A smoothing approach for the optimal parameter selection problem with continuous inequality constraint
dc.contributor.author | Feng, Z. | |
dc.contributor.author | Yiu, K. | |
dc.contributor.author | Teo, Kok Lay | |
dc.date.accessioned | 2017-01-30T10:35:49Z | |
dc.date.available | 2017-01-30T10:35:49Z | |
dc.date.created | 2013-11-11T20:00:31Z | |
dc.date.issued | 2013 | |
dc.identifier.citation | Feng, Z.G. and Yiu, K.F. and Teo, K.L. 2013. A smoothing approach for the optimal parameter selection problem with continuous inequality constraint. Optimization Methods and Software. 28 (4): pp. 689-705. | |
dc.identifier.uri | http://hdl.handle.net/20.500.11937/4017 | |
dc.identifier.doi | 10.1080/10556788.2013.775282 | |
dc.description.abstract |
In this paper, we consider a class of optimal parameter selection problems with continuous inequality constraints. By introducing a smoothing parameter, we formulate a sequence of KKT (Karush-Kuhn-Tucker) systems of this problem and then transform it into a system of constrained nonlinear equations. Then, the first- and second-order gradients formulae of the cost functional and the constraints are derived. On this basis, a smoothing projected Newton-type algorithm is developed to solving this system of nonlinear equations. To illustrate the effectiveness of the proposed method, some numerical results are solved and presented. | |
dc.publisher | Taylor & Francis | |
dc.subject | optimal parameter selection problem | |
dc.subject | KKT system | |
dc.subject | projected Newton-type algorithm | |
dc.subject | continuous inequality constraint | |
dc.title | A smoothing approach for the optimal parameter selection problem with continuous inequality constraint | |
dc.type | Journal Article | |
dcterms.source.volume | 28 | |
dcterms.source.number | 4 | |
dcterms.source.startPage | 689 | |
dcterms.source.endPage | 705 | |
dcterms.source.issn | 10556788 | |
dcterms.source.title | Optimization Methods and Software | |
curtin.department | ||
curtin.accessStatus | Fulltext not available |