## Seminars and Colloquia by Series

Friday, November 21, 2014 - 15:00 , Location: Skiles 006 , Jennifer Park , McGill University , Organizer: Joseph Rabinoff
Using the ideas of Poonen and Stoll, we develop a modified version of Chabauty's method, which shows that a positive proportion of hyperelliptic curves have as few quadratic points as possible.
Wednesday, November 19, 2014 - 15:00 , Location: Skiles 006 , Jennifer Park , McGill University , Organizer: Joseph Rabinoff
Faltings' theorem states that curves of genus g> 1 have finitely many rational points. Using the ideas of Faltings, Mumford, Parshin and Raynaud, one obtains an upper bound on the number of rational points, but this bound is too large to be used in any reasonable sense. In 1985, Coleman showed that Chabauty's method, which works when the Mordell-Weil rank of the Jacobian of the curve is smaller than g, can be used to give a good effective bound on the number of rational points of curves of genus g > 1. We draw ideas from nonarchimedean geometry and tropical geometry to show that we can also give an effective bound on the number of rational points outside of the special set of the d-th symmetric power of X, where X is a curve of genus g > d, when the Mordell-Weil rank of the Jacobian of the curve is at most g-d.
Friday, November 14, 2014 - 14:00 , Location: Skiles 202 , Ralph Morrison , Berkeley , Organizer: Joseph Rabinoff
Smooth curves in the tropical plane correspond to unimodulartriangulations of lattice polygons. The skeleton of such a curve is ametric graph whose genus is the number of lattice points in the interior ofthe polygon. In this talk we report on work concerning the followingrealizability problem: Characterize all metric graphs that admit a planarrepresentation as a smooth tropical curve. For instance, about 29.5 percentof metric graphs of genus 3 have this property. (Joint work with SarahBrodsky, Michael Joswig, and Bernd Sturmfels.)
Tuesday, November 4, 2014 - 16:00 , Location: Skiles 005 , Arul Shankar and Wei Zhang , Harvard University and Columbia University , Organizer:
The Joint Athens-Atlanta Number Theory Seminar meets once a semester, usually on a Tuesday, with two talks back to back, at 4:00 and at 5:15. Participants then go to dinner together.
Monday, November 3, 2014 - 15:00 , Location: Skiles 005 , François Charles , MIT and Paris-Sud , Organizer: Kirsten Wickelgren
We will discuss a proof of the following result: if E and E' are two elliptic curves over a number field, there exist infinitely many places p of k such that the reduction of E and E' modulo p are isogenous. We will explain the relationship with the dynamics of Hecke correspondences on modular curves and the heuristics behind such results.
Monday, October 20, 2014 - 15:05 , Location: Skiles 005 , , University of Utah , Organizer: Anton Leykin
I will discuss recent work of Bhargav Bhatt, myself and Shunsuke Takagi relating several open problems and generalizing work of Mustata and Srinivas. First: whether a smooth complex variety is ordinary after reduction to characteristic $p > 0$ for infinitely many $p$. Second: that multiplier ideals reduce to test ideals for infinitely many $p$ (regardless of coefficients). Finally, whether complex varieties with Du Bois singularities have $F$-injective singularities after reduction to infinitely many $p > 0$.
Tuesday, October 7, 2014 - 03:05 , Location: Skiles 005 , Ngoc Mai Tran , UT Austin , Organizer: Josephine Yu
This talk surveys the connection between economics and tropical geometry, as developed in the paper of Baldwin and Klemperer (Tropical Geometry to Analyse Demand). I will focus on translating concepts, theorems and questions in economics to tropical geometry terms.
Monday, October 6, 2014 - 15:05 , Location: Skiles 005 , , UT Austin , Organizer: Josephine Yu
Polytropes are both ordinary and tropical polytopes. Tropical types of polytropes in \R^n are in bijection with certain cones of a specific Gr\"obner fan in \R^{n^2-n}. Unfortunately, even for n = 5 the entire fan is too large to be computed by existing software. We show that the polytrope cones can be decomposed as the cones from the refinement of two fans, intersecting with a specific cone. This allows us to enumerate types of full-dimensional polytropes for $n = 4$, and maximal polytropes for $n = 5$ and $n = 6$. In this talk, I will prove the above result and describe the key difficulty in higher dimensions.
Friday, October 3, 2014 - 15:00 , Location: Skiles 006 , Ján Minac , University of Western Ontario , Organizer: Kirsten Wickelgren
Similar to the glamour of Las Vegas, the excitement and drama of winning in casinos and falling under the spell of such legends as Frank Sinatra and Dean Martin; is the search for revealing the mystery of absolute Galois groups and their special properties among other profinite groups. The recent, spectacular proof of the Bloch-Kato conjecture by Rost and Voevodsky, with Weibel's patch, and some current and interesting developments involving Massey products, hold great promise and new challenges on the road to understanding the structure of absolute Galois groups. This talk will provide an overview of the subject, and then explain some recent results obtained with Nguyen Duy Tan.
Monday, September 29, 2014 - 15:05 , Location: Skiles 005 , , IST Austria , Organizer: Anton Leykin
Markov bases have been developed in algebraic statistics for exact goodness-of-fit testing.  They connect all elements in a fiber (given by the sufficient statistics) and allow building a Markov chain to approximate the distribution of a test statistic by its posterior distribution.  However, finding a Markov basis is often computationally intractable. In addition, the number of Markov steps required for converging to the stationary distribution depends on the connectivity of the sampling space.In this joint work with Caroline Uhler and Sarah Cepeda, we compare different test statistics and study the combinatorial structure of the finite lattice Ising model. We propose a new method for exact goodness-of-fit testing. Our technique avoids computing a Markov basis but builds a Markov chain consisting only of simple moves (i.e. swaps of two interior sites). These simple moves might not be sufficient to create a connected Markov chain.  We prove that when a bounded change in the sufficient statistics is allowed, the resulting Markov chain is connected. The proposed algorithm not only overcomes the computational burden of finding a Markov basis, but it might also lead to a better connectivity of the sampling space and hence a faster convergence.