{"id":"c3e280ad-c98a-4af2-9def-f94d0aec2856","arxiv_id":"1908.01982","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"high","formal_verification":"none","parameter_count":2,"one_line_summary":"For quadratic fields with Cl3(C9 x C9), Artin patterns determine the metabelianization and often the length of the 3-class field tower, but the claimed infinite families of tower groups are only verified up to a computational bound.","lead":"This paper uses computer data about ideal classes to identify the Galois groups of 3-class field towers for quadratic number fields with 3-class group C9 x C9. It extends a project begun by Arnold Scholz in 1930, but some claims about infinite families rest on an unproved conjecture.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Theorems 8.6 and 8.9 assert an infinite family of tower groups, but the family comes from Conjecture 6.1, verified only up to u=30; as stated, the main theorem exceeds what is proved.","rationale":"The reader correctly identifies the load-bearing point. My reading of the proof of Theorems 8.6 and 8.9 confirms the explicit reliance on Conjecture 6.1, so the assertion of an infinite family of possible tower groups is not currently a theorem. I do not see an internal inconsistency in the finite calculations themselves; the concern is the strength of the theorem statement. The paper's computational evidence is substantial and the finite identifications are plausible, so a conditional verdict remains appropriate. Restricting the infinite-family theorems to the verified range and adding an explicit completeness argument would make the core claims solid; as written, they exceed the verified evidence. I therefore agree with the reader's weakest assumption and recommend no change to the conditional verdict.","tokens_in":17329,"tokens_out":6807,"duration_ms":79509,"concrete_test":"Recompute the p-group generation tree for u = 31 (order 3^101, just beyond the 3^100 verification horizon) using an independent implementation, and compare the unique sigma-child/nuclear-rank statistics and the metabelianization M with the pattern required by Theorem 6.1; any deviation at the first unverified index would falsify the infinite-family assertion. A definitive settlement additionally requires an inductive proof of the periodicity, since no finite computation can establish \"infinitely many\".","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim under stress is the assertion in Theorems 8.6 and 8.9 that a quadratic field with Artin pattern (8.6) or (8.8) has a 3-class tower group G isomorphic to one of the infinitely many non-metabelian Schur sigma-groups T_{i,k} (i >= 1, 1 <= k <= 3). The only support for the infinite family is Theorem 6.1, and its proof is a Magma verification \"up to u = 30 and logarithmic order 100\"; the statement itself says \"For any upper bound u <= 30\". Conjecture 6.1 then explicitly asserts that the same periodic trifurcation remains true for u > 30. The simultaneous proof of Theorems 8.6, 8.8, and 8.9 states that the claims were \"verified ... as described in the proof of Theorem 6.1 and expressed in Conjecture 6.1.\" Thus the words \"infinitely many\" encode an unproved conjecture. If the trifurcation terminates or alters after u = 30, the list of candidate groups is finite or different, and the theorem statements fail as written. A related completeness gap is that pattern recognition over a finite descendant subtree does not by itself prove that no deeper Schur sigma-group with the same Artin pattern exists. Both issues are repairable by restricting the theorem to the verified range or by supplying an inductive proof of the periodicity.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper determines the isomorphism type of the Galois group G of the finite 3-class field tower of quadratic fields K with 3-class group Cl_3(K) isomorphic to C_9 x C_9. The method is pattern recognition: arithmetic Artin patterns (transfer kernels and transfer targets) are computed by Magma and matched by group-theoretic Artin patterns of candidate 3-groups generated from the root R = <243,2> using the p-group generation algorithm. The paper gives finite classifications for several two-stage towers (Theorems 8.1-8.5, 8.7, 10.1), introduces a notion of harmonically balanced capitulation, and states that certain Artin patterns force a tower of length at least 3 with Galois group isomorphic to one of an infinite family of non-metabelian Schur sigma-groups T_{i,k} (Theorems 8.6 and 8.9). A stage separation criterion using second-order abelian quotient invariants is proposed in Theorem 8.8.","tokens_in":17677,"tokens_out":8798,"duration_ms":86909,"significance":"If the main claims hold, the paper gives a fairly complete and explicit classification of 3-class tower groups for a natural family of quadratic fields, with concrete discriminant examples and falsifiable predictions. The finite