No label defined (Q4406446)

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 1936602
Language Label Description Also known as
default for all languages
No label defined
    English
    No label defined
    scientific article; zbMATH DE number 1936602

      Statements

      0 references
      2002
      0 references
      linear ill-posed problems
      0 references
      collocation method
      0 references
      parameter estimation problems
      0 references
      boundary integral method
      0 references
      Laplace equation
      0 references
      regularization methods
      0 references
      spline functions
      0 references
      singular integral equation
      0 references
      inverse problem
      0 references
      recursive smoothing method
      0 references
      parabolic equation
      0 references
      algorithm
      0 references
      convergence
      0 references
      numerical experiments
      0 references
      For the solution of linear ill-posed problems with a perturbed right-hand side and an approximately given operator, a collocation scheme with spaces of spline functions is considered. The author provides conditions which guarantee the nonsingularity of the collocation matrix. For the construction of the right-hand side of the collocation equations a recursive smoothing method is proposed, where the number of smoothing steps is considered as a regularization parameter. As an example a two-dimensional integral equation of the first kind with a singular kernel is considered, with an application to a boundary integral approach for the Laplace equation with Dirichlet and Neumann data. In the second part of the paper, a parameter estimation problem for a two-dimensional parabolic equation of the form \[ \frac{\partial}{\partial x} \Big( T(x,y) \frac{\partial q(x,y,t)}{\partial x}\Big) + \frac{\partial}{\partial y} \Big( T(x,y) \frac{\partial q(x,y,t)}{\partial y}\Big) + P(x,y) \frac{\partial q(x,y,t)}{\partial t} = Q(x,y,t) \tag{*} \] is considered, where \( t \) belongs to a bounded time interval, and \( (x,y) \) belongs to some rectangle \(\Omega_0 \subset \mathbb{R}^2\). An algorithm is considered which consists of three steps. At first for a finite number of times \( t \) the given perturbed values of \(q\) and \( Q \) are interpolated in \( \Omega_0 \) to obtain data on a uniform grid in \( \Omega_0 \), respectively. Secondly, for each of the considered finite number of times \( t \), a recursive smoothing method with cubic splines is applied to the data generated on the uniform grid in \( \Omega_0 \), respectively. Finally, at each point \( (x,y) \) on the uniform grid, approximations for \( \frac{\partial T(x,y)}{\partial x}, \frac{\partial T(x,y)}{\partial y}, T(x,y) \) and \( P(x,y) \) are obtained by considering equation \( (*) \) with the smoothed data corresponding to \( (x,y) \) and for four different time levels. Convergence of this algorithm is shown and numerical experiments are included.
      0 references

      Identifiers

      0 references
      0 references
      0 references
      0 references
      0 references
      0 references
      0 references
      0 references
      0 references
      0 references
      0 references