Fictitious component method of solving the schemes of the finite element method for elliptic boundary value problems with nonlocal boundary conditions in multiply connected domains (Q4222222)

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 1229815
Language Label Description Also known as
default for all languages
No label defined
    English
    Fictitious component method of solving the schemes of the finite element method for elliptic boundary value problems with nonlocal boundary conditions in multiply connected domains
    scientific article; zbMATH DE number 1229815

      Statements

      Fictitious component method of solving the schemes of the finite element method for elliptic boundary value problems with nonlocal boundary conditions in multiply connected domains (English)
      0 references
      0 references
      0 references
      25 November 1999
      0 references
      multiply connected domains
      0 references
      fictitious domain
      0 references
      iterative solution
      0 references
      finite elements
      0 references
      fictitious component method
      0 references
      elliptic boundary value problem
      0 references
      convergence
      0 references
      The authors are interested in the elliptic boundary value problem in a multiply-connected two-dimensional region \(\Omega\). The boundary of \(\Omega\) consists of several disjoint curves \(S_0, S_1, \ldots, S_m\), and boundary conditions for the unknown function \(u\) are given as \(u\bigl|_{S_0}=0\bigr.\); \(\int_{S_k}\partial u/\partial n\;ds=g_k\) and \(u\bigl|_{S_k}=u_k\bigr.\) for \(k=1,\ldots,m\), where \(u_k\) are unknown constants and \(g_k\) are specified constants. The problem is discretized using piecewise linear triangular finite elements on a mesh equivalent to a union of squares, each one of which is a union of two right triangles. Formulating the boundary conditions requires coupling mesh points that may not lie on adjacent finite elements. The so-called fictitious component method consists of embedding the region \(\Omega\) in a larger rectangular region \(\Omega_0\). The coefficients of the original problem are extended to \(\Omega_0\) in a natural manner, and the solution \(u\) is extended as zero outside \(\Omega\) and as constant inside the boundary curves \(S_k\). An elliptic boundary value problem is thus constructed on \(\Omega_0\) and this problem does not involve coupling mesh points other than those in adjacent finite elements. The resulting matrix equation has smaller bandwidth and can be easier to solve than the original matrix equation. Denote by \(A\) the matrix associated with the original problem but expressed on \(\Omega_0\), and by \(\Lambda\) the matrix associated with the problem extended to the region \(\Omega_0\). The authors formulate an iterative method based on solution of matrix equations involving \(\Lambda\) on the left side and \(A\) on the right. Convergence of this iterative method depends on ratios of eigenvalues of \(A\) and \(\Lambda\) that the authors prove using ratios of quadratic forms.
      0 references

      Identifiers

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