Variants of the reduced Newton method for nonlinear equality constrained optimization problems

Abstract

Following an elementary substitution approach efficient quadratically convergent algorithms for nonlinear equality constrained optimization problems are derived, which reduce the dimension of the problem and allow to exploit special structures of the constraints. The algorithms obtained are compared with those of the Lagrange-Newton technique.

Keywords:

• Nonlinear optimization
• equality constraints
• iterative methods
• local -quadratic convergence

AMS 1980 Subject Classification:

• Primary:65 K 05
• Secondary:65 H 10