Uniform convergence and preconditioning method for mortar mixed element method for nonselfadjoint and indefinite problems (Q1587963)

From MaRDI portal





scientific article; zbMATH DE number 1538640
Language Label Description Also known as
English
Uniform convergence and preconditioning method for mortar mixed element method for nonselfadjoint and indefinite problems
scientific article; zbMATH DE number 1538640

    Statements

    Uniform convergence and preconditioning method for mortar mixed element method for nonselfadjoint and indefinite problems (English)
    0 references
    0 references
    0 references
    0 references
    19 March 2001
    0 references
    The Dirichlet problem on a bounded polygonal domain \(\Omega\) in \(\mathbb{R}^2\) is considered for linear ellipic equation of general form with the right-hand term \(f\). The authors assume that the problem has a unique solution in \(H^1_0(\Omega)\) for any given \(f\in L_2(\Omega)\) -- the same is assumed for the dual problem as well. They give a review of grid methods with discontinuous approximations (mortar element methods) and pay special attention to the analysis of similar variants of mixed finite element methods ``in case of nonselfadjoint and indefinite problems''. The analysis is given only for sufficiently small \(h<h_0\) (such a condition is necessary) and leads to estimates \[ \|u-u_h\|_0+\|\sigma-\sigma_h\|_0\leq C\Biggl[\sum_{k}h_k^2(\|f\|^2_{L_2(\Omega_k)}+\|u\|_{H^2(\Omega_k)})\Biggr]^{1/2} \] if the partition \(\bar\Omega=\bar\Omega_1\cup\bar\Omega_2\) is used in the construction of the method; index 0 corresponds to the norm in \(L_2(\Omega)\), \(\sigma\equiv -(a\nabla u+bu)\). (The estimates are called uniform by the authors.) The authors considered also construction of preconditioners on the base of additive Schwarz method. It should be noted that in the study of numerical methods for general elliptic problems the most natural approach is connected with the classical Fredholm theory in the corresponding energy space. The conditions in the paper imply that the corresponding bounded operator \(L\) has \(\text{Ker} L=0\) and \(\|L^{-1}\|<\infty\). Hence, there is no need to make assumptions about \(L^*\). Also it enables one to obtain similar estimates for grid approximations and corresponding matrices (for \(h<h_0\)) in a fairly natural way -- it was done by the reviewer for difference and conformal finite element approximations in the seventies; the references, estimates of convergence and corresponding iterative methods with preconditioning can be found in the book \textit{E. G. D'yakonov} [Optimization in solving elliptic problems. Boca Raton, CRC Press (1966; Zbl 0852.65087)].
    0 references
    correct elliptic problems
    0 references
    mortar mixed element methods
    0 references
    additive Schwarz preconditioners
    0 references
    Dirichlet problem
    0 references
    linear elliptic equation
    0 references
    nonselfadjoint and indefinite problems
    0 references
    additive Schwarz method
    0 references

    Identifiers

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