Searcharxiv⌕ Search

arXiv subjects

N. I. Shepherd-Barron

Publications and source records attributed to N. I. Shepherd-Barron.

15 recordsLinked to original sources

A non-linear differential equation for the periods of elliptic surfaces

Suppose that $f:X\to C$ is a general Jacobian elliptic surface over the complex numbers. Then the primitive cohomology $H^{1,1}_{prim}(X)$ has, up to a sign, a natural orthonormal basis $(η_i)_{i\in [1, N]}$ given by certain meromorphic $2$-forms $η_i$ of the second kind, one for each ramification point of the classifying morphism $ϕ$ from $C$ to the stack of generalized elliptic curves. (Here $N$ is any one of $h^{1,1}_{prim}(X)$, the number of moduli of $X$ and the degree of the ramification of $ϕ$; these numbers are equal.) A choice of local co-ordinate on the stack of elliptic curves provides, via the branch locus of $ϕ$, an {é}tale local co-ordinate system $(t_i)_{i\in [1, N]}$ on the stack of Jacobian elliptic surfaces. The main result here is that truncation of the Gauss--Manin connexion yields the system $$\{\partial_i H=(\partial_i η_i\wedgeη_i)H\}_{i\in [1, N]}$$ of non-linear pde satisfied by $H=[η_1,\ldots, η_N]$, where $\partial_i =\partial/\partial t_i$ and the skew tensor $\partial_i η_i\wedgeη_i$ of rank $2$ is the ecliptic of $η_i$ (the plane in which the particle $η_i$ is instantaneously moving with respect to $t_i$). Moreover, after rigidification of the integral cohomology, $H$ can be interpreted as providing a period map for these surfaces with values in the complex orthogonal group $O_N$, and we prove a generic infinitesimal Torelli theorem for this map. For rational elliptic surfaces this can be calculated explicitly.

math.AG↗

Recognizing flag varieties and reductive groups

Fix a flat and projective morphism $X\rightarrowΣ$ of schemes. We show, first, that any set of $\mathbb{P}^1$-fibrations on $X$ defines a set of simple roots, a set of simple coroots and a Cartan matrix $C$. Second, $X$ is an étale $F$-bundle over some projective $Σ$-scheme, where $F$ is the flag variety of the adjoint Chevalley group over the integers defined by $C$. In particular, if the simple roots generate the Néron--Severi group of $X$ relative to $Σ$ and $X$ is cohomologically flat in degree zero over $Σ$ then $X$ is a form of $F$. When $X$ is a smooth Fano variety over the complex numbers all of whose extremal rays are accounted for by these fibrations this is due to Occhetta, Solá-Conde, Watanabe and Wiśniewski. Third, we recover, in a uniform way, the isomorphism and isogeny theorems of Chevalley and Demazure: over any base a pinned reductive group is determined by its pinned root datum, and a $p$-morphism of pinned root data determines a unique homomorphism of the corresponding groups.

math.AG↗

On exceptional Enriques surfaces

We give a complete description of all classical Enriques surfaces with non-zero global vector fields. In particular we show that there are such surfaces. The obtained result also applies to supersingular Enriques surfaces fulfilling a rather special condition. During this classification we study some properties of genus 1-fibrations special to characteristic as well as make a close study of the genus 1-fibration on the surfaces that we classify.

math.AG↗

Asymptotic period relations for Jacobian elliptic surfaces

We find an asymptotic description of the period locus of simply connected Jacobian elliptic surfaces and of the period locus of hyperelliptic curves. The two descriptions are essentially the same, and are given by the alkanes of organic chemistry.

math.AG↗

Tropes, Torelli and theta characteristics

Our main result is an effective version of the Torelli theorem in genus $3$ and any characteristic not $2$: the configuration of the odd theta characteristics of a curve $C$ of genus $3$ determines a del Pezzo surface $S$ of degree two and that $C$ is recovered as the normalization of a certain curve on $S$. This surface is not the one usually associated to $C$; we describe the distinction. We also give an explicit geometrical description of the quadratic twist (observed by Serre) that arises in the statement of the Torelli theorem.

math.AG↗

Some effectivity questions for plane Cremona transformations

We give a way of estimating quickly the length of the hyperbolic isometry associated to a plane Cremona transformation that is presented as the product of two involutions. We also show, using the ideas of Cantat and Lamy, that, provided that the ground field is either algebraic of characteristic not 2 or of characteristic zero, the normal closure of a high power of such a transformation is a proper subgroup of the Cremona group; this gives an effective instantiation of some of their results. Moreover, we show that any hyperbolic Cremona transformation is rigid and give a simple criterion for it to be tight. This version is thoroughly revised from the previous version. In particular, it corrects an error in an earlier version that was pointed out by Lonjou.

math.AG↗

Weyl group covers for Brieskorn's resolutions in all characteristics and the integral cohomology of $G/P$

We unify results of Artin, Brieskorn, Slodowy and others by showing that, in all characteristics, the Artin component of the deformation space of a rational surface singularity has a ramified cover where simultaneous resolution exists and the Galois group of this cover is the Weyl group determined by the configuration of $(-2)$-curves in the minimal resolution. This verifies a conjecture made by Burns and Rapoport. We use an integral version of this to show that certain actions of Weyl groups on polynomial rings over the integers give rings of invariants that are also polynomial, and deduce that the integral cohomology rings of complete flag varieties $G/B$ of types $A,D$ or $E$ can be described as the corresponding rings of co-invariants.

math.AG↗

"Del Pezzo surfaces as Springer fibres for exceptional groups"

We show that simultaneous log resolutions of simply elliptic singularities can be constructed inside suitable stacks of principal bundles over elliptic curves. In particular, we give a direct geometrical construction of del Pezzo surfaces from the corresponding exceptional simple algebraic groups.

math.AG↗

The non-existence of stable Schottky forms

Let $A_g^S$ be the Satake compactification of the moduli space $A_g$ of principally polarized abelian $g$-folds and $M_g^S$ the closure of the image of the moduli space $M_g$ of genus $g$ curves in $A_g$ under the Jacobian morphism. Then $A_g^S$ lies in the boundary of $A_{g+m}^S$ for any $m$. We prove that $M_{g+m}^S$ and $A_g^S$ do not meet transversely in $A_{g+m}^S$, but rather that their intersection contains the $m$th order infinitesimal neighbourhood of $M_g^S$ in $A_g^S$. We deduce that there is no non-trivial stable Siegel modular form that vanishes on $M_g$ for every $g$. In particular, given two inequivalent positive even unimodular quadratic forms $P$ and $Q$, there is a curve whose period matrix distinguishes between the theta series of $P$ and $Q$.

math.AG↗

Moduli and periods of simply connected Enriques surfaces

We describe a period map for those simply connected Enriques surfaces in characteristic 2 whose canonical double cover is K3. The moduli stack for these surfaces has a Deligne-Mumford quotient that is an open substack of a $\mathbb P^1$-bundle over the period space. We also give some general results relating local and global moduli for algebraic varieties and describe the difference in their dimensions in terms of the failure of the automorphism group scheme to be reduced.

math.AG↗

Perfect forms and the moduli space of abelian varieties

Perfect quadratic forms give a toroidal compactification of the moduli space of principally polarized abelian g-folds that is Q-factorial and whose ample classes are characterized, over any base. In characteristic zero it has canonical singularities if g is at least 5, and is the canonical model (in the sense of Mori and Reid) if g is at least 12.

math.AG↗

Families of K3 surfaces

We use automorphic forms to prove that a compact family of Kaehler K3 surfaces with constant Picard number is isotrivial.

alg-geom↗