On almost interpolation (Q1379654)

In this paper, the author investigates the concept of ``almost interpolation'' in finite-dimensional spaces of real functions on a topological space. An almost interpolation (AI) set is a configuration of points such that any choice of neighbourhoods of these points contains another point set which admits Lagrange interpolation. In this article, AI sets are characterized with the help of so-called neighbourhood and local dimensions of their subsets, which includes establishing several properties of these dimension concepts. Almost Chebyshev (AT) spaces are then introduced as \(n\)-dimensional function spaces, where any choice of \(n\) points forms an AI set. With some assumptions on the underlying topological space, AT spaces of generalized Chebyshev splines are considered, where AI sets can be characterized by a condition corresponding to the Schoenberg-Whitney condition for univariate polynomial splines.