{
  "case_id": "case-7af01c06f34fe5a7428c979c",
  "target_id": "theo",
  "original_proof_status": "defective",
  "statement_status": "refuted",
  "statement_argument": "As literally stated, Theorem 2.5 (theo) fails for several parameter values expressly permitted by its hypotheses: (1) in family M_3(j,n), the stated domain 0 < j <= n-2 (j odd) allows j = n-2 (e.g., n=11, j=9), which makes the denominator 2n-2j-4 = 0 in the displayed formula for beta and makes Q_{n-2}(z) = 2(z^{n-2}-1) non-monic; (2) in family N_4(j,k,n), the stated condition (j,k+1)=1 and 0 < j < (n-3)/2 with ell*k = n-2 allows ell > 1 and j >= k+1 (e.g., n=11, k=1, ell=9, j=3), for which k+1-j = -1 < 0 and Q_{n-2,k}(z) = (z^{27}-1)(z^{-9}-1)/(z^9-1) = z^{-9} - z^{18} is not a polynomial of degree n-2; (3) in family M_1(j,n), the displayed middle coordinates (n-1)/(2n) and (n+1)/(2n) of alpha have even numerators since n is odd, so their exponentials are roots of z^n - 1 rather than P_n(z) = z^n + 1 (they should be (n-2)/(2n) and (n+2)/(2n)); and (4) the completeness claim for n > 9 omits pairs obtained via Theorem 2.2(2) by substituting z -> z^ell into sporadic primitive finite interlacing pairs of degree k <= 7 and then applying the Lemma 2.4 modifications to produce primitive hyperbolic pairs in dimension n > 9 (for example, substituting z -> z^3 into the degree-4 sporadic interlacing pair P_0(z)=Phi_{12}(z), Q_0(z)=(z^2-1)Phi_6(z) and applying the Case 2 modification yields a primitive integral hyperbolic pair in dimension n=13). These counterexamples refute the literal text as written, without precluding a suitably corrected classification with tightened parameter domains, corrected formulas, and treatment of sporadic substitutions.",
  "summary": "The proof and literal statement of Theorem 2.5 (theo) are defective. Specifically: (1) the parameter ranges in M_3 and N_4 permit boundary/divisor values where denominators vanish or polynomial exponents become negative, and M_1, M_2, M_3, N_1, N_2, N_3 omit necessary coprimality and upper-bound conditions on j; (2) the displayed formula for M_1(j,n) lists (n-1)/(2n) and (n+1)/(2n) instead of (n-2)/(2n) and (n+2)/(2n), which are not roots of z^n+1 for odd n, and the proof's formula for M_3 has a denominator typo 2n-2 in its final terms; (3) Definition 2.3 and the signature formula (2.2) assume strictly distinct parameters in [0,1), whereas Lemma 2.4 and five of the seven families in Theorem 2.5 rely on triple repetitions of 0 or 1/2 without establishing a multiplicity-inclusive definition and reduction lemma; (4) the proof conflates reducibility of the reflection subgroup H_r in Theorem 2.2(2) with reducibility of H itself, while Theorem 2.2(2) internally states that H_r is both reducible and primitive and Section 2.1 misstates the reflection generators as A^k B A^{-k} instead of B^k C B^{-k}; and (5) the exclusion of sporadic examples for n > 9 overlooks that Theorem 2.2(2) permits substitutions P(z^ell), Q(z^ell) into sporadic primitive pairs, which can increase the dimension to ell*k + 1 or ell*k + 2 > 9 and become primitive after the root modification. A sixth point regarding unrestricted rational scalar shifts is a cosmetic imprecision in the theorem statement given the standing cyclotomicity/integrality assumption in Section 2.1.",
  "decisions": [
    {
      "finding_id": "finding-e5e69a8233b7118e538a",
      "outcome": "confirmed_error",
      "severity": "substantive",
      "reasoning": "In Theorem 2.5 item 3), the family M_3(j,n) specifies 0 < j <= n-2 with j odd. Since n is odd, n-2 is odd, so j = n-2 is expressly permitted (e.g., n=11, j=9). However, substituting j = n-2 into the displayed beta vector gives denominator 2n - 2j - 4 = 0, leaving the fractions undefined; correspondingly, in (2.3) for m = n-2, Q_{n-2}(z) = (z^{n-2}-1)(z^0+1) = 2(z^{n-2}-1) is not a monic product of cyclotomic polynomials. Similarly, in item 7), N_4(j,k,n) requires only (j,k+1)=1 and 0 < j < (n-3)/2 where ell*k = n-2. Instead of bounding j by (k+1)/2 (or k) as in Theorem 2.2(1), the bound (n-3)/2 depends on n rather than k. Consequently, when ell > 1, one can choose parameters such as n=11, k=1, ell=9, j=3 satisfying (3,2)=1 and 0 < 3 < 4, for which k+1-j = -1 < 0 and Q_{9,1}(z) = (z^{27}-1)(z^{-9}-1)/(z^9-1) = z^{-9} - z^{18} is not a polynomial. In addition, coprimality conditions from Theorem 2.2(3) (such as gcd(j,n)=1 in M_1 and gcd(j,n-2)=1 in M_3) and upper bounds 1 <= j <= k in N_1, N_2, N_3 are omitted from the theorem statement.",
