|
CPSC 660: Computational
Linear Algebra (Spring 08) |
|
Instructor |
Vivek Sarin, 309C HRBB, 458-2214, sarin@cs.tamu.edu |
|
Lecture |
TR, 12:45pm-2:00pm, ZACH 227A |
|
Office Hours |
TR, 11:30am-12:30pm |
|
Prerequisites |
CPSC 442 or MATH 417 or equivalent |
|
Textbook |
Numerical Linear Algebra by Lloyd N. Trefethen and David Bau, III |
|
Reference |
Matrix Computations by Gene H. Golub
and Charles V. Loan |
DescriptionThis course will focus on algorithms for matrix computations; the following broad topics will be covered: ·
Matrix
fundamentals ·
Solution of
systems of equations ·
Orthogonalization and
Least-Squares problems ·
Eigenvalue
problems ·
Iterative
methods for systems of equations Computational techniques will be accompanied by error and
stability analysis. Homework will consist of theoretical problems as well as
programming exercises using MATLAB. You are expected to learn basic MATLAB
during the early part of the course (this is not at all difficult). Outline·
Fundamentals: Vectors, Norms, Matrices
·
QR Factorization and Least Squares: Gram-Schmidt orthogonalization, Householder reflections and Givens
rotations, QR factorization of a matrix, Least Squares problems
·
Conditioning and Stability: Conditioning of a
problem, Floating point arithmetic, Stability of algorithms
·
Systems of Equations: Gaussian elimination, Pivoting, Stability
of Gaussian elimination, Cholesky factorization,
·
Sparse linear systems: Factorization techniques for sparse
matrices
·
Eigenvalues:
Eigenvalue problems; Reduction to Hessenberg
or tridiagonal form; Rayleigh quotient, power method, inverse iteration;
QR method; Other algorithms (Jacobi, bisection,
divide-and-conquer, SVD)
·
Iterative Methods: Stationary iterative methods (Jacobi, Gauss-Seidel, SOR); GMRES;
Conjugate gradients method; Preconditioning for convergence; Other iterative
methods |
HomeworkHomework assignments will be given once every couple of weeks. They will be posted in the Lecture Schedule section below. MATLABYou are expected to learn basic MATLAB during the early part of the course. CS majors should use their accounts on department machines to access MATLAB. Non-CS majors with no access to MATLAB in their respective departments should contact the CS system administrators for accounts. MATLAB documentation can be found at the MathWorks web site. GradingThe final grade will depend upon your performance on
homework and three exams as follows:
Homework policy: 1 day late: -10%, 2 day late: -20%, 3 day late: -40%, 4 or more days late homeworks will not accepted Exams: All exams are open-book/open-notes Scholastic DishonestyUniversity regulations (Section 42) define scholastic dishonesty to include acquiring answers from any unauthorized source, working with another person when not specifically permitted, observing the work of other students during any exam, providing answers when not specifically authorized to do so, informing any person of the contents of an exam prior to the exam, and failing to credit sources used. Disciplinary actions range from grade penalty to expulsion. |
Handouts, homework assignments, solutions, etc., will be
available here.
|
Final Exam: May 7, 8:00am-10:00am (Wed.) |
Last updated by Vivek Sarin on November
16, 2007