An Inexact Trust-Region SQP Method with Applications to PDE-Constrained Optimization
We present a trust-region sequential quadratic programming (SQP) method with iterative linear system solves, for the solution of smooth nonlinear equality-constrained optimization problems. Stopping criteria for iterative solvers are selected by the optimization algorithm, based on its overall progress. Global convergence is ensured and unnecessary oversolving of linear systems is eliminated. The algorithm is applied to several PDE-constrained optimization problems.
KeywordsSequential Quadratic Programming Linear Solver Iterative Solver Augmented System Minimum Norm Solution
Unable to display preview. Download preview PDF.