{"id":"dfc4a2bb-33d4-4fba-ac2f-108da7edfbab","arxiv_id":"2509.03982","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"For every level N, the moduli space of pointed stable curves with a dormant PGL_2^{(N)}-oper is irreducible when nonempty, and the space of p^N-nilpotent opers is connected.","lead":"This paper proves that, for any level N, the moduli space of stable curves equipped with a dormant PGL_2^{(N)}-oper (a special flat bundle in characteristic p) is irreducible, that is, it is a single piece. The result generalizes Mochizuki's classical N=1 theorem and is a step toward understanding how these arithmetic objects vary as the level N grows.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Proposition 3.10's one-sentence edge-numbering lift from level N to N+1 is the step on which the induction rests; a defect in the cited [Wak14, Prop. 10.18] at this point would collapse Lemma 4.5.","rationale":"I find the paper's internal chain coherent after correcting the evident level-index typos (e.g., Lemma 4.5's erho should lie in Xi^r_{2,N+1}; Corollaries 3.20 and 3.24 have N/N+1 shifted). The proof of Theorem 4.6 is a well-organized double induction, and the gluing diagrams in Lemma 4.5 do connect components assuming the level-lifting machinery. But that machinery is not self-contained: Proposition 3.10's key assertion is a single sentence referring to [Wak14, Prop. 10.18], and Theorem 3.8(iii) is similarly cited. Since the paper is a continuation of the author's own program, disagreement with consensus is not the issue; the concern is verification burden. The reader's CONDITIONAL verdict already captures this, so I would not change it. A targeted combinatorial check for p=3, N=1 to 2 would resolve whether the level-lifting step is valid.","tokens_in":56224,"tokens_out":15674,"duration_ms":151663,"concrete_test":"For p=3, N=1 to 2, compute explicitly for the trivalent dual graphs of types (0,4) and (1,1) the sets of balanced (p,N)- and (p,N+1)-edge numberings from [Wak14, Def. 10.17 / Prop. 10.18]. Check that the map sending a balanced (p,N)-numbering (a_e) to the same integer labels at level N+1, with leaf radii (2a_{e_i}+1)/2, lands in the balanced (p,N+1) set, and that every balanced (p,N+1)-numbering reduces to a balanced (p,N)-numbering. If either fails for one example, Proposition 3.10 is false as stated and Theorem 4.6 is not established.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central induction (Lemma 4.5 -> Theorem 4.6) steps from level N to N+1 only through Theorem 3.23, whose hypotheses are supplied by Proposition 3.14, which in turn uses Proposition 3.10. Proposition 3.10 is not really proved: after passing to a totally degenerate curve, it invokes [Wak14, Prop. 10.18] to identify dormant PGL_2^{(N)}-opers with balanced (p,N)-edge numberings, and then asserts in one sentence that the same integer edge-collection (a_e) is a balanced (p,N+1)-edge numbering with lifted radii (2a_{e_i}+1)/2. That assertion is exactly the level-N to N+1 lifting theorem needed for the induction; it is not derived here. If the balance condition at level N+1 involves p^{N+1}-adic congruences not implied by level N, or if the bijection of Prop. 10.18 is not compatible with the natural level-reduction map, then for some nonempty Op^{Zzz...}_{N,rho,g,r} there may be no nonempty Op^{Zzz...}_{N+1,erho,g,r}. Proposition 3.14, Theorem 3.23, and Lemma 4.5 would then lose their lifting step. This is a black-box dependency, not an internal contradiction, but it is the load-bearing point: all dimension-counting and surjectivity arguments in Section 4 sit on it.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper studies moduli stacks of pointed stable curves equipped with PGL_n^{(N)}-opers in characteristic p, i.e. flat PGL_n-bundles with an action of differential operators of level N-1. It introduces a higher-level Hitchin-Mochizuki morphism defined by the characteristic polynomial of the p^N-curvature, proves that this morphism is finite and faithfully flat under suitable hypotheses (Theorems 3.21 and 3.23), and then uses degeneration and induction on (N,dim) to prove the central result: for n = 2, the stack Op^{Zzz...