{"id":"c0d49c07-6de8-48e3-a538-e5e6a34bfe0a","arxiv_id":"2509.03994","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"For primes p ≤ 7 the count of dormant PGL_n-opers on any pointed curve is now explicit, because hypergeometric dormant opers are rigid and the lone exception at (7,3) is fixed by a conservation identity.","lead":"This paper proves that PGL_n-opers coming from hypergeometric equations in prime characteristic are rigid, uniquely determined by their local data, and uses that rigidity to produce the complete counting tables (a 2d TQFT) for dormant opers on pointed curves for primes up to 7. The payoff is the first effective enumeration in a previously unreachable range, with one genuinely exceptional value at p = 7.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The claimed exhaustion O_{7,3}\\Hyp_{7,3} = {(w5,w5,w5)} in §3.3.6 is not proved; the residual 4 = Σ N_ρ N_{ρ⊻} allows other decompositions, so the complete p=7 tables are conditional.","rationale":"The reader identified two fragile points: the uniqueness step in Theorem 2.13 and the exhaustion O_{7,3}\\Hyp_{7,3}={(w5,w5,w5)}. I agree with the reader's overall CONDITIONAL verdict, but I would elevate the exhaustion to the primary load-bearing concern. The uniqueness of a logarithmic connection with two poles and fixed companion-form residues is plausible and likely fillable from the almost non-logarithmic extension construction; the text's one-sentence justification is terse but not evidently wrong. By contrast, the exhaustion is a concrete logical gap in the final computation: the residual 4 is compatible with many configurations of nonzero N_ρ, and the paper gives no argument excluding them. Because the claimed complete TQFT for (7,3) and, by duality, (7,4) rests directly on this unproved assertion, the central claim is conditional pending either a direct enumeration of dormant opers or an additional TQFT consistency argument. I do not find grounds for rejection: the arithmetic consistency of the total 56 and the duality relations are real supporting evidence, and the missing step may be repairable. My disagreement with the reader is only about which fragile premise is most load-bearing, not about the verdict itself.","tokens_in":27826,"tokens_out":14610,"duration_ms":148604,"concrete_test":"Set up the linear system of TQFT factorization identities (3.3)–(3.4) in the 125 unknowns N_{7,3,ρ,0,3}, imposing the known value N_{7,3,∅,2,0}=56, the duality N_ρ = N_{ρ⊻}, and N_ρ=1 for ρ∈Hyp_{7,3}. If this integer system admits any solution with N_ρ>0 for some ρ∉Hyp_{7,3}∪{(w5,w5,w5)}, then the asserted exhaustion in §3.3.6 is not a consequence of the stated axioms and the complete tables are unjustified.","verdict_should_be":"CONDITIONAL","load_bearing_attack":"Section 3.3.6 asserts that O_{7,3}\\Hyp_{7,3} = {(w5,w5,w5)}. The only evidence offered is arithmetic: from N_{7,3,∅,2,0}=56 and N=1 on the 52 hypergeometric triples, the non-hypergeometric contribution is Σ_{ρ∉Hyp} N_ρ N_{ρ⊻} = 4. The text then concludes that (w5,w5,w5) is the only possible element and that N_{7,3,(w5,w5,w5),0,3}=2. But the equation Σ N_ρ N_{ρ⊻} = 4 does not rule out other decompositions: e.g., two distinct dual pairs with N=1 each, or four self-dual triples with N=1 each, or one other self-dual triple with N=2 together with further zero/nonzero combinations. No independent enumeration, obstruction argument, or reference is supplied to show that every other triple in Ξ_{7,3}^3 has N=0. The statement 'only the possible element' is therefore unsupported. Since the final value N=2 at (w5,w5,w5) and all dual (7,4) entries depend on this exhaustion, the central claim of a complete and explicit TQFT for p≤7 is not fully established.