Interpolation of data by smooth nonnegative functions (Q520724): Difference between revisions

The authors, and especially Charles Fefferman, have published a large number of most significant papers on the problem of interpolating a function given on a domain or on a discrete set. The main point is here that the interpolant is required to have a specified minimal smoothness or other side conditions such as nonnegativity and that the said domains may be required to have very general geometric features such as lack of smoothness of their boundaries. And of course a goal is then to establish certain bounds on the interpolants and their derivatives on the whole of \(n\)-dimensional real spaces. An example of the most interesting results in this paper is the statement that, given a nonnegative interpolant on a finite set \(E\) in \(n\) dimensions, such that for all subsets \(S\) of that finite set \(E\) with at most a certain, large enough number of elements, an interpolant on \(S\) with the maximum of all partial derivatives up to degree \(m\) bounded by \(1\) exists which is nonnegative, then there is an interpolant with the maximum of all partial derivatives up to degree \(m\) bounded, and this interpolant is nonnegative everywhere and is interpolating the function we stated with on the initial finite set \(E\).