Date: Tue, 14 Jan 1997 20:02:34 GMT
Server: NCSA/1.5.2
Last-modified: Thu, 01 Feb 1996 17:13:18 GMT
Content-type: text/html
Content-length: 2967
Computer Science 510 - Numerical Analysis
Computer Science 510 - Numerical Analysis
Instructor:
Professor Andrew Gelsey, (gelsey@cs.rutgers.edu)
402 Hill Center, Busch Campus, 5-4869
Office Hours: Wed 11am-noon
TA:
Daqing Li, (daqing@paul.rutgers.edu)
429 Hill Center, Busch Campus, 5-3766 (ext. 40)
Office Hours: Fri 2pm-4pm
Course Objective: derivation, analysis, and implementation of algorithms for
numerical problems.
Optional text books (should be available at Rutgers Bookstore):
- S. D. Conte, Carl de Boor, Elementary Numerical Analysis: an
Algorithmic Approach, 3rd edition.
- Gene Golub, James M. Ortega, Scientific Computing : an
Introduction with Parallel Computing.
Additional references:
- Kendall E. Atkinson, An Introduction to Numerical Analysis,
2nd edition.
- Germund Dahlquist and Ake Bjorck, Numerical Methods.
(All of these books are on reserve for CS 510 in the Hill Center
Math library.)
Prerequisites: Calculus, Linear Algebra, ability to program in a high
level language (preferably Fortran or C)
Grading:
- written homework(approx 10) + computer programs(approx 4) -- approx 25%
- midterm approx 30%
- final approx 45%
Course Outline:
- Floating point numbers and roundoff error: (brief treatment)
- Solution of nonlinear algebraic equations:
bisection method, regula falsi (binary and interpolation
search), fixed point iteration, Newton's method,
convergence rates (linear, quadratic), secant method,
systems of nonlinear equations --- Newton's method.
- Solution of linear algebraic systems:
Gaussian elimination/ LU decomposition, pivoting schemes,
complexity, matrix inversion,
variants of elimination for symmetric and banded matrices,
sparse matrices (connections to graph theory),
norms, condition number,
error analysis,
iterative methods (SOR, convergence rates),
parallel solution of linear systems.
- Interpolation, approximation of functions:
the interpolating polynomial (construction and error term),
piecewise polynomial interpolation, splines,
multidimensional interpolation,
least squares approximation, orthogonal polynomials,
trigonometric approximation, fast Fourier transform.
- Numerical differentiation and integration:
derivation of polynomial-based quadrature formulas, error terms,
adaptive quadrature, Romberg integration, Gaussian quadrature,
numerical differentiation, error terms.
- Numerical solution of ordinary differential equations:
basic methods (e.g., Euler's method, quadrature-based methods),
truncation error,
higher order equations, systems.