classifications for two-stage towers are plausible and internally coherent, and the group-theoretic analysis of harmonically balanced capitulation is a useful contribution. The paper's concrete strengths are the explicit lists of small discriminants with their Artin patterns (Examples 8.1-8.5 and 10.1), the use of standard computational tools (Magma SmallGroups and p-group generation), and the clear separation of metabelian from non-metabelian cases via relation ranks. However, the central infinite-family assertions are not fully proved: they depend on an explicit conjecture about periodic trifurcations that has only been verified up to a finite bound, and the exhaustiveness of the finite descendant-tree searches is asserted rather than demonstrated.","major_comments":[{"comment":"Theorems 8.6 and 8.9 state that the tower group G is isomorphic to one of the infinitely many non-metabelian Schur sigma-groups T_{i,k} with i >= 1 and 1 <= k <= 3. The existence of T_{i,k} for all i is not proved: Theorem 6.1 is verified only for u <= 30 by a Magma computation, and Conjecture 6.1 explicitly asserts that the periodic trifurcation continues for u > 30. The simultaneous proof of Theorems 8.6, 8.8, and 8.9 says only that the claims were verified as described in the proof of Theorem 6.1 and expressed in Conjecture 6.1. Thus the words 'infinitely many' and the parameter range i >= 1 exceed the proven content. The statements should be restricted to the verified range i <= 30, or an inductive proof of the periodicity should be supplied, before the infinite-family claims are presented as theorems.","section":"Section 6, Theorem 6.1 and Conjecture 6.1; Section 8, Theorems 8.6, 8.8, 8.9"},{"comment":"The proofs of the finite classifications construct a descendant tree with 27,222 vertices and sift it by pattern recognition with respect to transfer targets and relation rank. No argument is given in the manuscript that every metabelian 3-group M arising from a quadratic field with the displayed Artin pattern must occur among the vertices of this finite tree. The related assertion in Section 9 that all quadratic fields with Cl_3(K) = C9 x C9 have higher 3-class groups in the descendant tree of R = <243,2> is stated without proof or reference. Without such an exhaustiveness statement, the conclusions of the form 'G is isomorphic to one of ...' are conditional on the completeness of the search, not merely on the correctness of individual group computations. The author should supply the missing structural theorem or state explicitly that the classification is restricted to the finite searched subtree.","section":"Section 8, proofs of Theorems 8.1-8.5; Section 9"},{"comment":"The stage separation criterion is stated as an 'if and only if', but its proof is the same simultaneous verification used for Theorems 8.6-8.9 and is therefore doubly conditional: on Conjecture 6.1 and on the exhaustiveness of the descendant-tree search. In particular, the claim that the second-order invariants tau_{2,1} = ([(511)^3,331]^12; [(322)^4,(331)^9]) force length at least 3 and one of the infinite family T_{i,k} is not unconditional as written. The theorem should be reformulated to distinguish the unconditional finite part from the parts that depend on Conjecture 6.1 and on the search bound.","section":"Section 8, Theorem 8.8"}],"minor_comments":[{"comment":"The paper does not include the Magma code, scripts, or output logs used for the computations in Theorem 6.1 and the descendant-tree searches. Given that the central claims rest on computational verification, making the code available as supplementary material would greatly improve verifiability.","section":"General reproducibility"},{"comment":"The notation tau_4 = (2+i, 2, 2) should be explained explicitly; in the logarithmic type-invariant convention the entry '2+i' is presumably a formal exponent, but the paper does not define this notation at the point of use.","section":"Theorems 8.6 and 8.9"},{"comment":"The DOI given for reference [17] is 10.4236/apm.2018.62008, but the bibliographic data report Adv. Pure Math. 6 (2016), No. 2; the DOI appears to be for a different volume or year and should be corrected.","section":"Reference [17]"},{"comment":"In the proof of Theorem 8.1, the three sibling groups that are said to have different second-layer Artin patterns are not identified by SmallGroups identifiers; giving their identifiers would make the verification easier.","section":"Proof of Theorem 8.1"}],"recommendation":"major_revision","confidential_remarks":"This is a single-author computational classification paper that relies heavily on the author's own earlier work and on extensive Magma computations that are not independently checkable from the manuscript. The main obstruction to acceptance is not the plausibility of the finite classifications but the fact that the headline infinite-family theorems depend on an explicitly unproved conjecture. I would recommend asking the author to restate those theorems in the verified range or prove the periodicity, and to provide the computational scripts or at least a precise algorithmic description of the search and its completeness conditions."