Title :
Gradient projection techniques for large-scale optimization problems
Author_Institution :
Div. of Math. & Comput. Sci., Argonne Nat. Lab., IL, USA
Abstract :
A description is given of the impact of the identification properties of the gradient projection method on algorithms for the solution of large-scale optimization problems with bound constraints. In particular, the author describes results of J.J. More and G. Toraldo (Rep. MCS-P77-0589, Argonne Nat. Lab., 1989), which show that the gradient projection method can be combined with the conjugate gradient method to solve an important class of large-scale (number of variables between 5000 and 15000) bound constrained quadratic programming problems with a few (less than 15) iterations
Keywords :
iterative methods; optimisation; bound constrained quadratic programming problems; conjugate gradient method; gradient projection method; iterative methods; large-scale optimization; Computer science; Constraint optimization; Contracts; Costs; Gradient methods; Laboratories; Large-scale systems; Mathematics; Optimization methods; Quadratic programming;
Conference_Titel :
Decision and Control, 1989., Proceedings of the 28th IEEE Conference on
Conference_Location :
Tampa, FL
DOI :
10.1109/CDC.1989.70141
