Projected combination direction method for a class of large scale nonlinear programming

Abstract

This paper presents a new practical algorithm (i.e. Projected Combination Direction Algorithm) for large scale optimization with simple bound constraints, and study some properties of this method, particularly, the global convergence property is shown. The numerical results are reported for 3 modified forms

Keywords:

• Nonlinear programming
• combination direction method
• Armijo rule
• global convergence

C.R.Categories:

• G 1.6 optimization