Nonlinear Programming Methods: Iterative solution, Preconditioning
              and Regularization
              
              Jorge Nocedal
           Northwestern University
              
 
 We first discuss the relative strengths of interior and active
 set methods for three classes of problems: (i) problems in
 which the Hessian is not sparse or is not available; (ii) very
 large problems; (iii) problems that do not satisfy constraint
 qualifications. Some attention will be given to a new active
 set approach that uses linear programming techniques to identify
 the active set. The second part of the talk focuses on the use
 of iterative linear algebra techniques in interior and active-set 
 methods. Preconditioning and regularization are essential to achieve
 robustness over a wide range of applications.
 
 

====================================================
Prof. Jorge Nocedal          (847) 491-5038   office
ECE Department               (847) 467-4144   fax
Northwestern University      nocedal@ece.northwestern.edu
Evanston, IL 60208-3118      http://www.ece.nwu.edu/~nocedal