The proximal normal formula in Hilbert space

From MaRDI portal





Let H be a real Hilbert space, and let C be a nonempty closed subset of H. A vector \(v\in H\) is called a proximal normal to C at a point \(c\in C\) if \(\| tv\| =d_ C(c+tv)\) for some \(t>0\) (where \(d_ C\) denotes the distance function from C). Let \(N_ C(c)\) denote the (Clarke) normal cone to C at c [for the definition, see \textit{F. H. Clarke}, Optimization and nonsmooth analysis (1983; Zbl 0582.49001), p. 51]. The aim of this paper is to prove the following proximal normal formula: \(N_ C(c)=\overline{co}\{w\)-lim \(v_ i:\) \(v_ i\) is a bounded sequence of proximal normals to C and \(c_ i\), \(c_ i\to c\}.\) An analogous formula was proved, by using other methods, by \textit{J. M. Borwein} and \textit{H. M. Strojwas} [Can. J. Math. 38, 431-452 (1986; Zbl 0577.46011)] for a more general case of a reflexive Banach space whose norm is Fréchet differentiable away from 0 and Kadec.











This page was built for publication: The proximal normal formula in Hilbert space

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q3791765)