1 / 59

Visualization

Physical Phenomena. Mathematical Model. Visualization. Computational Science. Computing. Numerical Method. Software. Randverdiproblem i 1D u’’(x) = f(x), u(0) = u(1) = 0. D x. x=0. x=1. Deler intervallet [0,1] i N+1 like deler med størrelse D x = 1/(N+1).

lel
Télécharger la présentation

Visualization

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Physical Phenomena Mathematical Model Visualization Computational Science Computing Numerical Method Software

  2. Randverdiproblem i 1D u’’(x) = f(x), u(0) = u(1) = 0. Dx x=0 x=1 Deler intervallet [0,1] i N+1 like deler med størrelse Dx = 1/(N+1). Med xi = i Dx, fås numeriske approksimasjon ui≈u(xi). En Taylor-rekke gir at u”(xi) ≈ (ui+1 – 2ui + ui-1)/Dx2 .

  3. Ved å bruke randbetingelse, u0 = uN+1 = 0, fås likningsystemet Dette kan skrives på formen Au = b Merk båndstrukturen på matrisen!

  4. Ustrukturert grid ”High lift configuration” CRAY T3E – 1450 prosessorer, 25 millioner gridceller University of Wyoming (1998)

  5. Værmelding 4 km oppløsning horisontalt 300 x 500 x 38 gridpunkter tidskritt på 1 min Simulerer 60 timer Bestemmer parametre i 20.5 mrd punkter Roar Skålin, IT-Direktør, met.no

  6. Tsunamien – 26 desember 2004, indiske hav AMRCLAW – adaptiv gridforfining Jan Olav Langseth Dave George Randy LeVeque ”Mesh level 1” 111 km x 111 km ”Mesh level 3” 1.7 km x 1.7 km ”Mesh level 4” 25 m x 25 m

  7. Fakta om simuleringen til venstre... • En million CPU-timer • Et tusen prosessorer • 100.000 GB med data... Joe Werne, Colorado Research Associates DivisionNorthWest Research Associates, Inc.

  8. Numerisk løsning av turbulent miksing forårsaket av en KH-instabilitet. (Rødt/gult – viskøs dissipasjon, blå – termisk dissipasjon) - NWRA/CoRA

  9. Blood Flow Simulations in the Circle of Willis Martin Sandve Alnæs Tor Ingebrigtsen Jørgen Isaksen Kent-Andre Mardal Ola Skavhaug

  10. Navier-Stokes equations are solved with the Finite Element Method, using Featflow Grids are created from a parameterization, using custom written software

  11. Knut Andreas Lie, Sintef anvendt matematikk

  12. Knut Andreas Lie, Sintef anvendt matematikkTo-fase flyt; reservoar

  13. Modelling Geometry Process

  14. Geometry • MR-images • Manual segmentation • Smooth approximation • Computational Mesh

  15. Geometry reconstruction from medical images Goal: - generate grids from MRdata - suitable for FEM - feature/organ sensitive Challenges: - in vivo measurements - the heart beats - image quality - segmentation

  16. Two data sets are generated

  17. Two data sets are generated

  18. Two data sets are generated

  19. A typical data set of torso: 512 x 320 x 40 (x,y,z) images. Body surface, left lung and right lung. A typical data set of heart: 256 x 256 x 10 x 35 (x,y,z,t) images. Heart surface, left ventricle and right ventricle

  20. Raw MR data Manually digitized slices Continuous model 3D grid

  21. A model for human ventriculartissue K. H. W. J. ten Tusscher,1 D. Noble,2 P. J. Noble,2 and A. V. Panfilov1,31Department of Theoretical Biology, Utrecht University, 3584 CH Utrecht, The Netherlands; and 2University Laboratory of Physiology, University of Oxford, Oxford OX1 3PT; and 3Division of Mathematics, University of Dundee, Dundee DD1 4HN, United Kingdom

  22. Figuren kommer fra det å løse store lineære likningssystemer Ax = b som kommer fra PDE-er. ”Improved algorithms and libraries have contributed as much to increases in capability as have improvements in hardware.”

  23. Top 500

  24. LINPACK Benchmarks • Solve a dense NxN system of linear equations, Ax=b • 2/3·N3 + 2·N2 floating point operations • Measure performance in Floating point Operations Per Second (FLOPS) • Maximum performance Rmax for problem size Nmax • Nmax varies between systems.

More Related