Polynomiality of the Generalized Verschiebung Degree

Siqing Zhang · 2026

Read Polynomiality of the Generalized Verschiebung Degree by Siqing Zhang (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2606-26070.txt.

Introduction

Siqing Zhang works in algebraic geometry on curves over fields of positive characteristic. Pulling bundles back along the Frobenius map defines a rational map between moduli spaces. The map, called the generalized Verschiebung, links spaces of rank-two vector bundles with trivial determinant. Its generic degree varies with the field's characteristic. Kondo and Wakabayashi recently showed that variation follows a quasi-polynomial law.

Zhang proves the law is simpler than it looked. The quasi-polynomial is in fact a genuine polynomial, for every genus from two upward. The paper writes the polynomial out explicitly. Earlier cases fit the pattern: genus two gives a known cubic-style formula in the characteristic. Kondo and Wakabayashi had verified genus three by direct computation. The general proof replaces case-checking with one argument. Degree counts for Frobenius pullbacks across all genera now have a closed form.

Catalog record

CREATORSiqing Zhang
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE27 KB
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2606.26070

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Polynomiality of the Generalized Verschiebung Degree Siqing Zhang Abstract . For a general curve in positive characteristic, taking the Frobenius pullback induces a generically finite rational map V V on the moduli space of rank 2 vector bundles with trivial determinant. Recently, Kondo–Wakabayashi show that the generic degree of V V , considered as a function on the characteristic of the base field, is a quasi-polynomial. In this paper, we show that this quasi-polynomial is indeed a polynomial, and we write out this polynomial explicitly. 1. Introduction Let X be a smooth curve in characteristic p>2 . Let F_{X}:X\to X^{\prime} be the relative Frobenius morphism over k . Let M_{2}(X) and M_{2}(X^{\prime}) be the moduli space of stable rank 2 vector bundles of trivial determinant on X and X^{\prime} respectively. Taking the Frobenius pullback E\mapsto F_{X}^{*}E defines a generically finite rational map V:M_{2}(X^{\prime})\dashrightarrow M_{2}(X) , called the generalized Verschiebung [ 13 , Theorem A.6] . The behavior of this rational map has been studied extensively; see, e.g. [ 2 , 8 , 9 , 13 , 6 , 7 , 5 , 4 ] . When X is a general curve of genus g=2 , the generic degree of V is \deg(V)=\frac{p^{3}+2p}{3} , as shown by Osserman [ 13 , Theorem 1.3] and Lange–Pauly [ 7 , Corollary] . For higher genus, Kondo–Wakabayashi recently expressed the generic degree in terms of certain edge-numberings on a graph [ 3 , Theorem 5.3] . They also show that \deg(V) , as a function of the characteristic p , is given by a quasi-polynomial in p of degree 3g-3 [ 3 , Theorem 5.5] . They then use direct computation to show that this quasi-polynomial is a polynomial for g=3 [ 3 , Theorem C] . The first result of this paper shows that the quasi-polynomial \deg(V) is always a polynomial in p for arbitrary genus g\geq 2 . Theorem 1 (Polynomiality) . Let X be a general curve of genus g\geq 2 over an algebraically closed field of characteristic p>2 . Let R_{g}(t)\in\mathbb{Q}[t] be the following polynomial: R_{g}(t)=1-\frac{1}{2^{g}}+\frac{1}{2^{g}}\sum_{r=0}^{g-1}\frac{(-1)^{r-1}2^{4r}B_{2r}}{(2r)!