Numerical methods (NM) - KAIN6M05

  • Number of hours

    • Lectures 13.5
    • Projects -
    • Tutorials -
    • Internship -
    • Laboratory works 15.0
    • Written tests 2.5

    ECTS

    ECTS 0.5

Goal(s)

The course presents numerical algorithms, their implementations on modern processors (multicore). We will focus mainly on classical algorithms of linear algebra. The performances of these numerical algorithms will also be studied, in terms of floating operations per second and use of the memory hierarchy.

Content(s)

  • Computations on polynomials
  • Matrix and vector computations (sparse and dense structures)
  • BLAS Library (Basic Linear Algebra Subprograms)
  • Performance evaluation (Floating-point Operation per Second)
  • Parallelizing of numerical algorithms (OpenMP directives)
  • Vector computation units, vectorization (MMX, SSE, AVX)

3 TPs are realized:

  • TP1 on polynomial computations
  • TP2 on a library of linear algebra (a subset of BLAS Library)
  • TP3 on OpenMP parallelization and vectorization of the library
    of linear algebra.

Prerequisites

L2 level in Mathematics (analysis and algebra), matrix calculus seen in the core curriculum of Semester 5

Test

  • 40% CC
  • 60% EXAM
    • in person
    • written test on paper
    • 1h30
    • adaptation to disability: extended test time
    • no document allowed
    • electronic devices: not permitted

Calendar

The course exists in the following branches:

  • Curriculum - INFO - Semester 6

Additional Information

Course ID : KAIN6M05
Course language(s): FR

You can find this course among all other courses.

Bibliography

Alfio Quarteroni, Ricardo Sacco, Fausto Saleri, Méthodes Numériques: Algorithmes, analyse et applications, Springer, 2007