}_{N,ρ,g,r} of dormant PGL_2^{(N)}-opers of radius ρ is irreducible whenever nonempty, and Op^{nilp}_{N,ρ,g,r} is connected whenever nonempty (Theorem 4.6 / Theorem A). Applications are given to Frobenius-destabilized vector bundles and to deformations of dormant opers to prime-power characteristic, and an appendix relates dormant generic Miura opers on elliptic curves to Igusa structures.","tokens_in":1610,"tokens_out":1628,"duration_ms":173344,"significance":"If correct, the main theorem is a substantial characteristic-p analogue of Mochizuki's N=1 irreducibility result and gives a higher-genus analogue of Igusa curves. The paper is well structured, with explicit dimension counts, a clear induction on (N,d), and genuine applications to Frobenius pull-backs and flat bundles. Credit is due for the detailed proofs of the auxiliary cohomological statements (Lemma 3.13, Proposition 3.11) and for stating the nonemptiness hypotheses explicitly. The main caveat is that several load-bearing inputs are quoted from the author's own prior work [Wak14], and at least one step in the present text—the level-N to level-(N+1) lifting in Proposition 3.10—is asserted rather than proved.","major_comments":[{"comment":"This proposition is the hinge for the whole induction: it produces a level-(N+1) lift eρ of a nonempty dormant moduli stack. The proof passes to a totally degenerate curve and applies [Wak14, Prop. 10.18], obtaining a balanced (p,N)-edge numbering (a_e). It then asserts in one sentence that the same (a_e) 'can be regarded' as a balanced (p,N+1)-edge numbering with radii (2a_{e_i}+1)/2 in (Z/p^{N+1}Z)^×/{±1}. That assertion is exactly the compatibility between the level-N and level-(N+1) classifications; [Wak14, Prop. 10.18] is a bijection at each fixed level and does not, by itself, imply that the level-N numbering is admissible at level N+1. If the balance condition at level N+1 involves congruences modulo p^{N+1} not implied by level N, then Proposition 3.14, Theorem 3.23, and Lemma 4.5 lose their lifting step. Please prove this compatibility explicitly or quote a precise statement fro","section":"§3.3, Proposition 3.10"},{"comment":"The proof of the local section property uses flatness of the composite (3.31), justified by the phrase 'both stacks are smooth and of the same dimension' together with [Har, III, Exercise 10.9]. That exercise requires control of fiber dimensions; the required quasi-finiteness is not stated. One can repair the argument by observing that the fiber over a point of Op^{Zzz...}_{N,ρ,g,r} is contained in the finite fiber of Π'_{eρ,g,r} over M_{g,r}, so the composite is proper and quasi-finite, hence finite, and then use Miracle Flatness. As written, however, the proof skips this step. Since Proposition 3.14 is used to supply the hypotheses of Theorem 3.23, this gap is load-bearing and should be fixed by expanding the argument.","section":"§3.5, Proposition 3.14"}],"minor_comments":[{"comment":"The statement says 'let eρ be an element of Ξ^r_{2,N} satisfying Op^{nilp}_{N+1,eρ,g,r} ≠ ∅'. The radii eρ should lie in Ξ^r_{2,N+1}, not Ξ^r_{2,N}; otherwise Λ_{eρ,g,r} is ill-typed. The same typo appears in Theorem 3.23 ('Ξ^r_{n,N}' should be 'Ξ^r_{n,N+1}').","section":"§4.3, Lemma 4.5 statement"},{"comment":"The statement contains 'Op^{nilp}_{N+1,ρ,0,3}' where the subscript should be eρ; the proof itself uses the correct notation. Please correct the typo.","section":"§3.6, Corollary 3.24"},{"comment":"There is a typo 'ζ ∈ Ξ_{2,Ξ}' in the g≥1 case; it should be 'ζ ∈ Ξ_{2,N}'. Also Op^{Zzz}_{2,N+1,eρ,g,r} appears in the proof of Proposition 3.14 and should be Op^{Zzz}_{N+1,eρ,g,r}.","section":"§4.3, proof of Lemma 4.5"},{"comment":"The proof invokes [Wak4, Proposition 4.55] for the existence of a normal dormant (GL_2^{(N)},ϑ)-oper at level N, while the cited statement is formulated for N=1. If it generalizes to all N, a remark to that effect would be helpful; if not, a separate proof is needed.","section":"§4.2, Lemma 4.4"},{"comment":"The proof says the assertion follows from 'a standard argument' using Theorem 4.6 and Theorem 3.8. Since Proposition 4.7 is used in the applications, a slightly more explicit argument (or a reference to the exact combination of finiteness, flatness, and generic