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, Oct 16

Powering Next-Generation MRI with Physics, Information, and Computation: An Integrated Encoding and Decoding Approach

Friday, Oct 16, 2020 from 9:30AM to 10:30AM | Zoom Meeting

Important Update: NOTE: Seminar begins earlier than usual start time.
  • Additional Information

    Hosted by Stefan Henneking

    Sponsor: Oden Institute Virtual Seminar - Babuška Forum series

    Speaker: Bo Zhao

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

  • Abstract

    Magnetic resonance imaging (MRI) is a powerful and versatile imaging technology, which has provided unprecedented capabilities to probe the structural, functional, and metabolic information of living systems. Since its inception, the MR imaging process has been formulated as a “communication” problem – i.e., it involves both encoding and decoding. The encoding process maps an underlying image function that depends on physical, physiological, and experimental parameters into sensory data utilizing spin physics; and the decoding process reconstructs this desired image function from the measured data. This long-standing encoding/decoding model often results in poor trade-offs between image resolution, signal-to-noise ratio, and data acquisition speed, which limits the practical utility of high-dimensional MRI.

    In this talk, I will present a novel imaging framework to tackle these challenges, by using an integrated encoding and decoding paradigm. The proposed framework leverages advanced mathematical models and algorithms to tightly integrate the encoding and decoding processes. It exploits the synergistic interactions between spin physics, statistical inference, and machine learning to help overcome major technical barriers of the existing MRI techniques. I will illustrate the power of this framework using two concrete approaches, i.e., subspace imaging and statistical imaging, and will highlight their impacts on applications in cardiovascular imaging and quantitative neuroimaging. Finally, I will discuss some exciting new opportunities with this framework that leverage the rapid development of advanced computing and machine learning technologies.

    Bio
    Bo Zhao is an Assistant Professor at the Oden Institute for Computational Engineering and Sciences and the Department of Biomedical Engineering. His research is in the general area of computational imaging and medical imaging, which lies at the intersection of imaging science and data science. Specifically, his group focuses on developing novel mathematical models, computational algorithms, and data acquisition schemes to address inverse problems in magnetic resonance (MR) imaging. His group takes unique approaches that synergistically integrate spin physics, information processing, and advanced computing to push the performance limits of MR imaging systems.

    (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: 1. Matt Smith and 2. Cyrus Neary

    Speaker Affiliation: CSEM PhD Program, Oden Institute, UT Austin

  • Abstract

    Speaker 1: Matt Smith
    Title: Hidden Markov Models for Protein Fluorosequencing
    Abstract: A research group in Dr. Edward Marcotte's lab is working on a new technology called protein fluorosequencing that depends partly on the fast and accurate identification of small strings of amino acids based on time series data. In this talk, we develop a Hidden Markov Model (HMM) to perform this classification and compare it with other methods. We also explore the possibility of combining the HMM with faster methods to increase speed while retaining accuracy for large scale problems.
    Bio: Matt Smith is a third year student in the CSEM program, advised by Dr. Edward Marcotte. His research is in the computational aspects of protein fluorosequencing.

    Speaker 2: Cyrus Neary
    Title: Reward Machines for Cooperative Multi-Agent Reinforcement Learning
    Abstract: In cooperative multi-agent reinforcement learning, a collection of agents learns to interact in a shared environment to achieve a common goal. We propose the use of reward machines (RM) --- Mealy machines used as structured representations of reward functions --- to encode the team's task. The proposed novel interpretation of RMs in the multi-agent setting explicitly encodes required teammate interdependencies and independencies, allowing the team-level task to be decomposed into sub-tasks for individual agents. We define such a notion of RM decomposition and present algorithmically verifiable conditions guaranteeing that distributed completion of the sub-tasks leads to team behavior accomplishing the original task. Experimental results in three discrete settings exemplify the effectiveness of the proposed RM decomposition approach, which converges to a successful team policy two orders of magnitude faster than a centralized learner and significantly outperforms hierarchical and independent q-learning approaches.
    Bio: Cyrus Neary is a third-year CSEM PhD student. He works with Dr. Ufuk Topcu in the Autonomous Systems Group. His research currently focuses on structured task representations for reinforcement learning.