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Daniel Mayer's paper is the first systematic attempt to determine 3-class tower groups for quadratic fields with Cl3 ≅ C9×C9, the case Scholz outlined in 1930. The finite classifications for two-stage towers—Theorems 8.1–8.5, 8.7, and 10.1—are concrete, internally consistent, and useful. The new notions are genuinely new: harmonically balanced capitulation gives a clean way to organize transfer kernels, and the periodic trifurcation in the descendant tree of ⟨243, 2⟩ is a striking structural observation even if its infinite version remains conjectural.\n\nThe soft spot is exactly the one flagged in the stress test. Theorems 8.6 and 8.9 state that a field with certain Artin patterns has tower group isomorphic to one of 'infinitely many' non-metabelian Schur σ-groups Ti,k with i ≥ 1. That infinite family is supplied by Theorem 6.1, which is proved only for u ≤ 30 via a Magma run; Conjecture 6.1 then asserts the rest. The simultaneous proof of 8.6/8.8/8.9 explicitly invokes Theorem 6.1 'and Conjecture 6.1.' So as stated, the 'infinitely many' exceeds what is proved. This is a load-bearing gap, not a cosmetic one.\n\nA related completeness issue: the pattern recognition sifts a finite descendant subtree. For the two-stage theorems the search depth is fixed and the result is exhaustive within that finite set, so those are fine. For the infinite-family claims, the search bound leaves open the possibility of a deeper σ-group with the same Artin pattern that the truncation never reached. This compounds the conjecture problem.\n\nBoth issues are repairable. Restrict Theorems 8.6 and 8.9 to i ≤ 30 or prove the periodicity; state the search bound explicitly; and ship the Magma code and computed data, which the arXiv version does not include. The author is honest about the verification limits, and the prior self-citations are to relevant work, so the citation pattern is not a real problem here.\n\nI would send this to a serious referee. The finite classifications deserve publication, and the conjectural infinite family is worth a clear conditional statement. The right outcome is a revision, not a desk reject. A reading group could profit from this as a case study in p-group descendant trees meeting class field theory, so I'd say maybe bring it.","headline":"The finite results are solid and useful; the 'infinitely many' claim relies on a conjecture, but the paper deserves serious refereeing with requested revisions.","tokens_in":18169,"tokens_out":3388,"would_cite":true,"duration_ms":33320,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["20D15","20E18","20E22","20F05","20F12","20F14","20-04","11R37"],"pacs":[],"model":"deepseek-v4-flash","headline":"Artin patterns force non-metabelian 3-class towers for (9,9) quadratic fields.","keywords":["finite 3-groups","Artin transfers","Artin pattern","harmonically balanced capitulation","Hilbert 3-class field tower","quadratic fields","Schur sigma-groups","p-group generation algorithm"],"falsifier":"Run the descendant-tree computation of the roots $\\langle 6561,23\\rangle$ and $\\langle 6561,25\\rangle$ past logarithmic order 100 and look for the first index $i>30$ where $V_i$ lacks a unique $\\sigma$-child $S_{i+1}$ of step size 3 or where $S_{i+1}$ has fewer than three non-metabelian Schur $\\sigma$-children of step size 2; such a break would disprove Conjecture 6.1 and remove the infinite-family conclusions. Alternatively, search imaginary quadratic fields of type $(9,9)$ for an Artin pattern (8.6) or (8.8) whose fourth-layer target $\\tau_4$ or second-order invariant $\\tau_{2,1}$ is not of the form produced by any $T_{i,k}$.","tokens_in":17094,"feed_emoji":"🧮","tokens_out":11392,"duration_ms":105235,"temperature":0.7,"pith_summary":"The paper sets out to identify the Galois group $G$ of the finite 3-class field tower for a quadratic field $K$ whose 3-class group is $C_9 \\times C_9$, using the Artin pattern: the collection of kernels and targets of transfer maps from $K$ to its unramified abelian 3-extensions. For most explicit Artin patterns, it concludes that the tower has length $\\ell_3(K)=2$ and that $G$ is one of a short list of metabelian Schur $\\sigma$-groups of order $3^8$. For two patterns, it concludes that $\\ell_3(K)\\ge 3$ and that $G$ belongs to an infinite periodic family of non-metabelian Schur $\\sigma$-groups. The classification is carried by a group-theoretic phenomenon called harmonically balanced capitulation, in which every transfer kernel is itself an intermediate subgroup of the tower group, permuted by a fixed permutation. A sympathetic reader should