Constrained Optimization - Stanford University

Now we have n+ ^m+2mequations and two main possibilities for each inequality constraint k: s k>0: the k-th constraint is inactive, and k= 0. s k= 0: the k-th constraint is active, and k6= 0. k must then be non-negative, otherwise from the rst equations, the gradient of objective and gradient of constraint point in the same direction. ^ ................
................