Skip to main content
U.S. Department of Energy
Office of Scientific and Technical Information

Numerical solution of bound constrained problems

Technical Report ·
OSTI ID:6118678
This paper considers the application of projected gradient techniques to the numerical solution of bound constrained optimization problems. An advantage of algorithms based on projected gradients is that they have the ability to drop and add many constraints at each iteration, and to quickly identify the optimal active constraints. This ability is of importance in large scale problems. In contrast, active set strategies restrict the change in the dimension of the working subspace by only dropping or adding one constraint at each iteration. We introduce and motivate the gradient projection method, and survey some of the recent numerical and theoretical results. In particular, we discuss algorithms that combine trust region methods and projected gradient methods. 16 refs.
Research Organization:
Argonne National Lab., IL (USA). Mathematics and Computer Science Div.
DOE Contract Number:
W-31109-ENG-38
OSTI ID:
6118678
Report Number(s):
ANL/MCS-TM-96; ON: DE88000653
Country of Publication:
United States
Language:
English