Seminars and Colloquia by Series

Matrix completion and tensor codes

Series
Algebra Seminar
Time
Monday, September 23, 2024 - 11:30 for 1 hour (actually 50 minutes)
Location
Skiles 005
Speaker
Matt LarsonPrinceton University and the Institute for Advanced Study

Please Note: There will be a pre-seminar at 10:55 am in Skiles 005.

The rank r matrix completion problem studies whether a matrix where some of the entries have been filled in with generic complex numbers can be completed to a matrix of rank at most r. This problem is governed by the bipartite rigidity matroid, which is a matroid studied in combinatorial rigidity theory. We show that the study of the bipartite rigidity matroid is related to the study of tensor codes, a topic in information theory, and use this relation to understand new cases of both problems. Joint work with Joshua Brakensiek, Manik Dhar, Jiyang Gao, and Sivakanth Gopi.

The Small Quasikernel Conjecture

Series
Combinatorics Seminar
Time
Friday, September 20, 2024 - 15:15 for 1 hour (actually 50 minutes)
Location
Skiles 005
Speaker
Sam SpiroRutgers University

Given a digraph $D$, we say that a set of vertices $Q\subseteq V(D)$ is a quasikernel if $Q$ is an independent set and if every vertex of $D$ can be reached from $Q$ by a path of length at most 2.  The Small Quasikernel Conjecture of P.L. Erdős and Székely from 1976 states that every $n$-vertex source-free digraph $D$ contains a quasikernel of size at most $\frac{1}{2}n$.  Despite being posed nearly 50 years ago, very little is known about this conjecture, with the only non-trivial upper bound of $n-\frac{1}{4}\sqrt{n\log n}$ being proven recently by ourself.  We discuss this result together with a number of other related results and open problems around the Small Quasikernel Conjecture.

Digital Twins in the era of generative AI — Application to Geological CO2 Storage

Series
GT-MAP Seminar
Time
Friday, September 20, 2024 - 15:00 for 2 hours
Location
Skiles 006
Speaker
Felix J. HerrmannGT CSE, ECE, and EAS

Please Note: Felix J. Herrmann Georgia Research Alliance Eminent Scholar Chair in Energy Seismic Laboratory for Imaging and Modeling Schools of Earth & Atmospheric Sciences, Computational Science & Engineering, Electrical and Computer Engineering Georgia Institute of Technology https://slim.gatech.edu Felix J. Herrmann is a professor with appointments at the College of Sciences (EAS), Computing (CSE), and Engineering (ECE) at the Georgia Institute of Technology. He leads the Seismic Laboratory for Imaging and modeling (SLIM) and he is co-founder/director of the Center for Machine Learning for Seismic (ML4Seismic). This Center is designed to foster industrial research partnerships and drive innovations in artificial-intelligence assisted seismic imaging, interpretation, analysis, and time-lapse monitoring. In 2019, he toured the world presenting the SEG Distinguished Lecture. In 2020, he was the recipient of the SEG Reginald Fessenden Award for his contributions to seismic data acquisition with compressive sensing. Since his arrival at Georgia Tech in 2017, he expanded his research program to include machine learning for Bayesian wave-equation based inference using techniques from simulation-based inference. More recently, he started a research program on seismic monitoring of Geological Carbon Storage, which includes the development of an uncertainty-aware Digital Twin. In 2023, the manuscript entitled “Learned multiphysics inversion with differentiable programming and machine learning” was the most downloaded paper of 2023 in Society of Exploration Geophysicist’s The Leading Edge.

As a society, we are faced with important challenges to combat climate change. Geological Carbon Storage, during which gigatonnes of super-critical CO2 are stored underground, is arguably the only scalable net-negative negative CO2-emission technology that is available. Recent advances in generative AI offer unique opportunities—especially in the context of Digital Twins for subsurface CO2-storage monitoring, decision making, and control—to help scale this technology, optimize its operations, lower its costs, and reduce its risks, so assurances can be made whether storage projects proceed as expected and whether CO2 remains underground.

During this talk, it is shown how techniques from Simulation-Based Inference and Ensemble Bayesian Filtering can be extended to establish probabilistic baselines and assimilate multimodal data for problems challenged by large degrees of freedom, nonlinear multiphysics, and computationally expensive to evaluate simulations. Key concepts that will be reviewed include neural Wave-Based Inference with Amortized Uncertainty Quantification and physics-based Summary Statistics, Ensemble Bayesian Filtering with Conditional Neural Networks, and learned multiphysics inversion with Differentiable Programming.

This is joint work with Rafael Orozco.

