Pontryagin's maximum principle for multiple integrals

From MaRDI portal





This paper deals with necessary optimality conditions for multidimensional problems of optimal control of Dieudonné-Rashevsky type: minimize \(\int_ \Omega r(t,x(t),u(t))dt\) on \(W^{1,n}_ p(\Omega)\times L^ \nu_ \infty(\Omega)\) for \(\Omega\subset E^ m\), \(p> m\), subject to the state equations \(x_ t= g(t,x,u)\), control restrictions \(u(t)\in V\subset E^ \nu\) and fixed boundary values of \(x\). It is shown that the (for \(m=1\)) well-known local form of Pontryagin's maximum principle cannot be transferred to the case \(m>1\). However, for convex \(r(t,\xi,.)\), linear \(g(t,.,.)\) and compact \(V\) one can prove (by separation theorems) at least an \(\varepsilon\)-maximum principle in integrated form. For the entire collection see [Zbl 0778.00028].











This page was built for publication: Pontryagin's maximum principle for multiple integrals

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q3139152)