    (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 15

Crossing domain-specific boundaries for performance and portability

Thursday, Oct 15, 2020 from 3:30PM to 5PM | Zoom Meeting

  • Additional Information

    Hosted by Robert van de geijn

    Sponsor: Oden Institute Virtual Seminar

    Speaker: Tze Meng Low

    Speaker Affiliation: Professor, Electrical and Computer Engineering, Carnegie Mellon University

  • Abstract

    New applications domains are emerging due to the increasing computational capabilities of modern architectures. At the same time, these architectures are becoming more complex; making them more difficult to program. These two trends make it more essential that expert developers can quickly design high performance implementations to support the community needs. Our approach is to capture expert knowledge In the form of design patterns and analytical models that tie hardware features to software parameters. These patterns and models are then applied in other domains and to new architectures to quickly develop high performance implementations for the new domain and on newer architectures. We demonstrate our approach through a variety of examples from different applications, including population genomics, graph algorithms and machine learning.

    Bio
    Tze Meng Low is an assistant research professor in the Electrical and Computer Engineering Department at Carnegie Mellon University. He graduated from the University of Texas at Austin with a Ph.D. in Computer Science in 2013. His research focuses on the development of systematic approaches and analytical models to achieve the vertical integration of high performance algorithms, software, and hardware. His current interest is in the use of models and insights from one application domain to develop new high performance implementations in other domains such as bioinformatics, signal processing, deep learning and graph algorithms.

  • Multimedia

     Event Stream Link: Click Here to Watch


Tuesday, Oct 13

Multilevel Norms for Negative Order Sobolev Spaces

Tuesday, Oct 13, 2020 from 3:30PM to 5PM | Zoom Meeting

  • Additional Information

    Hosted by Leszek Demkowicz

    Sponsor: Oden Institute Virtual Seminar

    Speaker: Thomas Führer

    Speaker Affiliation: Assistant Professor, Facultad de Matemáticas, Pontificia Universidad Católica de Chile

  • Abstract

    In this talk, I present some recent results on multilevel decompositions of piecewise constants on simplicial meshes that are stable in negative order Sobolev spaces. Our findings can be applied to define local multilevel diagonal preconditioners that lead to bounded condition numbers (independent of the mesh-sizes and levels) and have optimal computational complexity. We discuss multilevel norms based on local (quasi-)projection operators that allow the efficient evaluation of negative order Sobolev norms.
    Finally, some extensions and possible further applications will conclude the talk.

    Bio
    since 07/2017 Assistant Professor position at Facultad de Matemáticas, Pontificia Universidad Católica de Chile
    01/2017-06/2017 PostDoc position at Institute for Analysis and Scientific Computing, Vienna University of Technology
    01/2015-10/2016 PostDoc position at Facultad de Matemáticas, Pontificia Universidad Católica de Chile
    06/2014 Ph.D. graduation in Mathematics, Vienna University of Technology
    03/2011 Bachelor of Science graduation in Technical Physics (B.Sc.), Vienna University of Technology
    10/2010 Diploma in Technical Mathematics, Vienna University of Technology

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

     Event Stream Link: Click Here to Watch


Friday, Oct 9

  • Additional Information

    Hosted by Stefan Henneking

    Sponsor: Oden Institute Virtual Seminar - Babuška Forum series

    Speaker: Ufuk Topcu

    Speaker Affiliation: Associate Professor, Oden Institute & Department of Aerospace Engineering, UT Austin

  • Abstract

    Autonomous systems are emerging as a driving technology for countless applications. Numerous disciplines tackle the challenges toward making these systems agile, adaptable, reliable, user-friendly and economical. On the other hand, the existing disciplinary boundaries delay and possibly even obstruct progress. I argue that the non-conventional problems that arise in the design and verification of autonomous systems require hybrid solutions at the intersection of learning, formal methods, and controls. I will present examples of such hybrid solutions in several problems in autonomy at varying levels of detail.

    Bio
    Ufuk Topcu joined the Department of Aerospace Engineering at the University of Texas at Austin in Fall 2015. He received his Ph.D. degree from the University of California at Berkeley in 2008. He held research positions at the University of Pennsylvania and California Institute of Technology. His research focuses on the theoretical, algorithmic and computational aspects of design and verification of autonomous systems through novel connections between formal methods, learning theory, and controls.

    (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.)

