Errors and perturbations in Vandermonde systems (Q4024983)

From MaRDI portal

!

This is the item page for this Wikibase entity, intended for internal use and editing purposes. Please use the normal view instead:

scientific article; zbMATH DE number 120278
Language Label Description Also known as
default for all languages
No label defined
    English
    Errors and perturbations in Vandermonde systems
    scientific article; zbMATH DE number 120278

      Statements

      Errors and perturbations in Vandermonde systems (English)
      0 references
      22 February 1993
      0 references
      effective well-conditioning
      0 references
      Vandermonde systems
      0 references
      UL factorization
      0 references
      Vandermonde matrix
      0 references
      numerical tests
      0 references
      perturbations
      0 references
      0 references
      Although Vandermonde systems of equations can be very poorly conditioned, the algorithm of \textit{Å. Björck} and \textit{V. Pereyra} [Math. Comput. 24(1970), 893-903 (1971; Zbl 0221.65054)] may still produce very accurate results. The algorithm is essentially a \(LU\) factorization of the inverse of the \(n\times n\) Vandermonde matrix \(V\) (the \((i,j)\) element of \(V\) is \(\alpha_ j^{i-1}\)). In a recent series of papers, starting with [Numer. Math. 50, 613-632 (1987; Zbl 0595.65029)], \textit{N. J. Higham} has shown that the error in the computed solution can be small when the \(\{\alpha_ j\}\) are positive and in increasing order. The current author shows that Higham's analysis provides a good predictor of more general error behaviour and illustrates this result with a large set of numerical tests. The computational error is related to that caused by perturbations in the matrix elements. The relationship between these and the ``effective well-conditioning'' of \textit{T. F. Chan} and \textit{D. E. Foulser} [SIAM J. Sci. Stat. Comput. 9, No. 6, 963-969 (1988; Zbl 0664.65041)], who pointed out that the sensitivity of the solution \(x\) of a linear system \(Ax=b\) to perturbations in \(b\) does depend on \(b\), is also discussed.
      0 references

      Identifiers