A class of best simultaneous approximation problems (Q1921230)

Let \((X, |\cdot |)\) be a normed linear space. Let \(\rho\) be a norm on \(\mathbb{R}^2\). For \(w= (x,y) \in X \times X= X^2\), \(\beta (t)= \rho(|x|, |y|)\) is a norm on \(X^2\). Let \(S \subseteq X\) and \(\widetilde S = \{(s,s) \in X^2: s\in S\}\). Given an \(F= (x,y) \in X^2\), the problem of best simultaneous approximation involves detailing the properties of the set of best approximations \(P\eta (F-) = \inf_{t \in \widetilde S} \beta (F-t)\}\). This paper contributes theorems on characterization and uniqueness to this interesting area. The theorems require restrictions on the norms involved and on the set \(S\). Applications are made to examples involving \(X= C[0,1]\), \(S\) equal the polynomials of degree 2 and 3, and \(\rho\) an \(l^2_p\) norm.