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 - ICES Student Forum Series
Friday, Mar 6, 2015 from 11AM to 12PM
Representing model inadequacy: A stochastic operator approach
by Rebecca Morrison
"The ICES Student Forums are presentations given by current students in the CSEM program to their peers. The aim of the forums is to expose students to each other's research and encourage collaboration. Seminar credit will be given to first and second year CSEM students."
ICES, The University of Texas at Austin
We investigate model inadequacy of a reaction mechanism model for hydrogen combustion. In a typical reaction, the complete mechanism is either not well-understood, or too complex to effectively use as part of a larger combustion problem, necessitating a reduced model. To account for the discrepancy between the full model and its reduced version, we propose an additive, linear, stochastic operator. This representation is encoded in a random matrix, whose entries are calibrated using a hierarchical Bayesian scheme. In particular, this formulation is designed to respect certain physical constraints, but also be flexible enough to apply to multiple reactions.
ICES Seminar - CCS Series
Tuesday, Mar 10, 2015 from 1:30PM to 2:30PM
Computational Modeling of Cardiac Tissues
by Frank Sachse
Bioengineering Department and a faculty investigator, Nora Eccles Cardiovascular Research and Training Institute
Computational models play an important role in studies of cardiac tissue physiology and pathophysiology. Several types of models have been developed applying knowledge from histological and physiological studies. For instance, the prevalent monodomain and bidomain models describe cardiac electrical conduction based on studies of cellular electrophysiology and intercellular electrical coupling. An assumption underlying most models of cardiac conduction is that muscle cells (myocytes) are the predominant cell type in cardiac tissue. In contrast our multidomain model represents cardiac tissue as a mixture of various cell types. We developed approaches for deriving parameters for multidomain models from three-dimensional reconstructions of cardiac tissue at sub-micrometer resolution. We created the reconstructions using image data from fluorescent labeling and scanning confocal microscopy. In this talk I will give an overview of our approaches for multidomain modeling and microscopy-based derivation of model parameters. I will present findings from our studies of cellular remodeling in cardiac infarction and fibrosis. I will discuss applications of the developed approaches in computational studies of tissue remodeling in disease and restoration after therapy.
Frank B. Sachse received a diploma in computer science in 1992 and a PhD in electrical engineering in 1997 from the Universität Karlsruhe (TH), Germany. In 2002, he habilitated in biomedical engineering with the thesis “Mathematical Modeling of the Mammalian Heart”. Since 2003, he works at the University of Utah, Salt Lake City, USA. Currently, he is an Associate Professor in the Bioengineering Department and a faculty investigator at the Nora Eccles Cardiovascular Research and Training Institute. His research group applies microscopic imaging and computational approaches to gain insights into structure and function of cardiac tissues, cells and proteins. Areas of research include cellular remodeling in heart failure, computational modeling, and confocal microscopy. Dr. Sachse taught classes on biomedical engineering, modeling of the cardiovascular system, and biomedical imaging at universities in Germany, Finland, Spain and the USA. He received several awards for his work, for instance, from the Federal Ministry of Education and Research, Germany, and a visiting fellowship at the Isaac Newton Institute For Mathematical Sciences, Cambridge, UK.
Hosted by Michael Sacks
Joint ICES/ASE/EM Seminar
Thursday, Mar 12, 2015 from 3:30PM to 5PM
Massively Parallel Simulations of Patient- Specific Hemodynamics
by Amanda Randles
Lawrence Fellow, Livermore National Laboratory
The recognition of the role hemodynamic forces have in the localization and development of disease has motivated large-scale efforts to enable patient-specific simulations. When combined with computational approaches that can extend the models to include physiologically accurate hematocrit levels in large regions of the circulatory system, these image-based models yield insight into the underlying mechanisms driving disease progression and inform surgical planning or the design of next generation drug delivery systems. Building a detailed, realistic model of human blood flow, however, is a formidable mathematical and computational challenge. The models must incorporate the motion of fluid, intricate geometry of the blood vessels, continual pulse-driven changes in flow and pressure, and the behavior of suspended bodies such as red blood cells. In this talk, I will discuss the development of HARVEY, a parallel fluid dynamics application designed to model hemodynamics in patient-specific geometries. I will cover the methods introduced to reduce the overall time-to solution and enable near-linear strong scaling on up to 1,572,864 core of the IBM Blue Gene/Q supercomputer. Finally, I will present the expansion of the scope of projects to address not only vascular diseases, but also treatment planning and the movement of circulating tumor cells in the bloodstream.
