Past Events

Seminars are held Tuesdays and Thursdays in POB 6.304 from 3:30-5:00 pm, unless otherwise noted. Speakers include scientists, researchers, visiting scholars, potential faculty, and ICES/UT Faculty or staff. Everyone is welcome to attend. Refreshments are served at 3:15 pm.

Friday, Feb 15

  • Additional Information

    Hosted by Irene Gamba

    Sponsor: ICES Seminar-Numerical Analysis Series

    Speaker: Theirry Magin

    Speaker Affiliation: The von Karman Institute for Fluid Dynamics

  • Abstract

    Solving nonconservative hyperbolic systems is a difficult problem encountered in a variety of applications from plasma to two-phase flows. In particular, the numerical simulation of two-temperature fluid models exhibits a nonconservative product in the electron energy equation. We derive jump conditions based on travelling wave solutions and propose an original numerical treatment in order to avoid non-physical shocks for the solution, which remains valid in the case of coarse-resolution simulations. A key element for the numerical scheme proposed is the presence of diffusion in the electron variables. A scaling obtained from dimensional analysis allows us to derive a fluid model following a multiscale Chapman-Enskog expansion method. The numerical strategy is assessed for a solar physics test case. The computational method is able to capture the travelling wave solutions in both the highly- and coarsely-resolved cases.

Thursday, Feb 14

High Order Immersed Finite Element Methods for Interface Problems

Thursday, Feb 14, 2019 from 3:30PM to 5PM | POB 6.304

  • Additional Information

    Hosted by Ivo Babuska

    Sponsor: ICES Seminar

    Speaker: Slimane Adjerid

    Speaker Affiliation: Professor, Mathematics, Rensselaer Polytechnic

  • Abstract

    Many physical phenomena such as heat conduction and wave propagation in inhomogeneous media is modeled by partial differential equations with discontinuous coefficients referred to as interface problems. We introduce and motivate the immersed finite element approach for solving interface problems. The immersed finite element methods allow elements to be cut by the interface leading to special piecewise polynomial finite element spaces and modified weak formulations.

    A brief historical review of immersed finite element methods will be presented. We will show how to construct high order immersed finite element spaces and weak Galerkin formulations for high accuracy computations. We will present computational results for several applications from acoustics and fluid dynamics and conclude with a list of open questions and future research projects.

    Professor Adjerid received his PhD in mathematics from RPI in 1985. In 1998-2005 he was an Associate Professor of Mathematics and Virginia Tech. In 2005, he was Professor of Mathematics, Virginia Tech. His principal area of research is the dinite element methods for PDEs.

Tuesday, Feb 12

Control of Uncertain Autonomous Systems with Intermittent Feedback

Tuesday, Feb 12, 2019 from 11AM to 12:30PM | POB 2.402 (Electronic)

Important Update: Please note: Different Location, Time
  • Additional Information

    Hosted by Ufuk Topcu

    Sponsor: ICES Seminar

    Speaker: Warren Dixon

    Speaker Affiliation: Newton C. Ebaugh Professorship, Mechanical and Aerospace Engineering Department, University of Florida

  • Abstract

    Autonomous systems use closed-loop feedback of sensed or communicated information to meet desired objectives. Meeting such objectives is more challenging when autonomous systems are tasked with operating in uncertain complex environments with intermittent feedback. This presentation explores different analysis methods that quantify the effects of intermittent feedback with respect to stability and performance of the autonomous agent. Various scenarios are considered where the intermittency results from natural phenomena or adversarial actors, including purposeful intermittency to enable new capabilities. Specific examples include intermittency due to occlusions in image-based feedback and intermittency resulting from various network control problems.

