Gap functions and error bounds for generalized mixed vector equilibrium problems (Q495725): Difference between revisions

For generalized vector equilibrium problems of the form \[ \forall y\in K, \;\text{find} \;x\in K \;\text{such that} \;f(x,y)+\langle T(x),y-x\rangle +\Phi(x,y)-\Phi(x,x) \notin -\mathrm{int}\,\mathbb{R}^m_+ , \tag{1} \] where \(K\) is a nonempty, closed and convex set in \(\mathbb {R}^n,\) \(f\) is strongly monotone and convex in the second argument, and \(\Phi\) fulfills a kind of quadratic Hölder condition, the authors give a scalar-valued gap function \(\psi\) without scalarizing parameters: \[ \psi(x) = \sup_{y\in K} \min_{1\leq i\leq m} \{-f_i(x,y)+\langle T_i(x),x-y\rangle +\Phi_i(x,x)-\Phi_i(x,y)\}. \] Then, it is shown that a regularized form \(\psi_{\Theta}\) of \(\psi\) given by \[ \psi_{\Theta}(x) = \sup_{y\in K} \min_{1\leq i\leq m} \{-f_i(x,y)+\langle T_i(x),x-y\rangle +\Phi_i(x,x)-\Phi_i(x,y)-\frac{\Theta}{2} \| y-x\|^2\}, \] where \(\Theta > 0\) and \(f_i\) and \(\Phi_i\) are convex in the second argument, is well-defined and also a gap function for (1). With the help of it, and if all the mentioned conditions are satisfied, an error bound result (without projections and in explicit form) for the distance from any point \(x \in K\) to the solution set (if nonvoid) of the given equilibrium problem (1) is proved. The special case \(\Phi(x,y) = 0\) for all \(x,y\in K\) is handled separately.