care because it converts finite arithmetic data about class groups into a description of the maximal unramified pro-3 extension above the field, including cases where the tower does not stop after two stages.","feed_headline":"Artin patterns force 3-stage towers for (9,9) fields","feed_subtitle":"Transfer kernels identify a periodic family of non-metabelian Galois groups in quadratic fields with class group C9×C9.","key_machinery":"The central object is the Artin pattern $\\mathrm{AP}(K)=(\\kappa(K),\\tau(K))$: for each unramified abelian extension $K\\le N\\le F_3^1(K)$, it records the kernel $\\ker(T_{K,N}:\\mathrm{Cl}_3(K)\\to\\mathrm{Cl}_3(N))$ and the target $\\mathrm{Cl}_3(N)$. By Artin reciprocity these match the group-theoretic Artin transfers $T_{G,S}:G/G'\\to S/S'$ of the tower group, so a finite computation on the base field prescribes the pattern the group must fit. The second load-bearing mechanism is the descendant tree of the metabelian root $R=\\langle 243,2\\rangle$, extended beyond the small-groups database by the p-group generation algorithm; its periodic trifurcation produces a chain $V_i$ of groups with nuclear rank 3, each with a unique $\\sigma$-child $S_{i+1}$ and three non-metabelian Schur $\\sigma$-group children $T_{i+1,j}$. The third ingredient is the Shafarevich cohomology bound, which for imaginary quadratic fields forces relation rank $d_2=d_1=2$, so the tower group must be a balanced Schur group; for real quadratic fields it allows $d_2\\le3$, which explains the different two-stage results.","core_discovery":"On the paper's own terms, the discovery is that Artin patterns determine the isomorphism type of the 3-class tower group for quadratic fields of type $(9,9)$. For six Artin patterns, stated in Theorems 8.1–8.5, 8.7, and 10.1, the tower group is metabelian, has order $3^8$, and the tower length is exactly 2. For the two patterns (8.6) and (8.8), Theorems 8.6 and 8.9 assert that the tower length is at least 3 and the tower group is one of the infinitely many non-metabelian Schur $\\sigma$-groups $T_{i,k}$ with $i\\ge1$, $1\\le k\\le3$, living in the descendant tree of the metabelian root $\\langle 6561,25\\rangle$ or $\\langle 6561,23\\rangle$; these groups have variable fourth-layer target $\\tau_4=(2+i,2,2)$, coclass $5+3i$, and metabelianization $S_1$ for $i=1$ and a fixed child of $V_1$ for $i\\ge2$. The paper also proves a stage separation criterion: for pattern (8.6)/(8.7), the second-order abelian quotient invariants $\\tau_{2,1}$ decide whether the tower has length 2 or at least 3. All non-metabelian groups obtained carry explicitly described harmonically balanced capitulation permutations. The 'infinitely many' part depends on a periodic-trifurcation theorem verified computationally up to logarithmic order 100 and conjectured to continue (Conjecture 6.1).","pith_inferences":["If Conjecture 6.1 holds, the periodic trifurcation provides a model for infinite non-metabelian p-class tower families, and a similar descendant-tree search could be attempted for fields with $\\mathrm{Cl}_p(K)\\cong C_{p^2}\\times C_{p^2}$ for primes $p>3$, where the paper notes that more capitulation permutations become admissible.","The stage separation invariant $\\tau_{2,1}$ could serve as a practical arithmetic sieve: it uses only subfields of relative degree up to 27 and could be computed before deciding whether a field warrants the expensive full tower computation.","The explicit permutation description of harmonically balanced capitulation suggests a possible route to proving Conjecture 6.1: if the balanced kernel equations force the trifurcation pattern at each step purely group-theoretically, the computational verification up to order $3^{100}$ could be replaced by an inductive argument."],"forward_implications":["Any imaginary quadratic field with Artin pattern (8.6) or (8.8) has a 3-class field tower of length at least 3, so its Galois group is not metabelian and the second Hilbert 3-class field cannot be the whole tower.","For fields with pattern (8.6) or (8.7), the second-order invariant $\\tau_{2,1}$ decides between a two-stage tower and a tower of length at least 3 without computing the full tower group.","The two non-metabelian families exhibit tower groups of coclass $5+3i$ and fourth-layer target $(2+i,2,2)$ for $i\\ge1$, so the possible tower groups for type $(9,9)$ are not a finite set but an infinite periodic collection.","The known real quadratic fields of type $(9,9)$ with pattern (10.1) have metabelian two-stage towers whose groups have relation rank 3, showing that the imaginary-field Schur condition $d_2=2$ is what forces the non-metabelian cases."],"supporting_citations":[{"why":"Supplies the overall strategy of pattern recognition via Artin transfers that the paper adapts to the (9,9) case.","marker":"[22]"},{"why":"Provides the computational algebra environment used for class group invariants, Artin