Thursday, Mar 19, 2015 from 3:30PM to 5PM
Evaluation of Fixed-Point and Quasi-Newton Approaches for Parallel Multiphysics Coupling
by Miriam Mehl
Professor, University of Stuttgart Germany, Institute for Parallel and Distributed Systems
With increasing compute power and higher efficiency of numerical algorithms, very accurate models for technical or natural systems consisting of several components describing different physical, chemical, biological, processes have become feasible for simulation purposes. If we aim at being flexible in terms of using existing solvers for single phenomena, exchanging or enhancing methods, implementations, and the involved fields, a partitioned simulation approach is the optimal choice. In order to achieve a numerical accuracy compatible to the model accuracy, we have to efficiently implement the resulting simulation environment consisting of several independent software components (solvers) on massively parallel computer architectures. Besides, of course, the mathematical complexity and robustness of iterative solvers for the overall coupled system is an issue. In the presentation, various possibilities to formulate the coupling in terms of fixed-point equations more or less suitable for parallel computing and to solve the fixed-point equations using different variants of quasi-Newton accelerated fixed-point iterations are presented and compared. We show that good combinations of these two ingredients yield efficient, scalable and robust partitioned coupling methods that can even be extended to a essententially arbitrarily many fields which is not the case for many standard state-of-the-art approaches.
After receiving her diploma in mathematics from TUM in 1997 ('summa cum laude'), Miriam Mehl completed her doctorate ('summa cum laude') in the interdisciplinary area of biological waste water treatment in 2001. She finished her habilitation in the TUM Department of Computer Science in summer 2010, and is involved in several projects in the fields of computational fluid dynamics (CFD), simulation of fluid-structure interactions, and high-performance computing. From 2002 onwards, she was the head of the CFD Group at the Department for Scientific Computing in Computer Science at TUM. This group works with a strong focus on numerically and hardware-efficient algorithms for PDE solvers on high-performance computing architectures.
Since 2013, Miriam Mehl has held the position of a professor at the University of Stuttgart’s Institute for Parallel and Distributed Systems. She is head of the Institute for the Simulation of Large Systems.
Hosted by George Biros
Thursday, Mar 26, 2015 from 3:30PM to 5PM
Efficient Methods for 3D- Matching Problems
by Antje Vollrath
Professor, Institute of Computational Mathematics, TU Braunschweig
Matching problems deal with the analysis of 3D objects by means of efficient and fast algorithms chosen correspondingly to the chosen mathematical representation of the objects.
In industrial applications, the main focus on efficiently tackling the matching problem is cheap and fast data acquisition and real time evaluation of the 3D object poses. As an alternative to 3D scanners, one finds various techniques to acquire 2D normal maps of scenes instead of 3D data, like photometric stereo. But despite the comparatively easy single shot acquisition, the data processing is complex and many assumptions are needed for accurate depth estimation. For this reason, normal maps have previously been rarely used for pose estimation of 3D Objects. We propose an approach to use normal maps for pose estimation using fast spherical and rotational nonuniform Fourier transforms to evaluate correlation integrals. We also discuss uniform samplings on the sphere and the rotation group to achieve improved results. Using the estimated orientation, accurate monocular translation estimation techniques are discussed to build a complete 6D pose estimation.
Based on these observations, we will discuss a related application from computational structural biology. Here, detecting local extrema of a correlation can be used to reconcile available protein structure data from different sources like X-ray crystallography or cryo-electron microscopy resulting in a refined protein model that combines the finer resolution information in the former with the native-state information at lower resolution in the latter. The main focus in solving the matching problem here is to efficiently incorporate additional knowledge about the proteins to improve results and to allow flexible motions like shear and hinge bending.
Hosted by Chandrajit Bajaj
ICES Seminar-Molecular Biophysics Series
Monday, Mar 30, 2015 from 2PM to 3PM
Geometric and Topological Modeling of Biomolecules
by Guowei Wei
Michigan State University
A major feature of biological sciences in the 21st Century is their transition from phenomenological and descriptive disciplines to quantitative and predictive ones. However, the emergence of complexity in self-organizing biological systems poses fabulous challenges to their quantitative description because of the excessively high dimensionality. A crucial question is how to reduce the number of degrees of freedom, while preserving the fundamental physics in complex biological systems. We discuss a multiscale and multiphysics paradigm for biomolecular systems. We describe macromolecular systems by a number of approaches, including macroscopic electrostatics and elasticity and/or microscopic molecular mechanics and quantum mechanics; while treating the aqueous environment as a dielectric continuum or electrolytic fluids. We use differential geometry theory of surfaces to couple various microscopic and macroscopic domains on an equal footing. Based on the variational principle, we derive the coupled Poisson-Boltzmann, Nernst-Planck, Kohn-Sham, Laplace-Beltrami, Newton, elasticity and/or Navier-Stokes equations for the structure, function, dynamics and transport of protein, protein-ligand binding and ion-channel systems. Finally, we briefly introduce multiscale and multidimensional topology to simplify biomolecular data.