étaleness) would improve readability.","section":"§4.4, Proposition 4.7"}],"recommendation":"major_revision","confidential_remarks":"The paper is part of a long chain of the author's own works, and the decisive black box is [Wak14], itself a large preprint. The refereeing difficulty is not the internal consistency of the present text, which is generally careful, but the unproved compatibility in Proposition 3.10. Acceptance should be conditional on making that step independently verifiable: either by a proof in this paper or by an explicit quotation of the precise statement in [Wak14] that implies it. I would suggest requiring this before publication, as the induction in Theorem 4.6 rests on it."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Dear [Colleague],\n\nHere is my read on Wakabayashi's arXiv:2509.03982.\n\nThe headline: this is a serious paper with a genuinely new result—irreducibility and connectedness for the moduli stacks of dormant and p^N-nilpotent PGL_2^{(N)}-opers at every level N, for n=2—and it introduces a new tool, the level-N Hitchin-Mochizuki morphism (Definition 3.22). The main theorem (Theorem 4.6) extends Mochizuki's N=1 irreducibility, and the induction strategy is transparent: base cases from Mochizuki plus the dimension 0/1 cases, then Lemma 4.5 steps up dimension via gluing, using the new morphism to get finite flat level-reduction maps. The proof of the central chain is for the most part detailed and coherent; the applications to Frobenius-destabilized bundles and to lifting dormant opers to prime-power characteristic are natural and useful.\n\nNow the soft spots, in proportion.\n\nThe load-bearing step is Proposition 3.10, which lifts a dormant PGL_2^{(N)}-oper to level N+1. The proof is one sentence: after passing to a totally degenerate curve, it invokes the bijection from [Wak14, Prop. 10.18] and asserts that the same edge-index collection is a balanced (p,N+1)-edge numbering with lifted radii. Everything downstream—Proposition 3.14, Theorem 3.23, and hence Lemma 4.5 and the induction—rests on that assertion. It is not derived here, and it is exactly the level-lifting theorem needed. If the balance condition at level N+1 involves congruences not implied by level N, or if the [Wak14] bijection is not compatible with level reduction, the induction loses its step. This is a black-box dependency, not an internal contradiction, but it is exactly where a referee needs to dig.\n\nThere are also several auxiliary proofs that are sketches: Lemma 3.16's proof is a paragraph, and Proposition 4.7 is described as 'standard.' I found level-index inconsistencies in Lemma 4.5, Theorem 3.23, and Corollary 3.20; the surrounding text makes the intended reading clear, so those look like typos or rendering artifacts, not mathematical errors. The paper is also heavily self-referential, drawing on the author's own [Wak14] for the foundational geometry of these moduli spaces; that is not intrinsically bad, since [Wak14] is a real preprint and much of it is likely verifiable, but it does concentrate the risk.\n\nWho is this for? Specialists in p-adic Teichmüller theory and characteristic-p opers. If the lifting step checks out, this is a solid and important contribution. It deserves a serious referee—the main theorem is new, the strategy is traceable, and the soft spots are concentrated in known places.\n\nRecommendation: send to peer review, but ask the referee to verify Proposition 3.10 against [Wak14] carefully.","headline":"A genuinely new higher-level irreducibility theorem that is well worth refereeing, but its level-lifting step is currently a one-sentence black box.","tokens_in":57200,"tokens_out":2578,"would_cite":true,"duration_ms":23708,"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":"The moduli stack of pointed stable curves with dormant PGL_2^(N)-opers is irreducible whenever it is nonempty, for all levels N, all genera, and all admissible radii.","keywords":["dormant oper","PGL_2-oper","moduli stack","Hitchin-Mochizuki morphism","p-curvature","irreducibility","Frobenius-destabilized bundles","positive characteristic"],"falsifier":"Check the one-dimensional base cases directly: for a fixed level N≥2 and nonempty radius tuple ρ, compute