patterns, and p-group generation.","marker":"[12]"},{"why":"The database of groups of small order that is sifted to identify candidate metabelianizations.","marker":"[5]"},{"why":"Introduces the p-group generation method used to build descendant trees beyond the database.","marker":"[23]"},{"why":"Describes the p-group generation algorithm actually used to compute the descendant trees and trifurcations.","marker":"[24]"},{"why":"Proves that the tower group of an imaginary quadratic field is a Schur sigma-group, the constraint that selects balanced groups.","marker":"[10]"},{"why":"Shows that higher p-class groups are sigma2-groups, justifying the relator-inverting automorphism requirement.","marker":"[26]"},{"why":"Establishes the transfer kernel type and transfer target type formalism that underlies the Artin pattern.","marker":"[15]"},{"why":"Proves that all higher p-class groups share a common Artin pattern, used for the stage separation criterion.","marker":"[17]"},{"why":"Provides the identification of the group-theoretic Artin pattern with the arithmetic Artin pattern and the child/fork topology concepts.","marker":"[18]"}],"fun_headline_variants":["Artin patterns determine tower groups for (9,9) fields","Artin patterns resolve (9,9) field towers: metabelian or not","Harmonically balanced capitulation yields infinite non-metabelian towers","Periodic trifurcations produce infinite Schur σ-groups","Artin patterns pin down tower groups for (9,9) fields"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The entire 'infinitely many' statement rests on Conjecture 6.1: the periodic trifurcation of $\\sigma$-groups verified by computation up to $u=30$ (orders up to $3^{100}$) continues for every $u>30$; if the chain of parents or the three non-metabelian grandchildren ever stops, Theorems 8.6 and 8.9 no longer yield infinite families.","fun_headline_variants_meta":{"raw":{"variants":["Artin patterns determine tower groups for (9,9) fields","Artin patterns resolve (9,9) field towers: metabelian or not","Harmonically balanced capitulation yields infinite non-metabelian towers","Periodic trifurcations produce infinite Schur σ-groups","Artin patterns pin down tower groups for (9,9) fields"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000785,"raw_usage":{"total_tokens":3538,"prompt_tokens":1091,"completion_tokens":2447,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":707,"completion_tokens_details":{"reasoning_tokens":2351}},"tokens_in":707,"tokens_out":2447,"duration_ms":18809,"temperature":1.0,"reasoning_tokens":2351,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T14:58:03.501942+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Run the descendant-tree computation of the roots $\\langle 6561,23\\rangle$ and $\\langle 6561,25\\rangle$ past logarithmic order 100 and look for the first index $i>30$ where $V_i$ lacks a unique $\\sigma$-child $S_{i+1}$ of step size 3 or where $S_{i+1}$ has fewer than three non-metabelian Schur $\\sigma$-children of step size 2; such a break would disprove Conjecture 6.1 and remove the infinite-family conclusions. Alternatively, search imaginary quadratic fields of type $(9,9)$ for an Artin pattern (8.6) or (8.8) whose fourth-layer target $\\tau_4$ or second-order invariant $\\tau_{2,1}$ is not of the form produced by any $T_{i,k}$.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the overall strategy of pattern recognition via Artin transfers that the paper adapts to the (9,9) case."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides the computational algebra environment used for class group invariants, Artin patterns, and p-group generation."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"The database of groups of small order that is sifted to identify candidate metabelianizations."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Introduces the p-group generation method used to build descendant trees beyond the database."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Describes the p-group generation algorithm actually used to compute the descendant trees and trifurcations."},{"cited_title":"Koch und B","cited_arxiv_id":null,"evidence_quote":"Proves that the tower group of an imaginary quadratic field is a Schur sigma-group, the constraint that selects balanced groups."},{"cited_title":"Schoof, Infinite class field towers of quadratic fields , J","cited_arxiv_id":null,"evidence_quote":"Shows that higher p-class groups are sigma2-groups, justifying the relator-inverting automorphism requirement."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Proves that all higher p-class groups share a common Artin pattern, used for the stage separation criterion."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides the identification of the group-theoretic Artin pattern with the arithmetic Artin pattern and the child/fork topology concepts."}],"review_version":1}