A globally convergent version of a general recursive algorithm for nonlinear programming

Abstract

This paper discusses a general recursive algorithm for the minimization of a nonlinear objective function subject to nonlinear inequality and equality constraints. The proposed method includes both Wilson's and Robinson's quadratically convergent methods. The considerations are influenced by a paper of Schittkowsex. His idea of using a differentiable augmented Lagrange function for forcing the global convergence can be generalized by extending the principle to a whole class of locally convergent algorithms.

Keywords:

• Nonlinear Programming
• Recursive Methods
• Superlinear Rate of Convergence
• Global Convergence

AMS 1980 Subject Classifications:

• Primary:65K05
• Secondary90C30
• 49D15