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

     Event Stream Link: Click Here to Watch


Thursday, Oct 8

Numerical Methods for Predicting Coastal Flooding With Uncertainty

Thursday, Oct 8, 2020 from 3:30PM to 5PM | Zoom Meeting

  • Additional Information

    Hosted by Clint Dawson

    Sponsor: Oden Institute Virtual Seminar

    Speaker: Kyle T. Mandli

    Speaker Affiliation: Professor, Department of Applied Physics and Applied Mathematics, Columbia University

  • Abstract

    Coastal hazards related to strong storms are one of the most ubiquitous of hazards to coastal communities throughout the world. In particular storm surge, the rise of the sea surface in response to wind and pressure forcing from these storms, can have a devastating effect at the coastline. Changes to the climate only compound the need for predictive tools that can also handle the uncertainty inherent in climate predictions. Computational approaches are of course the go-to tool for dealing with these difficulties but it is a far from trivial problem. The problem is inherently multi-scale, the uncertainties difficult to represent, and the hyperbolic structure of the most well-used set of representative equations, the shallow water equations, presents additional issues when looking for low-rank approximations.

    This talk will describe many of these difficulties, where they come from, and what research efforts are attempting to address them. This includes extensions to the shallow water equations, techniques for representing the uncertainty and measuring sensitivity in the problem, and finally how reduce order modeling may help to produce low-rank approximations to hyperbolic equations in general.

    Bio
    Kyle Mandli is Associate Professor of Applied Mathematics in the department of Applied Physics and Applied Mathematics and affiliated with the Columbia Data Science Institute. Before Columbia he was at the University of Texas at Austin where he was a Research Associate at the Institute for Computational and Engineering Sciences working in the computational hydraulics group. He received his Ph.D. in Applied Mathematics in 2011 from the University of Washington studying multi-layered flow as it applies to storm-surge simulation. His research interests involve the computational and analytical aspects of geophysical shallow mass flows such as storm-surge, tsunamis, and other coastal flooding. This also includes the development of advanced computational approaches, such as adaptive mesh refinement, leveraging novel computational technologies, such as accelerators, and the application of good software development practices as applied more generally to scientific and engineering software.

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

     Event Stream Link: Click Here to Watch


Tuesday, Oct 6

Conforming Galerkin Schemes Via Traces and Applications to Plate Bending

Tuesday, Oct 6, 2020 from 3:30PM to 5PM | Zoom Meeting

  • Additional Information

    Hosted by Leszek Demkowicz

    Sponsor: Oden Institute Virtual Seminar

    Speaker: Norbert Heuer

    Speaker Affiliation: Professor, Department of Mathematics, Pontificia Universidad Católica de Chile

  • Abstract

    In recent years the DPG method has raised some attention. It is a discontinuous Petrov-Galerkin method where the selection of special test functions guarantees discrete stability. In this way, for a given well-posed problem, any well-posed variational formulation is appropriate to set up a Galerkin approximation. Practical and theoretical reasons suggest to use ultraweak variational formulations. In this case, field variables are considered in L_2 so that test functions carry all the appearing derivatives. Transferring derivatives to test functions by integrating by parts, this gives rise to trace terms and thus, trace operators. In the ultraweak case, trace operators carry all the regularity weight of the problem. They have to be defined in appropriate spaces with corresponding images. They also carry the burden of conformity, when and where wanted. Independently of the ultraweak formulation and implied DPG scheme, the conformity of trace approximations is essential to understand and characterize the conformity of Galerkin schemes in general. We discuss this relation, and strategies and arising difficulties of this approach in the case of plate bending models.

    Bio
    Phd 1992, Habilitation 1998, both at University of Hanover, Germany
    1992-2000 Research Associate at U of Bremen, Germany
    2000-2004 Full Professor at U of Concepcion, Chile
    2004-2008 Full Professor at Brunel University, UK
    since 2008 Full Professor at Pontical Catholic University of Chile

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

     Event Stream Link: Click Here to Watch


  • Additional Information

    Hosted by Shane McQuarrie

    Sponsor: Oden Institute Virtual Seminar - Student Forum Series

    Speaker: Shane McQuarrie & Amelia Henriksen

    Speaker Affiliation: McQuarrie is a third-year CSEM PhD student & Amelia Henriksen is fifth-year PhD candidate, Oden Institute, UT Austin

  • Abstract

    Speaker 1: Shane McQuarrie
    Title: An Introduction to Nonintrusive Model Reduction via Operator Inference
    Abstract: This talk is a gentle introduction to projection-based model reduction and a technique for learning reduced-order models from data. We compare and contrast the procedure to generic finite-element methods and show numerical results for a combustion application.
    Bio: Shane McQuarrie is a third-year CSEM PhD student. He works with Dr. Karen Willcox on parametric reduced-order modeling for a variety of applications, including combustion, additive manufacturing, and plasma physics.