    Prof. Warren Dixon received his Ph.D. in 2000 from the Department of Electrical and Computer Engineering from Clemson University. He worked as a research staff member and Eugene P. Wigner Fellow at Oak Ridge National Laboratory (ORNL) until 2004, when he joined the University of Florida in the Mechanical and Aerospace Engineering Department where he currently holds the Newton C. Ebaugh professorship. His main research interest has been the development and application of Lyapunov-based control techniques for uncertain nonlinear systems. His work has been recognized by a number of early career, best paper, and student mentoring awards. He is a Fellow of ASME and IEEE for his contributions to control of uncertain nonlinear systems.

Monday, Feb 11

Drug Discovery with Computers

Monday, Feb 11, 2019 from 3PM to 4:30PM | POB 2.402 (Electronic)

Important Update: NOTE: seminar to be held in POB 2.402
  • Additional Information

    Hosted by Ron Elber

    Sponsor: ICES Seminar-Molecular Biophysics Series

    Speaker: Jeremy Smith

    Speaker Affiliation: Governor’s Chair Professor, BCMB Director, University of Tennessee, ORNL Center for Molecular Biophysics

  • Abstract

    The availability of large numbers of 3D protein structures means that structure-based computation can take center stage in drug discovery efforts. We show how computational drug design protocols benefit considerably from a dynamic, rather than just a static, description of the protein to be modulated. We describe how taking these motions into account in virtual high-throughput screening has led to the discovery of lead compounds for a variety of diseases.

Friday, Feb 8

Explorations in Computational Molecular Biology

Friday, Feb 8, 2019 from 10AM to 11AM | POB 6.304

  • Additional Information

    Hosted by Tom O'Leary-Roseberry

    Sponsor: ICES Seminar-Babuska Forum Series

    Speaker: John Hawkins

    Speaker Affiliation: Postdoctoral researcher. CES

  • Abstract

    Molecular biology is currently in a golden age. Cheap DNA sequencing and cheap DNA editing with CRISPR proteins are both only a few years old, and in that short time they have revolutionized seemingly every line of inquiry across the whole of life sciences. Cheap DNA printing is just starting to take off as well. The dramatic increase in scale of data and the scale of inquiry has opened up opportunities for those with the computational and mathematical skills to help forge new paths and make sense of the deluge of data.

    During this talk, I will share a few of the projects from my corner of the space, including large-scale characterization of CRISPR genome-editing proteins and storing cat gifs in DNA.

Thursday, Feb 7

Static condensation, hybridization and the devising of the HDG methods. (Title change)

Thursday, Feb 7, 2019 from 3:30PM to 5PM | POB 6.304

Important Update: NOTE: Title and Abstract Change
  • Additional Information

    Hosted by Tan Bui-Thanh

    Sponsor: ICES Seminar

    Speaker: Bernardo Cockburn

    Speaker Affiliation: Minnesota Supercomputing Institute, University of Minnesota at Twin Peaks

  • Abstract

    In the framework of steady-state diffusion problems, we show how the ideas of static condensation and hybridization lead to the introduction of the hybridizable discontinuous Galerkin methods.

    Professor Cockburn received his Ph.D from University of Chicago in 1986 under the direction of Jim Douglas, Jr. He has spent all his academic career at University of Minnesota where he is now Distinguished McKnight University Professor. His research interests include the development of Discontinuous Galerkin methods for nonlinear conservation laws, second-order elliptic problems, electro-magnetism, wave propagation and elasticity.

Tuesday, Feb 5

Optimal Operator and Experimental Design for Uncertain Systems

Tuesday, Feb 5, 2019 from 3:30PM to 5PM | POB 6.304

  • Additional Information

    Hosted by Omar Ghattas

    Sponsor: ICES Seminar

    Speaker: Edward R. Dougherty

    Speaker Affiliation: Chair and Distinguished Professor, Department of Electrical and Computer Engineering, and Scientific Director of the Center for Bioinformatics and Genomic Systems Engineering, Texas A&M University

