1: \begin{abstract}

2:   In the vicinity of a solution of a nonlinear programming problem at

3:   which both strict complementarity and linear independence of the

4:   active constraints may fail to hold, we describe a technique for

5:   distinguishing weakly active from strongly active constraints.  We

6:   show that this information can be used to modify the sequential

7:   quadratic programming algorithm so that it exhibits

8:   superlinear convergence to the solution under assumptions weaker

9:   than those made in previous analyses. 

10: %  

11: %   Other constrained optimization algorithms based on the primal-dual

12: %   interior-point approach and the method of multipliers also can be

13: %   modified to enhance their local convergence properties

14: %

15: \end{abstract}

16: