CMSC 477: OPTIMIZATION
Catalog Description
Linear programming including the simplex algorithm and dual linear
programs, convex sets and elements of convex programming,
combinatorial optimization, integer programming.
Listed also as MAPL 477.
Prerequisites
CMSC/MAPL 460, 466, or 467.
Topics
- Convex sets (1 week)\\
Definitions of hull, cone, hyperplane, polytope, extreme points;
separating hyperplane theorem; supporting hyperplane theorem.
- Linear programming (5 weeks)\\
Examples, standard form, geometric and algebraic structure,
revised Simplex algorithm, duality, affine transformation algorithms.
- Unconstrained optimization (5 weeks)\\
Necessary and sufficient conditions for minimizers, order of
convergence, one dimensional algorithms (Fibonacci, golden section,
curve fitting, practical line search algorithms), multidimensional
algorithms (steepest descent, Newton's method, conjugate gradients,
quasi-Newton methods).
- Constrained optimization (3 weeks)\\
Necessary and sufficient conditions for minimizers, inequality
constraints, gradient projection algorithms with active
set strategy.
Course Text
David G. Luenberger, Linear and Nonlinear Programming,
Addison-Wesley, 1984.
Typical Grading and Workload
- 60% - 2-3 exams
- 40% - 4-5 homeworks and projects