      "hypothesis_check": "Checked against the hypotheses of Theorem 2.5 (n >= 1 odd, P_m, Q_m, P_{m,k}, Q_{m,k} as in (2.3)-(2.4) with ell*k = m). The counterexamples n=11, j=9 for M_3 and n=11, k=1, ell=9, j=3 for N_4 satisfy all stated hypotheses.",
      "affected_claims": [
        "theo",
        "narrative:4083",
        "narrative:32909",
        "narrative:47545"
      ],
      "revised_status": "not_checked",
      "revised_reasoning": "No later edition was supplied for comparison.",
      "source_quote": "where $0<j\\leq n-2$ is an odd integer.",
      "source_span": {
        "char_start": 36468,
        "char_end": 36506,
        "match_mode": "exact"
      },
      "revised_quote": "",
      "revised_span": null
    },
    {
      "finding_id": "finding-16c6ac914f26510a018e",
      "outcome": "confirmed_error",
      "severity": "substantive",
      "reasoning": "The family M_1(j,n) is constructed in Case 1 of the proof from the roots of P_n(z) = z^n + 1 by replacing the root at alpha'_k = 1/2 = n/(2n) with 0. The roots of z^n + 1 = 0 are exp(2*pi*i*(2r-1)/(2n)), which all have odd numerators over 2n. Because n is odd, the odd numerators adjacent to n are n-2 and n+2 (as correctly written later in Theorem 4.5(2)), whereas n-1 and n+1 are even. Thus exp(2*pi*i*(n-1)/(2n)) and exp(2*pi*i*(n+1)/(2n)) have n-th power +1 rather than -1 and are not roots of z^n + 1. Separately, in Case 3 of the proof of Theorem 2.5, the displayed formula for alpha in M_3 ends with (2n-5)/(2n-2), (2n-3)/(2n-2), which has a denominator typo 2n-2 instead of 2n-4.",
      "hypothesis_check": "Under the standing hypothesis of Theorem 2.5 that n >= 1 is odd, n-1 and n+1 are always even, so (n-1)/(2n) and (n+1)/(2n) never correspond to roots of P_n(z) = z^n + 1.",
      "affected_claims": [
        "theo"
      ],
      "revised_status": "not_checked",
      "revised_reasoning": "No later edition was supplied for comparison.",
      "source_quote": "\\frac{n-1}{2n}, \\frac{n+1}{2n}",
      "source_span": {
        "char_start": 35197,
        "char_end": 35227,
        "match_mode": "exact"
      },
      "revised_quote": "",
      "revised_span": null
    },
    {
      "finding_id": "finding-fc224ab63b59d7ce3aae",
      "outcome": "confirmed_error",
      "severity": "substantive",
      "reasoning": "Definition 2.3 (almostinterlacing) and the setup in Section 1 and Section 2.1 (equation (2.2)) require the elements within each sequence alpha and beta to be strictly increasing: 0 <= alpha_1 < ... < alpha_n < 1 and 0 <= beta_1 < ... < beta_n < 1. Nevertheless, options (2), (3), and (4) in step (g) of the proof of Lemma 2.4 (interlacingtypes) and families M_2, M_3, N_2, N_3, and N_4 in Theorem 2.5 contain triple repetitions of 0 or 1/2. Moreover, Lemma 2.4 is stated only for n >= 7 as a necessary condition on almost interlacing sequences and contains index/value typos in its statement (alpha_0=0) and step (d) (alpha_m=0 and beta_1=1/2). Applying Definition 2.3, equation (2.2), and Lemma 2.4 to families with repeated roots requires a multiplicity-inclusive formulation of the definition, signature formula, and reduction lemma.",
      "hypothesis_check": "Checked against Section 1, Section 2.1 (equation (2.2)), Definition 2.3, and Lemma 2.4. The strict inequalities in Definition 2.3 exclude the repeated-root configurations in M_2, M_3, N_2, N_3, and N_4.",
      "affected_claims": [
        "interlacingtypes",
        "theo"
      ],
      "revised_status": "not_checked",
      "revised_reasoning": "No later edition was supplied for comparison.",
      "source_quote": "Two sequences $0\\le \\alpha_1<\\ldots < \\alpha_n <1$ and $0\\le \\beta_1<\\ldots < \\beta_n <1$ are called almost interlacing cyclotomic sequences",
      "source_span": {
        "char_start": 27245,
        "char_end": 27385,
        "match_mode": "exact"
      },
      "revised_quote": "",
      "revised_span": null
    },
    {
      "finding_id": "finding-e7cb2d83553bc81dabf4",
      "outcome": "confirmed_error",
      "severity": "substantive",