Hosted by Dmitrii Makarov
Tuesday, Apr 7, 2015 from 3:30PM to 5PM
High-Order Methods for Turbulent Flow Simulations on Deforming Domains
by Per-Olof Persson
University of California at Berkeley
It is widely believed that high-order accurate numerical methods, for example discontinuous Galerkin (DG) methods, will eventually replace the traditional low-order methods in the solution of many problems, including fluid flow, solid dynamics, and wave propagation. In this talk I will present some of the recent developments in our work on efficient and robust DG schemes for real-world problems with deforming domains. Topics include high-quality unstructured curved mesh generation, high-order compact and sparse numerical schemes, artificial viscosity based stabilization of underresolved features such as shocks and turbulence models, scalable preconditioners for parallel iterative solvers, and implicit-explicit schemes for the partitioning of coupled fluid-structure interaction problems. The methods will be demonstrated on important practical problems, including the inverse design of energetically optimal flapping wings and large eddy simulation of vertical axis wind turbines.
Hosted by Tan Bui-Thanh
Tuesday, Apr 14, 2015 from 3:30PM to 5PM
Spiraled Boreholes: An Expression of 3D Directional Instability of Drilling Systems
by Emmanuel Detournay
Department of Civil, Environmental, and Geo-Engineering, University of Minnesota
Occurrence of borehole spiraling is predicted by analyzing the delay-differential equations governing the propagation of a borehole. These evolution equations for the borehole inclination and azimuth are obtained from consideration involving: (i) a bit/rock interaction law that relates the force and moment acting on the bit to its penetration into the rock; (ii) kinematic relationships that describe the local borehole geometry in relation to the bit penetration; and (iii) a beam model for the bottom-hole assembly (BHA) that can be used to express the force and moment at the bit from the external loads applied on the BHA and the geometrical constraints arising from the stabilizers conforming to the borehole geometry. The analytical nature of the propagation equations makes it possible to conduct a systematic stability analysis in terms of a key dimensionless group that controls the directional stability of the drilling system. This group depends on the downhole weight on bit (WOB), on properties of the BHA, on the bluntness of the bit, and on parameters characterizing its response. The directional stability of a particular drilling system can be assessed by comparing the magnitude of this group with a bifurcation value that depends only on the BHA configuration and the bit walk. If this dimensionless group, which depends on the actual drilling conditions, is less than the bifurcation value, the system is directionally unstable, and borehole spiraling is likely. Stability curves for an ideal BHA with two stabilizers are shown to depend on the bit walk, which tends to enhance conditions for spiraling. An application to a field case is discussed. Simulations conducted by integrating the equations of borehole propagation also are presented. They illustrate that, for unstable systems, the model predicts spiraled boreholes with a pitch comparable to what is observed in the field.
Dr Detournay joined in 1993 the Department of Civil Engineering of the University of Minnesota, where he is now the Theodore W Bennett Chair in Mining Engineering and Rock Mechanics. Prior to joining the UMN, he was Senior Research Scientist at Schlumberger Cambridge Research in England. His expertise is in Petroleum Geomechanics, with two current focuses: drilling mechanics (bit-rock interaction, self-excited drilling vibrations, directional drilling) and mechanics of fluid-driven fractures (asymptotic analysis, scaling, numerical modeling).
Hosted by Greg Rodin
Tuesday, Apr 21, 2015 from 3:30PM to 5PM
Equation-Oriented Flowsheet Simulation and Optimization Using Pseudo-Transient Models
by MIchael Baldea
Assistant Professor, Chemical Engineering, UT Austin (and ICES Affiliated Faculty)
In this presentation, we introduce a new process modeling and simulation framework that relies on pseudo-transient continuation to improve the convergence properties of chemical process flow-sheet models. The steady-state models of unit operations are reformulated as differential-algebraic equation (DAE) systems that have the same steady-state solution as the original model. We present a generic model reformulation algorithm and a library of pseudo-transient models for the most widely-used unit operations. We then develop a flow-sheet optimization strategy by seamlessly integrating these models with a time-relaxation optimization algorithm implemented in gPROMS. Several industrial case studies will be discussed to illustrate the theoretical ideas.
Hosted by Robert Moser