  • Abstract

    The most basic aspect of modern engineering is the design of an operator to act on a physical system in an optimal manner relative to a desired objective – for instance, designing a control policy to autonomously direct a system or designing a classifier to make decisions regarding the system. In the classical paradigm, knowledge regarding the system model is assumed to be certain; however, in practice, especially with complex systems, knowledge is uncertain and operators must be designed while taking this uncertainty into account. An intrinsically Bayesian robust (IBR) operator is optimal relative to both the operational objective and the partial knowledge, as quantified by a prior distribution over an uncertainty class of possible models. An objective-based experimental design procedure is naturally related to optimal operator design because, if we are to select among a collection of experiments, then we would like to perform an experiment that maximally reduces objective-related uncertainty. This uncertainty is quantified via the mean objective cost of uncertainty (MOCU), which measures the expected cost of applying an IBR operator instead of actually optimal operators across the uncertainty class.

    Edward R. Dougherty is the Robert M. Kennedy ‘26 Chair and Distinguished Professor in the Department of Electrical and Computer Engineering at Texas A&M University, and is Scientific Director of the Center for Bioinformatics and Genomic Systems Engineering. He holds a Ph.D. in mathematics from Rutgers University and the Doctor Honoris Causa from the Tampere University of Technology.

Tuesday, Jan 22

  • Additional Information

    Hosted by Karen Willcox

    Sponsor: ICES Seminar

    Speaker: Joshua Chang

    Speaker Affiliation: Assistant Professor, Neurology and Population Health, Dell Medical School.

  • Abstract

    The use of electrical stimulation as a therapeutic modality is rapidly expanding its reach into all disciplines in medicine. These electroceuticals are being used to combat everything from Parkinsonian tremors, epilepsy to depression and inflammatory bowel disease. Compared to the field of pharmaceuticals where much is known about the pharmacodynamics and kinetics of molecular compounds, very little is known in this field regarding the electrical dynamics and kinetics of complex neurological networks. Most research done in this field has been through the use of rudimentary rectangular biphasic stimulus waveforms. In this talk, I will review some of the computational strategies that have been used to optimize the shape of the stimulus waveform, and I will discuss some potential ideas for future exploration.

    Joshua Chang is an assistant professor of the Departments of Neurology and Population Health at Dell Medical School at The University of Texas at Austin. He received his B.S. and M.Eng at MIT in Electrical Engineering and Computer Science, with a focusing on signal processing, control theory and artificial intelligence. After working a couple years as a software engineer and IT consultant, he pursued an MD/PhD at the University of Massachusetts Medical School, where he completed his thesis in Quantitative Health Sciences under the supervision of Dr. David Paydarfar in the design and development of evolutionary algorithms to optimize stimulus waveforms for implantable electrical devices. He continues his research here as an investigator of the Clayton Foundation for Research.

Tuesday, Jan 22

Differentially Private Linear-Quadratic Control

Tuesday, Jan 22, 2019 from 11AM to 12PM | POB 6.304

Important Update: Please NOTE: different time
  • Additional Information

    Hosted by Ufuk Topcu

    Sponsor: ICES Seminar

    Speaker: Matthew Hale

    Speaker Affiliation: Assistant Professor, Mechanical and Aerospace Engineering, University of Florida

  • Abstract

    As multi-agent systems grow and become increasingly data-driven, more and more personal data can be shared with unknown or unintended recipients. For example, self-driving cars may share position information for collision avoidance, and smart power grids may share power consumption data to optimize power generation. Even seemingly innocuous data can be very revealing about users, and new data-driven technologies must therefore protect sensitive user data while still allowing networks of agents to function. To address this need, I will present a differentially private implementation for multi-agent tracking control. This talk will use the classic linear-quadratic (LQ) tracking problem to give a broadly applicable problem formulation, and I will cover a recent privacy implementation that integrates a centralized cloud computer into an otherwise decentralized network. The agents add noise to all data sent to the cloud in order to enforce differential privacy, which gives each agent strong, rigorous privacy guarantees. In contrast to some existing approaches, the cloud does not need to be trusted and instead receives only private information from users, which it then uses to generate control values for them. Functions of private data are therefore fed back into the system. To characterize privacy in feedback, I will present numerical bounds on how difficult it is to compute control values using private user data. The end result of this work is a privacy implementation coupled with a method for quantitatively trading off individual privacy and aggregate performance in networks.

