Benjamin Hutz, Jamie Juu l, and Yu Yasufuku, A large ar- boreal Galois representation for a cubic postcritically finite polynom ial, Res","cited_arxiv_id":null,"evidence_quote":"Supplies the earlier cubic PCF analogue and the E∞ model whose quadratic counterpart this paper constructs."}],"review_version":1}