NTS ABSTRACT: Difference between revisions

A remarkable theorem of Christol from 1979 gives a criterion for detecting whether a power series over a finite field of characteristic p represents an algebraic function: this happens if and only if the coefficient of the n-th power of the series variable can be extracted from the base-p expansion of n using a finite automaton. We will describe a result that extends this result in two directions: we allow an arbitrary field of characteristic p, and we allow "generalized power series" in the sense of Hahn-Mal'cev-Neumann. In particular, this gives a concrete description of an algebraic closure of a rational function field in characteristic p (and corrects a mistake in my previous attempt to give this description some 15 years ago).

The epipelagic representations of Reeder-Yu, a generalization of the "simple supercuspidals" of Gross-Reeder, are certain low-depth supercuspidal representations of reductive algebraic groups G. Given a "stable functional" f, which is a suitably 'generic' linear functional on a vector space coming from a Moy-Prasad filtration for G, one can create such a representation. It is known that the representations created in this way are compactly induced from the fixer in G of f and it is important to identify explicitly all the elements that belong to this fixer. This work is in-progress.

Let X be a curve of genus g over a number field F of degree d = [F:Q]. The conjectural existence of a uniform bound N(g,d) on the number #X(F) of F-rational points of X is an outstanding open problem in arithmetic geometry, known to follow from the Bomberi-Lang conjecture. We prove a special case of this conjecture - we give an explicit uniform bound when X has Mordell-Weil rank r ≤ g-3. This generalizes recent work of Stoll on uniform bounds on hyperelliptic curves. Using the same techniques, we give an explicit, unconditional uniform bound on the number of F-rational torsion points of J lying on the image of X under an Abel-Jacobi map. We also give an explicit uniform bound on the number of geometric torsion points of J lying on X when the reduction type of X is highly degenerate. Our methods combine Chabauty-Coleman's p-adic integration, non-Archimedean potential theory on Berkovich curves, and the theory of linear systems and divisors on metric graphs. This is joint work with Joe Rabinoff and Eric Katz.

We'll present two new applications of Poonen's closed point sieve over finite fields. The first is that the obvious local obstruction to embedding a curve in a smooth surface is the only global obstruction. The second is a proof of a recent conjecture of Vakil and Wood on the asymptotic probability of hypersurface sections having a prescribed number of singularities.

Note: The day of this seminar is Tuesday.

Cohen-Lenstra heuristics posit the distribution of unramified abelian extensions of quadratic fields. A natural question to ask would be how to get an analogous heuristic for nonabelian groups. In this talk I take and extend on recent work in the area of unramified extensions of imaginary quadratic fields and bring it all together under one Cohen-Lenstra style heuristic.

I will discuss joint work in progress with Peter Scholze showing that torsion in the cohomology of certain compact unitary Shimura varieties occurs in the middle degree, under a genericity assumption on the corresponding Galois representation.

We establish an asymptotic formula (with power saving error term) for the number of rational points of bounded height for a certain cubic fourfold, thereby proving a strong form of Manin's conjecture for this algebraic variety by techniques of analytic number theory.

A result of Katznelson and Weiss states that given a suitably dense (measurable) subset of the Euclidean plane realizes every sufficiently large distance, that is, for every prescribed (sufficiently large) real number the set contains two elements whose distance is this number. The analogue of this statement for finding three equally spaced points on a line, i.e. for finding three term arithmetic progressions, in a given set is false, and in fact false in every dimension. In this talk we revisit the case of three term progressions when the standard Euclidean metric is replaced by other metrics.

We will discuss the problem of classifying the behavior of integral points on affine subsets of the projective plane. As an application, we will examine the problem of classifying endomorphisms of the projective plane with an orbit containing a Zariski dense set of integral points (with respect to some plane curve). This is joint work with Yu Yasufuku.

We revisit the recent result of Lior-Bary-Soroker. It deals with a function field analogue of Landau's classical result about the asymptotic density of numbers which are sums of two integer squares. The results obtained are just in the large characteristic and large degree regime. We obtain a characterization as qⁿ goes to infinity, which is the desired analogue of the result over the integers. We take a geometric perspective in computing the number of polynomials of degree n which are split in the extension F_q[T^1/2] / F_q[T] and we obtain a geometric explanation for the "mysterious" binomial coefficients appearing in his asymptotic.

In the original Gross-Zagier formula and Zhang's extension to Shimura curves, the modularity of the generating function

[math]\displaystyle{ \Sigma (Z(m) - \deg Z(m) [\xi])q^m \in CH^1(X)_0 \otimes \mathbb{C} }[/math]

is a very important step, where Z(m) are the Heeger divisors and [\xi] is the rational canonical divisor of degree 1 (associated to Hodge bundle). In the proof, they actually use arithmetic version in calculation: [\hat{\xi}] + \sum \hat{Z}(m) q^m \in \widehat{CH}^1(X) \otimes \C which is also modular. In this talk, we define a generalization of this arithmetic generating function to unitary Shimura variety of type (n, 1) and prove that it is modular. It has application to Colmez conjecture and Gross-Zagier type formula.

Coming soon...