the irreducible components of Op^{Zzz}_{N,ρ,1,1} and Op^{Zzz}_{N,ρ,0,4}. On a totally degenerate fiber, the balanced (p,N)-edge numberings give a finite explicit set of local components; if two such components survive as disjoint closed substacks, the theorem is false. This is exactly where the proof's Proposition 4.3 rules out a trivial line subbundle, so an explicit component count there would settle the claim.","tokens_in":55889,"feed_emoji":"","tokens_out":12805,"duration_ms":115753,"temperature":0.7,"pith_summary":"This paper proves that the moduli stack of pointed stable curves equipped with a dormant PGL_2^(N)-oper—a flat projective-linear bundle in prime characteristic whose level-N p-curvature vanishes identically—is irreducible whenever it is nonempty, for every level N and every stable genus with marked points. The larger p^N-nilpotent locus is connected, with any two irreducible components intersecting. The proof works by constructing a higher-level Hitchin-Mochizuki morphism from the characteristic polynomials of p^N-curvatures; its zero fiber is the nilpotent locus, and its level-reduction map is finite, flat, and surjective. From irreducibility the paper derives lifting and deformation consequences, including that dormant opers can be raised to higher level over a finite cover, and that maximally Frobenius-destabilized stable bundles are Frobenius pull-backs. If correct, these moduli spaces—higher-genus analogues of Igusa curves and p-adic indigenous bundles—each have exactly one component.","feed_headline":"Dormant opers of every level form one irreducible moduli space","feed_subtitle":"A Hitchin-Mochizuki map built from p^N-curvature proves uniqueness and yields Frobenius-lifting results","key_machinery":"The engine is the level-N Hitchin-Mochizuki morphism HM^{(N)}_{eρ}: Op_{N+1,eρ,g,r} → B^♭_{univ,0}, a map to a relative affine Hitchin base taking a level-(N+1) oper to the characteristic polynomial of its p^N-curvature. Its inverse image of the zero section is exactly the p^N-nilpotent locus Op^{nilp}_{N+1,eρ,g,r}; the induced level-reduction map Λ_{eρ,g,r}: Op^{nilp}_{N+1,eρ,g,r} → Op^{Zzz}_{N,ρ,g,r} is finite, faithfully flat, and surjective. This transfers connectedness and irreducibility from lower level to higher level and from smaller curves to larger ones through tree and loop gluing morphisms, making the induction on dimension work.","core_discovery":"The central claim (Theorem 4.6 = Theorem A): for n=2, the stack Op^{Zzz}_{N,ρ,g,r} of pointed stable curves equipped with dormant PGL_2^(N)-opers of radii ρ is irreducible whenever nonempty, and the p^N-nilpotent stack Op^{nilp}_{N,ρ,g,r} is connected, with any two irreducible components meeting. A dormant PGL_2^(N)-oper is a flat PGL_2-bundle with a level-N differential-operator action and vanishing p^N-curvature; ρ records local monodromy at marked points. The proof uses the known smooth proper structure of Op^{Zzz}_{N,ρ,g,r}, builds a level-N Hitchin-Mochizuki morphism whose zero fiber is the nilpotent locus and whose level-reduction map is finite, flat, surjective, then glues curves and","pith_inferences":["One consequence the paper does not spell out: the projective tower of level-reduction maps is a tower of finite flat surjections between irreducible stacks, so it is natural to ask whether its limit is a single p-adic analytic object; irreducibility at each finite level is compatible with, but does not by itself prove, uniqueness of the limit.","The same induction strategy might apply to PGL_n^(N)-opers for n>2 if the combinatorial edge-numbering classification and the dimension-1 base cases are supplied; the paper proves only n=2, and its appendix indicates the elliptic-curve case is governed by Igusa structures.","The finite flat generically étale forgetful maps now have a well-defined generic degree for every level and radius; combined with the 2d TQFT factorization mentioned in the introduction, irreducibility makes the total-count invariant of dormant opers on a general curve insensitive to the order of gluing, which could be tested explicitly in low genus."],"forward_implications":["For every nonempty Op^{Zzz}_{N,ρ,g,r}, all dormant opers of that level and radius lie on a