    Matthew Hale is an Assistant Professor of Mechanical and Aerospace Engineering at the University of Florida. He received his BSE in Electrical Engineering from the University of Pennsylvania in 2012, and his MS and PhD in Electrical and Computer Engineering from the Georgia Institute of Technology in 2015 and 2017, respectively. His research broadly pertains to designing coordination strategies for multi-agent systems under challenging conditions. Current research interests include privacy in control, asynchronous coordination of networks, and graph theory. He directs the Control, Optimization, and Robotics Engineering (CORE) Lab at the University of Florida, which houses a swarm robotics testbed for testing and validating algorithms developed by his group.

Tuesday, Jan 15

Computers and “Computing” in Medical Imaging

Tuesday, Jan 15, 2019 from 3:30PM to 5PM | POB 6.304

  • Additional Information

    Hosted by George Biros

    Sponsor: ICES Seminar

    Speaker: R. Nick Bryan, M.D.

    Speaker Affiliation: Chair, Department of Diagnostic Medicine, Dell Medical School

  • Abstract

    Prior to the advent of the x-ray CT scanner in 1973, there were no computers in radiology. Medical images were analog in origin and nature. Humans, often radiologists, analyzed and interpreted these ‘radiographs’ by direct visual observation of the analog image. By 2010 essentially all medical images were digital in origin and nature, but still analyzed and interpreted by humans, based on direct visual observation of the now digital images. Until very recently, any computing that was performed for image interpretation was performed by humans. Today, thanks to computer hardware and algorithmic advances, digital medical images are being analyzed and beginning to be interpreted by computers. I predict that within 10 years, no medical image will be viewed by a human until it has been analyzed and at least partially interpreted by a computer.

    The impact that computers have had on medical imaging cannot be overestimated and selective examples of their transformative impact on image creation, presentation, quantitative analysis and new research and clinical applications will be presented. However, it is the rapidly evolving AI/ML image analysis techniques that are of greatest contemporary interest. For radiologists the key question is, “Can a machine do what we do?” There are secondary, corollary questions with technological implications. “Can we teach a machine what we know and to do what we do,” “Can a machine by itself learn what we know and to do what we do,” or “Can a machine learn more than what we now know and use this new knowledge to make better clinical decisions?” Though computational technology is the topic of general interest, these specific questions relate the new technology to current human performance. Being a radiologist, and not a computer scientist, I will focus on what I think radiologists’ know and do in the image interpretative process. I will attempt to compare and contrast these human processes with what I think computer algorithms’ do, or might do when applied to similar task. Through this process I hope to support my self-serving bias that for the foreseeable future, medical image interpretations, including final diagnosis, will be a finely coordinated, joint product of the computer and radiologist, benefitting from their complementary computing strengths.

    R. Nick Bryan is the chair of the Department of Diagnostic Medicine for Dell Medical School. Bryan came to Austin from the Perelman School of Medicine at the University of Pennsylvania, where he served as the Eugene P. Pendergrass Professor and Chair of Radiology. A nationally known leader, thinker and innovator in the field, he also is past president of the Radiological Society of North America, the American Society of Neuroradiology, and the American Society of Head and Neck Radiology.

    Bryan previously served at the Johns Hopkins University School of Medicine, where he was professor of radiology and neurosurgery, director of the Neuroradiology Division and vice chairman of the Department of Radiology. He spent much of his childhood in Texas and earned degrees at the University of Texas Medical Branch in Galveston. He previously was director of neuroradiology at Houston Methodist and a professor of radiology at Baylor College of Medicine.