The Clarke coderivative of the frontier map in a multi-objective optimal control problem (Q6166355)

Given three Asplund spaces \(X,W,Z\), a closed convex set \(\Omega \subseteq Z\), a continuous mapping \(g:W\times Z\rightarrow X\) and a vector function \(f:W\times Z\rightarrow \mathbb{R}^{s}\), define for each \(w\in W\) the set \[ G(w):=\left \{ z\in Z:g\left( w,z\right) =0\right \}. \] Then one defines the frontier map \(\tilde{\mathcal{F}}:W\rightrightarrows \mathbb{R}^{s}\) by \[ \tilde{\mathcal{F}}(w):=\operatorname*{Min}\nolimits_{\mathbb{R}_{+}^{s} }\left \{ f(w,z):z\in G(w)\cap \Omega \right \} \] where \(\operatorname*{Min}\nolimits_{\mathbb{R}_{+}^{s}}A\) is the set of efficient points of a set \(A\). A basic result of this paper gives inner and outer estimates of the Clarke coderivative of the multifunction \(\tilde{\mathcal{F}}\). Based on this, formulae for upper and lower evaluation of the Clarke coderivatives of the extremum multifunction in a multi-objective parametric optimal control problem are obtained.