Seminars

Epi-convergence, asymptotic analysis and stability in set optimization problems.

Event Date: Aug 05, 2020 in Optimization and Equilibrium, Seminars

Abstract: We study the stability of set optimization problems with data that are not necessarily bounded. To do this, we use the well-known notion of epi-convergence coupled with asymptotic tools for set-valued maps. We derive characterizations for this notion that allows us to study the stability of vector and set type solutions by considering variations of the whole data (feasible set and objective map). We extend the notion of total epi-convergence to set-valued maps. * This work has been supported by Conicyt-Chile under project FONDECYT 1181368 Joint work with Elvira Hérnández,...

Read More

Crystallization of point configurations to the Square Lattice in 2D.

Event Date: Aug 04, 2020 in Differential Equations, Seminars

Abstract: We consider minimum-energy configurations x_1, …, x_N in euclidean space, minimizing the energy given by the sum of V( |x_i-x_j| ) for all pairs of distinct indices i,j, with V an interaction potential. Recent breakthroughs (Viazovska Ann. Math. 2017, Cohn-Kumar-Miller-Radchenko-Viazovska Ann. Math. 2017) led to a proof in arXiv:1902.05438 that special lattices in dimensions 8 and 24 form “universally optimal configurations”, i.e. optimize our energy at fixed density for a large class of V’s. This was recently extended (P.-Serfaty, Proc. AMS 2020) to include...

Read More

Maximal Quadratic-Free Sets.

Event Date: Jul 29, 2020 in ACGO, Seminars

Abstract: The intersection cut paradigm is a powerful framework that facilitates the generation of valid linear inequalities, or cutting planes, for a potentially complex feasible set S of an optimization problem. The key ingredient in this construction is an S-free set: a convex zone whose interior does not intersect S. Ideally, such S-free set would be maximal inclusion-wise, as it would generate a deeper cutting plane. In the case of integer programming, maximal lattice-free sets (i.e., when S corresponds to the integer lattice) have proven to be a powerful tool and have been carefully...

Read More

Learning to draw and drawing to teach: Considerations for teacher learning.

Event Date: Sep 01, 2020 in Education, Seminars

Abstract Mathematics education researchers and teachers generally agree that drawings support student learning, especially in elementary and middle school. However, drawings are not considered “mathematical;” symbols and equations are considered the main representations of mathematics. In this talk, I will argue why teachers and teacher educators should focus on using drawings as a legitimate form of mathematics. Additionally, I will also argue that a shift to using drawings entails a culture shift in how we think about and do mathematics. I will supplement the argument with a study where I...

Read More

Satisfying Instead of Optimizing in the Nash Demand Games.

Event Date: Jul 22, 2020 in Optimization and Equilibrium, Seminars

Abstract:   The Nash Demand Game (NDG) has been one of the first models (Nash 1953) that has tried to describe the process of negotiation, competition, and cooperation. This model is still subject to active research, in fact, it maintains a set of open questions regarding how agents optimally select their decisions and how they face uncertainty.   However, the agents act rather guided by chance and necessity, with a Darwinian flavor. Satisfying, instead of optimising. The Viability Theory (VT) has this approach. Therefore, we investigate the NDG under this point of view. In...

Read More

Online Algorithms with Machine Learned Advice.

Event Date: Jul 22, 2020 in ACGO, Seminars

Abstract: Machine-learned predictors, although achieving very good results for inputs resembling training data, cannot possibly provide perfect predictions in all situations. Still, decision-making systems that are based on such predictors need not only to benefit from good predictions but also to achieve a decent performance when the predictions are inadequate.   Assume that you are given some (machine-learned) information regarding an online problem. It would be desirable to design an online algorithm that incorporates this information in order to on one hand obtain an improved...

Read More