A path-following primal-dual augmented Lagrangian method for NEP
A new path-following primal-dual augmented Lagrangian method is proposed
for solving nonlinear equality constrained optimization problems (NEP).
At each iteration, a Newton-like method is used to solve a perturbed
optimality condition that defines a penalty trajectory parameterized by
both the penalty parameter and the estimated Lagrange multipliers.
We show that this method is globally convergent and has a quadratic
convergence rate in the limit. Finally, numerical experiments on
problems from the CUTEst test collection are are used to support the
Tuesday, May 15, 2018
11:00AM AP&M 2402
Center for Computational Mathematics9500 Gilman Dr. #0112La Jolla, CA 92093-0112Tel: (858)534-9813