    Speaker 2: Amelia Henriksen
    Title: Overview of "A Comparative Evaluation of Unsupervised Anomaly Detection Algorithms for Multivariate Data"
    Abstract: This paper explains foundational unsupervised algorithms for anomaly detection, the problems they solve, and their development over time. We analyze the pros and cons of the paper's empirical study, which offers insight into testing and verification for this critical area of research.
    Bio: Amelia Henriksen is a fifth-year PhD candidate advised by Dr. Rachel Ward. Her research is in theory and algorithm development for dimensionality reduction in the online setting.

    Note: The Oden Institute Student Forum Series provides CSEM Phd students the opportunity to present their work.

  • Multimedia

     Event Stream Link: Click Here to Watch


Thursday, Oct 1

Decision-making Across Scales: From Supply Chains to Materials Nanostructure

Thursday, Oct 1, 2020 from 3:30PM to 5PM | Zoom Meeting

  • Additional Information

    Hosted by Michael Baldea

    Sponsor: Oden Institute Virtual Seminar

    Speaker: Chrysanthos Gounaris

    Speaker Affiliation: Associate Professor, Department of Chemical Engineering and Center for Advanced Process Decision-making, Carnegie Mellon University

  • Abstract

    Modern chemical engineering contemplates topics across a wide span of scales, ranging from the need to understand and harness the chemistry that governs the performance of advanced compounds and materials, to designing industrial equipment and facilities of all kinds, to managing operations and defining corporate strategy at the enterprise level.

    In this talk, we discuss our work in employing mathematical optimization approaches to tackle decision-making in the context of multiple such scales. We start with some settings arising in the supply chain of the chemical industry, for which we develop custom-built mathematical optimization models and solution algorithms to design optimal plans for daily logistics operations. Turning our focus to project scheduling, we show how the management of a pharmaceutical company can optimally allocate R&D resources towards progressing their portfolio of drugs under development.

    We continue by discussing the design of process flowsheets, and present novel methods to ensure robustness of optimal process designs against uncertainties in the underlying physicochemical properties at play. Such methods have been incorporated in our tool PyROS, a Python-based implementation for robust optimization of highly nonlinear models. We conclude the talk by presenting MatOpt, our recently developed crystalline materials framework, which efficiently explores the combinatorics of how atoms may arrange themselves on lattices and identifies the specific microstructure that induces desirable properties in various materials related to energy applications.

    Bio
    Chrysanthos Gounaris is currently Associate Professor of Chemical Engineering at Carnegie Mellon University. He received a Dipl. in Chemical Engineering and an M.Sc. in Automation Systems from the National Technical University of Athens, as well as a Ph.D. in Chemical Engineering from Princeton University. After graduation, Chrysanthos worked as an Associate at McKinsey & Co. He returned to academia to pursue post-doctoral research at Princeton, before joining the Department of Chemical Engineering at Carnegie Mellon University in 2013. His research interests lie in the development of theory and quantitative methodologies for decision-making, with emphasis in supply chain optimization and distribution logistics, production planning and scheduling, project management, process design under uncertainty, microporous and nano structured materials design, as well as methods and tools for robust optimization and global optimization. Chrysanthos actively participates in the Center of Advanced Process Decision-making consortium, where he now directs its Enterprise-Wide Optimization special interest group. He serves as principal investigator for a number of academia-industry research collaborations, as well as participates in the leadership team of DOE’s Institute for the Design of Advanced Energy Systems (IDAES). Recent recognitions for Chrysanthos include his being named a “2020 MSDE Emerging Investigator”, his induction in the “2019 I&ECR Class of Influential Researchers”, the Glover-Klingman Prize, the CIT Dean’s Early Career Fellowship, and the Kun Li Award for Teaching Excellence. Chrysanthos has been an active member of the American Institute of Chemical Engineers, having served as Programming Chair for its Computing and Systems Technology Area 10C, while he is currently serving as co-Chair of the upcoming inaugural conference of the new Advanced Manufacturing & Processing Society, AMPc-2021. Chrysanthos is also a member of the Institute for Operations Research and the Management Sciences, being active in its Transportation Science & Logistics society.

