Numerical solution of integral equations in a space of distributions (Q1092647)

The basic integral equation of random field estimation theory is of the form (*) \(Rh=f\), where R is a linear integral operator which is an isomorphism between the spaces \(H_{-k}\) and \(H_ k\). Here \(H_ k=H_ k(D)\), \(k\geq 0\), is the Sobolev space and \(H_{-k}\) is the dual space of distributions with support in the closure of the domain D of signal processing. The author establishes convergence of the least squares method and a projection method for solving (*), and suggests the basis functions which describe the correct singularity in the approximate solutions.