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, Apr 13

An Optimal Control Framework for Efficient Training of Deep Neural Networks

Friday, Apr 13, 2018 from 1PM to 2PM | POB 6.304

  • Additional Information

    Hosted by Kui Ren

    Sponsor: ICES Seminar-Numerical Analysis Series

    Speaker: Lars Ruthotto

    Speaker Affiliation: Department of Mathematics and Computer Science, Emory University

  • Abstract

    One of the most promising areas in artificial intelligence is deep learning, a form of machine learning that uses neural networks containing many hidden layers. Recent success has led to breakthroughs in applications such as speech and image recognition. However, more theoretical insight is needed to create a rigorous scientific basis for designing and training deep neural networks, increasing their scalability, and providing insight into their reasoning.

    In this talk, we present a new mathematical framework that simplifies designing, training, and analyzing deep neural networks. It is based on the interpretation of deep learning as a dynamic optimal control problem similar to path-planning problems. We will exemplify how this understanding helps design, analyze, and train deep neural networks.

    First, we will focus on ways to ensure the stability of the dynamics in both the continuous and discrete setting and on ways to exploit discretization to obtain adaptive neural networks. Second, we will present new multilevel and multiscale approaches, derived from he continuous formulation. Finally, we will discuss adaptive higher-order discretization methods and illustrate their impact on the optimization problem.


Friday, Apr 13

  • Additional Information

    Hosted by Ivana Escobar and Sheroze Sheriffdeen

    Sponsor: ICES Seminar-Student Forum Series

    Speaker: Gopal R. Yalla

    Speaker Affiliation: ICES, UT Austin

  • Abstract

    We present a novel coarse scale solver for the parareal computation of dynamical systems. The coarse scale solver can be de ned through interpolation or as the output of a neural network, and accounts for slow scale motion in the system. When pairing this coarse solver with a ne scale solver that corrects for fast scale motion through a parareal scheme, we are able to achieve the accuracy of the ne solver at the ef ciency of the coarse solver. Successful tests for smaller but challenging problems are presented, which cover both highly oscillatory solutions and problems with strong forces localized in time. The results suggest signi cant speed up can be gained for multiscale problems when using a parareal scheme with this new coarse solver as opposed to the traditional parareal setup.


Thursday, Apr 12

  • Additional Information

    Hosted by Tan Bui-Thanh

    Sponsor: ICES Seminar

    Speaker: Krishna Garikipati

    Speaker Affiliation: Professor, Mechanical Engineering, and Mathematics; Director, Michigan Institute for Computational Discovery & Engineering, University of Michigan

  • Abstract

    A central question in developmental biology is how size and position are determined. The genetic code carries instructions on how to control these properties in order to regulate the pattern and morphology of structures in the developing organism. Transcription and protein translation mechanisms implement these instructions. However, this cannot happen without some manner of sampling of epigenetic information on the current patterns and morphological forms of structures in the organism. Any rigorous description of space- and time-varying patterns and morphological forms reduces to one among various classes of spatio-temporal partial differential equations. Reaction-transport equations represent one such class. Starting from simple Fickian diffusion, the incorporation of reaction, phase segregation and advection terms can represent many of the patterns seen in the animal and plant kingdoms. Morphological form, requiring the development of three-dimensional structure, also can be represented by these equations of mass transport, albeit to a limited degree. The recognition that physical forces play controlling roles in shaping tissues leads to the conclusion that (nonlinear) elasticity governs the development of morphological form. In this setting, inhomogeneous growth drives the elasticity problem. The combination of reaction-transport equations with those of elasto-growth makes accessible a potentially unlimited spectrum of patterning and morphogenetic phenomena in developmental biology. This perspective talk is a survey of the partial differential equations of mathematical physics that have been proposed to govern patterning and morphogenesis in developmental biology. Several numerical examples will be shown to illustrate these equations and the corresponding physics, with the intention of providing physical insight wherever possible.

    Bio
    Krishna Garikipati obtained his undergraduate degree from IIT Bombay in 1991, and his Masters and PhD in 1992 and 1996, respectively, from Stanford University. He joined the faculty at University of Michigan in 2000, where he has been a professor since 2012. Since 2016, he has been the Director of the Michigan Institute for Computational Discovery & Engineering (MICDE), which is the focus of research, education and outreach in computational science and engineering at University of Michigan. His work in computational science draws upon applications in biophysics, mathematical biology and materials physics. Of particular interest now are patterning and morphogenesis in developmental biology, and mechano-chemically driven phase transformations in materials. Data-driven modeling is a recent methodological interest, also. He has been awarded the Department of Energy Early Career Award for Scientists and Engineers, the Presidential Early Career Award for Scientists and Engineers, and the Alexander von Humboldt Foundation's Research Fellowship.