 

The Heilbronn triangle problem

Series
Additional Talks and Lectures
Time
Friday, September 20, 2024 - 14:00 for 1 hour (actually 50 minutes)
Location
Skiles 005
Speaker
Cosmin PohoataEmory University

The Heilbronn triangle problem is a classical problem in discrete geometry with several old and new connections to various topics in extremal and additive combinatorics, graph theory, incidence geometry, harmonic analysis, and projection theory. In this talk, we will give an overview of some of these connections, and discuss some recent developments. Based on joint work with Alex Cohen and Dmitrii Zakharov.

Pseudo-Maximum Likelihood Theory for High-Dimension Rank-One Inference

Series
Stochastics Seminar
Time
Thursday, September 19, 2024 - 15:30 for 1 hour (actually 50 minutes)
Location
Skiles 006
Speaker
Justin KoUniversity of Waterloo

We consider the task of estimating a rank-one matrix from noisy observations. Models that fall in this framework include community detection and spiked Wigner models. In this talk, I will discuss pseudo-maximum likelihood theory for such inference problems. We provide a variational formula for the asymptotic maximum pseudo-likelihood and characterize the asymptotic performance of pseudo maximum likelihood estimators. We will also discuss the implications of these findings to least squares estimators. Our approach uses the recent connections between statistical inference and statistical physics, and in particular the connection between the maximum likelihood and the ground state of a modified spin glass.

Based on joint work with Curtis Grant and Aukosh Jagannath.

Homology cobordism and Heegaard Floer homology

Series
School of Mathematics Colloquium
Time
Thursday, September 19, 2024 - 11:00 for 1 hour (actually 50 minutes)
Location
Skiles 005
Speaker
Jen HomGeorgia Tech

Under the operation of connected sum, the set of three-manifolds form a monoid. Modulo an equivalence relation called homology cobordism, this monoid (of homology spheres) becomes a group. What is the structure of this group? What families of three-manifolds generate (or don’t generate) this group? We give some answers to these questions using Heegaard Floer homology. This is joint work with (various subsets of) I. Dai, K. Hendricks, M. Stoffregen, L. Truong, and I. Zemke.

Galois groups of reciprocal polynomials and the van der Waerden-Bhargava theorem

Series
Number Theory
Time
Wednesday, September 18, 2024 - 15:30 for 1 hour (actually 50 minutes)
Location
Skiles 005
Speaker
Evan O'DorneyCarnegie Mellon University

Given a random polynomial f of degree n with integer coefficients each drawn uniformly and independently from an interval [-H, H], what is the probability that the Galois group of the roots of f is NOT the full symmetric group Sₙ? In 1936, van der Waerden conjectured that the answer should be of order 1/H, with the dominant contribution coming from f with a rational root. This conjecture was finally resolved by Bhargava in 2023. In this project (joint w/ Theresa Anderson), we ask the same question for reciprocal (a.k.a. palindromic) polynomials, which arise for instance as the characteristic polynomials of symplectic matrices. Using a suitably modified variant of the Fourier-analytic methods of Bhargava and others, we find that polynomials with non-generic Galois group appear with frequency O(log H/H) and, unlike in van der Waerden's setting, almost all of these polynomials are irreducible.

Finite-time blowup for the Fourier-restricted Euler and hypodissipative Navier-Stokes model equations

Series
PDE Seminar
Time
Tuesday, September 17, 2024 - 15:30 for 1 hour (actually 50 minutes)
Location
Skiles 006
Speaker
Evan MillerUniversity of Alabama in Huntsville

In this talk, I will introduce the Fourier-restricted Euler and hypodissipative Navier–Stokes equations. These equations are analogous to the Euler equation and hypodissipative Navier–Stokes equation, respectively, but with the Helmholtz projection replaced by a projection onto a more restrictive constraint space. The nonlinear term arising from the self-advection of velocity is otherwise unchanged. I will prove finite time-blowup when the dissipation is weak enough, by making use of a permutation symmetric Ansatz that allows for a dyadic energy cascade of the type found in the Friedlander-Katz-Pavlović dyadic Euler/Navier–Stokes model equation.

 On orientations of graphs with forbidden out-degrees (Owen Henderschedt, Auburn)

Series
Graph Theory Seminar
Time
Tuesday, September 17, 2024 - 15:30 for 1 hour (actually 50 minutes)
Location
Skiles 005
Speaker
Owen HenderschedtAuburn University

 When does a graph admit an orientation with some desired properties? This question has been studied extensively for many years and across many different properties. Specifically, I will talk about properties having to do with degree restrictions, and progress towards a conjecture of Akbari, Dalirrooyfard, Ehsani, Ozeki, and Sherkati dealing with a list-type of degree restriction. This is all joint work with my PhD advisor Jessica McDonald.

Pages