MS-E1651: Numerical Matrix Computations

Course name Numerical Matrix Computations
Course code MS-E1651
Abbreviation “NMC”
Period I
Lecturer Antti Hannukainen

Description

The core subjects of the course are factorization of matrices numerically and error bound analysis. The course serves as an introduction to dealing with matrices numerically.

Course material

Official material

Lecture notes as a PDF and the book “Trefethen, Bau: Numerical Linear Algebra”.

Extra material

Contents and workload

Workload feels similar to Bachelor courses.

Overall workload

Weekly contents

Week Topics
1 Introduction, Permutation matrices, Block matrix notation, Gaussian elimination, Back-substitution, LU-factorisation
2 Cholesky factorisation, fill-in, fill-in reducing permutation, minimum degree ordering
3 Numerical stability analysis, Backward error analysis, floating-point representation, floating-point arithmetic model, round-off error
4 Numerical stability analysis, Backward error analysis, Back-substitution, Cholesky factorisation, QR-factorisation, Givens Rotation
5 Iterative solution method, Fixed-point iteration, Convergence, Conjugate Gradient method, Line search method, Gradient Descend
6 Iterative solution method, Conjugate Gradient method, orthogonal projection, error estimate, Krylov subspace

Practicalities

The course can be completed by only participating in the exam or by a combination of an exam, homework submissions and participation in the exercise sessions.

Official prerequisites

Recommended prerequisites: MS-A00XX, MS-A01XX, MS-A02XX, MS-C134X. The courses MS-A03XX and MS-C1540 may also be useful.

Additional prerequisites

Sufficient Matlab knowledge.

More like this

MS-E1653: Finite Element Method, MS-E1654: Computational Inverse Problems.

Trivia