sysnetwebmailadmin

Upcoming Seminars

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.

 

ICES Seminar-Babuska Forum Series
Friday, Feb 24, 2017 from 10AM to 11AM
POB 6.304

Mixed regression: Fast Algorithms and Optimal Rates
by Constantine Caramanis

Department of Electrical & Computer Engineering at The University of Texas at Austin

Details

Mixture models represent the superposition of statistical processes, and are natural and powerful modeling tools in machine learning and statistics. We consider the basic problem of mixed regression. In the worst-case, the problem is hard. But despite the prevalence of mixture models in practice, even in the average case, little is known in the realm of efficient algorithms with strong statistical guarantees.

We provide a convex formulation for this problem, and derive matching minimax lower bounds. An interesting phenomenon our results show, is that the minimax rates of convergence are different at low and high SNRs -- something that does not happen in standard regression. On the algorithmic side, we provide conditions that guarantee convergence of fast EM-like algorithms in low and high dimensions, and give the first algorithm that is polynomial in the number of components, with sample complexity scaling linearly with dimension.

Speaker Bio: Dr. Constantine Caramanis is an Associate Professor and holds the Dawson Fellowship in the Department of Electrical & Computer Engineering at The University of Texas at Austin. He received his PhD in EECS from The Massachusetts Institute of Technology, in the Laboratory for Information and Decision Systems (LIDS), and his AB in Mathematics from Harvard University. His research interests center on decision-making in large-scale complex systems, with a focus on learning and computation. Specifically, he is interested in robust and adaptable optimization, and high dimensional statistics.

Hosted by Amir Gholaminejad





 

ICES Seminar-Molecular Biophysics Series
Wednesday, Mar 1, 2017 from 2PM to 3PM
POB 6.304

Event Update Notification: Note that seminar is on Wednesday (not the usual Monday)

Bridging the gap: Single-molecule studies of DNA double strand break repair
by Joseph J. Loparo

Harvard University

Details

NHEJ is the primary repair pathway for double-strand breaks in human cells, yet we lack a detailed understanding of how broken DNA ends are brought together, enzymatically processed, and joined. To better understand these questions, we have developed single-molecule assays to visualize NHEJ in Xenopus egg extracts. Egg extract is a powerful system to study NHEJ because it contains the complete soluble proteome and robustly carries out efficient repair in a manner that depends on all of the core NHEJ factors. Using single-molecule colocalization and FRET assays to watch the association of DNA ends in real time, we recently showed that the NHEJ machinery passes through at least two distinct synaptic states (Graham et al. Mol Cell 2016). DNA ends are initially tethered >50 Å apart in a relatively unstable “long-range” complex that depends on the NHEJ factors Ku and DNA-PKcs. DNA-PKcs kinase activity and the NHEJ factors responsible for ligation (XLF, XRCC4, and LIG4) are then required to transition to a stable “short-range” complex in which the ends are brought into close alignment prior to ligation. In this talk I will describe how we are extending these studies to better understand how the NHEJ machinery assembles on DNA ends and how DNA end processing is regulated to maximize the fidelity of repair.

Hosted by Ron Elber





 

ICES Seminar
Tuesday, Mar 28, 2017 from 3:30PM to 5PM
POB 6.304

Analysis of the DG method applied to an elliptic problem with nonlinear Newton boundary conditions
by Miloslav Feistauer

Professor, Charles University, Faculty of Mathematics and Physics Prague, Czech Republic

Details

The lecture will be devoted to the analysis of the discontinuous Galerkin method (DGM) for the numerical solution of an elliptic boundary value problem with a nonlinear Newton boundary condition in a two-dimensional polygonal domain. Problems of this type appear in the modeling of electrolysis of aluminium, radiation heat transfer problems or in nonlinear elasticity. Using the monotone operator theory, it is possible to prove the existence and uniqueness of the exact weak solution and the approximate DG solution. The main attention is paid to the study of error estimates. To this end, the regularity of the weak solution is investigated and it is shown that due to the singular boundary points, the solution looses regularity in a vicinity of these points. It comes out that the error estimation depends essentially on the opening angle of the corner points and the nonlinearity in the boundary term. It also depends on the parameter defining the nonlinear behaviour of the Newton boundary condition. Important tools in the analysis are the results of A. Buffa and C. Ortner on compact embeddings of broken Sobolev spaces. Theoretical results are accompanied by numerical experiments showing nonstandard behaviour of the error in the L2-norm and the H1-seminorm.

