Godela Scherer (Biography)
Cesma, Universidad Simon Bolivar
Caracas, Venezuela

“Large Scale Least Squares Scattered Data Fitting”

Abstract

The least squares approximation by tensor products of B-splines of large sets of scattered data is considered. This ill-conditioned or even singular problem requires special techniques, some of which are described in this paper.

The performance of a block Truncated Singular Value Decomposition algorithm and several Lanczos based algorithms for sparse LSQ is evaluated. Appropriate regularization methods are discussed and their implementations compared.