{"id":"10488430-1ef2-4815-afe9-775562d612ac","arxiv_id":"2509.03993","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"For a general genus-3 curve in characteristic p, the generic degree of the Frobenius pull-back map on rank-2 stable bundles is (2p^6 + 5p^4 + 38p^2)/45.","lead":"In characteristic p, pulling back rank-2 vector bundles by Frobenius defines a map on the moduli space of stable bundles. This paper computes its generic degree for genus 3 and connects it to a combinatorial count on a trivalent graph.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Theorem C relies on unproved generic étaleness of Π_N ([Wak7, Thm C(i)]) and irreducibility of Op_{N,g} ([Wak8]); if either fails, the degree-to-fiber-count identification and hence the formula for deg(Ver_1^2) collapse.","rationale":"The reader's weakest-assumption identification matches my own reading: the proof of the main theorem is conditional on the second author's unpublished results [Wak7] and [Wak8]. The present paper gives substantial internal contributions—the maximally Frobenius-destabilized/dormant oper correspondence (Prop 4.3), the deformation-space isomorphism (Thm 4.5), and the formal-completion analysis (Prop 4.6–4.9)—but these do not circumvent the need for the generic étaleness of Π_N and the irreducibility of Op_{N,g}. The paper's recovery of the known genus-2 formula is a healthy consistency check but does not prove the missing étaleness. I considered other potential issues, such as the lower-semicontinuity argument in the proof of Theorem 4.10, which appears to conflate lower and upper semicontinuity; however, the equality part for general X (the one used later) does not rely on that argument. The dominant risk remains the external dependence. Therefore, I do not change the CONDITIONAL verdict; the result should be accepted only after [Wak7] and [Wak8] are publicly available and verified.","tokens_in":26251,"tokens_out":18353,"duration_ms":174875,"concrete_test":"For g=2, where deg(Ver_1^2)=(p^3+2p)/3 is independently known, compute ♯Ed_{p,2,G} by direct enumeration using Definition 5.2 for a genus-2 trivalent graph and compare it with the prediction from Corollary 4.11: ♯Ed_{p,2,G} = deg(Π_2) = deg(Π_1)·deg(Ver), with deg(Π_1) = p^{2-1}/2^{3}·Σ_{θ=1}^{p-1} csc^{2}(πθ/p). If the counts disagree for any odd prime p (e.g., p=3,5,7), the generic étaleness of Π_N or the combinatorial degree formula [Wak7, Thm E] is incorrect.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central equality deg(Ver_1^2) = ♯Ed_{p,2,G}/♯Ed_{p,1,G} depends on Corollary 4.11, which identifies deg(Π_N) with deg(Ver_1^2)^{N-1} times a constant. This identification requires that Π_N be generically étale over Mg and étale at the totally degenerate curve X_G used for the combinatorial count. Both facts are explicitly quoted from the second author's preprint [Wak7, Theorems B and C(i) and Theorem E] (see §3.2 after (3.7), and §5.1 where 'the projection Π_N is étale over all the points of totally degenerate curves'). Additionally, the surjectivity of Π_{N⇒N'} and the constancy of the degree use the irreducibility of Op_{N,g}, which is cited as [Wak8, Theorem A(ii)] and marked 'in preparation'. None of these results is proved or independently verified in the present paper. If generic étaleness fails, the fiber size of Π_N would not equal its degree, so the comparison deg(Π_N) = ♯Ed_{p,N,G} and the subsequent formula for deg(Ver_1^2) would break. This is the most load-bearing assumption; no internal proof in this paper rescues it.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper studies the generic degree of the generalized Verschiebung map Ver^2_1 on moduli spaces of rank-2 stable bundles on a general genus-g curve in characteristic p>2. It establishes a comparison, for general curves, between deg(Ver^2_1) and the degree of the projection Π_N from the moduli stack of dormant PGL_2-opers of level N to M_g (Corollary 4.11). Combining this with a combinatorial description of Π_N on totally degenerate curves, the authors express deg(Ver^2_1) as a ratio of counts of balanced (p,N)-edge numberings on a trivalent graph (Theorem 5.3). They show that this ratio is given by a quasi-polynomial Q(p) of degree 3g-3 with known leading coefficient (Theorem 5.5), and for genus 3 they compute Q(p)= (2p^6+5p^4+38p^2)/45 (Theorem 5.6).","tokens_in":26598,"tokens_out":10651,"duration_ms":101476,"significance":"The paper gives a new attack on a long-standing problem: it connects the coarse invariants of Frobenius pull-back on moduli of bundles to a finite combinatorial count via dormant opers. If its hypotheses are satisfied, it produces the first explicit value of deg(Ver^2_1) beyond the known genus-2 case, and the quasi-polynomial statement gives structural information for all genera. The local equivalence in Theorem 4.5, comparing deformation spaces of maximally Frobenius-destabilized bundles and dormant opers, is a strong