Oden Institute Past Seminars

In-person seminars are held Tuesdays and Thursdays in POB 6.304 from 3:30-5:00 pm, unless otherwise noted. Zoom seminars usually occur during the same time period. Speakers include scientists, researchers, visiting scholars, potential faculty, and Oden Institute/UT Faculty or staff. Everyone is welcome to attend. Refreshments are served at in-person seminars at 3:15 pm.

Friday, Nov 6

Tensor decomposition in data science

Friday, Nov 6, 2020 from 10AM to 11AM | Zoom Meeting - Babuška Forum series

  • Additional Information

    Hosted by Stefan Henneking

    Sponsor: Oden Institute Virtual Seminar - Babuška Forum series

    Speaker: Joe Kileel

    Speaker Affiliation: Assistant Professor, Oden Institute, Department of Math, UT Austin

  • Abstract

    Tensors are higher-order matrices, and decomposing tensors can reveal structure in datasets. In recent years, tensor decomposition has found applications in statistics, computational imaging, signal processing, and quantum chemistry. In this talk, we will present a new numerical method for low-rank symmetric tensor decomposition, building on the usual power method and ideas from classical algebraic geometry. The approach achieves a speed-up over the state-of-the-art by roughly one order of magnitude. We will also describe an “implicit” variant of the algorithm for the case of moment tensors which avoids the explicit formation of higher-order moments, analogously to matrix-free techniques in linear algebra. Time permitting, we will mention various open problems in the subject, concerning numerical stability, non-convex optimization, random behavior and challenging applications. This is based on joint works with Joao Pereira, Tammy Kolda and Timo Klock.

    Bio
    Joe Kileel is an Assistant Professor of the Oden Institute and Mathematics Department at UT Austin, since August 2020. Prior to this, he was a Simons postdoctoral fellow at the Program in Applied and Computational Mathematics, Princeton University and he obtained a PhD in Mathematics from UC Berkeley in 2017. His research interests are in mathematics of data, computational algebra, tensor methods, inverse problems and non-convex optimization.

    (The Babuška Forum series was started by Professor Ivo Babuška several years ago to expose students to interesting and curious topics relevant to computational engineering and science with technical content at the graduate student level (i.e. the focus of the lectures is on main ideas with some technical content). Seminar credit is given to those students who attend.)

  • Multimedia

     Event Stream Link: Click Here to Watch


Thursday, Nov 5

Inverting tumor angiogenesis with fluid flow

Thursday, Nov 5, 2020 from 3:30PM to 5PM | Zoom Meeting

  • Additional Information

    Hosted by Thomas Yankeelov

    Sponsor: Oden Institute Virtual Seminar

    Speaker: Hector Gomez

    Speaker Affiliation: Professor, School of Mechanical Engineering, Weldon School of Biomedical Engineering, Purdue Center for Cancer Research, Purdue University

  • Abstract

    Angiogenesis, the growth of blood vessels from pre-existing ones, plays a key role in cancer progression. Cancerous tumors release pro-angiogenic growth factors into the extracellular matrix that promote vessel growth once they reach the pre-existing vasculature. The neovasculature provides nutrients to the tumor, usually accelerating its growth. Therefore, the understanding and control of angiogenesis are critical to combat cancer. Until very recently, the literature had systematically assumed that the interstitial flows unavoidably occurring in the extracellular matrix had little impact on angiogenesis. Surprisingly, recent experimental evidence has shown that even very mild flows like those likely occurring in the human body can significantly alter vascular growth patterns. However, the mechanisms whereby fluid flow alters angiogenesis remain unknown; and different experiments show opposite effects of fluid flow on angiogenesis. In this seminar, I will present our recent modeling work to investigate the influence of fluid flow in tumor angiogenesis. Our model demonstrates the key role of interstitial flow in angiogenesis and reconciles two seemingly contradicting experiments: one showing more prominent angiogenic growth against the flow and another other showing more prominent growth with the flow. The model suggests that fluid flow may be used to invert the direction of angiogenic growth when combined with the adequate isoform of the growth factors.

