It has been estimated that over 70% of computer models in science and engineering rely in part at least on some type of linear algebraic method. The impact of computational linear algebra is also being felt in the analytical realm. The habits of the field reliance on orthogonal matrices and the study of problem sensitivity under perturbation have spilled over into diverse areas. A prime example is the increased use of the singular value decomposition (SVD) as an analytical tool in control engineering, sta tistics, signal processing, image compression, and other disciplines. Another facet of numerical linear algebra is that participants require skills from both mathematics and computer science in order to provide effective software tools. Moreover, the blossoming of new multiprocessor architectures is making the field ever more pertinent and dynamic. The course aims to give an overview of the field with focus on high-level algorithmic concepts as they pertain to new computer architectures. The presentation will be machine independent and at the conceptual level in order to emphasize linear algebraic techniques rather that real life applications on specific machines. However, some small scale illustrative computing should be expected.

Time: Tuesday & Thursday 2:30-3:45 PM, Room: CBC C113

Instructor: Professor George Miel, Office: CBC B509, ph: 895-0360

Text: Gene H. Golub and Charles F. Van Loan, Matrix Computations, The John Hopkins University Press, second edition, 1989.

Prerequisites: Graduate standing and prior training in numerical analysis or consent of the instructor.

[an error occurred while processing this directive]