Markoff triples were introduced in 1879 and have a rich history spanning many branches of mathematics. In 2016, Bourgain, Gamburd, and Sarnak answered a long standing question by showing there exist infinitely many composite Markoff numbers. Their proof relied on showing the connectivity for an infinite family of graphs associated to Markoff triples modulo p for infinitely many primes p. In this talk we discuss what happens for the projective analogue of Markoff triples, that is surfaces W in $P^1$x$P^1$x$P^1$ cut out by the vanishing of a (2,2,2)-form that admit three non-commuting involutions and are fixed under coordinate permutations and double sign changes. Inspired by the work of B-G-S we investigate such surfaces over finite fields, specifically their orbit structure under their automorphism group. For a specific one-parameter subfamily $W_k$ of such surfaces, we construct finite orbits in $W_k(C)$ by studying small orbits that appear in  $W_k$($F_p$) for many values of p and k. This talk is based on joint work with E. Fuchs, J. Silverman, and A. Tran.

We motivate the study of coactions, developing our intuition by taking a tour through topological dynamics. We reinforce this intuition by exploring the particular example of skew product topological quivers - a subject of recent study by the speaker.

The self-dual Chern-Simons-Schrödinger model is a gauged cubic NLS on the plane with self-duality, i.e., energy minimizers are given by a first-order Cauchy-Riemann-type equation, rather than a second-order elliptic equation. While this equation shares all formal symmetries with the usual cubic NLS on the plane, the structure of solitary waves is quite different due to self-duality and nonlocality (which stems from the gauge structure). In accordance, this model possesses blow-up and global dynamics that are quite different from that of the usual cubic NLS. The goal of this talk is to present some recent results concerning the blow-up and global dynamics of this model, with emphasis on a few surprising features of this model such as the impossibility of a "bubble-tree'' blow-up and a nonlinear rotational instability of pseudoconformal blow-ups. This talk is based on joint work with Kihyun Kim (IHES) and Soonsik Kwon (KAIST).

In the late 1980’s, Krasauskas and Fiedorowicz-Loday independently developed the theory of crossed simplicial groups, which generalize Connes’ cyclic category. Of particular interest is the Dihedral category, which has recently been used to develop the theory of Real topological Hochschild homology, a first approximation to Grothendieck-Witt groups.

In the first part of my talk, I will discuss ongoing joint work with Mona Merling and Maximilien Péroux on a topological analogue of the homology of crossed simplicial groups. As a special case, we recover the theory of Real topological Hochschild homology.

In the second part of my talk, I will discuss joint work with Teena Gerhardt and Mike Hill. We provide a norm model for Real topological Hochschild homology, prove a multiplicative double coset formula for Real topological Hochschild homology, and we construct the Real Witt vectors of rings with anti-involution.

In this talk, we will explore several combinatorial objects whose existence depends on some (relatively) straightforward divisibility conditions. In each of these case, perhaps somewhat surprisingly, these necessary divisibility conditions are in fact sufficient. The talk will conclude by mentioned what a complete resolution is and will not require any background knowledge in combinatorics.

Convergence analysis of accelerated first-order methods for convex optimization problems are presented from the point of view of ordinary differential equation (ODE) solvers. We first take another look at the acceleration phenomenon via A-stability theory for ODE solvers and present a revealing spectrum analysis for quadratic programming. After that, we present the Lyapunov framework for dynamical system and introduce the strong Lyapunov condition. Many existing continuous convex optimization models, such as gradient flow, heavy ball system, Nesterov accelerated gradient flow, and dynamical inertial Newton system etc, are addressed and analyzed in this framework. Then we present convergence analyses of optimization algorithms obtained from implicit or explicit methods of underlying dynamical systems.

We study minimizers of a family of functionals arising in combustion theory, which converge, for infinitesimal values of the parameter, to minimizers of the one-phase free boundary problem. We prove a $C^{1,\alpha}$ estimate for the "interfaces'' of critical points (i.e. the level sets separating the burnt and unburnt regions). As a byproduct, we obtain the one-dimensional symmetry of minimizers in the whole $\mathbb{R}^N$ for $N \le 4$, answering positively a conjecture of Fernández-Real and Ros-Oton. Our results are to the one-phase free boundary problem what Savin's results for the Allen-Cahn equation are to minimal surfaces. This is a joint work with J. Serra (ETHZ).

Modern learning algorithms such as deep neural networks operate in regimes that defy the traditional statistical learning theory. Neural networks architectures often contain more parameters than training samples. Despite their huge complexity, the generalization error achieved on real data is small. In this talk, we aim to study the generalization properties of algorithms in high dimensions. Interestingly, we show that algorithms in high dimensions require a small bias for good generalization. We show that this is indeed the case for deep neural networks in the over-parametrized regime. In addition, we provide lower bounds on the generalization error in various settings for any algorithm. We calculate such bounds using random matrix theory (RMT). We will review the connection between deep neural networks and RMT and existing results. These bounds are particularly useful when the analytic evaluation of standard performance bounds is not possible due to the complexity and nonlinearity of the model. The bounds can serve as a benchmark for testing performance and optimizing the design of actual learning algorithms. (Joint work with Ofer Zeitouni)

The theory of unipotent flows plays a central role in homogeneous dynamics. Time-changes are a simple perturbation of a given flow. In this talk, we shall discuss the rigidity of time-changes of unipotent flows. More precisely, we shall see how to utilize the branching theory of the complementary series, combining it with the works of Ratner and Flaminio-Forni to get to our purpose.

In this talk, I will explain a theorem of Bhatt-Scholze: the equivalence between prismatic $F$-crystal and $\mathbb Z_p$-lattices inside crystalline representation, and how to extend this theorem to allow more general types of base ring like Tate algebra ${\mathbb Z}_p\langle t^{\pm 1}\rangle$.  This is a joint work with Heng Du, Yong-Suk Moon and Koji Shimizu.

This is a talk in integral $p$-adic Hodge theory.  So in the pre-talk, I will explain the motivations and base ideas in integral $p$-adic Hodge theory.
 

Stochastic reaction network models are a popular tool for studying the effects of dynamical randomness in biological systems. Such models are typically analysed by estimating the solution of Kolmogorov's forward equation, called the chemical master equation (CME), which describes the evolution of the probability distribution of the random state-vector representing molecular counts of the reacting species. The size of the CME system is typically very large or even infinite, and due to this high-dimensional nature, accurate numerical solutions of the CME are very difficult to obtain. In this talk we will present a novel deep learning approach for computing solution statistics of high-dimensional CMEs by reformulating the stochastic dynamics using Kolmogorov's backward equation. The proposed method leverages superior approximation properties of Deep Neural Networks (DNNs) to reliably estimate expectations under the CME solution for several user-defined functions of the state-vector. Our method only requires a handful of stochastic simulations and it allows not just the numerical approximation of various expectations for the CME solution but also of its sensitivities with respect to all the reaction network parameters (e.g. rate constants). We illustrate the method with a number of examples and discuss possible extensions and improvements. 

This is joint work with Prof. Christoph Schwab (Seminar for Applied Mathematics, ETH Zürich) and Prof. Mustafa Khammash (Department of Biosystems Science and Engineering, ETH Zürich)

Reference:  Gupta A, Schwab C, Khammash M (2021) DeepCME: A deep learning framework for computing solution statistics of the chemical master equation. PLoS Comput Biol 17(12): e1009623. https://doi.org/10.1371/journal.pcbi.1009623