Abstract
We develop a nonoverlapping Additive Schwarz Preconditioner for the p-version finite element with \(H^2\)-conforming elements on triangles based on Argyris elements. With a particular choice of basis functions corresponding to the \(C^2\) degrees of freedom (dofs), we give a preconditioner consisting of eliminating the interior dofs, global solves of the \(C^0\) and \(C^1\) vertex dofs, diagonal solves of the \(C^2\) vertex dofs, and block diagonal solves of the edge dofs. We show that the condition number of the preconditioner system grows at most like \({\mathscr {O}}(1+\log ^3 p)\) independent of the mesh size h. The analysis permits the use of inexact interior solves which lends itself to a more efficient implementation while maintaining the same \({\mathscr {O}}(1 + \log ^3 p)\) asymptotic growth of the preconditioned system.
Similar content being viewed by others
References
Ainsworth, M.: A preconditioner based on domain decomposition for h-p finite-element approximation on quasi-uniform meshes. SIAM J. Numer. Anal. 33(4), 1358–1376 (1996)
Ainsworth, M., Jiang, S.: Preconditioning the mass matrix for high order finite element approximation on triangles. SIAM J. Numer. Anal. 57(1), 355–377 (2019)
Ainsworth, M., Parker, C.: \(H^2\)-stable polynomial liftings on triangles. SIAM J. Numer. Anal. 58(3), 1867–1892 (2020)
Argyris, J.H., Fried, I., Scharpf, D.W.: The TUBA family of plate elements for the matrix displacement method. Aeronaut. J. 72(692), 701–709 (1968)
Babuška, I., Craig, A., Mandel, J., Pitkäranta, J.: Efficient preconditioning for the p-version finite element method in two dimensions. SIAM J. Numer. Anal. 28(3), 624–661 (1991)
Bartels, S.: Numerical Approximation of Partial Differential Equations, vol. 64. Springer, New York (2016)
Bernadou, M., Boisserie, J.M.: The Finite Element Method in Thin Shell Theory: Application to Arch Dam Simulations. Springer, New York (1982)
Bernardi, C., Dauge, M., Maday, Y.: Polynomials in the sobolev world (2007). https://hal.archives-ouvertes.fr/hal-00153795/file/BeDaMa07b.pdf
Bramble, J.H., Pasciak, J.E., Schatz, A.H.: The construction of preconditioners for elliptic problems by substructuring. I. Math. Comput. 47(175), 103–134 (1986)
Bramble, J.H., Zhang, X.: Multigrid methods for the biharmonic problem discretized by conforming \(C^1\) finite elements on nonnested meshes. Numer. Funct. Anal. Optim. 16(7–8), 835–846 (1995)
Cahn, J.W., Hilliard, J.E.: Free energy of a nonuniform system. I. interfacial free energy. J. Chem. Phys. 28(2), 258–267 (1958)
Chan, T.F., Mathew, T.P.: Domain decomposition algorithms. Acta numer. 3, 61–143 (1994)
Ciarlet, P.G.: The Finite Element Method for Elliptic Problems, vol. 40. Siam, Philadelphia (2002)
Girault, V., Raviart, P.A.: Finite Element Methods for Navier-Stokes Equations: Theory and Algorithms, vol. 5. Springer, New York (2012)
Grisvard, P.: Elliptic Problems in Nonsmooth Domains, vol. 69. SIAM, Philadelphia (2011)
Landau, L., Lifshitz, E., Sykes, J., Reid, W., Dill, E.H.: Theory of elasticity: vol. 7 of course of theoretical physics. Phys. Today 13, 44 (1960)
Morgan, J., Scott, R.: A nodal basis for \(c^1\) piecewise polynomials of degree \(n \ge 5\). Math. Comput. 29(131), 736–740 (1975)
Necas, J.: Direct Methods in the Theory of Elliptic Equations. Springer, New York (2011)
Pestana, J., Muddle, R., Heil, M., Tisseur, F., Mihajlovic, M.: Efficient block preconditioning for a \(C^1\) finite element discretization of the Dirichlet biharmonic problem. SIAM J. Sci. Comput. 38(1), A325–A345 (2016)
Raviart, P.A., Thomas, J.M.: A mixed finite element method for 2-nd order elliptic problems. In: Mathematical Aspects of Finite Element Methods, pp. 292–315. Springer, New York (1977)
Smith, B., Bjorstad, P., Gropp, W.: Domain Decomposition: Parallel Multilevel Methods for Elliptic Partial Differential Equations. Cambridge University Press, Cambridge (2004)
Suri, M.: The \( p \)-version of the finite element method for elliptic equations of order \(2 l\). ESAIM Math. Model. Numer. Anal-Modél. Math. Anal. Numérique 24(2), 265–304 (1990)
Szegö, G.: Orthogonal Polynomials. American Mathematical Soc, Rhode Island (1939)
Timoshenko, S.P., Woinowsky-Krieger, S.: Theory of Plates and Shells. McGraw-hill, New York (1959)
Toselli, A., Widlund, O.: Domain Decomposition Methods-Algorithms and Theory, vol. 34. Springer, New York (2006)
Zhang, X.: Studies in domain decomposition: multi-level methods and the biharmonic dirichlet problem. Ph.D. thesis, New York Univeristy: Courant Institute of Mathematical Sciences (1991)
Zhang, X.: Multilevel Schwarz methods for the biharmonic Dirichlet problem. SIAM J. Sci. Comput. 15(3), 621–644 (1994)
Zhang, X.: Two-level Schwarz methods for the biharmonic problem discretized conforming \(C^1\) elements. SIAM J. Numer. Anal. 33(2), 555–570 (1996)
Author information
Authors and Affiliations
Corresponding author
Additional information
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Appendices
A \(H^2\) Polynomial extensions
Let \(I = (0,1)\). We use the standard Sobolev spaces \(H^s(I)\), \(s \in {\mathbb {N}}\), where \(H^0(I) = L^2(I)\). For spaces of vector fields we use \({\varvec{H}}^s(I)\). We shall also need the fractional order spaces \(H^{1/2}(I)\), \(H^{1/2}_{00}(I)\), \(H^{3/2}(I)\), and \(H^{3/2}_{00}(I)\) equipped with the norms
The corresponding spaces \(H^{s}(\gamma )\) and \({\varvec{H}}^{s}(\gamma )\) are defined on an element edge \(\gamma \in \{ \gamma _1, \gamma _2, \gamma _3 \}\) in a similar fashion.