    Bio
    Hector Gomez is currently a Professor in the School of Mechanical Engineering at Purdue University. Prof. Gomez specializes in computational mechanics with particular emphasis in isogeometric modeling and analysis, interfacial mechanics of multiphysics systems and simulation at the interface of engineering and medicine. Prof. Gomez’s research has been recognized with multiple awards including the Juan C. Simo Award from the Spanish Society of Computational Mechanics, the MIT Innovators Under 35 (Spain section), the Young Investigator Award from the Royal Academy of Engineering of Spain, the Gallagher Young Investigator Award and the Princess of Girona Scientific Research Award (the award is presented by the King of Spain to the best young researcher in all fields of science, engineering and humanities). Prof. Gomez has published over 80 journal papers and made over 130 contributions to conferences.

  • Multimedia

     Event Stream Link: Click Here to Watch


Tuesday, Nov 3

A Fast Approach to Optimal Transport: the Back-and-Forth Method

Tuesday, Nov 3, 2020 from 3:30PM to 5PM | Zoom Meeting

  • Additional Information

    Hosted by Richard Tsai

    Sponsor: Oden Seminar

    Speaker: Matt Jacobs

    Speaker Affiliation: Assistant Adjunct Professor, Department of Mathematics, UCLA

  • Abstract

    Given two probability measures and a transportation cost, the optimal transport problem asks to find the most cost efficient way to transport one measure to the other. Since its introduction in 1781 by Gaspard Monge, the optimal transport problem has found applications in logistics, economics, physics, PDEs, and more recently data science. However, despite sustained attention from the numerics community, solving optimal transport problems has been a notoriously difficult task.

    In this talk I will introduce the back-and-forth method, a new algorithm to efficiently solve the optimal transportation problem for a general class of strictly convex transportation costs. Given two probability measures supported on a discrete grid with n points, the method computes the optimal map in O(n log(n)) operations using O(n) storage space. As a result, the method can compute highly accurate solutions to optimal transportation problems on spatial grids as large as 4096 x 4096 and 384 x 384 x 384 in a matter of minutes. If time permits, I will demonstrate an extension of the algorithm to the simulation of a class of gradient flows.

    This talk is joint work with Flavien Leger.

    Bio
    Matt Jacobs earned his Ph.D. in Mathematics from the University of Michigan, 2017. His research interests include Calculus of variations, optimization, numerical methods, PDEs, optimal transport, fluid mechanics, free boundary problems, GANs.

         **Note:  Please join this Zoom seminar online with the "Audio Only" function (no video)**
    
  • Multimedia

     Event Stream Link: Click Here to Watch


Friday, Oct 30

3D Soft Tissue Simulations Using a Neural Network PDE Approach

Friday, Oct 30, 2020 from 10AM to 11AM | Zoom Meeting - Babuška Forum series

  • Additional Information

    Hosted by Stefan Henneking

    Sponsor: Oden Institute Virtual Seminar - Babuška Forum series

    Speaker: Michael Sacks

    Speaker Affiliation: Professor, Oden Institute, UT Austin

  • Abstract

    The ability to fully characterize and simulate the three-dimensional (3D) mechanical behavior of soft tissues is essential in understanding their function in health and disease. The complex 3D hierarchical structure of soft tissues results in their highly anisotropic mechanical behaviors, and it particular the spatial variations in fiber structure gives rise to structural and mechanical heterogeneity. To addresses issues in a full 3D context, we have developed a novel numerical-experimental approach to determine the optimal model form and parameter estimation for continuum constitutive models of soft tissues, as applied to the myocardium. This approach utilizes optimal experimental design of the full 3D kinematic (triaxial) experiments coupled to an inverse model that incorporated local fibrous structure to perform robust parameter estimation. However, in-silico implementation of such complex 3D continuum soft tissue constitutive models to obtain the responses of varying boundary conditions and fibrous structures requires the solution of the associated hyper-elasticity problem, which remains impractical in translational clinical time frames. To alleviate the associated substantial computational costs at the time of simulation, we have developed a neural network-based method that can simulate the 3D mechanical behavior of soft tissues. A physics-informed approach was employed to train the neural network to give physically correct solution by minimizing the potential energy without any labelled training datasets. The extensibility of the neural network for problems with fiber structures and loading path conditions is also discussed.

