Computation of the singular value expansion (Q1094850)

A method for computing the singular values and singular functions of real square-integrable kernels is presented. The analysis shows that a ``good'' discretization always yields a matrix whose singular value decomposition is closely related to the singular value expansion of the kernel. This relationship is important in connection with the solution of ill-posed problems since it shows that regularization of the algebraic problem, derived from an integral equation, is equivalent to the integral equation itself.