Abstract
This paper describes a new technique to find the minimum norm solution of a linear program. The main idea is to reformulate this problem as an unconstrained minimization problem with a convex and smooth objective function. The minimization of this objective function can be carried out by a Newton-type method which is shown to be globally convergent. Furthermore, under certain assumptions, this Newton-type method converges in a finite number of iterations to the minimum norm solution of the underlying linear program.
Original language | English |
---|---|
Pages (from-to) | 333-345 |
Number of pages | 13 |
Journal | Journal of Optimization Theory and Applications |
Volume | 116 |
Issue number | 2 |
DOIs | |
Publication status | Published - 1 Feb 2003 |
Keywords
- finite termination
- Linear programs
- minimum norm solution
- Newton method
- unconstrained minimization
ASJC Scopus subject areas
- Control and Optimization
- Management Science and Operations Research
- Applied Mathematics