High Performance Computing II - summer term 2019
We will focus on a realistic model problem (rotor) in order to introduce the main topics of numerical methods for High Performance Computing.
First exercise class on Wednesday, April 24! Please enroll in moodle for this course!
Questions concerning the exercises/lab or the lecture, please contact Constantin Greif.
Contents
- Architecture of Parallel Computers
- Parallelization, MPI, Cuda
- The Cluster Pacioli
- Parallel Finite Elemente Method
- Parallel Numerical Methods for Linear Systems of Equations
- Domain Decomposition
- Parallel Preconditioning
- Parallel Multigrid Methods
- Symmetric Eigenvalue Problems
- Storage Formats for Sparse Matrices
Information
| Responsible |
| ||||
Type | Lecture, Exercises, Lab (2/2/2), i.e. 8 credits (Master) | ||||
| Lecture | Monday 8:00 - 10:00; Heho18 - 120 |
| Exercises and Lab | Wednesday 16:00 - 20:00; Heho18 - E60
|
| Courses |
|
Exam The exam will be a project at the end of the semester.
Lectures
-
April 29: PDE model problem.
-
May 06: FEM realization in MATLAB.
-
May 13: Sparse Matrix data formats and uniform refinement.
-
May 20: Direct & Iterative Methods.
-
May 27: Multiscale methods 1.
-
June 03: Multiscale methods 2.
-
June 17: Partitioning and coloring methods.
-
June 24: Domain decomposition 1.
-
July 01: Domain decomposition 2.
-
July 08: Parallel solvers.
-
July 15: Domain decomposition 3.
Resources
Literature:
- G. Alefeld, I. Lenhardt, H. Obermaier, Parallele numerische Verfahren, Springer 2002
Online:
- C++ Language Tutorial
- C++ Online References
- stackoverflow.com
- The Homepage of Bjarne Stroustrup
- "Thinking in C++" by Bruce Eckel. Here with Examples as HTML, here as PDF.
Contact
Prof. Dr. Stefan Funken
Helmholtzstr. 20
Room 1.08
Prof. Dr. Karsten Urban
Helmholtzstr. 20
Room 1.12
M. Sc. Constantin Greif
Helmholtzstr. 20
Room 1.28