    Bio
    Michael Sacks is professor of biomedical engineering and holder of the W. A. "Tex" Moncrief, Jr. Endowment in Simulation-Based Engineering and Sciences Chair No. 1. He is also director of the Oden Institute's Willerson Center for Cardiovascular Modeling and Simulation. Sacks formerly held the John A. Swanson Chair in the Department of Bioengineering at the University of Pittsburgh. He earned his B.S. and M.S. in engineering mechanics from Michigan State University, and his Ph.D. in biomedical engineering (biomechanics) from The University of Texas Southwestern Medical Center at Dallas. He is a world authority on cardiovascular biomechanics, with a focus on the quantification and simulation of the structure-mechanical properties of native and engineered cardiovascular soft tissues. He is a leading authority on the mechanical behavior and function of heart valves, including the development of the first constitutive models for these tissues using a structural approach. He is also active in the biomechanics of engineered tissues, and in understanding the in-vitro and in-vivo remodeling processes from a functional biomechanical perspective. His research includes multiscale studies of cell/tissue/organ mechanical interactions in heart valves and he is particularly interested in determining the local stress environment for heart valve interstitial cells. His recent research has included developing novel constitutive models of right ventricular myocardium that allow for the individual contributions of the myocyte and connective tissue networks.

    (The Babuška Forum series was started by Professor Ivo Babuška several years ago to expose students to interesting and curious topics relevant to computational engineering and science with technical content at the graduate student level (i.e. the focus of the lectures is on main ideas with some technical content). Seminar credit is given to those students who attend.)

  • Multimedia

     Event Stream Link: Click Here to Watch


  • Additional Information

    Hosted by Shane McQuarrie

    Sponsor: Oden Institute Virtual Seminar - Student Forum series

    Speaker: Jon Kelley

    Speaker Affiliation: CSEM student, Oden Institute, UT Austin

  • Abstract

    This talk presents the path to adapt and integrate a recently-developed physics-based low-rank approximation algorithm for hierarchical-matrix (H-matrix) blocks, which has so far been demonstrated only for accelerating integral kernels (point sources and observers), to an iterative method of moments (Mom) solution of the surface electric field integral equation (S-EFIE) and combined field integral equation (S-CFIE) using divergence-conforming basis/testing functions.

    Bio
    Jon Kelley is a third-year CSEM student advised by Dr. Ali Yilmaz. His research is in computational electromagnetics, focusing on physics-based low-rank approximation methods and scattering problems with aerospace applications.

    (The CSEM Student Forum is a seminar series given by current CSEM graduate students to their peers. The aim of the forum is to expose students to each other's research, encourage collaboration, and provide opportunities to practice presentation skills. First- and second-year CSEM students receive seminar credit for attending.)

  • Multimedia

     Event Stream Link: Click Here to Watch


Thursday, Oct 29

Post-hoc Uncertainty Quantification for Remote Sensing Observing Systems

Thursday, Oct 29, 2020 from 3:30PM to 5PM | Zoom Meeting

  • Additional Information

    Hosted by Tan Bui-Thanh

    Sponsor: Oden Institute Virtual Seminar

    Speaker: Amy Braverman

    Speaker Affiliation: Principal Statistician, Jet Propulsion Laboratory, California Institute of Technology

  • Abstract

    The ability of spaceborne remote sensing data to address important Earth and climate science problems rests crucially on how well the underlying geophysical quantities can be inferred from these observations. Remote sensing instruments measure parts of the electromagnetic spectrum and use computational algorithms to infer the unobserved true physical states. However, the accompanying uncertainties, if they are provided at all, are usually incomplete. There are many reasons why including but not limited to unknown physics, computational artifacts and compromises, unknown uncertainties in the inputs, and more. In this talk I will describe a practical methodology for uncertainty quantification of physical state estimates derived from remote sensing observing systems. The method we pro-pose combines Monte Carlo simulation experiments with statistical modeling to approximate conditional distributions of unknown true states given point estimates produced by imperfect operational algorithms. Our procedure is carried out post-hoc; that is, after the operational processing step because it is not feasible to redesign and rerun operational code. I demonstrate the procedure using four months of data from NASA’s Orbiting Carbon Observatory-2 mission, and compare our results to those obtained by validation against data from the Total Carbon Column Observing Network where it exists.

    Keywords: Uncertainty quantification, remote sensing, carbon cycle science.

