Best interpolation with convex constraints (Q1801554): Difference between revisions

Suppose \(X\) and \(Y\) are two locally convex spaces; \(A: X\to Y\) is a continuous linear map and \(f: X\to R_ +\cup \{\infty\}\) a proper lower semicontinuous convex function. For \(d\in Y\) fixed and for the closed convex set \(C\subseteq X\) one considers the minimization problem: \(\min\{f(x): x\in K=C\cap A^{-1} d\}\). In order to characterize a solution \(x_ 0\) of this problem the author obtains Kuhn-Tucker type conditions in the case when \(f\) is Gateaux differentiable or only continuous at \(x_ 0\). In the particular case when \(X\) is a real Banach space and \(f(x)=\| x-z\|\), \(z\) being fixed in \(X\) one obtains a best approximation problem with convex constraints. A characterization of the projection \(P_ k(z)\) is given (more explicitely) in the case when \(X\) is a Hilbert space and a characterization for the solution to the corresponding nonnegative best interpolation in the \(L_ p\) norm is also produced. Finally, known results in this direction are reproduced or generalized as particular cases.