and potentially reusable result. However, the main theorems are not self-contained: the crucial generic-etaleness, finiteness-flatness, irreducibility, and total-degeneracy fiber-count assertions are quoted from the second author's unpublished preprint [Wak7] and the in-preparation [Wak8]. The proof as written is conditional on those external inputs.","major_comments":[{"comment":"The central formula deg(Ver^2_1)=#Ed_{p,2,G}/#Ed_{p,1,G} relies on [Wak7, Thms B and C(i)] and [Wak8, Thm A(ii)] for the assertions that Op_{N,g} is smooth, proper, of dimension 3g-3, that Π_N is finite, faithfully flat, and generically étale, and that Π_N is étale at the totally degenerate curve X_G (stated in §5.1 immediately before Theorem 5.3). None of these facts is proved in the present paper, and [Wak8] is marked 'in preparation'. If generic étaleness fails, the fiber size of Π_N does not equal deg(Π_N), so the equality deg(Π_N)=#Ed_{p,N,G} and the subsequent formula for deg(Ver^2_1) do not follow. The authors should either prove the necessary étaleness/finiteness statements or clearly present the main results as conditional on published versions of [Wak7] and [Wak8].","section":"§3.2, (3.7); §4.4, Cor. 4.11; §5.1"},{"comment":"The equality in Theorem 4.10 depends on Proposition 4.9, which asserts that the formal fiber of Ver^2_{N⇒N'} over a maximally F-destabilized point splits as deg(Ver^2_{N⇒N'}) copies of the formal neighborhood. The proof of Proposition 4.9 uses [Wak7, Theorem C(i)] to conclude that Π_N is étale at a general point, and hence that Ver is étale at maximally Frobenius-destabilized points. The internal deformation argument is plausible, but the étaleness input is exactly the external fact that is load-bearing for the degree comparison. This step needs a proof or a complete published reference.","section":"§4.3, Prop. 4.9; §4.4, Thm. 4.10"},{"comment":"The explicit values of #Ed_{p,2,G} for genus 3 (p=1,3,5,...,25) are asserted without derivation, with only a reference to [Wak7, §10.5]. These values are the data used to fit the quasi-polynomial Q(t) and to obtain the closed formula for deg(Ver^2_1). This is a load-bearing computation for Theorem C. A reproducible count, or at least a tabulation of the polytope decomposition (5.2) and the resulting systems of linear equations, should be supplied so that the numerical result can be independently checked.","section":"§5.2, values before Thm. 5.6"}],"minor_comments":[{"comment":"The definitions 'a' := Π_{N⇒N'}(a)' and 'b' := Ver^2_{N⇒N'}(b)' are swapped: a is in SU_X^(N), so should be mapped by Ver, and b is in Op_{N,g}, so should be mapped by Π. This contradicts the displayed diagram (1.3) and the surrounding text.","section":"Thm. 4.5"},{"comment":"In the definition of balanced (p,N)-edge numbering, the notation '(a_{ζ(b)})_{b∈B_v}' is inconsistent: ζ(b) is a vertex, while the numbering is indexed by edges. The tuple should be indexed by the edge containing the branch b, e.g. '(a_{e(b)})_{b∈B_v}'.","section":"Def. 5.2"},{"comment":"Property (a) is typeset as 'P1(a) ⊇ P1(a) ⊇ P1(a) \\ ∂P1(a)', which is tautological as printed. It should presumably read '\\overline{P_1(a)} ⊇ P_1(a) ⊇ \\overline{P_1(a)} \\setminus ∂P_1(a)' (and similarly for P_2(a)).","section":"§5.2, Prop. 5.4 proof"},{"comment":"'Vershiebung' is a typo for 'Verschiebung'.","section":"Key words"},{"comment":"Several occurrences of 'characteristic pN' should read 'characteristic p^N'; the superscript is missing (see the discussion of Gauss hypergeometric differential operators).","section":"§5.1, Remark 5.1"},{"comment":"The text 'For each2 ∈ {T, η, s}' is missing a symbol; it should read 'For each 2 ∈ {T, η, s}'.","section":"Prop. 4.6 proof"}],"recommendation":"major_revision","confidential_remarks":"The main theorems are heavily dependent on the second author's own unpublished work, including [Wak7], which is an arXiv preprint, and [Wak8], which is listed as in preparation. Even if the internal arguments are sound, the paper is not currently self-contained for its central claim. I would recommend requiring either a complete proof of the needed étaleness/irreducibility statements in an appendix or the appearance of [Wak7] and [Wak8] in publishable form before acceptance."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The paper gives the first explicit generic degree for the rank-two generalized Verschiebung in genus 3, namely (1/45)(2p^6+5p^4+38p^2), and a general procedure for any genus by counting balanced edge numberings on a trivalent graph. That is a real result, not just a repackaging. The local comparison in Theorem 4.5 — the tangent isomorphism between the Verschiebung and level reduction of dormant opers — is proved in detail and is the genuine engine of the paper. The genus-3 formula is consistent with the known g=2 case and with the listed counts, which gives me some confidence that the machinery is working.