Monday, Apr 9

Learning biology from the information encoded in sequences

Monday, Apr 9, 2018 from 2PM to 3:30PM | POB 6.304

  • Additional Information

    Hosted by Dmitrii Makarov

    Sponsor: ICES Seminar-Molecular Biophysics Series

    Speaker: Ryan Cheng

    Speaker Affiliation: Rice University

  • Abstract

    Modern biology is characterized by the abundance of large sequence datasets, particularly genomic data, while structural data remains more limited. All of this one-dimensional sequence information can be leveraged to learn about diverse biological phenomena, ranging from the three-dimensional structure of large biomolecules to cell phenotypes.
    For example, one can predict the 3D structure of a protein or complex from coevolutionary information. At the systems level, one can gain insights on the interaction network of bacterial two-component signaling (TCS) by quantifying the determinants of interaction specificity and using this information to connect mutations to organism phenotypes. Similarly, we can once again exploit sequence information to study chromosomal organization. We show how it is possible to combine energy landscape theory with biochemical information to predict genome architecture using the sequences of epigenetic marking patterns that decorate DNA. Finally, we combine aspects of the first two examples to predict the bacterial condensin protein complex, which is one of the molecular determinants of genome architecture.


Friday, Apr 6

CSEM Challenges in Space Situational Awareness and Space Traffic Management

Friday, Apr 6, 2018 from 10AM to 11AM | POB 6.304

Important Update: Note Date Change
  • Additional Information

    Hosted by Federico Fuentes and Sriram Nagaraj

    Sponsor: ICES Seminar-Babuska Forum Series

    Speaker: Moriba K. Jah

    Speaker Affiliation: Department of Aerospace Engineering & Engineering Mechanics, Institute for Computational Engineering and Sciences (ICES)

  • Abstract

    60 years after the launch of the first US Satellite, Explorer 1, the U.S. Department of Defense currently tracks nearly 23000 so-called Resident Space Objects (RSOs) ranging from the size of a softball to a school bus. Of these, only about 1500 are working satellites while the remainder is space debris. Not all sensor detections result in matches to known RSOs. Part of that is due to an absence of a globally shared data lake, and all of the RSOs are modeled as uniform spheres. Moreover, there is a new space race being fought in commercial landscapes given the wealth to be made via space-based platforms, from human-based activity monitoring and global internet to asteroid mining. However, there are no global laws regulating activities in space; we are absent any norms of behavior to help guarantee orbital safety and the long-term sustainability of the space environment for future generations. The Advanced Sciences and Technology Research in Astronautics (ASTRIA) program is led by Dr Moriba Jah, Associate Prof ASE/EM and ICES Affiliate Faculty. Dr Jah will explain these issues and provide methods for ICES/CSEM students to join this exciting and impactful research endeavor.

    Bio: Dr. Moriba Jah is an Associate Professor in the Aerospace Engineering and Engineering Mechanics Department at the University of Texas at Austin, and directs the ASTRIA research program. His research interests are in non-gravitational astrodynamics and advanced/non-linear multi-sensor/object tracking, prediction, and information fusion. His expertise is in space object detection, tracking, identification, and characterization, as well as spacecraft navigation. Prior to being at UT Austin, Dr. Jah was the Director of the University of Arizona’s Space Object Behavioral Sciences with applications to Space Domain Awareness, Space Protection, Space Traffic Monitoring, and Space Debris research to name a few. Preceding that, Dr. Jah was the lead for the Air Force Research Laboratory’s (AFRL) Advanced Sciences and Technology Research Institute for Astronautics (ASTRIA) and a Principal Investigator for Detect/Track/Id/Characterize Program at AFRL’s Space Vehicles Directorate. Before joining AFRL he was a spacecraft navigator for NASA’s Jet Propulsion Laboratory (JPL) in Pasadena, CA, serving on Mars Global Surveyor, Mars Odyssey, Mars Express (joint mission with ESA), Mars Exploration Rovers, Hayabusa (joint mission with JAXA), and the Mars Reconnaissance Orbiter. Dr. Jah has served as a member of the U.S. delegation to the United Nations Committee on the Peaceful Uses of Outer Space (UN-COPUOS), provided formal expert testimony to congress, and is the chair of the NATO SCI-279-TG activity on defining a Common NATO Space Domain Awareness Operating Picture. Dr. Jah founded the American Astronautical Society’s (AAS) Space Surveillance Technical Committee and is the Chair of the AIAA Astrodynamics Technical Committee. He is a member of the Astrodynamics Technical Committee of the International Astronautical Federation (IAF) and a permanent member of the Space Debris Technical Committee of the International Academy of Astronautics (IAA). Dr. Jah is a Fellow of the International Association for the Advancement of Space Safety (IAASS), the AFRL, the AAS and the Royal Astronomical Society (RAS), as well as an AIAA Associate Fellow, IEEE Senior Member, Associate Editor of Elsevier’s Advances in Space Research Journal.


