Seminar Numerische Mathematik / Numerical mathematics seminars
aktuelles Programm / current program

Donnerstag, 01. 11. 2007, 14:00 Uhr (ESH)

Prof. J. Stockie   (Simon Fraser University, Department of Mathematics, Canada)
Multiscale modelling of the fuel cell catalyst layer


Donnerstag, 25. 10. 2007, 14:00 Uhr (ESH)

Dr. O. Schenk   (University of Basel, Department of Computer Science)
General-purpose sparse matrix building blocks on Graphics Processing Units

Abstract:
We report on our experience with integrating and using graphics processing units (GPUs) as fast parallel floating-point co-processors to accelerate two fundamental computational scientific kernels on the GPU: sparse direct factorization and nonlinear interior-point optimization. Since a full re-implementation of these complex kernels is typically not feasible, we identify e.g. the matrix-matrix multiplication as a first natural entry-point for a minimally invasive integration of GPUs. We investigate the performance on the NVIDIA GeForce 8800 multicore chip. We exploit the architectural features of the GeForce 8800 GPU to design an efficient GPU-parallel sparse matrix solver. A prototype approach to leverage the bandwidth and computing power of GPUs for these matrix kernel operation is demonstrated resulting in an overall performance of over 110 GFlops/s on the desktop for large matrices. We use our GPU algorithm for PDE-constrained optimization problems and demonstrate that the commodity GPU is a useful co-processor for scientific applications.


Donnerstag, 20. 9. 2007, 14:00 Uhr (ESH)

Prof. Dr. K. Krischer   (Technische Unitersität München, Physik-Department E19)
Dynamic instabilities in fuel cell relevant reactions

Abstract:
Dynamic instabilities, such as bistability, oscillations or traveling wave patterns, commonly occur at the electrode|electrolyte interface. Many of the systems are of great practical or technological importance. For example, spiral waves emerge during galvanic electrodeposition of alloys, or regular arrays of nanopores arise during the electropolishing of Si surfaces. Also oscillating cell voltages are often observed in fuel cells. In this talk I will discuss a general equation describing the spatio-temporal evolution of the interfacial potential. Coupled to the evolution equations of the respective reacting chemical species, the resulting model reproduces many of the instabilities observed in electrochemical reactions. In the talk I will focus on instabilities and pattern formation occurring in fuel cells when the feed gas (H2) is contaminated with CO. An outlook will be given how the dynamic instabilities might be exploited to maximize the fuel cell performance in the presence of CO contaminations.

Donnerstag, 20. 9. 2007, 15:00 Uhr (ESH)

Prof. B. Zaltzman   (Ben-Gurion University of the Negev, Jacob Blaustein Institute for Desert Research, Israel)
Electroosmotic flows and electroconvection - from theory to experiment

Abstract:
Electric conduction from an electrolyte solution into a charge selective solid such as ion exchange membrane or electrode becomes unstable when the electrolyte concentration near the interface approaches zero due to diffusion limitation. The sequence of events leading to instability is as follows upon the decrease of the interface concentration the electric double layer at the interface transforms from its common quasi-equilibrium structure to a different non-equilibrium one. The key feature of this new structure is an extended space charge added to the usual one of the quasi-equilibrium electric double layer. The non-equilibrium electroosmotic slip related to this extended space charge renders the quiescent conductance unstable. A unified asymptotic picture of the electric double layer under current encompassing all regimes from quasi-equilibrium to the extreme non-equilibrium one is developed and employed for derivation of a universal electroosmotic slip formula This formula is used for a linear stability study of quiescent electric conduction yielding the precise parameter range of instability compared with that in the full electroconvective formulation. The physic al mechanism of instability is traced both kinematically in terms of non-equilibrium electroosmotic slip and dynamically in terms of forces acting in the electric double layer.


Montag, 12. 2. 2007, 14:00 Uhr (ESH)

Dr. C. Wolters   (Westfälische Wilhelms-Universität Münster, Institut für Biomagnetismus und Biosignalanalyse)
Numerical approaches for finite element method based EEG and MEG source analysis

Abstract:
The inverse problem in Electro- and Magneto-EncephaloGraphy (EEG/MEG) aims at reconstructing the underlying current distribution in the human brain using potential differences and/or magnetic fluxes that are measured non-invasively directly, or at a close distance, from the head surface. Not only do various head tissues exhibit different conductivities, some of them are also anisotropic conductors as, e.g., skull, brain white and also gray matter. In my talk, techniques of multimodal Magnetic Resonance Imaging (MRI) are presented in order to generate high-resolution realistically shaped anisotropic volume conductor models. The Finite Element (FE) method in combination with an EEG and MEG lead field basis approach and a parallel algebraic multigrid solver yields a highly efficient solution of the anisotropic forward problem. Different FE approaches for the modeling of the dipole singularity and their accuracy in multilayer-sphere models will be discussed. The influence of anisotropy in realistic head models will then be presented. It was found that for EEG, the presence of tissue anisotropy both for the skull and white matter compartment substantially compromises the forward potential computation and therefore the inverse source reconstruction. For the MEG, only the anisotropy of the white matter compartment has an effect. The deeper the source and the more it is surrounded by anisotropic fiber bundles, the larger the effect is. In summary, high-resolution anisotropic FE forward modeling is crucial for an accurate solution of the inverse problem in EEG and MEG.