}\left[z^{-2r}\right]\csc^{2g-2}(z)\,t^{r}, where B_{2r} denotes the 2r -th Bernoulli number, and \left[z^{-2r}\right]\csc^{2g-2}(z) denotes the coefficient of z^{-2r} in the Laurent expansion of \csc^{2g-2}(z) at z=0 . Then \deg(V)=p^{g-1}R_{g}(p^{2}). The proof relies on a new combinatorial result in Wakabayashi’s enumerative geometry of dormant opers , see [ 17 ] and [ 18 , §1] for an illuminating introduction of the related ideas. Namely, given a finite connected trivalent graph G of genus g , and two natural numbers P\geq 3 and N\geq 1 , Kondo–Wakabayashi define a finite set \operatorname{Ed}_{P,N,G} of certain edge-numberings on G and show that (1) \deg(V)=\frac{|\operatorname{Ed}_{p,2,G}|}{|\operatorname{Ed}_{p,1,G}|}. The set \operatorname{Ed}_{P,N,G} is recalled in Definition 8 . When N=1 , the set \operatorname{Ed}_{P,1,G} , implicit already in works of S. Mochizuki, is given by the lattice points in a dilation of a polytope defined by Liu-Osserman in [ 10 , Definition 2.3] . Our combinatorial result reduces the level- N edge-numbering sets to the level-1 counterparts. Theorem 2 (Level-reduction) . Let P\geq 3 be an odd integer and N\geq 1 be an integer, and let G be a finite trivalent graph. Let \operatorname{Ed}_{P,N,G} be the set of (P,N) -edge-numberings of G as in Definition 8 . Then, there is a natural bijection \operatorname{Ed}_{P,N,G}\cong\operatorname{Ed}_{P,1,G}\times\operatorname{Ed}_{2P,1,G}^{N-1}. Combinatorial background In his work on p -adic Teichmüller Theory [ 11 ] , S. Mochizuki linked the nowadays so-called dormant opers to the uniformization problem of p -adic curves, and showed that the number of dormant opers in characteristic p is given by |\operatorname{Ed}_{p,1,G}| , see Liu-Osserman’s summary in [ 10 , Theorem 3.9] . By Ehrhart Theory [ 15 , Theorem 4.6.8] , the number |\operatorname{Ed}_{P,1,G}| of elements of \operatorname{Ed}_{P,1,G} , considered as a function of P , is a priori a quasi-polynomial . That is, a function f:\mathbb{Z}\to\mathbb{R} of the form f(n)=\sum_{i=0}^{M}c_{i}(n)n^{i} , where each c_{i}(n) is a periodic function for n . Liu-Osserman show the polynomiality of |\operatorname{Ed}_{P,1,G}| for odd P in [ 10 , Corollary 3.6] . In [ 1 , Theorem 5] , the polynomiality of |\operatorname{Ed}_{P,1,G}| for even P is shown by Fernandes–de Pina–Ramírez Alfonsín–Robins. The derivation of the explicit formulas for the odd and even polynomials have a different flavor. The polynomial for |\operatorname{Ed}_{P,1,G}| when P is odd is given by Wakabayashi in [ 16 , Theorem A] as |\operatorname{Ed}_{P,1,G}|=\frac{P^{g-1}}{2^{2g-1}}\sum_{\theta=1}^{P-1}\csc^{2g-2}\!\left(\frac{\pi\theta}{P}\right)\hskip 20.00003pt(P\ \mathrm{odd}). His proof passes through Quot schemes and Gromov-Witten Theory. In the proof of Theorem 1 (see ( 13 )), we give an elementary derivation of the corresponding formula for even P : |\operatorname{Ed}_{P,1,G}|=\left(\frac{P}{2}\right)^{g-1}+\left(\frac{P}{4}\right)^{g-1}\sum_{\theta=1}^{P/2-1}\csc^{2g-2}\!