A Black-Box Multigrid Preconditioner for the Biharmonic by Silvester D. J., Mihajlovic M. D.

By Silvester D. J., Mihajlovic M. D.

We research the convergence features of a preconditioned Krylov subspace solver utilized to the linear structures coming up from low-order combined finite point approximation of the biharmonic challenge. the foremost characteristic of our process is that the preconditioning could be learned utilizing any "black-box" multigrid solver designed for the discrete Dirichlet Laplacian operator. This results in preconditioned structures having an eigenvalue distribution together with a tightly clustered set including a small variety of outliers. Numerical effects express that the functionality of the method is aggressive with that of specialised quick new release tools which have been built within the context of biharmonic difficulties.

Show description

Read Online or Download A Black-Box Multigrid Preconditioner for the Biharmonic Equation PDF

Similar mathematics books

Complex Variables and Applications (9th Edition)

Complicated Variables and purposes, 9e will serve, simply because the past variations did, as a textbook for an introductory path within the thought and alertness of features of a fancy variable. This new version preserves the elemental content material and elegance of the sooner versions. The textual content is designed to advance the idea that's favorite in functions of the topic.

Additional resources for A Black-Box Multigrid Preconditioner for the Biharmonic Equation

Example text

Its determinant PM is thus well-defined, because K(X) is a commutative integral domain with a unit element. One calls PM the characteristic polynomial of M. Substituting 0 for X, one sees that the constant term in PM is simply (−1)n det M. The term corresponding to the permutation σ = id in the computation of the determinant is of degree n (it is ∏i (X − mii )) and the products corresponding to the other permutations are of degree less than or equal to n − 2, therefore one sees that PM is of degree n, with PM (X) = X n − n ∑ mii X n−1 + · · · + (−1)n det M.

Let β , β be two bases of E, and γ, γ two bases of F. Let M be the matrix of u associated with the bases (β , γ), and M be that associated with (β , γ ). Finally, let P be the matrix of the change of basis β → β and Q that of γ → γ . We have P ∈ GLm (K) and Q ∈ GLn (K). With obvious notations, we have n fk = ∑ qik fi , ej = i=1 We have u(e j ) = m ∑ p je . n n k=1 i,k=1 ∑ mk j fk = ∑ =1 mk j qik fi . On the other hand, we have u(e j ) = u m ∑ p je =1 = m m =1 i=1 ∑ p j ∑ mi fi . Comparing the two formulæ, we obtain m ∑ mi =1 p j= n ∑ qik mk j , ∀1 ≤ i ≤ n, 1 ≤ j ≤ m.

20 Let us assume that the characteristic polynomial of M ∈ Mn (K) splits over K. Then M decomposes as a sum M = D + N, where D is diagonalizable, N is nilpotent, DN = ND, and Sp(D) = Sp(M). In addition, D and N are unique, and are polynomials in M. The formula M = D + N is the Dunford decomposition of M. Proof. There remains to prove the uniqueness and polynomiality. Polynomiality. It is sufficient to prove that D and N are polynomials in D + N . This is done if we find one polynomial R, such that λ Inλ = R(λ Inλ +Nλ ) for every λ , because then we have D = R(M) and N = (X − R)(M).

Download PDF sample

Rated 4.98 of 5 – based on 44 votes