      "reasoning": "Theorem 2.2 (finiteH) classifies finite hypergeometric groups H(alpha',beta') into three cases: (1) H_r is primitive, (2) H_r acts reducibly on C^n, and (3) H is imprimitive and H_r is irreducible. In Case (2) of Theorem 2.2, the text states 'The case where H_r acts reducibly on C^n: in this case Theorem 5.3 of [bh] implies that H_r is primitive', which is internally contradictory since a reducible group cannot be primitive. Furthermore, in the proof of Theorem 2.5 (Cases 1, 2, and 3), the authors repeatedly refer to H(alpha',beta') itself as 'reducible' (e.g., 'there is only one family up to scalar shift of (alpha',beta') for which H(alpha',beta') is reducible'), conflating reducibility of the reflection subgroup H_r with reducibility of H and folding the ell=1 primitive case (Case 1 of Theorem 2.2) into the statement that H(alpha',beta') is reducible. In addition, Section 2.1 defines H_r as generated by {A^k B A^{-k}}, whereas in Levelt's notation (Theorem 2.1) B is not a reflection and C = A^{-1}B is the reflection (as used in Section 2.4).",
      "hypothesis_check": "Checked against Section 2.1 (Theorem 2.1 and Theorem 2.2) and Section 2.3 (proof of Theorem 2.5). Under the standing condition alpha'_i != beta'_j mod 1, H(alpha',beta') is irreducible by Beukers-Heckman, so the proof's references to H(alpha',beta') being reducible misstate the classification cases.",
      "affected_claims": [
        "finiteH",
        "theo"
      ],
      "revised_status": "not_checked",
      "revised_reasoning": "No later edition was supplied for comparison.",
      "source_quote": "Suppose $H(\\alpha',\\beta')$ is finite and reducible or primitive and irreducible.",
      "source_span": {
        "char_start": 39937,
        "char_end": 40018,
        "match_mode": "exact"
      },
      "revised_quote": "",
      "revised_span": null
    },
    {
      "finding_id": "finding-19bf18492a278d755b03",
      "outcome": "confirmed_error",
      "severity": "substantive",
      "reasoning": "At the end of the proof of Theorem 2.5, the authors conclude that M_1, M_2, M_3, N_1, N_2, N_3, N_4 exhaust all primitive hyperbolic H(alpha,beta) in dimension n > 9 'since there are no sporadic such interlacing pairs in dimensions n>7'. This inference has a substantive gap: Theorem 2.2(2) explicitly states that finite interlacing pairs (alpha',beta') with reducible H_r are obtained as P(z^ell), Q(z^ell) 'where P and Q are as in case (1) or correspond to one of the sporadic examples mentioned there'. While substituting z -> z^ell into the infinite family of Case (1) produces the families N_1, N_2, N_3, N_4, the proof ignores substituting z -> z^ell into the sporadic primitive examples of Case (1). Even though the sporadic primitive examples have degree k <= 7, substituting z -> z^ell gives finite interlacing cyclotomic pairs of arbitrarily large degree m = ell*k. When modified at 0 and 1/2 via Lemma 2.4, the resulting pair (alpha,beta) in dimension n = ell*k, ell*k+1, or ell*k+2 > 9 need no longer be invariant under shifts by 1/ell (because the modified multiplicities at 0 and 1/2 break the z -> e^{2*pi*i/ell}z symmetry) and thus can be primitive and irreducible! For instance, taking the degree-4 sporadic interlacing pair P_0(z)=Phi_{12}(z), Q_0(z)=(z^2-1)Phi_6(z), substituting z -> z^3 (degree 12), and applying the Case 2 modification yields a primitive, irreducible integral hyperbolic group in dimension n = 13 > 9 outside the seven families. Moreover, the proof provides no criterion or verification for the asserted primitivity of the modified pairs.",
      "hypothesis_check": "Checked against Theorem 2.2(1)-(2), Lemma 2.4, and the hypotheses of Theorem 2.5 for n > 9.",
      "affected_claims": [
        "theo",
        "narrative:4083",
        "narrative:32909",
        "narrative:47545"
      ],
      "revised_status": "not_checked",
      "revised_reasoning": "No later edition was supplied for comparison.",
      "source_quote": "since there are no sporadic such interlacing pairs in dimensions $n>7$",
      "source_span": {
        "char_start": 47234,
        "char_end": 47304,
        "match_mode": "exact"
      },
      "revised_quote": "",
      "revised_span": null
    },
    {
      "finding_id": "finding-012b18b45accbe4db5bf",
