On algebras and applications of operators with pseudosparse matrices (Q1358084): Difference between revisions

We introduce and study new classes of linear operators in \(l_p\), the algebras of operators with pseudosparse matrices (PSM). A number of problems in various fields of mathematics lead naturally to studying PSM. First, these are the allocation of problems of computational mathematics (the finite difference method, the finite element method, interpolation by multidimensional splines, etc.) which lead to high-order systems of linear algebraic equations with sparse matrices. We can heuristically define a PSM as a matrix all whose entries may be nonzero but an overwhelming majority of which are very small in magnitude. We devote in the present article to rigour definitions, examples, and the proof of invariance of algebras of PSMs under basic operations on matrices. Second, these are some problems of harmonic analysis. The first class of such problems relates to generalizations of the classical theorem by Wiener of absolute summability of Fourier series of functions \(g(x)\) and \(1/g(x)\). Using the theory of PSMs, we show that if the Fourier coefficients \(\xi_k\) of a function \(g(x)\) satisfy estimates of the form \[ |\xi_k|\leq Cf(|k|),\tag{1} \] where \(f\) is a decreasing function in some rather wide class, then the Fourier coefficients \(\eta_k\) of the function \(1/g(x)\) satisfy similar estimates. The second class of such problems relates to studying the rate of decay for the Fourier coefficients of the proper and adjoint function of differential operators on the condition that the Fourier coefficients of the operator coefficients are known to satisfy estimates like (1). Third, these are the problems of estimating the norms of orthogonal and interpolation projections to spaces of spline functions in the operator norms of the spaces \(C(\Omega)\) and \(l_\infty(\Omega)\).