    This is joint work by Amy Braverman, Jonathan Hobbs, Joaquim Teixeira, and Michael Gunson Jet Propulsion Laboratory, California Institute of Technology. Amy.Braverman@jpl.nasa.gov

    Bio
    Dr. Amy Braverman is a Principal Statistician at the Jet Propulsion Laboratory in Pasadena, California. She received her doctorate in statistics from the University of California, Los Angeles (UCLA), a masters in Mathematics from UCLA, and a B.A. degree in economics from Swarthmore College, Swarthmore, PA, in 1982. Her research interests include information-theoretic approaches for the analysis of massive data sets, data fusion methods for combining heterogeneous, spatial and spatio-temporal data, and statistical methods for the evaluation and diagnosis of climate models, particularly by comparison to observational data. Dr. Braverman focuses on the use of remote sensing data, and has designed and analyzed new Level 3 data products for MISR and other NASA missions.

  • Multimedia

     Event Stream Link: Click Here to Watch


Tuesday, Oct 27

Online Nonnegative Matrix Factorization for Markovian and Other Real Data

Tuesday, Oct 27, 2020 from 3:30PM to 5PM | Zoom Meeting

  • Additional Information

    Hosted by Rachel Ward

    Sponsor: Oden Institute Virtual Seminar

    Speaker: Deanna Needell

    Speaker Affiliation: Professor, Department of Mathematics, UCLA

  • Abstract

    Online Matrix Factorization (OMF) is a fundamental tool for dictionary learning problems, giving an approximate representation of complex data sets in terms of a reduced number of extracted features. Convergence guarantees for most of the OMF algorithms in the literature assume independence between data matrices, and the case of dependent data streams remains largely unexplored. In this talk, we present results showing that a non-convex generalization of the well-known OMF algorithm for i.i.d. data converges almost surely to the set of critical points of the expected loss function, even when the data matrices are functions of some underlying Markov chain satisfying a mild mixing condition. As the main application, by combining online non-negative matrix factorization and a recent MCMC algorithm for sampling motifs from networks, we propose a novel framework of Network Dictionary Learning that extracts `network dictionary patches' from a given network in an online manner that encodes main features of the network. We demonstrate this technique on real-world data and discuss recent extensions and variations.

    Bio
    Deanna Needell earned her PhD from UC Davis before working as a postdoctoral fellow at Stanford University. She is currently a full professor of mathematics at UCLA. She has earned many awards including the IEEE Best Young Author award, the Hottest paper in Applied and Computational Harmonic Analysis award, the Alfred P. Sloan fellowship, an NSF CAREER and NSF BIGDATA award, and the prestigious IMA prize in Applied Mathematics. She has been a research professor fellow at several top research institutes including the Mathematical Sciences Research Institute and Simons Institute in Berkeley. She also serves as associate editor for IEEE Signal Processing Letters, Linear Algebra and its Applications, the SIAM Journal on Imaging Sciences, and Transactions in Mathematics and its Applications as well as on the organizing committee for SIAM sessions and the Association for Women in Mathematics.

         **Note:  Please join this Zoom seminar online with the "Audio Only" function (no video)**
    
  • Multimedia

     Event Stream Link: Click Here to Watch


Friday, Oct 23

Using Continuum Limits To Understand Data Clustering And Classification

Friday, Oct 23, 2020 from 10AM to 11AM | Zoom Meeting

  • Additional Information

    Hosted by Stefan Henneking

    Sponsor: Oden Institute Virtual Seminar - Babuška Forum series

    Speaker: Franca Hoffmann

    Speaker Affiliation: Bonn Junior Fellow, University of Bonn, Germany

  • Abstract

    Graph Laplacians encode geometric information contained in data, via the eigenfunctions associated with their small eigenvalues. These spectral properties provide powerful tools in data clustering and data classification. When a large number of data points are available one may consider continuum limits of the graph Laplacian, both to give insight and, potentially, as the basis for numerical methods. We summarize recent insights into the properties of a family of weighted elliptic operators arising in the large data limit of different graph Laplacian normalizations, and propose an inverse problem formalism for continuous semi-supervised learning algorithms, making use of these differential operators. This is joint work with Bamdad Hosseini (Caltech), Assad A. Oberai (USC) and Andrew M. Stuart (Caltech).

