A nondegenerate fuzzy optimality condition for constrained optimization problems without qualification conditions (Q714465)

The author provides a nondegenerate fuzzy optimality condition for constrained optimization problems of the form \[ \text{minimize }\varphi_0(x), \] \[ \text{subject to }\varphi_k(x)\leq 0\;(k= 1,\dots,m),\;\varphi_k(x)= 0\;(k= m+1,\dots, m+r),\;x\in\Omega, \] where \(\varphi_k: X\to\overline{\mathbb{R}}\) \((k= 0,\dots, m+r)\) are extended real-valued lower-semicontinuous functions on an Asplund space \(X\) and \(\Omega\) is a closed set in \(X\). By means of the Fréchet subdifferentials \(\widehat\partial\varphi_k(x)\) \((k= 0,\dots, m+r)\) and the normal cone \(\widehat N(x;\Omega)\) and without constraint qualification it is shown that for any local solution \(\overline x\), for any number \(\varepsilon> 0\) and for any weak\(^*\) neighborhood \(V^*\) of the origin in \(X^*\) there are vectors \(x_k\in\mathbb{B}_\varepsilon(\overline x)\) \((k= 0,\dots, m+r)\), \(\widehat x\in\Omega\cap \mathbb{B}_\varepsilon(\overline x)\) and multipliers \(\lambda_k> 0\) \((k= 1,\dots, m+ r)\) such that \[ |\varphi_0(x)- \varphi_0(\overline x)|\leq\varepsilon,\quad \lambda_k|\varphi_k(x_k)|\leq \varepsilon\;(k=1,\dots, m+r), \] and \[ 0\in\widehat\partial\varphi_0(x_0)+ \sum^m_{k=1} \lambda_k\widehat\partial\varphi_k(x_k)+ \sum^{m+r}_{k=m+1} \lambda_k(\widehat\partial\varphi_k(x_k)\cup \widehat\partial(-\varphi_i))+\widehat N(\widehat x; \Omega)+ V^*. \]