Solution of KKT systems in constrained optimization

Project: Research

Description

Efficiently and robustly solving the Karush-Kuhn-Tucker
(KKT) systems arising in optimization methods to solve large scale nonlinear programming problems is a big
challenge that requires a good insight into both the linear algebra and optimization fields. Our aim is to
contribute to the development of appropriate preconditioners for iterative methods designed to solve such
saddle-point systems arising in constrained optimization, exploiting their special structure and properties.
StatusActive
Effective start/end date3/10/11 → …