Variété projective | EPFL Graph Search

En géométrie algébrique, les variétés projectives forment une classe importante de variétés. Elles vérifient des propriétés de compacité et des propriétés de finitude. C'est l'objet central de la géométrie algébrique globale. Sur un corps algébriquement clos, les points d'une variété projective sont les points d'un ensemble algébrique projectif. Définition On fixe un corps (commutatif) k.

Algèbre homogène. Soit B le quotient de k[T_0,\ldots, T_n] par un idéal homogène ( idéal engendré par des polynômes homogènes). C'est alors une algèbre graduée B=\oplus_{d\ge 0} B_d,
:où B_d est l'ensemble des classes modulo I des polynômes homogènes de degrés d. Les éléments de B_d sont appelés des éléments homogènes de degré d. Un idéal homogène de B est un idéal engendré par des éléments homogènes. Un idéal homogène particulier est B_+,

Motivic invariants of moduli spaces of rank 2 Bradlow-Higgs triples

Riccardo Grandi

In the present thesis we study the geometry of the moduli spaces of Bradlow-Higgs triples on a smooth projective curve C. There is a family of stability conditions for triples that depends on a positive real parameter Ï. The moduli spaces of Ï-semistable triples of rank r and degree d vary with Ï. The phenomenon arising Ï from this is known as wall-crossing. In the first half of the thesis we will examine how the moduli spaces and their universal additive invariants change as Ï varies, for the case r = 2. In particular we will study the case of Ï very close to 0, for which the moduli space relates to the moduli space of stable Higgs bundles, and Ï very large, for which the moduli space is a relative Hilbert scheme of points for the family of spectral curves. Some of these results will be generalized to Bradlow-Higgs triples with poles. In the second half we will prove a formula relating the cohomology of the moduli spaces for small and odd degree and the perverse filtration on the cohomology of the moduli space of stable Higgs bundles. We will also partially generalize this result to the case of rank greater than 2.

EPFL2016

Vanishing theorems in positive characteristic

Emelie Kerstin Arvidsson

The topic of this thesis is vanishing theorems in positive characteristic. In particular, we use "the covering trick of Ekedahl" to investigate the vanishing of

H^1(X, \mathcal{O}_X(-D))

for a big and nef Weil divisor

D

on a normal projective variety with

-K_X

nef. In dimension two, we show that on a surface of log del Pezzo type over a perfect field of characteristic

p>5

this vanishing holds. More generally, using techniques of the \emph{Minimal model program} we prove the Kawamata--Viehweg vanishing theorem in this setting. We also construct a counter-example in characteristic five, showing that our result is optimal. We discuss the relationship (due to Hacon--Witaszek) between this vanishing theorem and properties of threefold klt-singularities. We investigate if a similar relationship exists between threefold lc-singularities and a certain vanishing theorem for higher direct images of elliptic fibrations. This leads to a counter-example to a theorem of Koll'ar over the complex numbers, in every positive characteristic.

EPFL2021

Homology of the three flag Hilbert Scheme

Daniele Boccalini

We prove the existence of an affine paving for the three-step flag Hilbert scheme that parametrizes flag of three 0-dimensional subschemes of length, respectively, n, n+1 and n+2 that are supported at the origin of the affine plane. This is done by showing that the space stratifies in smooth subvarieties, the Hilbert-Samuel's strata, each of which has an affine paving with cells of known dimension, indexed by marked Young diagrams. The affine pavings of the Hilbert-Samuel's strata allow us to prove that the Poincaré polynomials for our spaces satisfy a generating function. In the process of proving the formula for the generating function we relate combinatorially the homology of our spaces with that of known smooth subspaces of another Hilbert scheme of flags, this time of length n and n+2. As a corollary we find an affine paving and a combinatorial formula for the Poincaré of these last ambient spaces.