No label defined (Q4428070)

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 1979694
Language Label Description Also known as
default for all languages
No label defined
    English
    No label defined
    scientific article; zbMATH DE number 1979694

      Statements

      0 references
      14 September 2003
      0 references
      interval linear system
      0 references
      outer problem
      0 references
      algebraic approach
      0 references
      Newton method
      0 references
      interval arithmetic
      0 references
      numerical example
      0 references
      algorithm
      0 references
      computational efficiency
      0 references
      The classical ``outer'' problem for the interval linear system \({\mathbf A}x={\mathbf b}\) is under consideration, here \({\mathbf A}\) is an interval matrix, \({\mathbf b}\) is an interval vector, the problem is to find ``good'' coordinate-wise estimates for the set of all solutions of all linear systems \(Ax=b\) with a matrix \(A\in {\mathbf A}\) and a vector \(b\in {\mathbf b}\). It is known that to find a coordinate-wise estimates with any given ratio or absolute error and even to determine is there a solution or not are the problems of class~NP. However, there is a criterion for the existence of a unique algebraic solution in terms of the spectral radii. The author provides a new algebraic approach which is based on the extension of interval arithmetic to the algebraic system with better algebraic properties, in particular, it is a commutative group under the addition. The problem under consideration is reduced to the solving a non-interval equation in the Euclidean space of double dimension. A special algorithm (subdifferential Newton method with a special initial approximation) is constructed. It is proved that under some natural assumptions on the interval \({\mathbf A}\) this algorithm converges to the unique algebraic solution. A modified algorithm which gives estimates of the solution set in more general situation is given and tested. The results of its numerical tests are presented. They show that the proposed algorithm combines high computational efficiency with the good estimates of the solution set.
      0 references

      Identifiers

      0 references
      0 references
      0 references
      0 references