On a general class of multivariate linear smoothing operators (Q1112278): Difference between revisions

The purpose of this paper is to introduce and study a very general class of multivariate linear smoothing operators, which may not be positive, using both summation and integration. Let D be a bounded region in \(R^ s\), \(L_ p(D)\) \((1\leq p<\infty)\) the usual Banach space with norm \(\| \cdot \|_ p\). Denote \(\Delta_ n\) a subset of \(Z^ s\) with cardinality n, \(\lambda_{nk}\) real-valued functions in \(L_ 1(D)\), and \(\omega_{nk}\) real-valued functions in \(L_ 1(D)\) with unit mean, where \(k\in \Delta_ n\), \(n\in Z_+\). Consider the linear operator \(\lambda_ n\) defined by \(A_ n(f)(x)=\int_{D}S_ n(x,t)f(t)dt,\) where \(S_ n(x,t)=\sum_{k\in \Delta_ n}\lambda_{nk}(x)\omega_{nk}(t)\) (x,t\(\in D)\) and f: \(D\to R\) belongs to a certain class of functions. General conditions on the kernels \(S_ n\) of \(\lambda_ n\) so that \(\lambda_ n(f)\to f\) as \(n\to \infty\) for all \(f\in L_ p(D)\) \((1\leq p<\infty)\), or C(F) where F is any closed subset of D, as well as the degree of approximation of f by \(\lambda_ n(f)\) in the norms \(\| \cdot \|_ p\) and \(\| \cdot \|_ F\) respectively are investigated. Some general convergence theorems and quantitative estimates are obtained.