The Fritz John and Kuhn-Tucker optimality conditions in continuous-time nonlinear programming (Q1083380): Difference between revisions

The continuous-time programming problem under consideration consists in minimizing \[ \phi (x)=\int^{T}_{0}f(x(t),t)dt \] subject to g(x(t),t)\(\leq 0\) a.e. in [0,T], \(x\in X\) where X is a nonempty open convex subset of the Banach space \(L^ n_{\infty}[0,T]\) of the equivalence classes of the Lebesgue measurable essentially bounded n- vector functions on [0,T] with the form \(\|.\|_{\infty}=\max_{1\leq j\leq n}._{0\leq t\leq T}\sup | x_ j(t)|\) where \(x_ 1(t),...,x_ n(t)\) are the components of x(t). Here, \(\phi\) : \(X\to R^ 1\), and \(g(x(t),t)=\gamma (x)(t)\) where \(\gamma\) is a map from X into the normed space \(\Lambda^ m_ 1[0,T]\) of the equivalence classes of all Lebesgue measurable essentially bounded m-vector functions on [0,T] with the norm \(\|.\|_ 1\) defined by \[ \| y\|_ 1=\max_{1\leq j\leq m}\int^{T}_{0}| y_ j(t)| dt. \] The author establishes the first order necessary optimality conditions of both the Fritz John and Kuhn-Tucker type for the above problem and discusses the relationship of these conditions and the saddle point optimality conditions. As an auxiliary result, the continuous version of Gordon's theorem is proved.