Dipartimento di Informatica e Sistemistica - Università la Sapienza di Roma
23,24,30,31 marzo 2009
Optimality conditions for constrained optimization problems (KKT). The linear case. Second order conditions. Algorithms for unconstrained optimization: line search, Armijo’s method. Monotonic and nonmonotonic gradient methods. Newton-type methods. Constrained optmization algorithms. Conditional gradient (Frank-Wolfe). Projected gradient. Decomposition methods with simple constraints. Transformation into unconstrained optimization problems. Local Newton algorithms. Introduction to filter and RQP methods.