Bjorck-Pereyra-type methods and total positivity [article]

Jose-Javier Martinez
2018 arXiv   pre-print
The approach to solving linear systems with structured matrices by means of the bidiagonal factorization of the inverse of the coefficient matrix is first considered, the starting point being the classical Bjorck-Pereyra algorithms for Vandermonde systems, published in 1970 and carefully analyzed by Higham in 1987. The work of Higham showed the crucial role of total positivity for obtaining accurate results, which led to the generalization of this approach to totally positive Cauchy,
more » ... ermonde and generalized Vandermonde matrices. Then, the solution of other linear algebra problems (eigenvalue and singular value computation, least squares problems) is addressed, a fundamental tool being the bidiagonal decomposition of the corresponding matrices. This bidiagonal decomposition is related to the theory of Neville elimination, although for achieving high relative accuracy the algorithm of Neville elimination is not used. Numerical experiments showing the good behaviour of these algorithms when compared with algorithms which ignore the matrix structure are also included.
arXiv:1811.08406v1 fatcat:xvld2sfrrrdpxpvrzdkfw7b3tq