NUMERICAL METHODS FOR HIGH PERFORMANCE COMPUTING

Second cycle degree in MATHEMATICAL ENGINEERING - INGEGNERIA MATEMATICA (Ord. 2015) (discontinued)

Campus: PADOVA

Language: English

Teaching period: Second Semester

Lecturer: CARLO JANNA

Number of ECTS credits allocated: 6


Syllabus
Prerequisites: Numerical Methods for Differential Equations
Course unit contents: 1. Advanced numerical linear algebra: projection methods for non-symmetric systems (Bi-CG, QMR) and eigenproblems (Power Method, QR Method, Lanczos, DACG);
2. Multigrid;
3. Preconditioning techniques: ILU, approximate inverses, AMG;
4. Parallel numerical analysis: basic concepts, operations and communications, data structures;
5. Parallel programming paradigms: OpenMP and Message Passing Interface standards;
6. Parallel implementations: sparse linear algebra kernels, iterative methods, domain decomposition.