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-Molecular Biophysics Series
Monday, Oct 3, 2016 from 2PM to 3PM
Control of Water by Protein Functional Dynamics in Two Photoreceptors
by Jiali Gao
University of Minnesota/Jilin University,China
UVR8 and the LOV domain are photoreceptor proteins. Up on photo-excitation, the UVR8 dimer dissociates into monomers and the LOV domain undergoes conformation change, both triggering a range of cell functions. Computation studies show that water plays an integral role of the dimer dissociation process of UVR8 and the dark-state recovery of the covalently linked photoproduct in the LOV domain. In this talk, I will describe computational methods that have been developed to study charge transfer processes in the ultrafast photochemical processes, and analyses of molecular dynamics trajectories that reveal the interplay of protein conformational dynamics and access of water in the active center of the proteins.
Hosted by Ron Elber
ICES Seminar-Numerical Analysis Series
Friday, Oct 7, 2016 from 1PM to 2PM
Image Reconstruction Methods for Photoacoustic Computed Tomography in Heterogeneous Media with Application to Experimental Data
by Mark Anastasio
Professor, Washington University in St. Louis
Photoacoustic computed tomography (PACT) is an emerging biomedical imaging modality that has great potential for a wide range of preclinical and clinical imaging applications. It can be viewed as a hybrid imaging modality in the sense that it utilizes an optical contrast mechanism combined with ultrasonic detection principles, thereby combining the advantages of optical and ultrasonic imaging while circumventing their primary limitations. The goal of PACT is to reconstruct the distribution of an object's absorbed optical energy density from measurements of pressure wavefields that are induced via the thermoacoustic effect. This corresponds to an inverse source problem. In this talk, we review our recent advancements in image reconstruction approaches for PACT. Such advancements include physics-based models of the measurement process and associated inversion methods for reconstructing images from limited data sets in acoustically heterogeneous media. Applications of PACT to transcranial brain imaging and breast cancer detection will be presented.
Hosted by Kui Ren
Tuesday, Oct 11, 2016 from 3:30PM to 5PM
Schwarz Iterative Methods: Randomization and Acceleration
by Peter Oswald
University of Bonn
In the talk, I will briefly review the setup of Schwarz iterative methods and then outline the recently obtained convergence estimates for various randomized orderings, including block versions and acceleration.
Joint work with M. Griebel (Bonn) and W. Zhou (Marburg).
Hosted by Ivo Babuska
Thursday, Oct 13, 2016 from 3:30PM to 5PM
A Consistent Bayesian Approach for Stochastic Inverse Problems
by Tim M. Wildey
Computer Science Research Institute, Sandia National Laboratories
Uncertainty is ubiquitous in computational science and engineering. Often, parameters of interest cannot be measured directly and must be inferred from observable data. The mapping between these parameters and the measurable data is often referred to as the forward model and the goal is to use the forward model to gain knowledge about the parameters given the observations on the data. Statistical Bayesian inference is the most common approach for incorporating stochastic data into probabilistic descriptions of the input parameters. This particular approach uses data and an error model to inform posterior distributions of model inputs and model discrepancies. An explicit characterization of the posterior distribution is not necessary since certain sampling methods, such as Markov Chain Monte Carlo, can be used to draw samples from the posterior.
We have recently developed an alternative Bayesian solution to the stochastic inverse problem. We use measure-theoretic principles to prove that this approach produces a posterior probability density that is consistent with the model and the data in the sense that the push-forward of the posterior through the model will match the observed density on the data. Our approach requires approximating the push-forward of the prior through the computational model, which is fundamentally a forward propagation of uncertainty. We employ advanced approaches for forward propagation of uncertainty to reduce the cost of approximating the posterior density. Numerical results are presented to demonstrate the fact that our approach is consistent with the model and the data, and to compare our approach with the statistical Bayesian approach.
Tim Wildey is a Senior Member of the Technical Staff at the Computer Science Research Institute at Sandia National Laboratories in Albuquerque, NM. His research interests are finite element and finite volume methods, a posteriori error analysis and estimation, uncertainty quantification, adjoint methods, multiphysics and multiscale problems, operator splitting and decomposition, computational fluid dynamics, geomechanics, flow and transport in porous media, numerical linear algebra, domain decomposition, multilevel and multiscale preconditioners, and parallel computing.
Hosted by Tan Bui-Thanh
Thursday, Oct 20, 2016 from 3:30PM to 5PM
Uncertainty Quantification in the Prediction of Turbulent Flows
by Gianluca Iaccarino
William R. and Inez Kerr Bell Faculty Scholar Associate Professor, Mechanical Engineering & ICME, Stanford University
Numerical simulations based on Reynolds averaged Navier-Stokes (RANS) models are routinely used in the analysis and design of engineering devices subject to turbulent flows. It is however known that RANS closures are fundamentally limited in their ability to represent turbulent processes--introducing (epistemic) model-form uncertainties in the predictions. These are compounded by variability (aleatory uncertainties) introduced by limited information regarding the operating conditions of the system of interest or, for example, manufacturing tolerances.
