Stable and Efficient Algorithms for Structured Systems of Linear Equations (Q4389075)

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 1150977
Language Label Description Also known as
default for all languages
No label defined
    English
    Stable and Efficient Algorithms for Structured Systems of Linear Equations
    scientific article; zbMATH DE number 1150977

      Statements

      Stable and Efficient Algorithms for Structured Systems of Linear Equations (English)
      0 references
      0 references
      11 May 1998
      0 references
      structured systems
      0 references
      displacement equation
      0 references
      error analysis
      0 references
      fast algorithm
      0 references
      Toeplitz matrix
      0 references
      Hankel matrix
      0 references
      Cauchy-like matrix
      0 references
      Gaussian elimination
      0 references
      pivoting
      0 references
      numerical experiments
      0 references
      A Cauchy-like matrix has the form \[ C= \Biggl({a^T_i b_j\over \omega_i-\lambda_j}\Biggr)_{1\leq i,j\leq n}\qquad (a_i, b_j\in\mathbb{R}^\alpha), \] where \(\alpha\ll n\). It satisfied a so-called displacement equation, i.e., \[ \Omega C- C\Lambda= A\cdot B\qquad (A\in\mathbb{R}^{n\times\alpha}, B\in\mathbb{R}^{\alpha\times n}). \] The author studies variations of a fast Gaussian elimination algorithm with partial or complete pivoting for Cauchy-like matrices, working systematically with the displacement equation. He describes algorithms that avoid internal element growths and therefore have better backward error properties than the existing methods. The solution of Toeplitz, Hankel and Toeplitz-plus-Hankel matrices can be reduced (in more than one way) to solutions of linear systems with a Cauchy-like matrix. The author discusses implementation techniques that further reduce execution. He presents a careful error analysis, flow counts and extensive numerical experiments that show the superiority of the newly developed methods.
      0 references

      Identifiers