Where: Toll Physics Bldg 1412

Speaker: C. David Levermore (UMd) - http://math.umd.edu/~lvrmr

Where: Kirwan Hall 1310

Speaker: Dianne P. O'Leary () - http://www.cs.umd.edu/~oleary/

Where: Kirwan Hall 1311

Speaker: Test (UMD) - http://www.umd.edu

Abstract: $\sum_{i=0}^n i^2 = \frac{(n^2+n)(2n+1)}{6}$

Where: Kirwan Hall 1308

Speaker: Tamas Darvas (UMCP) -

Where: Kirwan Hall 1308

Speaker: Aynur Bulut (Princeton University) -

Abstract: Dispersive equations such as nonlinear Schrödinger and wave equations

arise as mathematical models in a variety of physical settings,

including models of plasma physics, the propagation of laser beams,

water waves, and the study of many-body quantum mechanics. They also

serve as model equations for studying fundamental issues in many

aspects of nonlinear partial differential equations. Key questions in

the analysis of these equations include issues of well-posedness (for

instance, existence of solutions, uniqueness of these solutions, and

their continuous dependence on initial data in appropriate topologies)

locally in time, long-time existence and behavior of solutions, and,

conversely, the possible existence of solutions which blow-up in

finite time.

In this talk, we will give an overview of several recent results

concerning the local and global (long-time) theory, including some

results where probabilistic tools are used to obtain estimates for

randomly chosen initial data which are not available in deterministic

settings. A recurring theme (and oftentimes obstacle) is the notion

of supercriticality arising from the natural scaling of the equation —

seeking to characterize long-time behavior of solutions when the

relevant scale-invariant norms are not controlled by the conserved

energy, or for initial data of very low regularity. The techniques

involved include input from several areas of mathematics, including

ideas arising in many areas of PDE, harmonic analysis, and

probability.

---

Where: Kirwan Hall 3206

Speaker: Dr. Anastasios Kyrillidis (UT Austin) -

Abstract: With the quantity of generated data ever-increasing in most research areas, conventional data analytics run into solid computational, storage, and communication bottlenecks. These obstacles force practitioners to often use algorithmic heuristics, in an attempt to convert data into useful information, fast. It is necessary to rethink the algorithmic design, and devise smarter and provable methods in order to flexibly balance the trade-offs between solution

accuracy, efficiency, and data interpretability.

In this talk, I will focus on the problem of low rank matrix inference in large-scale settings. Such problems appear in fundamental applications such as structured inference, recommendation systems and multi-label classification problems. I will introduce a novel theoretical framework for analyzing the performance

of non-convex first-order methods, often used as heuristics in practice. These methods lead to computational gains over classic convex approaches, but their analysis is unknown for most problems. This talk will provide precise theoretical guarantees, answering the long-standing question why such non-convex techniques behave well in practice for a wide class of problems. I will discuss implementation details of these ideas and, if time permits, show the superior

performance we can obtain in applications found in physical sciences and machine learning.

Where: Kirwan Hall 1310

Speaker: Susan Carter and Sara Taylor (NSA) -

Where: Kirwan Hall 3206

Speaker: Ingrid Daubechies (Duke University) - https://math.duke.edu/people/ingrid-daubechies

Abstract: Mathematical tools for image analysis increasingly play a role in helping art historians and art conservators assess the state of conversation of paintings, and probe into the secrets of their history. the talk will review several case studies, Van Gogh, Gauguin, Van Eyck among others.