Math 4420/5420 - Optimization - Spring 2017
Dr. Radu C. Cascaval

Home

Syllabus

Assignments

Exam Info

MATLAB

Links

Nonlinear Least Square

Gauss-Newtron Method gaussn.m
Levenberg-Marquandt Method levmar.m

Derivative Free Optimization

Nelder - Mead method nelder.m

Conjugate Gradient Method

- linearCG.m
- linearCG_ex1.m (Example of Ax=b with A random matrix)
- linearCG_ex2.m (Example of Ax=b with clustered eigenvalues)
- nonlinearCG.m

Application to Curve Fitting

Curve fitting with steepest descent curvefittingSD.m

Quasi-Newton methods

BFGS method BFGS.m

Step Length computation

- linesearch.m (backtracking algorithm)
- linesearch_symbolic.m (solves the one-variable optimization symbolically - WARNING: It is slow!)
- linesearch_secant.m (uses secant method for solving the one-variable optimization)
- linesearch_strongwolfe.m (accounts for the strong Wolfe conditions) 

Steepest Descent method

Example of steepest descent with backtracking: hmw1sample.m

- steepest_session.m [Includes a graphical representation for Rosenbrock function, needs rosenbrock.m]
- steepest_secant.m (requires linesearch_secant.m)
- steepest_symbolic.m (requires linesearch_symbolic.m)

Functions of Interest in Optimization

- Rosenbrock function rosenbrock.m
- Newton method newton_session.m - (needs newton.m and myfunc.m)