single irreducible component; the stack is smooth, proper, and of dimension 3g−3+r.","The nilpotent stack Op^{nilp}_{N,ρ,g,r} is connected: any two irreducible components meet, so the moduli space has no disconnected pieces.","Every dormant PGL_2^(N')-oper on a genus >1 curve can be lifted to level N (N ≥ N') after a finite, flat, generically étale base change (Corollary 4.8).","Every stable rank-2 degree-0 bundle that is maximally Frobenius-destabilized (its iterated Frobenius pull-back acquires a degree g−1 line subbundle) lies in the image of Frobenius pull-back, hence underlies an F-divided sheaf and carries a D^(∞)-module structure (Props. 5.2, 5.3).","For a general curve, dormant PGL_2-opers deform from characteristic p^{N'} to p^N, and second-order differential operators with full root-function sets lift to W_N (Props. 5.5, 5.6)."],"supporting_citations":[{"why":"Supplies the foundational structure theorem—smoothness, properness, dimension, finite faithfully flat generically étale projection—and the balanced edge-numbering classification of dormant opers on totally degenerate curves; the induction's base and lifting step both rest on it.","marker":"[Wak14]"},{"why":"Establishes the N=1 case of irreducibility and connectedness that serves as the base of the induction, and provides the p-adic Teichmüller framework being generalized.","marker":"[Moc2]"},{"why":"Provides the N=1 theory of dormant opers, the oper/flat-bundle dictionary, normal opers used in the dimension-1 argument, and the characteristic-polynomial Hitchin morphism being lifted to higher level.","marker":"[Wak4]"},{"why":"Gives the irreducibility of the moduli stack M_{g,r} used to conclude that nonempty Op^{Zzz} maps surjectively to M_{g,r}, needed for degeneration arguments.","marker":"[DeMu]"},{"why":"One of the sources of the level-one Hitchin-Mochizuki morphism that the paper generalizes to arbitrary level.","marker":"[BeTr]"},{"why":"Supplies the p-curvature/Hitchin base technique and the characteristic polynomial construction that the higher-level morphism extends.","marker":"[LasPa1]"},{"why":"Provides the Hitchin-Mochizuki morphism and Frobenius-destabilized bundle viewpoint that motivate the applications.","marker":"[JoPa]"}],"fun_headline_variants":["All dormant opers, any level, one irreducible moduli space","Irreducible moduli for every level of dormant opers","Vanishing p^N-curvature yields one irreducible moduli","Hitchin-Mochizuki map proves dormant opers irreducible"],"cache_read_input_tokens":2688,"weakest_assumption_plain":"The proof treats as a black box the earlier theorem that dormant PGL_2^(N)-opers form a smooth proper moduli stack of the expected dimension with generically étale forgetful map, together with the classification of such opers on totally degenerate curves by balanced edge numberings; if that theorem fails at any level, the induction has no base and no lifting step.","fun_headline_variants_meta":{"raw":{"variants":["All dormant opers, any level, one irreducible moduli space","Irreducible moduli for every level of dormant opers","Vanishing p^N-curvature yields one irreducible moduli","Hitchin-Mochizuki map proves dormant opers irreducible"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000983,"raw_usage":{"total_tokens":4009,"prompt_tokens":746,"completion_tokens":3263,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":490,"completion_tokens_details":{"reasoning_tokens":3189}},"tokens_in":490,"tokens_out":3263,"duration_ms":24207,"temperature":1.0,"reasoning_tokens":3189,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-05T10:31:38.117222+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Check the one-dimensional base cases directly: for a fixed level N≥2 and nonempty radius tuple ρ, compute the irreducible components of Op^{Zzz}_{N,ρ,1,1} and Op^{Zzz}_{N,ρ,0,4}. On a totally degenerate fiber, the balanced (p,N)-edge numberings give a finite explicit set of local components; if two such components survive as disjoint closed substacks, the theorem is false. This is exactly where the proof's Proposition 4.3 rules out a trivial line subbundle, so an explicit component count there would settle the claim.","supporting_citations":[],"review_version":1}