Bounded scalar convergence (Q1900645)

There are many notions of convergence for a sequence (or more generally a net) of closed convex subsets in a real normed vector space \(E\). Denoting by \(C(E)\) the family of all closed, convex and nonvoid subsets of \(E\), the authors provide a detailed study for the bounded scalar convergence -- abbreviated bS -- in \(C(E)\): a net \((A_i)\) in \(C(E)\) bS-converges to \(A\) if \(s(A_i\cap \rho U, x^*)\to s(A\cap \rho U, x^*)\), for each \(x^*\) in \(E^*\) and each \(\rho> 0\) such that \(A\cap \text{int}(\rho U)\neq \emptyset\), where \(U\) is the closed unit ball of \(E\) and \(s(A,\cdot)\) denotes the support functional of \(A\). The authors give many properties of this convergence, its associated lower and upper convergences and relations to other type of convergences. They show that: -- the bS-convergence is generated by a Hausdorff topology -- in finite-dimensional spaces it coincides with Kuratowski convergence -- for sequences in reflexive separable Banach spaces it coincides with weak Kuratowski convergence -- in locally uniformly convex Banach spaces bS-convergence and slice convergence coincide (the latter meaning \(\lim d(X, A_i)= d(X, A)\), for each bounded \(X\in C(E)\)).