Numerical linear algebra and some problems in computational statistics

Abstract

Numerical linear algebra and computational statistics have much in common. They share some common roots, which has influenced developments in both fields.

We have briefly recalled the role of moments in modern iterative methods. We have further presented several examples of mathematically identical problems and closely related methods and algorithms for their solution, in, e.g., matching moments model reduction, linear and orthogonal regression, regularization of ill-posed problems, and total least squares.