Skip to main content
U.S. Department of Energy
Office of Scientific and Technical Information

A robust trust-region algorithm with a non-monotonic penality parameter scheme for constrained optimization

Conference ·
OSTI ID:35981

An algorithm for solving the problem of minimizing a non-linear function subject to equality constraints is introduced. This algorithm is a trust-region algorithm. In computing the trial step, a projected-Hessian technique is used that converts the trust-region subproblem to a one similar to that for the unconstrained case. To force global convergence, the augmented Lagrangian is employed as a merit function. One of the main advantages of this algorithm is the way that the penalty parameter is updated. We introduce an updating scheme that allows (for the first time to the best of our knowledge) the penalty parameter to be decreased whenever it is warranted. The behavior of this penalty parameter is studied. A convergence theory for this algorithm is presented. It is shown that this algorithm is globally convergent and that the globalization strategy will not disrupt fast local convergence. The local rate of convergence is also discussed. This theory is sufficiently general that it holds for any algorithm that generates steps whose normal components give at least a fraction of Cauchy decrease in the quadratic model of the constraints and uses Fletcher`s exact penalty function as a merit function.

OSTI ID:
35981
Report Number(s):
CONF-9408161--
Country of Publication:
United States
Language:
English

Similar Records

Efficient proximal subproblem solvers for a nonsmooth trust-region method
Journal Article · Fri Jan 03 23:00:00 EST 2025 · Computational Optimization and Applications · OSTI ID:2505123

An active-set algorithm and a trust-region approach in constrained minimax problem
Journal Article · Sun Jul 15 00:00:00 EDT 2018 · Computational and Applied Mathematics · OSTI ID:22769298

Newton's method with a model trust-region modification
Technical Report · Mon Sep 01 00:00:00 EDT 1980 · OSTI ID:6836252