Some sufficient conditions for a KKT point of the constrained optimization problem to be a solution of GNCP are presented.

,under suitable hypotheses and for sufficiently large values of the penalty parameters but without requiring the penalty parameters go to infinity,every minimum point(KKT point) of the original constrained problem corresponds to a minimum point(stationary point) of an augmented Lagrangian on the product space of problem variables and multipliers.

Based on KKT condition,the possible change of support vector set is analyed after new samples are added to training set and a view named parallel potential data set is put forward.

This thesis studies the KKT condition for a class of two-level programming.

New model and algorithm for solving the KKT system of optimal power flow;

In this paper,by using the FischerBurmeister function,we reformulate the KKT system of a contrained optimization problem into an equivalent nonsmooth equation.

By reformulating the KKT system as a constrained equation,the algorithm generates the search direction by solving a linear equation at each iteration.

Aimed at the problem that the traditional support vector machine (SVM) cannot adapt itself to text database updating continuously from time to time, a text categorization algorithm with incremental SVM was presented after analyzing the KKT qualification of newly added text sets and thoroughly investigating the variation of support vector set due to text addition.