In this talk I will describe a numerical technique based on a polynomial chaos approach and a tailored quadrature rule to characterize and rank the aleatory uncertainty sources and a novel, non-probabilistic strategy to characterize the model-form errors. The present analysis enables us to break down the relative importance of epistemic and aleatory uncertainty and to build confidence intervals on the predictions. We consider simulations of various configurations including a turbine guide vane cascade, and assess the effect of combined uncertainties on the prediction of both attached and separated turbulent flows. Ideas and perspectives on design under uncertainty will conclude the talk.
Gianluca Iaccarino is an associate professor in Mechanical Engineering and co-Director of the Institute for Computational Mathematical Engineering at Stanford University. He received his PhD in Italy from the Politecnico di Bari (Italy) and spent several years at NASA Ames research center before joining the faculty at Stanford in 2007. In 2008 Prof. Iaccarino funded the Uncertainty Quantification Lab working on algorithms to assess the impact of tolerances and limited knowledge on the simulation of engineering systems. Since 2009, he has served as the director of the Stanford Thermal and Fluid Sciences Industrial Affiliate Program. In 2010 he received the Presidential Early Career Award for Scientists and Engineers (PECASE) award from the US Department of Energy.
Hosted by Tan Bui-Thanh
Tuesday, Oct 25, 2016 from 3:30PM to 5PM
Isogeometric Analysis of Fluids, Structures, and Coupled Problems: Some Recent Developments and Applications
by Yuri Bazilevs
University of California, San Diego
In this presentation some recent advances in fundamental developments and applications of Isogeometric Analysis (IGA) are presented. In the applications involving solids and structures, the use of a novel IGA thin-shell formulation enables the development of accurate and efficient techniques for damage prediction in composite laminates due to low-velocity impacts. In the applications involving fluids and turbulence, the use of divergence-conforming B-Splines, in combination with an appropriately formulated Variational Multiscale Large-Eddy Simulation (LES) turbulence models, gives an excellent combination of accuracy and computational efficiency for such simulations. In the regime of compressible flow, appropriately stabilized smooth IGA discretizations also deliver very good results. A novel immersed framework for air-blast-structure interaction (ABSI) involving IGA-based compressible flow is presented and verified on a set of challenging examples. The presentation concludes with examples that highlight effective uses of IGA in advanced applications.
Yuri Bazilevs is a Full Professor and Vice Chair in the Structural Engineering (SE) Department, and an Adjunct Full Professor in the Mechanical and Aerospace Engineering (MAE) Department, in the Jacobs School of Engineering at UCSD. Yuri completed his PhD and Postdoc training at UT Austin's Institute for Computational Engineering and Sciences (ICES) under the direction of Prof. Thomas J.R. Hughes. He joined UCSD as an Assistant Professor in July of 2008, was promoted to Associate Professor with tenure in July 2012, and, subsequently, to Full Professor in July 2014. Yuri develops sophisticated computational techniques and tools to build predictive models for a wide range of applications. His work addresses complex problems in the areas of medicine, such as blood flow in the heart and arteries, as well as in medical devices including blood pumps and artificial hearts; renewable energy, such as assessing damage to wind turbines due to extreme conditions in harsh offshore environments; and protecting infrastructure against man-made and natural disasters, such as assessing the response of civil structures like bridges and buildings to terrorist attacks. For his research contributions Yuri received several awards and honors. Most recently, he was included in the 2014, 2015, and 2016 Thomson-Reuters lists of Highly Cited Researchers and World's Most Influential Scientific Minds, both in the Engineering and Computer Science categories. More information about Yuri may be found at http://ristretto.ucsd.edu/~bazily.
Hosted by Leszek Demkowicz
Tuesday, Nov 15, 2016 from 3:30PM to 5PM
Multilevel Monte Carlo methods
by Mike Giles
Mathematical Institute, University of Oxford, UK
Monte Carlo methods are a standard approach for the estimation of the expected value of functions of random input parameters. However, to achieve improved accuracy often requires more expensive sampling (such as a finer timestep discretisation of a stochastic differential equation) in addition to more samples. Multilevel Monte Carlo methods aim to avoid this by combining simulations with different levels of accuracy. In the best cases, the average cost of each sample is independent of the overall target accuracy, leading to very large computational savings. This talk will introduce the key ideas, and survey the progress in the area.
M.B. Giles. 'Multilevel Monte Carlo methods'. Acta Numerica, 24:259-328, 2015.
Hosted by Thaleia Zariphopoulou