We consider the NLP optimization problem ?? and discuss the duality gap between P and ?? The convex problem D is in fact the dual of a ``relaxed'' version of P via ``randomization'' which permits to give a simple interpretation for the presence or absence of a duality gap in the general case. Several particular cases are also discussed and the case of homogeneous functions is given special attention.
@article{463, title = {Duality and randomization in nonlinear programming}, journal = {ANZIAM Journal}, volume = {48}, year = {2008}, doi = {10.21914/anziamj.v42i0.463}, language = {EN}, url = {http://dml.mathdoc.fr/item/463} }
Lasserre, Jean B. Duality and randomization in nonlinear programming. ANZIAM Journal, Tome 48 (2008) . doi : 10.21914/anziamj.v42i0.463. http://gdmltest.u-ga.fr/item/463/