General descriptionThe course will go through numerical algorithms for solving problems in linear algebra, including linear systems of equations, eigenvalue problems, singular value decomposition, and least squares problems. Beside classical topics (e.g. GE, LU, Cholesky, Jacobi, Gauss-Seidel, CG, GMRES, domain decomposition, multigrid, preconditioning, power method, QR, Arnoldi, Lanczos, FFT, etc...), we will also discuss more recent techniques (such as Hierarchical matrices and randomized low rank approximations). The focus will be on understanding stability, errors, complexity, and implementation issues for the algorithms. We will be using Matlab throughout the course.Syllabus (updated in April)
InstructorMohammad Motamed
Instructor's office hours1) Tu 14.00-15.30     2) Th 14.00-15.30     3) by appointment
Required literature1) D: James Demmel. Applied Numerical Linear Algebra. SIAM. 1997.     (Also note errata page.) 2) NMT: N. Halko, P.-G. Martinsson and J. A. Tropp. Finding Structure with Randomness: Probabilistic Algorithms for Constructing Approximate Matrix Decompositions, SIAM Review, 57(2):217-288, 2011. 3) BGH: S. Borm, L. Grasedyck and W. Hackbusch. Hierarchical Matrices, Lecture Notes 21, MPI Leipzig, 2003.
Lecture notesIntroduction:   notesLinear Systems (direct methods):   notes Linear Systems (iterative methods):  notes on classical methods,   notes on Krylov methods and multigrid Eigenproblems:   notes 1   notes 2 Singular Value Decomposition:   notes Least Squares Problems:   notes Low-rank approximation by Hierarchical Metrices:   notes
HomeworkHomework Report Format HW1     (due extended to Feb. 19th) Files:   eiffel1.mat   eiffel2.mat   eiffel3.mat   eiffel4.mat   trussplot.m HW2     (due extended to April 18th) Files:   cooling_flange.mat   convdiff.mat   lap.m HW3     (Email your report on May 9th before 23:59) Files:   illposed.mat   hmatrix.mat
NEWS motamed@math.unm.edu |