An approach to obtaining global extremums in polynomial mathematical programming problems

From MaRDI portal





Any mathematical programming problem having polynomial objective function and constraints is converted, after some simple transformations, into an equivalent problem (P) consisting in minimizing a quadratic function \(K_ 0(x)\) subject to quadratic equality constraints \(K_ i(x)=0\), \(i=1,...,m\). Let \(\Omega\) (resp. \({\bar \Omega}\)) be the set of vectors \(u=(u_ 1,...,u_ m)\) for which the symmetric matrix corresponding to the quadratic part of the Lagrange function \[ L(x,u)\quad =\quad K_ 0(x)\quad +\quad \sum^{m}_{i=1}u_ iK_ i(x) \] is positive (resp. non-negative) definite. A simple proof of the equality \[ \inf \{K_ 0(x)/K_ i(x)\quad =\quad 0,\quad i=1,...,m\}\quad =\quad \sup_{u\in \Omega}\min_{x}L(x,u), \] under the assumption that the latter supremum is attained at some \(u\in \Omega\), is given. Furthermore, it is proved that a convenient reformulation, as a problem of type (P), of the unconstrained minimization of a polynomial of one variable and even degree satisfies this equality and the supremum is attained at some \(u\in {\bar \Omega}\).




Cited in
(32)








This page was built for publication: An approach to obtaining global extremums in polynomial mathematical programming problems

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q3801357)