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper studies dormant PGL_n-opers attached to generalized hypergeometric differential operators in prime characteristic p. It claims that on the 3-pointed projective line, any two dormant PGL_n-opers with the same hypergeometric radii are isomorphic (Theorem 2.13), so the corresponding moduli stack is Spec(k) and the generic degree is 1 on the hypergeometric locus. Combining this rigidity with the author's earlier 2D TQFT formalism and with known formulas and duality, the paper then claims a complete determination of all numbers N_{p,n,\\rho,g,r} for primes p\\le 7, including the exceptional non-hypergeometric value N_{7,3,(w5,w5,w5),0,3}=2, obtained from the arithmetic identity 56=52+2^2.","tokens_in":27973,"tokens_out":25389,"duration_ms":245439,"significance":"If the main claims are correct, this is a substantial step: it gives the first explicit enumerative description of dormant opers in the range 2<n<p-2, using a parameter-free combination of Katz's hypergeometric classification, rigidity, and a previously constructed TQFT. The paper also has genuine strengths: the arithmetic self-check 56=52+2^2 is consistent; the TQFT machinery is imported from independent prior work rather than fitted to the tables; and the explicit finite lists in Section 3.3 are potentially verifiable. However, several load-bearing points are not established in the written proof, as detailed below.","major_comments":[{"comment":"Proposition 2.1 is false as stated. For n=2, p=5, take \\alpha=(4,1), \\beta=3. Then the chain (2.12) holds, so Corollary 2.2 predicts rank 2 for Ker(D_{\\alpha,\\beta}). But the displayed definition of T gives #T=1, since m'=1 and T is a subset of {1,...,m'}. A direct computation of the bidiagonal matrix R for this example gives a 2-dimensional kernel. The error is an off-by-one indexing: the proof's block decomposition has m'+1 blocks, and the rank defect of the first block R'_1 corresponds to an index-0 condition e\\beta_0=p>e\\alpha\\ge e\\beta_1, which is absent from T. The equality in the proposition should presumably involve a reindexed T (with e\\beta_0=p). Since Proposition 2.12 uses this corollary to identify dormant hypergeometric opers, this must be corrected and re-proved.","section":"Section 2.1, Proposition 2.1 and Corollary 2.2"},{"comment":"The proof asserts, after fixing the companion-form residue data A[0], A[\\infty], that 'such a log connection is uniquely determined'. This is the step that promotes equality of exponents to equality of the connections \\breve{\\nabla}_\\circ=\\breve{\\nabla}_\\bullet and, via Proposition 2.8, to rigidity of the dormant oper. No proof or reference is supplied. For n\\ge3, logarithmic connections on a trivial rank-n bundle over P^1 with two poles can carry accessory parameters even after the residue characteristic polynomials are fixed, so this uniqueness is not automatic. The authors need either a precise argument that the global trivialization forces the companion matrices to be determined by the exponents, or a citation to a theorem that does so. As written, Theorem 2.13 and the N_{p,n,\\rho,0,3}=1 values for all hypergeometric \\rho are not fully established.","section":"Theorem 2.13, final paragraph of proof"},{"comment":"The exhaustion O_{7,3}\\setminus Hyp_{7,3} = {(w5,w5,w5)} is not proved. The displayed numerical identity gives only \\sum_{\\rho\\notin Hyp_{7,3}} N_{7,3,\\rho}N_{7,3,\\rho^\\vee}=4. This equation is compatible with many other decompositions, for example two distinct dual pairs with N=1 each, or four self-dual triples with N=1 each, or one other self-dual triple with N=2 together with additional zero contributions. No independent enumeration, obstruction argument, or reference is supplied to rule out all other triples in \\Xi_{7,3}^3. The subsequent conclusion N_{7,3,(w5,w5,w5),0,3}=2, and hence all dual (7,4) entries, depends entirely on this unsupported exhaustion. The parenthetical construction of one oper via Sym^2 only gives existence of at least one such oper; it does not identify the full complement of Hyp_{7,3}.","section":"Section 3.3.6"}],"minor_comments":[{"comment":"The statement says 'rank(Ker(D_{\\alpha,\\beta}))' but the proof computes dim_{k(x^p)}(Ker). Please use consistent notation.","section":"Proposition 2.1"},{"comment":"In the definition of \\gamma, the expression '\\sum_{j=1}^n \\alpha_n' should presumably be '\\sum_{j=1}^n \\alpha_j'.","section":"Theorem 2.13"},{"comment":"The sentence beginning 'Next, let us compute the values N_{7,2,\\rho,0,3}' should read N_{7,3,\\rho,0,3}.","section":"Section 3.3.6"},{"comment":"'As in the case of p=2' should be 'p=5'.","section":"Section 3.3.5"},{"comment":"The notation '2 \\in \\{\\circ,\\bullet\\}' is a typo; the intended symbol is likely a placeholder such as '\\star'.","section":"Proposition 2.8"}],"recommendation":"major_revision","confidential_remarks":"This is a paper whose main claims are plausible and potentially correct, but the written support has a clear technical error in Proposition 2.1, an unproved uniqueness assertion in Theorem 2.13, and an unsupported exhaustion step in Section 3.3.6. The first two may be fixable by reindexing and by adding a rigorous argument or citation; the third requires a genuine independent enumeration or obstruction argument. I would be willing to revise if these points are addressed. The paper is not circular, but the novelty is concentrated in exactly the points that need repair."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Quick take: this is a useful note with a real new idea—Theorem 2.13, that hypergeometric dormant PGL_n-opers on the thrice-punctured line are rigid, and the consequent tables for p≤7. If the rigidity theorem holds, it's the first complete enumeration for 2<n<p−2, and the TQFT machinery from Wakabayashi's earlier work then produces all N-values. The paper is also honest: it credits Katz for the kernel-rank criterion and admits the missing reference before Prop 2.10.\n\nWhere it gets soft: Proposition 2.1's proof is arithmetically wrong. The block decomposition telescopes to p, not b_{m'}−b_0, and the claimed rank formula can fail on boundary blocks; on a p=3 example the kernel dimension is 2 while ♯T=1. Corollary 2.2 is rescued by Katz's theorem, so the dormancy criterion stands, but the paper's own proof doesn't supply it.\n\nThe bigger issue is Section 3.3.6. The assertion that O_{7,3}\\Hyp_{7,3} = {(w5,w5,w5)} is not demonstrated. The arithmetic N_{7,3,∅,2,0}=56 = 52 + 4 only tells you the residual sum of products is 4; it doesn't rule out two dual pairs with N=1 each or one other self-dual triple with N=2. No independent enumeration or obstruction is given. So the p=7 tables, including the exceptional N=2 at (7,3), are conditional on an unproved exhaustion.\n\nThird, the uniqueness step in Theorem 2.13 is one sentence: 'such a log connection is uniquely determined' by its companion-form residues. On P^1 with two log poles there is generally a positive-dimensional affine space of connections with fixed residues, so this step needs a real argument. If the residues are regular semisimple and the bundle is trivial, uniqueness may follow from a rigidity theorem, but it's not supplied.\n\nNone of these failures obviously kills the headline theorem. The rigidity result is plausible and the arithmetic consistency of the tables (56 = 52 + 2^2, duality pairing) is a good sign. But the paper as written does not establish the complete TQFT for p=7.\n\nBottom line: send it to a serious referee who knows the dormant-opers literature. The referee should push hard for a fix of Prop 2.1, a proof or citation for the exhaustion, and a detailed justification of the uniqueness in 2.13. It's worth one round of revision.","headline":"Rigidity theorem for hypergeometric dormant opers is plausible, but the p=7 completeness claim rests on an unsupported exhaustion and a flawed lemma proof.","tokens_in":28727,"tokens_out":7813,"would_cite":false,"duration_ms":73479,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["14H60","14G17"],"pacs":[],"model":"deepseek-v4-flash","headline":"Rigid hypergeometric opers: radii determine them, so all dormant-oper counts for p ≤ 7 are explicit — including a lone exceptional case with count 2.","keywords":["dormant oper","p-curvature","generalized hypergeometric equation","2d TQFT","characteristic p","PGL_n-opers","moduli of opers","rigidity"],"falsifier":"Enumerate directly in characteristic 7 all normal rank-3 logarithmic connections d + A on the trivial bundle over P¹ with poles only at 0 and ∞, A in companion form and with vanishing p-curvature, for each radius triple in Ξ³_{7,3}. The paper predicts exactly one solution per hypergeometric triple and exactly two for (w5,w5,w5); any deviation refutes Theorem 2.13 or the split 56 = 52 + 2². As an extension check, the residual N_{11,3,∅,2,0} − ♯Hyp_{11,3} must be expressible as a sum of squares if the method is to close at p = 11.","tokens_in":27471,"feed_emoji":"🧮","tokens_out":46980,"duration_ms":375922,"temperature":0.7,"pith_summary":"Dormant PGL_n-opers