    Bio
    Franca Hoffmann is a Bonn Junior Fellow at University of Bonn (Germany). After completing her PhD at the Cambridge Centre for Analysis at University of Cambridge (UK) in 2017, she held the position of von Karman instructor at California Institute of Technology (US) from 2017 to 2020. Her research is focused on the applied mathematics/data analysis interface, driven by the need to provide rigorous mathematical foundations for modeling tools used in applications. In particular, Franca is interested in the theory of nonlinear and nonlocal PDEs, as well as in developing novel tools for data analysis and mathematical approaches to machine learning.

    (The Babuška Forum series was started by Professor Ivo Babuška several years ago to expose students to interesting and curious topics relevant to computational engineering and science with technical content at the graduate student level (i.e. the focus of the lectures is on main ideas with some technical content). Seminar credit is given to those students who attend.)

  • Multimedia

     Event Stream Link: Click Here to Watch


Thursday, Oct 22

Large Deviations in Nanoscale Transport Phenomena

Thursday, Oct 22, 2020 from 3:30PM to 5PM | Zoom Meeting

  • Additional Information

    Hosted by Ron Elber

    Sponsor: Oden Institute Virtual Seminar

    Speaker: David Limmer

    Speaker Affiliation: Professor, College of Chemistry, Berkeley

  • Abstract

    In this talk, I will discuss some recent efforts to develop a molecular perspective on nanoscale transport, including elucidating the molecular motions that underlie nonlinear responses, and molecular simulation techniques to study such processes on a computer. This work leverages recent advancements in applied mathematics in the study of large deviations and control theory. Specific questions concerning anomalous heat transport in low dimensional lattices and nonlinear electrokinetic phenomena in ionic solutions will be addressed.

    Bio
    David Limmer is an Assistant Professor in the Department of Chemistry at University of California Berkeley, a Research Scientist in the Materials and Chemical Sciences Divisions of Lawrence Berkeley National Laboratory, and a Fellow of the Kavli Energy NanoSciences Institute. He received his B.S. in chemical engineering in 2008 from the New Mexico Institute of Mining and Technology, and his Ph.D. in chemistry from the University of California, Berkeley under the supervision of David Chandler. From 2013-2016, David was an independent fellow of the Princeton Center for Theoretical Science. David has been recognized as a Heising-Simons Fellow of the Kavli Foundation, a Scialog Fellow of the Research Corporation for Science and Gordon and Betty Moore Foundation, and a Hellman Fellow. In 2019, he was the recipient of the Department of Energy Early Career Award.

         **Note:  Please join this Zoom seminar online with the "Audio Only" function (no video)**
    
  • Multimedia

     Event Stream Link: Click Here to Watch


Tuesday, Oct 20

SuiteSparse:GraphBLAS: graph algorithms in the language of sparse linear algebra

Tuesday, Oct 20, 2020 from 3:30PM to 5PM | Zoom Meeting

  • Additional Information

    Hosted by Robert van de geijn

    Sponsor: Oden Institute Virtual Seminar

    Speaker: Tim Davis

    Speaker Affiliation: Professor, Department of Computer Science and Engineering, Texas A&M University

  • Abstract

    SuiteSparse:GraphBLAS is a full implementation of the GraphBLAS standard, which defines a set of sparse matrix operations on an extended algebra of semirings using an almost unlimited variety of operators and types. When applied to sparse adjacency matrices, these algebraic operations are equivalent to computations on graphs. GraphBLAS provides a powerful and expressive framework for creating graph algorithms based on the elegant mathematics of sparse matrix operations on a semiring. Key features and performance of the SuiteSparse implementation of GraphBLAS package are described. The implementation appear in Linux distros, and forms the basis of the RedisGraph module of Redis, a commercial graph database system. Graph algorithms written in GraphBLAS can rival the performance of highly-tuned specialized kernels, while being far simpler for the end user to write.

    Bio
    Tim Davis is a Professor in the Computer Science and Engineering Department at Texas A&M University. His primary scholarly contribution is the creation of widely-used sparse matrix algorithms and software (including x=A\b in MATLAB). Davis is a Fellow of SIAM, ACM, and IEEE.

  • Multimedia

     Seminar Slides: Click Here to View Slides

     Event Stream Link: Click Here to Watch