Nonlinear network flow problems with side constraints through projected Lagrangian methods

Publication TypeConference Paper
Year of Publication1997
AuthorsHeredia, F. J.
Conference NameEURO XV / INFORMS XXXIV Joint International Conference
Conference Date14-17/06/97
Conference LocationBarcelona, Spain
Type of WorkContributed oral presentation
Key Wordsnonlinear network flows; nonlinear side constraints; projected lagrangian methods; research
AbstractRecents numerical experiments show that the resolution of the Nonlinear network Flow problem with side Constraints (NFC) can be significantly sped up, when the side constraints are linear, by specialised codes based on a conjunction of primal partitioning techniques and active set methods. A natural extension of these methods one is to be used into a Projected Lagrangian Algorithm (PLA). A specialised (PLA) will solve the general (NFC) problem through the optimization of a sequence of (NFC) with linear side constraints, taking benefit of the efficiency of the linear side constraints codes. The description of this methodology will be presented together with the preliminary numerical results.
ExportTagged XML BibTex
AttachmentSize
Heredia_EUROXV-INFORMS24_97.pdf239.66 KB