Non-Linear Optimization Using Space Mapping |
| | Abstract | In this project the optimization method named Space Mapping is presented, implemented and tested. The project presents novel theoretical additions to the conventional Space Mapping theory, that links the Space Mapping framework to surrogate and direct optimization theory. The Space Mapping theory defines a framework for optimization of an accurate, expensive non-linear model by utilizing a less accurate, cheaper model, through a parameter mapping. The implementation is based on a traditional gradient based trust region method, used in sequential optimization of surrogates aproximating the expensive model. Three numerical tests show how the novel theoretical additions greatly enhance the Space Mapping method when considering more general problems. | | Type | Master's thesis [Academic thesis] | | Year | 1999 | | Publisher | Informatics and Mathematical Modelling, Technical University of Denmark, DTU | | Address | Richard Petersens Plads, Building 321, DK-2800 Kgs. Lyngby | | IMM no. | IMM-EKS-1999-23 | | Electronic version(s) | [pdf] | | BibTeX data | [bibtex] | | IMM Group(s) | Scientific Computing |
|