Friday, Apr 6

A generalized MBO diffusion generated method for constrained harmonic maps

Friday, Apr 6, 2018 from 1PM to 2PM | POB 6.304

  • Additional Information

    Hosted by Kui Ren

    Sponsor: ICES Seminar-Numerical Analysis Series

    Speaker: Braxton Osting

    Speaker Affiliation: Department of Mathematics, University of Utah

  • Abstract

    A variety of tasks in inverse problems and data analysis can be formulated as the variational problem of minimizing the Dirichlet energy of a function that takes values in a certain submanifold and possibly satisfies additional constraints. These additional constraints may be used to enforce fidelity to data or other structural constraints arising in the particular problem considered. I'll present a generalization of the Merriman-Bence-Osher (MBO) method for minimizing such a functional. I’ll give examples of how this method can be used for the geometry processing task of generating quadrilateral meshes, finding Dirichlet partitions, and constructing smooth orthogonal matrix valued functions. For this last problem, I'll prove the stability of the method by introducing an appropriate Lyapunov function, generalizing a result of Esedoglu and Otto to matrix-valued functions. I'll also state a convergence result for the method. I’ll conclude with some applications in inverse problems for manifold-valued data. This is joint work with Dong Wang, Ryan Viertel, and Todd Reeb.


Thursday, Apr 5

  • Additional Information

    Hosted by Tan Bui-Thanh

    Sponsor: ICES Seminar

    Speaker: Nicholas Zabaras

    Speaker Affiliation: Center for Informatics and Computational Science, University of Notre Dame

  • Abstract

    We are interested in the development of surrogate models for uncertainty quantification and propagation in problems governed by stochastic PDEs using a deep convolutional encoder-decoder network in a similar fashion to approaches considered in deep learning for image-to-image regression tasks. Since normal neural networks are data intensive and cannot provide predictive uncertainty, we propose a Bayesian approach to convolutional neural nets. A recently introduced variational gradient descent algorithm based on Stein's method is scaled to deep convolutional networks to perform approximate Bayesian inference on millions of uncertain network parameters. This approach achieves state of the art performance in terms of predictive accuracy and uncertainty quantification in comparison to other approaches in Bayesian neural networks as well as techniques that include Gaussian processes and ensemble methods even when the training data size is relatively small. To evaluate the performance of this approach, we consider standard uncertainty quantification benchmark problems including flow in heterogeneous media defined in terms of limited data-driven permeability realizations. The performance of the surrogate model developed is surprisingly very good even though there is no underlying structure shared between the input (permeability) and output (flow/pressure) fields as is often the case in the image-to-image regression models used in computer vision problems. Studies are performed with a stochastic input dimensionality up to $4,225$ where most other uncertainty quantification methods fail. Uncertainty propagation tasks are considered and the predictive output Bayesian statistics are compared to those obtained with Monte Carlo estimates.

    Bio
    Prof. Nicholas Zabaras joined Notre Dame in 2016 as the Viola D. Hank Professor of Computational Science and Engineering after serving as Uncertainty Quantification Chair and founding director of the Warwick Centre for Predictive Modeling (WCPM) at the University of Warwick. He was recently appointed Director of the interdisciplinary Center for Informatics and Computational Science (CICS) that aims to bridge the areas of data-sciences, scientific computing and uncertainty quantification for complex multiscale/multiphysics problems in science and engineering. He is also serving as the Hans Fisher Senior Fellow with the Institute for Advanced Study at the Technical University of Munich where recently was appointed "TUM Ambassador". Prior to this, he served for 23 years as a faculty at all academic ranks at Cornell University where he was the director of the Materials Process Design and Control Laboratory (MPDC). He received his Ph.D. in Theoretical and Applied Mechanics from Cornell, after which he started his academic career at the faculty of the University of Minnesota. Professor Zabaras' research focuses on the integration of computational mathematics, statistics, and scientific computing for the predictive modeling of complex systems. He has been honored with the Wolfson Research Merit Award from the Royal Society, the Michael Tien '72 Excellence in Teaching Prize from Cornell University, and the Presidential Young Investigator Award from the National Science Foundation.