\left(\frac{\pi\theta}{P}\right)\hskip 20.00003pt(P\ \mathrm{even}). Remark 3 . Theorem 2 and ( 1 ) give us \deg(V)=|\operatorname{Ed}_{2p,1,G}| . Therefore, we have the following heuristic observation, which we find amusing: The generic degree \deg(V) in characteristic p agrees with the number of dormant opers over a nonexistent “field of characteristic 2p ”. Formalization The combinatorial cores of this paper, namely, the proofs of Lemma 7 , Theorem 2 , and Lemma 10 , have been formalized and verified in Lean 4, see [ 20 ] . Outline We first prove Theorem 2 in § 2 . We then prove Theorem 1 in § 3 . Finally, we briefly recall the notion of dormant opers and record the related Corollary 11 in § 4 . 2. Proof of Theorem 2 The set \operatorname{Ed}_{P,N,G} used by Kondo–Wakabayashi in [ 3 ] is given by certain numberings of the edges of G , such that, at each vertex of G , the numbers are constrained by a set {}^{\dagger}C_{N}(P) . In § 2.1 , we recall the set {}^{\dagger}C_{N}(P) and prove a relation among the sets when we change N ’s and P ’s. In § 2.2 , we recall the set \operatorname{Ed}_{P,N,G} and use § 2.1 to prove Theorem 2 . 2.1. The set {}^{\dagger}C_{N}(P) as in [ 3 , §5.1] For integers P\geq 2 and N\geq 1 , let {}^{\dagger}C_{N}(P) denote the set of triples (a_{1},a_{2},a_{3})\in\mathbb{Z}_{\geq 0}^{3} such that (2) a_{1}+a_{2}+a_{3}\leq P^{N}-2,\hskip 10.00002pt|a_{2}-a_{3}|\leq a_{1}\leq a_{2}+a_{3}, and such that, for every 1\leq M\leq N , if we let [a_{i}]_{M} be the remainder of a_{i} modulo P^{M} , then there are choices (3) a_{i,M}\in\{[a_{i}]_{M},\,P^{M}-1-[a_{i}]_{M}\} such that (4) a_{1,M}+a_{2,M}+a_{3,M}\leq P^{M}-2,\hskip 10.00002pt|a_{2,M}-a_{3,M}|\leq a_{1,M}\leq a_{2,M}+a_{3,M}. When N=1 or when M=N , we can choose a_{i,N}=a_{i} and the condition ( 4 ) is redundant. Note that the definition of {}^{\dagger}C_{N}(P) is independent of the ordering of the triple (a_{1},a_{2},a_{3}) . We record a lemma that will be useful later on. The proof is straightforward. Lemma 4 . The condition ( 2 ) is equivalent to the following four inequalities \displaystyle 2P^{N}-(2a_{1}+1)-(2a_{2}+1)-(2a_{3}+1) \displaystyle>0, \displaystyle-(2a_{i}+1)+\sum_{j\neq i}(2a_{j}+1) \displaystyle>0,\hskip 10.00002pti=1,2,3. Example 5 . {}^{\dagger}C_{1}(3) is the singleton \{(0,0,0)\} . {}^{\dagger}C_{2}(3) is \{(0,0,0)\}\cup S_{3}\cdot(2,2,0)\cup\{(2,2,2)\}\cup S_{3}\cdot(3,2,2)\cup S_{3}\cdot(3,3,0), where S_{3}\cdot x means the orbit of x under the action of S_{3} by permuting the coordinates. {}^{\dagger}C_{1}(6) is \{(0,0,0)\}\cup S_{3}\cdot(1,1,0)\cup\{(1,1,1)\}\cup S_{3}\cdot(2,1,1)\cup S_{3}\cdot(2,2,0). Note that, in this example, we have a bijection {}^{\dagger}C_{2}(3)\cong{}^{\dagger}C_{1}(3)\times{}^{\dagger}C_{1}(6) . More explicitly, this bijection sends every instance of 2 in {}^{\dagger}C_{1}(6) to 3\times\frac{2}{2}+0 in {}^{\dagger}C_{2}(3) and every instance of 1 in {}^{\dagger}C_{1}(6) to 3\times\frac{1+1}{2}-1-0 . Below we show that this bijection generalizes. Definition 6 . For each (n,b)\in\mathbb{Z}^{2} , let \Phi(n,b)\in\mathbb{Z} be (5) \Phi(n,b)=\begin{cases}\frac{n}{2}P^{N-1}+b,&n\ \text{even},\\

Read the complete text