\n\nThe main soft spot is exactly what the reader flagged: Theorem A and Theorem B depend on results from [Wak7] (a preprint) and [Wak8] (in preparation). Specifically, the identification deg(Π_N) = #Ed_{p,N,G} requires that Π_N be étale over the totally degenerate curve X_G, and the comparison deg(Ver) = deg(Π_N)^{1/(N-1)} requires generic étaleness over M_g. Both are quoted from [Wak7, Theorem C(i)]. The irreducibility of Op_{N,g}, used to get surjectivity of the level-reduction maps, is cited from [Wak8]. None of these is proved or even stated in enough detail here for the reader to check independently. If generic étaleness fails, the fiber count does not equal the degree, and the main formula collapses. This is a load-bearing dependency, and it is not a minor footnote.\n\nThat said, the dependency is explicit and honest. The authors are not hiding it. The logic of the paper itself is not circular: the Verschiebung degree is computed by counting edge numberings, and the counting uses only the dormant-oper side, not the Verschiebung side. So the paper is best read as a conditional result: if [Wak7] is correct, then the rank-two conjecture is resolved and the genus-3 formula is a genuine new computation.\n\nWho is this for? People working on Frobenius pull-back of stable bundles, dormant opers, or the moduli of bundles in positive characteristic. A reader in that area will want to see this paper, but they should also have [Wak7] open on the desk. The paper deserves a serious referee — this is exactly the kind of conditional but potentially major step that peer review should handle, not desk reject. The referee's main task will be to verify that the cited statements from [Wak7] and [Wak8] really do imply what is claimed, and to assess whether the in-preparation [Wak8] is a tolerable dependency or a blocker. I would send it out with that instruction.","headline":"Ranks-two Verschiebung degree computed via dormant opers, but the load-bearing hypotheses live in the second author's unpublished preprints.","tokens_in":27051,"tokens_out":2334,"would_cite":true,"duration_ms":24425,"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":"For a general genus-3 curve in characteristic p>2, the generic degree of Frobenius pull-back is (1/45)(2p^6+5p^4+38p^2).","keywords":["generalized Verschiebung","stable bundles","positive characteristic","dormant opers","balanced edge numberings","quasi-polynomial","Frobenius pull-back","moduli of curves"],"falsifier":"For a general genus-3 curve in characteristic 3 the formula predicts deg(Ver^2_1)=49; an independent computation of this generic degree, or a direct verification that a genus-3 trivalent graph has exactly 49 balanced (3,2)-edge numberings and 1 balanced (3,1)-edge numbering, would settle the central claim.","tokens_in":1995,"feed_emoji":"🧮","tokens_out":2503,"duration_ms":98792,"temperature":0.7,"pith_summary":"The paper sets out to determine the generic degree of the generalized Verschiebung map: the rational map that pulls a rank-two stable bundle back along Frobenius and lands in the moduli space of stable bundles on the original curve. Only the genus-two case was previously known. The authors prove that, for a general curve of genus g over an algebraically closed field of characteristic p>2, this generic degree is exactly the ratio of two finite combinatorial counts attached to any trivalent graph of genus g: balanced (p,2)-edge numberings divided by balanced (p,1)-edge numberings. Because those counts come from lattice points in rational polytopes, the generic degree is a quasi-polynomial in p of degree 3g-3. For genus 3 the paper evaluates it explicitly as (1/45)(2p^6+5p^4+38p^2).","feed_headline":"Frobenius pull-back degree equals a ratio of edge-number counts","feed_subtitle":"A bridge from rank-two bundles to dormant opers turns the unknown degree into finite combinatorial data.","key_machinery":"The load-bearing comparison is the commutative square between the tangent map of the generalized Verschiebung map and the tangent map of level reduction of dormant PGL_2-opers: at corresponding points these tangent maps are canonically isomorphic. A dormant PGL_2-oper of level N is, roughly, a flat bundle of level N-1 with vanishing p^N-curvature and a line subbundle that generates the rank-two bundle under differential operators, taken modulo tensoring with flat line bundles. On a totally degenerate curve these opers correspond bijectively to balanced (p,N)-edge numberings, i.e. nonnegative integers on the edges of the dual trivalent graph such that every vertex triple satisfies triangle-ty","core_discovery":"The core claim is that the generic degree of the generalized Verschiebung map Ver^2_1 is controlled by the degree of the projection Π_N from the moduli stack of dormant PGL_2-opers of level N to the moduli of curves. For a general curve the paper establishes the identity deg(Π_N) = deg(Π_1) · deg(Ver^2_1)^{N-1}, and a corresponding inequality for non-general