Tuesday, Apr 3

Modeling, simulations an experiments in light filamentation

Tuesday, Apr 3, 2018 from 3:30PM to 5PM | POB 6.304

  • Additional Information

    Hosted by Irene Gamba

    Sponsor: ICES Seminar-Applied Mathematics Series

    Speaker: Alejandro Aceves

    Speaker Affiliation: Southern Methodist University

  • Abstract

    Since the first observation of a nonlinear process in light matter interaction, more powerful lasers have opened new ways to explore nonlinear phenomena, including intense light filament propagation in the atmosphere. Motivated by experiments and applications, in this talk I will present a brief overview of the basic principles leading to nonlinear Schrӧdinger-like equations, evolving into current challenges on modeling and simulations.

    Bio
    He earned his MS in Applied Mathematics at the California Institute of Technology in 1983 and his PhD in Applied Mathematics, University of Arizona in 1988. Between 1989 and 2008, he moved through the ranks from Assistant to Full Professor of Mathematics at the University of New Mexico, where he held the position of Chair of the Department of Mathematics and Statistics between 2004 and 2008. He is currently Professor and Department Chair of Mathematics at Southern Methodist University. He has had visiting positions at Brown University, Universita di Brescia Italy, University of Limoges and University of Rouen, France and Deusto Tech, Bilbao. He has been a visiting scientist at the Los Alamos National Laboratory and the US Air Force Laboratory. His main research area has been in modeling in Nonlinear Optics and Photonics. In 2016, he was elected Fellow of the Optical Society of America.


Monday, Apr 2

Spatially compatible meshfree discretization

Monday, Apr 2, 2018 from 11AM to 12PM | POB 6.304

  • Additional Information

    Hosted by Kui Ren

    Sponsor: ICES Seminar-Numerical Analysis Series

    Speaker: Nat Trask

    Speaker Affiliation: Sandia National Laboratories

  • Abstract

    While meshfree methods have long promised a natural means of handling problems with large deformation with little numerical dissipation, they have struggled to obtain properties that are often taken for granted in mesh-based methods. From the perspective of compatible discretization, the lack of a mesh means that there is no chain complex upon which to develop a discrete exterior calculus. For this reason, particle methods that are simultaneously able to achieve high-order accuracy and discrete conservation principles have remained elusive. In this talk, we will present recent work from the Compadre (compatible particle discretization) project where we seek to develop meshfree discretizations that achieve these properties. In the first part of the talk, we present a computationally efficient meshfree Gauss divergence theorem which assigns virtual notions of volume and area to particles. With a consistent summation-by-parts theorem in hand, we then develop a meshfree analogue to the finite volume method and demonstrate its robustness when considering Darcy flows with jumps in material properties. In the second part of the talk, we present a meshfree approach to remedy well-known issues with numerical discretizations of peridynamics. While the non-local continuum theory of peridynamics provides an attractive framework for studying fracture with reduced regularity restrictions, particle discretizations of peridynamics fail to obtain a notion of asymptotic compatibility in which the discrete non-local solution recovers the exact local solution as the non-local interaction is reduced. We present a new optimization-based strong form method constructed to enforce reproduction of a given class of nonlocal operators, for which we prove asymptotic compatibility and demonstrate its implementation in a standard engineering workflow.

    Bio
    Nat Trask is a senior member of technical staff at Sandia National Laboratories in the Center for Computation and Visualization, where he previously held the National Science Foundation MSPRF postdoctoral position working with Pavel Bochev. He completed his PhD in the division of applied mathematics at Brown University in 2015, where he worked with Martin Maxey and George Karniadakis. Prior to that, he obtained a masters and bachelors degree from the department of mechanical engineering at the University of Massachusetts Amherst, where he studied fuel injection and combustion.


Friday, Mar 30

An Efficient Sequential Optimal Transport method for Bayesian inverse problems

Friday, Mar 30, 2018 from 10AM to 11AM | POB 6.304

  • Additional Information

    Hosted by Ivana Escobar and Sheroze Sheriffdeen

    Sponsor: ICES Seminar-Student Forum Series

    Speaker: Aaron Myers

    Speaker Affiliation: CSEM, ICES, UT Austin

  • Abstract

    We present the Sequential Ensemble Transform (SET) method for generating approximate samples from a posterior distribution as a solution to Bayesian inverse problems. The method explores the posterior by solving a sequence of discrete, linear optimal transport problems, resulting in a series of transport maps which map prior samples to posterior samples. This allows us to efficiently characterize statistical properties of quantities of interest, quantify uncertainty, and compute moments. We present theory proving that the sequence of Dirac mixture distributions generated by the SET method converges to the true posterior. Numerically, we show this method can offer superior computational efficiency when compared to resampling-based Sequential Monte Carlo (SMC) methods in the regime of low mutation steps and small ensemble size; the regime where particle degeneracy is likely to occur.