The results were obtained in cooperation with Filip Roskovec and Anna-Margarete Sandig.

Bio
Prof. Dr. Miloslav Feistauer, DrSc., Dr.h.c., professor of mathematics at Charles University in Prague, was born in Náchod into a family of teachers. From an early age he was interested in mathematics and physics, and also devoted himself to music. After eleven years at high school in 1960, wondering whether to study violin or mathematics, he decided to study at the Faculty of Mathematics and Physics, Charles University. There he has spent all his professional life. After graduating in Applied Mathematics in 1965, he joined the Department of Applied Mathematics. After three years he was appointed lecturer and in the following year received the degree of Doctor of Natural Sciences (RNDr.). In 1972 he defended the scientific degree of Candidate of Sciences (Ph.D.) and in 1982 took a habilitation in mathematics. However, since Prof. Feistauer had From Pokroky Mat. Fyz. Astronom. 58 (2013), 73–75. Translated by Miluša Bubeníková 489 never been politically active, his appointment to the position of associate professor was held up till 1988. In 1990 he was awarded the degree of Doctor of Sciences (DrSc.) and shortly thereafter in 1991 was appointed full professor of mathematics in the field of approximate and numerical methods. In this position he has worked up to now. In the period of 1986–1994 he worked at the Mathematical Institute of Charles University and in 1994 he gained the post of the head of the Department of Numerical Mathematics at Charles University in open competition. In this position he served till 2006. Professionally, Prof. Feistauer's scientific activity has been in the areas of partial differential equations, numerical and computational mathematics and mathematical methods in fluid dynamics. More than 200 papers, 3 monographs, more than 130 invited lectures at universities, more than 180 lectures at conferences (including 68 invited main or plenary lectures and 15 lectures in Oberwolfach), organizing conferences in the Czech Republic and abroad (e.g., member of the Program Committee of the series of the conferences ENUMATH). Teaching activities of Prof. Feistauer are in accord with his successful research activities. In addition to courses on numerical mathematics he presents lectures on mathematical methods in fluid mechanics and mathematical modelling and supervises seminars on continuum mechanics and numerical mathematics. He has significantly contributed to the development of numerical analysis and mathematical modelling at the Faculty of Mathematics and Physics. His lectures are highly evaluated by students. He has trained a number of graduate and doctoral students who were awarded in various national and international competitions. Many of his former students became prominent experts, associate professors and full professors at universities in the Czech Republic and abroad where they continue his work and fulfil his aims and objectives. In the years 1994–2012 Prof. Feistauer was member of the Scientific Council of the Faculty of Mathematics and Physics. He has worked in various committees and bodies of the faculty, but also beyond. He is member of the committee for doctoral studies F11 and M6 at Charles University, he was member of the Scientific Council of the Faculty of Mechanical Engineering, currently he is member of the Scientific Council of the Faculty of Chemical Engineering at the Institute of Chemistry and Technology in Prague and for many years he worked in an advisory commission of the Grant Agency of the Czech Republic. He is member of numerous scientific
societies (GAMM, ISIM, AMS, EUROMECH, ECMI and others) and member of the editorial boards of five international journals. His research and teaching activities were awarded by the medal of the Faculty of Mathematics and Physics and by the silver medal of Charles University. In 2004 he was elected member of the Learned Society of the Czech Republic. In 2006, Prof. Feistauer received another prestigious award, the title of Honorary Doctor of the Technical University of Dresden. Among his more recent awards, Feistauer received the 2008 Memorial Medal of the Charles University and the 2013 Memorial Medal of the Czech Mathematical Society. His nonprofessional interests include the history of art, music: painting, playing violin, viola, and music composition.

