Norm-relaxed method of feasible directions for solving nonlinear programming problems (Q1338556): Difference between revisions

The paper describes a variation of a feasible direction method for solving nonlinear programming problems. The authors show that the method converges to a Fritz John point. The algorithm is a feasible direction method where the optimality of the solutions is determined according to a constrained quadratic program. This program places no explicit norm constraints on the direction vector. The step length is the standard step length function in the direction determined by the optimality function. The authors include some numerical experience with the algorithm.