On the local convergence of inexact Newton-type methods under residual control-type conditions (Q711244)

From MaRDI portal





scientific article; zbMATH DE number 5804879
Language Label Description Also known as
default for all languages
No label defined
    English
    On the local convergence of inexact Newton-type methods under residual control-type conditions
    scientific article; zbMATH DE number 5804879

      Statements

      On the local convergence of inexact Newton-type methods under residual control-type conditions (English)
      0 references
      0 references
      25 October 2010
      0 references
      The article deals the following iterative method \[ x_{n+1} = x_n - B_n^{-1}(F(x_n) + r_n)\tag{1} \] of approximate solving a nonlinear operator equation \(F(x) = 0\) with an operator \(F\) between Banach spaces \(X\) and \(Y\). It is assumed that \(x^*\) is a simple zero of \(F\) and the following condition \[ \|F'(x^*)^{-1}(F'(x) - F'(y))\| \leq l_p\|x - y\|^p \quad x, y \in B(x^*,R), \] is true, \(B_n\) is a sequence of invertible linear operators between the Banach spaces \(X\) and \(Y\), and \(r_n\) a sequence of errors of calculations. The authors formulate a series of conditions under which the iteration (1) with an initial \(x_0 \in B(x^*,R)\) does not leave this ball \(B(x^*,R)\) and converges to \(x^*\) with an error estimate of type \(\|x_n - x^*\| = O(\|x_n - x^*\|^{1+\beta}\), where \(\beta \in [0,p]\). In the end of the article, the authors consider three numerical examples.
      0 references
      inexact Newton-type method
      0 references
      local convergence
      0 references
      radius of convergence
      0 references
      error estimates
      0 references
      Banach space
      0 references
      Hölder condition
      0 references
      numerical examples
      0 references

      Identifiers