Georgia Institute of Technology College of Sciences

The Erdos-Szekeres (happy ending) theorem claims that among any N points in general position in the plane there are at least log_4(n) of them that are the vertices of a convex polygon. I will explain generalizations of this result that were discovered in the last 30 years involving pseudoline arrangements and families of convex bodies. After surveying some previous work I will present the following results: 1) We improve the upper bound of the analogue Ramsey function for families of disjoint and noncrossing convex bodies. In fact this follows as a corollary of the equivalence between a conjecture of Goodman and Pollack about psudoline arrangements and a conjecture of Bisztrinsky and Fejes Toth about families of disjoint convex bodies. I will say a few words about how we show this equivalence. 2) We confirm a conjecture of Pach and Toth that generalizes the previous result. More precisely we give suffcient and necesary conditions for the existence of the analogue Ramsey function in the more general case in which each pair of bodies share less than k common tangents (for every fixed k). These results are joint work with Andreas Holmsen and Michael Dobbins.

Judge John Dever gives the mathematically ubiquitous concept of adjunction a categorical definition. With the hom functor acting as an "inner product", categorical adjoints may be seen as the analogy of adjoint linear operators of a Hilbert space.

We consider optimal alignments of random sequences of length n which are i.i.d. For such alignments we count which letters get aligned with which letters how often. This gives as for every opitmal alignment the frequency of the aligned letter pairs. These frequencies expressed as relative frequencies and put in vector form are called the "empirical distribution of letter pairs along an optimal alignment". It was previously established that if the scoring function is chosen at random, then the empirical distribution of letter pairs along an opitmal alignment converges. We show an upper bound for the rate of convergence which is larger thatn the rate of the alignement score. the rate of the alignemnt score can be obtained directly by Azuma-Hoeffding, but not so for the empirical distribution of the aligned letter pairs seen along an opitmal alignment: which changing on letter in one of the sequences, the optimal alginemnt score changes by at most a fixed quantity, but the empirical distribution of the aligned letter pairs potentially could change entirely.

It is well known that solutions of compressible Euler equations in general form discontinuities (shock waves) in finite time even when the initial data is $C^\infty$ smooth. The lack of regularity makes the system hard to resolve. When the initial data have large amplitude, the well-posedness of the full Euler equations is still wide open even in one space dimenssion. In this talk, we discuss some recent progress on large data solutions for the compressible Euler equations in one space dimension. The talk includes joint works with Alberto Bressan, Helge Kristian Jenssen, Robin Young and Qingtian Zhang.