Hosted by Ivo Babuska





 

ICES Seminar - Computational Medicine Series
Tuesday, Mar 28, 2017 from 10AM to 11AM
POB 6.304

Abdominal Aortic Aneurysm Risk Assessment: Biomechanics or Geometry-based Criterion?
by Ender Finol

Associate Professor, The University of Texas at San Antonio

Details

The rupture of an abdominal aortic aneurysm (AAA) is believed to represent the culmination of a complex vascular mechanism partially driven by the forces exerted on the arterial wall. To prevent rupture, a diagnosed AAA is differentiated by its suitability for surgical or endovascular repair based on its maximum diameter or expansion rate measured over time during patient follow-up. At nearly all major hospitals in the U.S., subjects with aneurysms smaller than 5.5 cm (on average) are placed under clinical surveillance while those greater or equal than 5.5 cm or growing at a rate 3 0.5 cm/year are recommended for elective repair. Since AAA is a largely asymptomatic condition, the screening necessary to measure growth rate over time often cannot be completed. It is a known fact, however, that basing the clinical management of this disease on expansion rate and/or maximum diameter is not a reliable measure of individual rupture risk. This is evident by the number of small aneu
rysms that rupture prior to reaching the critical diameter of 5.5 cm and the many more that are diagnosed at an advanced stage of expansion having exceeded the threshold size for intervention and yet did not rupture. This inability to predict accurately the individual at-risk status of an AAA has led to extensive research into other potential indicators of rupture or equivalent evaluation criteria for assessing the need for repair. Our laboratory is using a combination of image processing and numerical methods to model the geometry and the biomechanical environment of patient-specific AAAs. In this talk, I will describe our efforts in accurately modeling these aneurysms at the organ and tissue scales and the useful quantitative information we can predict from clinical images. This research represents a contribution to the ultimate goal of developing a computational tool that can be used in a clinical setting to assess the individual aneurysm rupture potential on the same day
of AAA diagnosis.

Bio:
Dr. Finol received his B.E. degree in Mechanical Engineering from Universidad de Carabobo, Venezuela (1994), M.S. degree in Mechanical Engineering from University of Massachusetts Lowell (1997) and Ph.D. degree with a dual major in Mechanical and Biomedical Engineering (2002) from Carnegie Mellon University. He is currently an Associate Professor in the Department of Mechanical Engineering at University of Texas at San Antonio and previously a research faculty at Carnegie Mellon Universitys Institute for Complex Engineered Systems (ICES). Dr. Finol operates the Vascular Biomechanics and Biofluids Laboratory and has research interests that include non-destructive tissue mechanics, fluid and solid mechanics modeling of blood vessels, design and optimization of intravascular medical devices, and medical image analysis.

Hosted by Michael Sacks





 

ICES Seminar-Student Forum Series
Friday, Mar 31, 2017 from 10AM to 11AM
POB 6.304

A Stochastic Operator Approach to Model Inadequacy with Applications to Contaminant Transport
by Teresa Portone

ICES, UT Austin

Details

We present recent developments on the uncertainty quantification of models. Models are imperfect representations of complex physical processes, hence exploring representations of the model inadequacies is crucial. We introduce an inadequacy model in the form of a linear operator acting on the model solution and explore methods for incorporating knowledge of model shortcomings and relevant physics. This representation is developed in the context of scalar dispersion in porous media, but the methods presented are applicable for other models.

Hosted by Gopal Yalla





 

ICES Seminar
Tuesday, Apr 11, 2017 from 3:30PM to 5PM
POB 6.304

TBA
by Ioannis Kevrekidis

TBA

Details

TBA

Hosted by Greg Rodin