The method of least squares, discovered by Gauss in 1795, is a principal tool for reducing the influence of errors when fitting a mathematical model to given observations. Applications arise in a great number of areas in sciences and engineering. The increased use of automatic data capturing frequently leads to large-scale least squares problems. Such problems can be solved by using recent developments in preconditioned iterative methods and in sparse QR factorization. The first edition of Numerical Methods for Least Squares Problem was the leading reference on least squares problems for many years. The updated second edition stands out compared to other books on the topic because: it provides an in-depth and up to date treatment of direct and iterative methods for solving different types of least squares problems and for computing the singular value decomposition; covers generalized, constrained, and nonlinear least squares problems as well as partial least squares and regularization methods for discrete ill-posed problems; and contains a bibliography with more than 1100 historical and recent references, providing a unique survey of past and present research in the field.