For a function f defined on \(\partial K\), we denote \(f_i = f|_{\gamma _i}\) where the edge \(\gamma _i\) is parameterized by arc length in the tangential direction. The following norm equivalence [15] holds on \(H^{1/2}(\partial K)\), the space traces of \(H^1(K)\) functions,
where
and \(\delta = \min _{i\in \{ 1,2,3 \} }(|\gamma _i|)\). We define \({\varvec{H}}^{1/2}(\partial K)\) analogously.
In order to show the existence of functions satisfying (3.7) and hence complete the proof of (4.2), we need the following consequence of [3, Corollary 2.3]:
Corollary A.1
Let \(f,g : \partial K \rightarrow {\mathbb {R}}\) be such that \(f_i \in {\mathbb {P}}_p(\gamma _i)\) and \(g_i \in {\mathbb {P}}_{p-1}(\gamma _i)\) for \(p \ge 1\). Assume further that f and g satisfy the compatibility conditions at the vertices:
for \(i \in \{1,2,3\}\). Then, there exists a polynomial \(U \in {\mathbb {P}}_p(K)\) such that
-
1.
U has trace (f, g): \(U |_{\gamma _i} = f_i\), \(\partial _n U |_{\gamma _i} = g_i\), and
-
2.
U depends continuously on f and g:
$$\begin{aligned} \begin{aligned}&h_K^{-1} \Vert U\Vert _{L^2(K)} + |U|_{H^1(K)} + h_K |U|_{H^2(K)} \\&\qquad \le C \left( h_K^{-1/2} \Vert f\Vert _{L^2(\partial K)} + h_{K}^{1/2} \Vert {\varvec{\sigma }}(f,g) \Vert _{L^2(\partial K)} + h_K|{\varvec{\sigma }}(f,g)|_{H^{1/2}(\partial K)} \right) \end{aligned} \end{aligned}$$(A.1)where \( {\varvec{\sigma }}(f,g) = (\partial _t f) {\varvec{n}} - g {\varvec{t}},\) and C is independent of p and \(h_K\).
Proof
The existence of a polynomial U satisfying the required conditions is given by [3, Corollary 2.3] where the estimate (A.1) is also established in the case \(h_K = 1\). Employing the usual scaling argument gives the estimate (A.1). \(\square \)
B Auxilliary univariate polynomials
The following lemma defines some univariate functions which are later used to construct \(C^2\) vertex functions satisfying the condition (3.7).
Lemma B.1
Let \(m,p \in {\mathbb {N}}\), \(\beta > -1\). Define \(\gamma := 2m-1+\beta \) and
Then, \(\psi _{p,m,\beta }^{(j)}(\pm 1) = 0\) for \(j = 0, 1, \ldots , m-2\), \(\psi _{p,m,\beta }^{(m-1)}(1) = 0\), and \(\psi _{p,m,\beta }^{(m-1)}(-1) \ne 0\). Moreover,
Proof
Let \(\psi _p := \psi _{p,m,\beta }\). We first bound
Now, using the identity [23, p. 4]:
we get
with
Using this identity and orthogonality, there holds
First note that
We now claim
For \(p = 0\), we have
and (B.2) holds as an equality in the case \(p = 0\).
Now suppose the result holds true for \(p = q-1\). Then, by (B.1),
Direct computation gives
and (B.2) follows by induction.
Again using identities from [23, p. 4]
Thus,
By Stirling’s formula, \(\frac{p!}{\varGamma (p+2\gamma +1)} \sim p^{-2(2m-1+\beta )}\). \(\square \)
C Traces of \(C^2\) vertex functions
Thanks to Corollary A.1, we only need to specify the boundary values of \(\varPhi _{n_i}^{\alpha }\) on each \(K \in {\mathscr {T}}\). We summarize them below: First, we label a general element K as in Fig. 1. Let \(F : [-1,1] \rightarrow {\mathbb {R}}\) and \(G : [-1,1] \rightarrow {\mathbb {R}}\) be given by
where \(\psi _{p,m,\beta }\) is given by Lemma B.1. Now, we define the traces f and g as follows: \(\alpha = (2,0)\):
\(\alpha = (1,1)\):
\(\alpha = (0,2)\):
Since F and G vanish to order 2 and 1, respectively, f and g satisfy the compatibility conditions in Corollary A.1. Let \({\bar{\varPhi }}_{n_i}^{\alpha }\) be the extension of f and g given by Corollary A.1. First note that we have the desired interpolation properties:
for all \(|\beta | \le 2\) and \(n_j \in {\mathscr {N}}\). Moreover, for \(m \in \{0,1,2\}\) there holds,
Now, thanks to inverse estimates [8] and Lemma B.1, we have
By interpolation,
Rights and permissions
About this article
Cite this article
Ainsworth, M., Parker, C. Preconditioning high order \(H^2\) conforming finite elements on triangles. Numer. Math. 148, 223–254 (2021). https://doi.org/10.1007/s00211-021-01206-7
Received:
Revised:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s00211-021-01206-7