Instructor: Emre Mengi Office: SCI 113 E-mail: emengi@ku.edu.tr Phone: (212) 338-1658 Office Hours: Tue. 9:45 - 10:45, or Thur. 16:00 - 17:00, or by appointment Teaching Assistant: Rifqi Aziz Office: SCI 130 E-mail: raziz14@ku.edu.tr Office Hours: Mon. 14:00 - 15:00 | Course Description: Majority
of the numerical algorithms rely on the solutions of linear algebra
problems. In the end whether these algorithms are worthy for common use
translates into whether the associated linear algebra problems can be
solved efficiently and accurately. The main purpose of the course is to feature modern numerical approaches for the solutions of linear algebra problems including systems of linear equations, eigenvalue problems, singular value problems. But applied linear algebra tools, such as matrix factorizations, projectors, reflectors without which modern numerical algorithms cannot exist, will also be introduced throughout. Both direct approaches and iterative approaches will be covered. Thorough analyses of how rounding errors affect the accuracy will be presented for direct approaches. The course concludes with extensions of iterative methods to systems of nonlinear equations, where Newton's and quasi-Newton methods as well as their convergence are discussed. Announcements:
|