ERC FUNDED PROJECTS

"We are addressing the analysis and numerical methods for the tractable simulation and the optimal control of dynamical systems which are modeling the behavior of a large number N of complex interacting agents described by a large amount of parameters (high-dimension). We are facing fundamental challenges: - Random projections and recovery for high-dimensional dynamical systems: we shall explore how concepts of data compression via Johnson-Lindenstrauss random embeddings onto lower-dimensional spaces can be applied for tractable simulation of complex dynamical interactions. As a fundamental subtask for the recovery of high-dimensional trajectories from low-dimensional simulated ones, we will address the efficient recovery of point clouds defined on embedded manifolds from random projections. -Mean field equations: for the limit of the number N of agents to infinity, we shall further explore how the concepts of compression can be generalized to work for associated mean field equations. - Approximating functions in high-dimension: differently from purely physical problems, in the real life the ”social forces” which are ruling the dynamics are actually not known. Hence we will address the problem of automatic learning from collected data the fundamental functions governing the dynamics. - Homogenization of multibody systems: while the emphasis of our modelling is on “social” dynamics, we will also investigate methods to recast multibody systems into our high-dimensional framework in order to achieve nonstandard homogenization by random projections. - Sparse optimal control in high-dimension and mean field optimal control: while self-organization of such dynamical systems has been so far a mainstream, we will focus on their sparse optimal control in high-dimension. We will investigate L1-minimization to design sparse optimal controls. We will learn high-dimensional (sparse) controls by random projections to lower dimension spaces and their mean field limit."

1 123 000 €

Start date: 2012-12-01, End date: 2017-11-30

The Atiyah-Singer index theorem was one of the most spectacular achievements of mathematics in the XXth century, connecting the analytic and topological properties of manifolds. The Baum-Connes conjecture is a hugely successful approach to generalizing the index theorem to a much broader setting. It has remarkable applications in topology and analysis. For instance, it implies the Novikov conjecture on the homotopy invariance of higher signatures of a closed manifold and the Kaplansky-Kadison conjecture on the existence of non-trivial idempotents in the reduced group C*-algebra of a torsion-free group. At present, the Baum-Connes conjecture is known to hold for a large class of groups, including groups admitting metrically proper isometric actions on Hilbert spaces and Gromov hyperbolic groups. The Baum-Connes conjecture with certain coefficients is known to fail for a class of groups, whose Cayley graphs contain coarsely embedded expander graphs. Nevertheless, the conjecture in full generality remains open and there is a growing need for new examples of groups and group actions, that would be counterexamples to the Baum-Connes conjecture. The main objective of this project is to exhibit such examples. Our approach relies on strengthening Kazhdan’s property (T), a prominent cohomological rigidity property, from its original setting of Hilbert spaces to much larger classes of Banach spaces. Such properties are an emerging direction in the study of cohomological rigidity and are not yet well-understood. They lie at the intersection of geometric group theory, non-commutative geometry and index theory. In their study we will implement novel approaches, combining geometric and analytic techniques with variety of new cohomological constructions.

880 625 €

Start date: 2016-08-01, End date: 2021-07-31

The aim of the proposed research is to study the homotopy theory of algebraic varieties and other algebraically defined geometric objects, especially over fields other than the complex numbers. A noticeable emphasis will be put on fundamental groups and on K(pi, 1) spaces, which serve as building blocks for more complicated objects. The most important source of both motivation and methodology is my recent discovery of the K(pi, 1) property of affine schemes in positive characteristic and its relation to wild ramification phenomena. The central goal is the study of etale homotopy types in positive characteristic, where we hope to use the aforementioned discovery to yield new results beyond the affine case and a better understanding of the fundamental group of affine schemes. The latter goal is closely tied to Grothendieck's anabelian geometry program, which we would like to extend beyond its usual scope of hyperbolic curves. There are two bridges going out of this central point. The first is the analogy between wild ramification and irregular singularities of algebraic integrable connections, which prompts us to translate our results to the latter setting, and to define a wild homotopy type whose fundamental group encodes the category of connections. The second bridge is the theory of perfectoid spaces, allowing one to pass between characteristic p and p-adic geometry, which we plan to use to shed some new light on the homotopy theory of adic spaces. At the same time, we address the related question: when is the universal cover of a p-adic variety a perfectoid space? We expect a connection between this question and the Shafarevich conjecture and varieties with large fundamental group. The last part of the project deals with varieties over the field of formal Laurent series over C, where we want to construct a Betti homotopy realization using logarithmic geometry. The need for such a construction is motivated by certain questions in mirror symmetry.

1 007 500 €

Start date: 2019-06-01, End date: 2024-05-31

Mathematical statistical physics has seen spectacular progress in recent years. Existing problems which were previously unattainable were solved, opening a way to approach some of the classical open questions in the field. The proposed research focuses on phenomena of localization and long-range order in physical systems of large size, identifying several fundamental questions lying at the interface of Statistical Physics, Probability Theory and Combinatorics. One circle of questions concerns the fluctuation behavior of random surfaces, where the PI has recently proved delocalization in two dimensions answering a 1975 question of Brascamp, Lieb and Lebowitz. A main goal of the research is to establish some of the long-standing universality conjectures for random surfaces. This study is also tied to the localization features of random operators, such as random Schrodinger operators and band matrices, as well as those of reinforced random walks. The PI intends to develop this connection further to bring the state-of-the-art to the conjectured thresholds. A second circle of questions regards long-range order in high-dimensional systems. This phenomenon is predicted to encompass many models of statistical physics but rigorous results are quite limited. A notable example is the PI’s proof of Kotecky’s 1985 conjecture on the rigidity of proper 3-colorings in high dimensions. The methods used in this context are not limited to high dimensions and were recently used by the PI to prove the analogue for the loop O(n) model of Polyakov’s 1975 prediction that the 2D Heisenberg model and its higher spin versions exhibit exponential decay of correlations at any temperature. Lastly, statistical physics methods are proposed for solving purely combinatorial problems. The PI has applied this approach successfully to solve questions of existence and asymptotics for combinatorial structures and intends to develop it further to answer some of the tantalizing open questions in the field.

1 136 904 €

Start date: 2016-01-01, End date: 2020-12-31