Continuous almost best approximations in \(C[0,1]\) (Q5944776)

The problem of almost best approximation in the space \({C[0,1]}\) is considered. Let \({\varepsilon>0}\). The operator of an almost best approximation, or the \({\varepsilon}\)-projection of an element \({x\in C[0,1]}\) on a subset \({L\subset C[0,1]}\), is defined as the set-valued mapping \[ x\mapsto P_{L,\varepsilon}(x)=\{z\in L:\| x-z\| \leq\rho(x,L)+\varepsilon\}. \] For \({\varepsilon=0}\), the operator \({P_{L,\varepsilon}}\) is the operator of metric projection. A continuous mapping \({x\mapsto z\in P_{L,\varepsilon}}\) is called a continuous \({\varepsilon}\)-selector. It is well-known that the set \(R_{m,n}\) of algebraic rational fractions is an existence and uniqueness set (Chebyshev set) for the operator of metric projection in \({C[0,1]}\). \textit{H. Maehly} and \textit{C. Witzgall} [Num. Math.2, No.~5, 293--307 (1960; Zbl 0131.29801)] showed that the metric projection on \(R_{m,n}\) is discontinuous. Nevertheless, \textit{Konyagin} [Mat. Zametki 44, No.~3, 404 (1988; Zbl 0694.41022)] proved that for each \({\varepsilon>0}\) there exists a continuous \(\varepsilon\)-selector on \(R_{m,n}\). In the paper under review the author defines the class of subsets \(K\) in \({C[0,1]}\) and prove the existence of a continuous \({\varepsilon}\)-selector on \({L\in K}\). The class \(K\) includes the set of algebraic rational fractions and the set of piecewise linear functions. Two other related statements are formulated. Let \(S^d_n\) be the set of all continuous piecewise polynomial functions \({l\in C[0,1]}\), that is, functions such that there exists a partition \({0=x_0<x_1<\dots <x_{n-1}<x_n=1}\) for which \({l| [x_{i-1},x_i]}\) is a polynomial of degree \(\leq d\) for all \({i=1,\dots ,n}\). {Theorem~2.} If \({n,d>1}\) and \({\varepsilon\geq 0}\), then there exists no continuous \(\varepsilon\)-selector on \(S^d_n\). {Theorem~3.} There exists a continuous \(0\)-selector on \(S^1_n\) if and only if \({n=1}\) or \({n=2}\). The paper contains only formulations and some comments on proofs. For full proofs, examples and references see the subsequent paper of \textit{E. D. Livshits} [Izv. Ross. Akad. Nauk, Ser. Mat. 67, No.~1, 99--130 (2003; Zbl 1079.41007)].