      "outcome": "confirmed_error",
      "severity": "cosmetic",
      "reasoning": "In Section 2.1, the authors state the standing assumption that P(z) and Q(z) factor as cyclotomic polynomials ('Since we are interested in integral monodromy groups, we will assume this throughout the article'), and also define a general scalar shift by d in Q/Z. In the first sentence of Theorem 2.5, they write that if (alpha,beta) belongs to one of the families 'or is a scalar shift of a pair in one of these families, then H(alpha,beta) subset O(n-1,1)', without repeating the cyclotomic/integrality restriction (which restricts non-trivial scalar shifts to d = 1/2, as used in the proof and in Tables 2 and 3). For an unrestricted rational shift d not in {0, 1/2}, the Levelt generators A, B are multiplied by exp(2*pi*i*d), which does not preserve reality or determinant +-1, so H(alpha,beta) lies in U(n-1,1) rather than O(n-1,1). Because the standing integrality assumption in Section 2.1 and the proof of Theorem 2.5 make clear that only shifts preserving cyclotomicity (namely d = 1/2) are intended, this omission in the theorem's phrasing is a cosmetic imprecision.",
      "hypothesis_check": "Checked against Section 2.1, Theorem 2.1, and Theorem 2.5.",
      "affected_claims": [
        "theo"
      ],
      "revised_status": "not_checked",
      "revised_reasoning": "No later edition was supplied for comparison.",
      "source_quote": "or is a scalar shift of a pair in one of these families",
      "source_span": {
        "char_start": 34726,
        "char_end": 34781,
        "match_mode": "exact"
      },
      "revised_quote": "",
      "revised_span": null
    }
  ],
  "chronology": {
    "review_policy": {
      "schema": 1,
      "cutoff_date": "2013-08-11",
      "basis": "journal_received",
      "evidence": "Publisher PDF p.1617 states Received August 11, 2013; publisher.html independently records Submitted 11 August 2013. Reviewed native source is arXiv v2 dated 2013-08-12; that does not advance the external proof-input cutoff. ArXiv v1 is 2013-05-03. Later sources may provide retrospective context, not proof inputs for the original paper.",
      "status": "recorded",
      "rule": "external-results-available-by-cutoff"
    },
    "note": "The recorded review policy cutoff is 2013-08-11 (journal received date). All six findings are adjudicated by self-contained algebraic and logical checks directly against the reviewed manuscript text (arXiv v2, 2013-08-12, matching the pre-cutoff submission content) and its internal definitions. No post-cutoff external results are imported.",
    "external_results": []
  },
  "origin": "independent_model_arbitration_not_formal_verification",
  "_meta": {
    "pipeline": "proof_arbitration",
    "model": "gemini-4-argon",
    "reasoning_effort": "high",
    "provider": "vertex",
    "batch_job_id": "projects/292312164194/locations/global/batchPredictionJobs/4724973786678427648",
    "custom_id": "arbitration-21550141c4e84c191f145059",
    "source_sha256": "7026f8e151de5996ffe6ff4218b67ce44200ad433fcc063ddc27150813644df2",
    "usage": {
      "candidatesTokenCount": 4483,
      "candidatesTokensDetails": [
        {
          "modality": "TEXT",
          "tokenCount": 4483
        }
      ],
      "promptTokenCount": 95983,
      "promptTokensDetails": [
        {
          "modality": "TEXT",
          "tokenCount": 95983
        }
      ],
      "thoughtsTokenCount": 19330,
      "totalTokenCount": 119796,
      "trafficType": "ON_DEMAND"
    },
    "max_output_tokens": 24576,
    "prompt_profile": "proof_arbitration-v2",
    "prompt_profile_sha256": "7eed840124321fd1b82530ec8d00ad039dd4f89b43ccc831ecbc578eb7bd6c56",
    "review_policy": {
      "schema": 1,
      "cutoff_date": "2013-08-11",
      "basis": "journal_received",
      "evidence": "Publisher PDF p.1617 states Received August 11, 2013; publisher.html independently records Submitted 11 August 2013. Reviewed native source is arXiv v2 dated 2013-08-12; that does not advance the external proof-input cutoff. ArXiv v1 is 2013-05-03. Later sources may provide retrospective context, not proof inputs for the original paper.",
      "status": "recorded",
      "rule": "external-results-available-by-cutoff"
    },
    "chronology_prompted": true
  }
}