curves. The proof works by canonically identifying the differential of Ver^2_{N⇒N'} at a maximally Frobenius-destabilized bundle with the differential of the level-reduction map Π_{N⇒N'} at the associated dormant oper; since the two maps are étale over the same points, their generic degrees coincide. Combining this with the existing fact","pith_inferences":["If the tangent-space comparison extends to rank n stable bundles, the same procedure would express deg(Ver^n_1) as a ratio of counts of dormant PGL_n-opers of levels 1 and 2, turning the higher-rank conjecture into a finite polytope count.","The ratio #Ed_{p,2,G}/#Ed_{p,1,G} being independent of the chosen trivalent graph is a strong constraint; testing both counts on non-isomorphic genus-g graphs would give a purely combinatorial check of the whole correspondence.","The quasi-polynomial form implies divisibility structure: for instance, the genus-3 formula is divisible by p^2 for every odd prime, so an independently computed degree for a single new characteristic would test the fitted coefficients.","One could formally evaluate the genus-3 formula at p=2, outside the stated hypothesis p>2, to obtain a predicted value that either extends the theorem or signals where the oper correspondence needs modification."],"forward_implications":["For every genus g and every odd prime p, the generic degree of Ver^2_1 can in principle be computed by counting balanced edge numberings on any trivalent graph of genus g.","The generic degree is a quasi-polynomial in p of degree 3g-3, with leading coefficient (-1)^g 2^{3g-4} B_{2g-2}/(2g-2)!, describing its asymptotic growth as p grows.","The rank-two case of the second author's conjecture is settled: for a general curve, deg(Ver^2_1) equals the large-N limit of deg(Π_N)^{1/N}.","The known genus-two formula (p^3+2p)/3 is recovered as a special case of the ratio formula.","For genus 3, the exact polynomial (1/45)(2p^6+5p^4+38p^2) gives concrete degree values for every odd prime characteristic."],"supporting_citations":[{"why":"Supplies the generic étaleness of the level-N oper moduli stack over M_g and the balanced-edge-numbering description of deg(Π_N); the main external input the paper builds on.","marker":"[Wak7]"},{"why":"Provides the foundational theory of dormant PGL_2-opers and the count behind the genus-2 formula.","marker":"[Moc]"},{"why":"Gives the explicit formula for deg(Π_1) used in the final equality of Corollary 4.11.","marker":"[Wak1]"},{"why":"Contains the earlier maximally Frobenius-destabilized/dormant SL_2-oper equivalence that Proposition 4.3 refines.","marker":"[Wak6]"},{"why":"Supplies the N=1 correspondence between dormant opers and Frobenius-destabilized bundles that the paper generalizes to higher level.","marker":"[JoXi]"},{"why":"Earlier analysis of the generalized Verschiebung map, including the genus-2 computation and the pull-back stability criterion used in Section 2.","marker":"[Oss1]"},{"why":"Gives the prior upper bound on deg(Ver^2_1) that the exact formula improves.","marker":"[HoWa]"},{"why":"Establishes generic étaleness of the Verschiebung map for general X, used to identify degrees with fiber counts.","marker":"[MeSu]"}],"fun_headline_variants":["Verschiebung degree via dormant opers count","Explicit rank-two Verschiebung from opers","Counting opers fixes Frobenius degree","Dormant opers unlock Verschiebung degree","Generic degree of Verschiebung: combinatorial"],"cache_read_input_tokens":28800,"weakest_assumption_plain":"The equality deg(Π_N)=deg(Π_1)deg(Ver^2_1)^{N-1} for general curves depends on the previously established generic étaleness of the oper stack Π_N over the moduli of curves, a result quoted from an earlier preprint rather than proved here; if that étaleness failed for some N, the fiber count would not equal the degree and the ratio formula would collapse.","fun_headline_variants_meta":{"raw":{"variants":["Verschiebung degree via dormant opers count","Explicit rank-two Verschiebung from opers","Counting opers fixes Frobenius degree","Dormant opers unlock Verschiebung degree","Generic degree of Verschiebung: combinatorial"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000277,"raw_usage":{"total_tokens":1442,"prompt_tokens":652,"completion_tokens":790,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":396,"completion_tokens_details":{"reasoning_tokens":729}},"tokens_in":396,"tokens_out":790,"duration_ms":8078,"temperature":1.0,"reasoning_tokens":729,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-05T10:28:29.336238+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"For a general genus-3 curve in characteristic 3 the formula predicts deg(Ver^2_1)=49; an independent computation of this generic degree, or a direct verification that a genus-3 trivalent graph has exactly 49 balanced (3,2)-edge numberings and 1 balanced (3,1)-edge numbering, would settle the central claim.","supporting_citations":[],"review_version":1}