are flat bundles with flags whose p-curvature — a characteristic-p invariant of connections — vanishes identically; counting how many such objects sit over a general curve is an enumerative problem organized as a 2-dimensional topological quantum field theory (2d TQFT). The paper proves that on the projective line marked at three points, every dormant oper arising from a generalized hypergeometric differential equation is rigid: for each admissible triple of radii (the residue data of the connection at the three marked points), there is exactly one such oper up to isomorphism. The proof works by a local surgery — the 'almost non-logarithmic extension' — that converts each oper into a logarithmic connection whose companion-form residue matrices are fixed by the radii, and a connection of that form is unique. Because the TQFT's factorization rules reduce every counting number to the three-pointed-line case, this rigidity yields the complete explicit table of all counts N_{p,n,ρ,g,r} for primes p ≤ 7 — the first effective computation in the previously open range of ranks between 2 and p − 2 with marked points — including the single genuinely non-hypergeometric case at (p,n) = (7,3), where the count is 2.","feed_headline":"Hypergeometric opers are rigid, settling every p ≤ 7 count","feed_subtitle":"For primes up to 7, counting dormant opers on any curve is now an explicit table — rigidity does the work.","key_machinery":"Three mechanisms carry the argument. (1) The generalized hypergeometric operator D_{α,β} = δ_x∏_{j<n}(δ_x+β_j−1) − x∏_{j≤n}(δ_x+α_j): viewed over k(x^p) its bidiagonal matrix has kernel dimension ♯(T_{α,β}), yielding the interlacing dormancy criterion. (2) The almost non-logarithmic extension: a local surgery turning each dormant oper into a filtered logarithmic connection on P¹ whose companion-form residue matrices are fixed by the exponents; its uniqueness is the rigidity step. (3) The 2d TQFT from prior work, whose factorization identities reduce every count to the (0,3) case; the duality involution (−)▼ and the closed genus-2 formula finish the p = 7 computation.","core_discovery":"The central claim is Theorem 2.13: any two dormant PGL_n-opers on the 3-pointed projective line with the same hypergeometric radii ρ^{α,β} are isomorphic, and the oper E♠_{α,β} is the unique one. The proof builds, for each oper, an 'almost non-logarithmic extension': a logarithmic connection on P¹ with companion-form residue matrices fixed by the exponents. Such a connection is unique, and Proposition 2.8 transfers this equality back to the opers. Hence N_{p,n,ρ,0,3} = 1 for every hypergeometric triple ρ. Then the TQFT factorization, the duality, and the genus-2 formula give all counts for p ≤ 7; at (7,3) the residual 56 − 52 = 4 forces the exceptional triple (w5,w5,w5) to carry two opers.","pith_inferences":["Arithmetic search at larger primes: since N_{p,3,∅,2,0} is known by closed formula and the hypergeometric triples are countable from the interlacing condition, the residual N_{p,3,∅,2,0} − ♯Hyp_{p,3} must decompose as a sum of pair-squares over exceptional triples for the method to close; whether the p = 11 residual is a sum of integer squares is a cheap test of how far the argument reaches.","The exceptional oper points to a general recipe: symmetric powers of dormant PGL_2-opers should yield dormant non-hypergeometric PGL_d-opers exactly when the weight profile (0, d−1, 2(d−1), …) is itself an admissible radius — at p = 7 this is the second symmetric power, and at larger p it would predict the first exceptional ranks.","The uniqueness step in the proof does not use the fact that the exponents come from a hypergeometric equation — only that companion-form data pin down the connection — so the open problem is purely linear-algebraic: classify which radius triples admit more than one normal companion connection with vanishing p-curvature in characteristic p."],"forward_implications":["Each hypergeometric triple of radii carries exactly one dormant oper on the 3-pointed projective line: the moduli stack Op^{Zzz...}_{n,ρ,0,3} is Spec(k), so N_{p,n,ρ,0,3} = 1.","All generic-degree values N_{p,n,ρ,g,r} for p ≤ 7 — every rank n, radius profile, genus g, and boundary r — are explicit, giving the first effective computation in the range 2 < n < p − 2 with r > 0.","Non-hypergeometric dormant opers exist: at (p,n) = (7,3) the triple (w5,w5,w5), w5 = [[0,2,4]], carries exactly two, one of them the second symmetric power of the unique dormant PGL_2-oper of radii ([[0,2]],[[0,2]],[[0,2]]); its dual (7,4) profile carries the same count 2.","The radius-profile tables O_{p,n} for p ≤ 7 are fixed, including the duality-symmetric pairs O_{7,2} ↔ O_{7,5} and O_{7,3} ↔ O_{7,4}, the trivial rank p−1 case with N = 1, and the genus-2 value N_{7,3,∅,2,0} = 56.","Every genus-g, boundary-r count with p ≤ 7 is a finite sum of products of the tabulated (0,3) values, so the 2d TQFTs Z_n are completely determined for these primes."],"supporting_citations":[{"why":"Supplies the criterion that the hypergeometric operator D_{α,β} has an n-dimensional kernel exactly when its parameters interlace as in (2.12); this classification defines the dormant hypergeometric set Hyp_{p,n}.","marker":"[NKa4]"},{"why":"Provides the framework of dormant PGL_n-opers, radii, the moduli stacks Op^{Zzz...}_{n,ρ,g,r}, the bijections (2.5)/(2.7), and the finiteness, generic-étaleness, and genus-2 formula results the paper invokes.","marker":"[Wak4]"},{"why":"Supplies the 2d TQFT structure (Theorem 3.3) and the factorization identities (3.3)–(3.4) that reduce every count N_{p,n,ρ,g,r} to the three-pointed-line values.","marker":"[Wak10]"},{"why":"Supplies the duality N_{p,n,ρ,g,r} = N_{p,p−n,ρ▼,g,r} used to transfer the (7,3) computations to (7,4) and the (7,2) computations to (7,5).","marker":"[Wak2]"},{"why":"Its Theorem 5.1 gives the canonical non-logarithmic connection on the kernel sheaf G, the technical core of the almost non-logarithmic extension used in the proof of Theorem 2.13.","marker":"[NKa1]"},{"why":"Provides the Cartier-operator criterion used in Proposition 2.9 to identify when the theta-characteristic connection attached to (α, β) has vanishing p-curvature.","marker":"[NKa2]"},{"why":"Supplies the fact that residue data of connections with vanishing p-curvature lie in F_p, used in the proof of Proposition 2.12.","marker":"[Oss]"},{"why":"Provides the classical rank-2 Gauss-hypergeometric dormancy criterion that underlies the p = 5 and p = 7 rank-2 lists the tables extend.","marker":"[Iha]"}],"fun_headline_variants":["Rigid hypergeometric opers: all p ≤ 7 counts solved","Counting dormant opers: rigidity cracks p ≤ 7","Hypergeometric rigidity pins down opers for primes ≤ 7","Rigidity unlocks dormant oper counts through p=7"],"cache_read_input_tokens":2688,"weakest_assumption_plain":"The rigidity theorem rests on the premise that a normal logarithmic connection on the trivial rank-n bundle over P¹ with poles only at 0 and ∞ is uniquely determined by its companion-form residue data — coincident exponents force identical connections — together with the separately asserted exhaustion claim that the residual 56 − 52 = 4 at (7,3) leaves (w5,w5,w5) as the only non-hypergeometric contribution.","fun_headline_variants_meta":{"raw":{"variants":["Rigid hypergeometric opers: all p ≤ 7 counts solved","Counting dormant opers: rigidity cracks p ≤ 7","Hypergeometric rigidity pins down opers for primes ≤ 7","Rigidity unlocks dormant oper counts through p=7"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.001015,"raw_usage":{"total_tokens":4084,"prompt_tokens":667,"completion_tokens":3417,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":411,"completion_tokens_details":{"reasoning_tokens":3343}},"tokens_in":411,"tokens_out":3417,"duration_ms":28178,"temperature":1.0,"reasoning_tokens":3343,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-05T10:33:37.019215+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Enumerate directly in characteristic 7 all normal rank-3 logarithmic connections d + A on the trivial bundle over P¹ with poles only at 0 and ∞, A in companion form and with vanishing p-curvature, for each radius triple in Ξ³_{7,3}. The paper predicts exactly one solution per hypergeometric triple and exactly two for (w5,w5,w5); any deviation refutes Theorem 2.13 or the split 56 = 52 + 2². As an extension check, the residual N_{11,3,∅,2,0} − ♯Hyp_{11,3} must be expressible as a sum of squares if the method is to close at p = 11.","supporting_citations":[],"review_version":1}