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

Tuesday, Sep 29

Integrating Machine Learning and Multiscale Modeling in Biomedical Sciences

Tuesday, Sep 29, 2020 from 3:30PM to 5PM | Zoom Meeting

  • Additional Information

    Hosted by Karen Willcox

    Sponsor: Oden Institute Virtual Seminar

    Speaker: George Em Karniadakis

    Speaker Affiliation: The Charles Pitts Robinson and John Palmer Barstow Professor of Applied Mathematics, Brown University; Research Scientist, at MIT & PNNL

  • Abstract

    Machine learning has emerged as a powerful approach for integrating multimodality/multifidelity data, and for revealing correlations between intertwined phenomena and cascades of scales. However, machine learning alone does not explicitly take into account the fundamental laws of physics and thermodynamics and can result in ill-posed problems or non-physical solutions. Many human diseases are multiscale in nature, e.g., the sickle cell anemia, first characterized as molecular disease by Linus Pauling in 1949. Multiscale modeling is an effective strategy to integrate multiscale/multiphysics data and uncover mechanisms that explain the emergence of function, from the protein level to the organ level. However, multiscale modeling alone may fail to efficiently combine multimodality and multifidelity datasets. We believe that machine learning and multiscale modeling can naturally complement each other to create robust predictive models that integrate the underlying biophysics to manage ill-posed problems and explore massive design spaces. To this end, we will present a new approach to develop a data-driven, learning-based framework for predicting outcomes of biological systems and for discovering hidden biophysics from noisy data. We will introduce a deep learning approach based on neural networks (NNs) and generative adversarial networks (GANs). We will also introduce the DeepOnet that learns functionals and nonlinear operators from functions and corresponding responses for system identification. Unlike other approaches that rely on big data, here we “learn” from small data by exploiting the information provided by the physical conservation laws, reactive transport and thermodynamics, which are used to obtain informative priors or regularize the neural networks. Our multidisciplinary perspective suggests that integrating machine learning and multiscale modeling can lead to creation of medical digital twins, hence, providing new insights into disease mechanisms, help discover new treatments, and inform decision making for the benefit of human health.

    Reference: M Alber, AB Tepole, WR Cannon, S De, S Dura-Bernal, K Garikipati, ..., GE Karniadakis, E. Kuhl, Integrating machine learning and multiscale modeling—perspectives, challenges, and opportunities in the biological, biomedical, and behavioral sciences, Nature Digital Medicine 2 (1), 1-11, 2020.

    Bio
    George Karniadakis (GS h-index 102) is from Crete. He received his S.M. and Ph.D. from Massachusetts Institute of Technology (1984/87). He was appointed Lecturer in the Department of Mechanical Engineering at MIT and subsequently he joined the Center for Turbulence Research at Stanford / Nasa Ames. He joined Princeton University as Assistant Professor in the Department of Mechanical and Aerospace Engineering and as Associate Faculty in the Program of Applied and Computational Mathematics. He was a Visiting Professor at Caltech in 1993 in the Aeronautics Department and joined Brown University as Associate Professor of Applied Mathematics in the Center for Fluid Mechanics in 1994. After becoming a full professor in 1996, he continues to be a Visiting Professor and Senior Lecturer of Ocean/Mechanical Engineering at MIT. He is an AAAS Fellow (2018-), Fellow of the Society for Industrial and Applied Mathematics (SIAM, 2010-), Fellow of the American Physical Society (APS, 2004-), Fellow of the American Society of Mechanical Engineers (ASME, 2003-) and Associate Fellow of the American Institute of Aeronautics and Astronautics (AIAA, 2006-). He received the Alexander von Humboldt award in 2017, the Ralf E Kleinman award from SIAM (2015), the inaugural J. Tinsley Oden Medal (2013), and the CFD award (2007) by the US Association in Computational Mechanics. His h-index is 102 and he has been cited over 51,000 times. https://www.brown.edu/research/projects/crunch/home.

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

     Event Stream Link: Click Here to Watch