A quasi-newton quadratic penalty method for minimization subject to nonlinear equality constraints

We present a modified quadratic penalty function method for equality constr ained optimization problems. The pivotal feature of our algorithm is that a t every iterate we invoke a special change of variables to improve the abil ity of the algorithm to follow the constraint level sets. This change of va riables gives rise to a suitable block diagonal approximation to the Hessia n which is then used to construct a quasi-Newton method. We show that the c omplete algorithm is globally convergent. Preliminary computational results are reported.