Optimal multilevel preconditioning of strongly anisotropic problems.part II: non-conforming FEM. p. 1/36

Size: px
Start display at page:

Download "Optimal multilevel preconditioning of strongly anisotropic problems.part II: non-conforming FEM. p. 1/36"

Transcription

1 Optimal multilevel preconditioning of strongly anisotropic problems. Part II: non-conforming FEM. Svetozar Margenov Institute for Parallel Processing, Bulgarian Academy of Sciences, Acad. G. Bonchev Str. Bl. 25-A, 1113 Sofia, Bulgaria Optimal multilevel preconditioning of strongly anisotropic problems.part II: non-conforming FEM. p. 1/36

2 CONTENTS 1. Introduction 2. Non-conforming Crouzeix-Raviart FEs 3. Two-level hierarchical decompositions 4. Optimal preconditioner for the block B 11 (FR) 5. Optimal preconditioner for the block Ã11 (DA) 6. Concluding remarks Optimal multilevel preconditioning of strongly anisotropic problems.part II: non-conforming FEM. p. 2/36

3 1. Introduction We study strategies to construct hierarchical basis functions (HB) multilevel preconditioners to solve algebraic systems arising from second order elliptic problems, discretized by the non-conforming Crouzeix-Raviart finite elements. To this end we follow the known framework for constructing HB preconditioners for conforming FEM. However, applying the latter framework and the existing theoretical results to non-conforming FEM is not straightforward as the classical construction of hierarchical preconditioner relies on a nested sequence of finite element spaces, in most cases related to nested grids, while: The non-conforming FEM on nested grids produces non-nested FE spaces. Optimal multilevel preconditioning of strongly anisotropic problems.part II: non-conforming FEM. p. 3/36

4 2. Crouzeix-Raviart FEs Consider the selfadjoint elliptic boundary value problem Lu (a(x) u(x)) = f(x) in Ω, u(x) = 0 on Γ D, (a(x) u(x)) n = 0 on Γ N, where Ω is a polygonal domain in R 2, f(x) is a given function in L 2 (Ω), a(x) = [a ij (x)] 2 i,j=1is a symmetric and uniformly positive definite matrix in Ω, n is the unit vector of outward normal to the boundary Γ = Ω, and Γ = Γ D Γ N. We assume that the entries a ij (x) are piece-wise smooth functions on Ω = Ω Ω. Optimal multilevel preconditioning of strongly anisotropic problems.part II: non-conforming FEM. p. 4/36

5 The weak formulation of the above problem reads as follows: for some given f, find u V H 1 D (Ω) = {v H1 (Ω) : v = 0 on Γ D }, which satisfies A(u, v) = (f, v) v H 1 D(Ω), where A(u, v) = Ω a(x) u(x) v(x)dx. Consider two partitionings of Ω: a coarse triangulation T H and a fine one T h, which is obtained by a regular refinement of T H. The partitioning T H is assumed to be aligned with the discontinuities of the coefficient a(x) so that over each element E T H the function a(x) is smooth. Optimal multilevel preconditioning of strongly anisotropic problems.part II: non-conforming FEM. p. 5/36

6 course nodes coarse nodes fine nodes fine nodes Crouzeix-Raviart non-conforming FEs We discretize the variational problem using the Crouzeix-Raviart FEs, i.e., we seek the solution in the finite dimensional space V h = {v L 2 (Ω) : v e is linear e T h, v is continuous at the midpoints of the edges ofe T h and v is zero at the midpoints on Γ D }. The nodal FE basis φ h i (i = 1,..., n h) in V h is naturally defined as φ h i being equal to unity at one midpoint m k in e and zero at the other two midpoints. Optimal multilevel preconditioning of strongly anisotropic problems.part II: non-conforming FEM. p. 6/36

7 The discrete formulation becomes: find u h V h, which satisfies A : 0 h (u h, v h ) = (f, v h ) v h V h, where A h (u h, v h ) = e T h e a(e) u h v h dx. Here a(e) is a piece-wise constant coefficient matrix, defined by the integral averaged values of a(x) over each triangle from the coarser triangulation T H. Let us note that in this way arbitrary large coefficient jumps across the boundaries between adjacent finite elements from T H are allowed. Using the nodal basis, the FEM problem reads as A h u h = b h, u h, b h R n h where A h = (a ij ), is the global stiffness matrix with entries a ij = A h (φ h i, φh j ). Optimal multilevel preconditioning of strongly anisotropic problems.part II: non-conforming FEM. p. 7/36

8 We pose no restrictions on the mesh and/or coefficient anisotropy. As it is known, to derive estimates for the CBS constant γ, it suffices to consider an isotropic problem in an arbitrary shaped triangle, T. Let us denote the angles in T by θ 1, θ 2 and θ 3 = π (θ 1 + θ 2 ), where a = cot θ 1, b = cot θ 2 and c = cot θ 3. Without loss of generality, for each triangle T, we assume that θ 1 θ 2 θ 3, and then denote α = a/c and β = b/c. A simple computation shows that the standard nodal basis element stiffness matrix for Crouzeix-Raviart non-conforming linear elements A CR e coincides with that for the conforming linear elements A cl e, up to a scalar factor 4: A CR e = 2 b + c c b 1 + β 1 β c a + c a = 2c α α. b a a + b β α α + β Optimal multilevel preconditioning of strongly anisotropic problems.part II: non-conforming FEM. p. 8/36

9 3. Two-level decompositions Consider now the two consecutive mesh refinements, T H and T h. To build a hierarchical preconditioner, we need a suitable decomposition of V h, V h = V 1 V 2. In the case of conforming FEM, one of the spaces (V 2 ) is naturally induced by the coarse mesh. For non-conforming FEM, V H V h and the direct construction with, say, V 2 V H is impossible. To overcome the latter difficulty, we consider two constructions of suitable hierarchical decompositions of the Crouzeix-Raviart spaces were introduced. Following the introduced terminology, Blaheta, Margenov, Neytcheva (2004), we refer them as: Two-level first reduce decomposition (FR); Two-level decomposition with differences and aggregates (DA). Optimal multilevel preconditioning of strongly anisotropic problems.part II: non-conforming FEM. p. 9/36

10 3.1. Two-level FR decomposition 7 II θ I θ 1 θ 2 8 III 9 Crouzeix-Raviart macro-element On macroelement level, we have the space V (E) = span {φ 1,..., φ 9 }. Let the basis functions φ i correspond to the midpoints m i, ordered as shown in the Figure. The FR splitting V (E) = V 1 (E) V 2 (E) is defined as follows V 1 (E) = span {φ 1, φ 2, φ 3, φ 4 φ 5, φ 6 φ 7, φ 8 φ 9 }, V 2 (E) = span {φ 4 + φ 5, φ 6 + φ 7, φ 8 + φ 9 }. Optimal multilevel preconditioning of strongly anisotropic problems.part II: non-conforming FEM. p. 10/36

11 Using the transformation matrix J F R, J F R = 1 2 2I J 1 J 2, J 1 = , J 2 = , 1 1 we introduce a corresponding hierarchical basis ϕ E = { φ i } 9 i=1 = J F R ϕ E. The hierarchical macroelement stiffness matrix is then computed as  E = J F R A E J T F R, and the related global stiffness matrix is obtained as Âh = E T ÂE. Optimal multilevel preconditioning of strongly anisotropic problems.part II: non-conforming FEM. p. 11/36

12 The hierarchical stiffness matrix Âh admits the 2 2 block structure  h = Â11  12  21  22 } V 1 } V 2, where V 1, V 2 are associated with the locally introduced red FR splitting. The matrix  h can be also seen as having a block 3 3 structure:  h = Ā 11 Ā (0) 12 Ā (0) 21 Ā (0) 31 Ā (0) 22 Ā (0) 32 Ā (0) 13 Ā (0) 23 Ā (0) 33 } interior basis functions } half-difference basis functions } half-sum basis functions ( V 2 ) Optimal multilevel preconditioning of strongly anisotropic problems.part II: non-conforming FEM. p. 12/36

13 For our purposes, however,  h is first decomposed into a 2 2 block form  h = Ā11 Ā 12 Ā 21 Ā 22 } interior basis functions, } rest. As a first step of the FR algorithm, the interior unknowns are eliminated and  h is reduced to its Schur complement B = Ā22 Ā21Ā 1 11 Ā12. Next we consider a two-level splitting of the matrix B, again in a block 2 2 form B = B 11 B 12 B 21 B 22 where the first block corresponds to the half-difference and the second block corresponds to the half-sum basis functions. The 2 2 block decomposition of B can now be used to construct two-level preconditioners, since the matrix B 22 is associated with the coarse grid.,. Optimal multilevel preconditioning of strongly anisotropic problems.part II: non-conforming FEM. p. 13/36

14 3.2. CBS constant of FR algorithm Let us consider the local eigenvalue problem S E v = λb E,22 v, v const = (c, c, c) T, c 0, where S E = B E,22 B E,21 B 1 E,11 B E,12. The minimum eigenvalue of B 1 E,22 S E is found to be of the form λ min (B 1 E,22 S E) = 5σ σ(σ 8αβ), σ = (α + 1)(β + 1)(α + β). 8σ The local problem is associated with the angles of the current triangle T, namely θ 1, θ 2 and θ 3 = π (θ 1 + θ 2 ), where a = cot θ 1, b = cot θ 2 and c = cot θ 3, assuming that θ 1 θ 2 θ 3, and then α = a/c and β = b/c. Optimal multilevel preconditioning of strongly anisotropic problems.part II: non-conforming FEM. p. 14/36

15 To estimate γ 2 E, we substitute σ, α, β by c i = cos θ i, using meanwhile s i = sin θ i, i = {1, 2, 3}. Applying a = (1 bc)/(b + c) in the form of ab + bc + ca = 1 we get (a + c)(b + c) = 1 + c 2 and therefore σc 3 = (1 + c 2 )(a + b). By definition, a + b = c 1 s 1 + c 2 s 2 = c 1s 1 +c 2 s 1 s 1 s 2 = s 3 s 1 s c 2 = 1 + c2 3 s 2. 3 Then σc 3 = 1 s 1 s 2 s 3 αβ σ = ab c 2 σ = abc c 3 σ = c 1c 2 c 3 s 1 s 2 s 3 c 3 σ =c 1c 2 c 3, that is γ 2 E= c1 c 2 c 3. Since c 1 c 2 c 3 > 1, 1 8c 1 c 2 c 3 < 9 which simply leads to γ 2 E < 3 4. Theorem 3.1. (Blaheta, Margenov, Neytcheva (2004)) The related FR constant in the strengthened CBS inequality is uniformly bounded with respect to both coefficient and mesh anisotropy, i.e., γ 2 < 3 4. Optimal multilevel preconditioning of strongly anisotropic problems.part II: non-conforming FEM. p. 15/36

16 3.2. Two-level DA decomposition This decomposition is referred to as (D)ifferentiation and (A)ggregation splitting. If φ 1,..., φ 9 are the standard nodal non-conforming linear finite element basis functions on the macroelement, then we define V (E) = span {φ 1,..., φ 9 } = V 1 (E) V 2 (E), V 1 (E) = span {φ 1, φ 2, φ 3, φ 4 φ 5, φ 6 φ 7, φ 8 φ 9 }, V 2 (E) = span {φ 1 + φ 4 + φ 5, φ 2 + φ 6 + φ 7, φ 3 + φ 8 + φ 9 }. II 7 θ θ 1 θ 2 8 III I 9 Optimal multilevel preconditioning of strongly anisotropic problems.part II: non-conforming FEM. p. 16/36

17 The related matrix J transforms the macroelement stiffness matrix into a hierarchical form à E = J E A E J T E = ÃE,11 à E,12 à E,21 à E,22 φ i V 1 (E) φ i V 2 (E). For the whole finite element space V h with the standard nodal finite element basis ϕ = {φ (i) h : i = 1,..., N h}, we can similarly construct a new hierarchical basis ϕ = ϕ 1 ϕ 2 ϕ 3 and a corresponding splitting V h = V 1 V 2, (i) V 1 = span{ φ h ϕ (i) 1 ϕ 2 }, V 2 = span{ φ h ϕ 3}, and respectively à h = JA h J T = Ã11 à 12 à 21 à 22 φ i V 1 φ i V 2. Optimal multilevel preconditioning of strongly anisotropic problems.part II: non-conforming FEM. p. 17/36

18 Again, the analysis of the related two-level method is performed locally, by considering the corresponding problems on macroelements. The obtained result is summarized in the following theorem, which is analogous to the estimate in the case of FR decomposition. Theorem 3.2. (Blaheta, Margenov, Neytcheva (2004)) Let us consider the two level DA splitting. Then the CBS constant is uniformly bounded with respect to both coefficients and mesh anisotropy, γ 2 3/4. The latter estimate is independent on the discretization (mesh) parameter h and possible coefficient jumps aligned with the finite element partitioning T H. Optimal multilevel preconditioning of strongly anisotropic problems.part II: non-conforming FEM. p. 18/36

19 Proof T T The reference coarse grid triangle and the macroelement Ê. Let V 1 (Ê), V 2 (Ê) be the two-level splitting for the reference macroelement and for u V 1 (Ê), v V 2 (Ê) denote d (k) = u Tk, δ (k) = v Tk. Then the relations between the function values in some nodal points, namely u(p 4 ) = u(p 5 ), u(p 6 ) = u(p 7 ), u(p 8 ) = u(p 9 ) and v(p 1 ) = v(p 4 ) = v(p 5 ), v(p 2 ) = v(p 6 ) = v(p 7 ), v(p 3 ) = v(p 8 ) = v(p 9 ), imply that d (1) + d (2) + d (3) + d (4) = 0, δ (1) = δ (2) = δ (3) = δ (4) = δ. T T Optimal multilevel preconditioning of strongly anisotropic problems.part II: non-conforming FEM. p. 19/36

20 Hence, A h, Ê (u, v) = 4 k=1 T k a u vdx = 4 k=1 = aδ, d (1) + d (2) + d (3) d (4) aδ (k), d (k) = 2 aδ, d (4) 2 δ a d (4) a where = area(t k ), x, y = x T y, and x a = ax, x. Further, 3 d (4) 2 a = d (1) + d (2) + d (3) 2 a 3 d (k) 2 a leads to and Thus, 4 A h,ê (u, u) = k=1 A h,ê (u,v) A h,ê (u,u) d (k) 2 a ( A h, Ê (v, v) = 4 δ 2 a. k= A h,ê (v,v)= 4 ) d (4) 2 a A h,ê (u,u) A h,ê (v,v). Optimal multilevel preconditioning of strongly anisotropic problems.part II: non-conforming FEM. p. 20/36

21 The following theorem is useful for extending the two-level to multilevel case. Theorem 3.3. (Blaheta, Margenov, Neytcheva (2004)) Let Ã22 be the stiffness matrix corresponding to the space V 2 with the basis ϕ 3 from the splitting DA and let A H be the stiffness matrix corresponding to the coarse discretization T H FE space V H, equipped with the standard nodal finite element basis {φ (k) H Proof. : k = 1,..., N H}. Then à 22 = 4 A H. Consider the nodal basis function φ (i) H V H and DA basis function Let both basis functions be equal to unity in the nodes belonging to the C-edges. Then for any macroelement E = 4 k=1 T k we get φ (i) h φ 3. d (1) i = d (2) i = d (3) i = d (4) i = 2 φ (i) H, where d(k) i = φ (i) h T k. Optimal multilevel preconditioning of strongly anisotropic problems.part II: non-conforming FEM. p. 21/36

22 4. Preconditioning of B 11 (FR) Recall that the top-left block of Ā 11 in FR algorithm is block-diagonal. After the elimination, one obtains a (6 6) element matrix B E, which constitutes the macroelement contribution to the matrix B. Next we split B E as: B E = B E,11 B E,12 B E,21 B E,22 }two-level half-difference basis functions }two-level half-sum basis functions The matrix block B E,11 in is found explicitly, namely, B E,11 = 2p q 3 q + 2(1 + β + β 2 ) q + 2 q 2 β 2 q q + 2(1 + α + α 2 ) q 2 α 2, q 2 β 2 q 2 α 2 3 q + 2 (α 2 + αβ + β 2 ) q = α + αβ + β, and p = 3(α + αβ + β) + 3(α 2 + αβ + β 2 ) + αβ(3α + 3β + 1). Optimal multilevel preconditioning of strongly anisotropic problems.part II: non-conforming FEM. p. 22/36

23 The following relations are known to hold or can be derived from Lemma 2.1. (r1) α + αβ + β = α(β + 1) + β = 1 c 2 0, c = cot θ 3, i.e., αβ α + β; (r2) α 2 + αβ + β 2 = α 2 + β(α + β) 0. Consider the off-diagonal elements of B E,11. It is easy to observe that 1. B E,11 (1, 2) = α + αβ + β + 2 > 0 for all valid values of α and β 2. B E,11 (1, 3) = α αβ β 2 β 2 < 0 3. B E,11 (2, 3) = α αβ β 2 α 2 < 0 and 4. B E,11 (2, 3) B E,11 (1, 3) B E,11 (1, 2), and B E,11 (1, 3) = B E,11 (1, 2) only for β = 1. Remark 3.1. Note that the direction of the strongest off-diagonal coupling of the macroelement matrix B E,11 is the same as of the related conforming FE. Optimal multilevel preconditioning of strongly anisotropic problems.part II: non-conforming FEM. p. 23/36

24 4.1. Additive preconditioner θ 3 2 e θ 1 3 θ E 3 1 (a) (b) Q (c) Figure: Dominating off-diagonal couplings for (a) the element stiffness matrix corresponding to e T h, (b) the macroelement matrix B E,11,and (c) the macroelement matrix B Q,11. Optimal multilevel preconditioning of strongly anisotropic problems.part II: non-conforming FEM. p. 24/36

25 We construct the additive preconditioner C E,11 to B E,11 by deleting the weakest off-diagonal couplings in B E,11, i.e., we let C E,11 = 2p q 3 q + 2(1 + β + β 2 ) q q q + 2(1 + α + α 2 ) q + 2 (α 2 + αβ + β 2 ) Then C 11 is obtained by assembling the modified element matrices C E,11. Lemma 3.1. For any element size and shape and for any anisotropy a(x), there holds that ( 1 ) 7/15 C E,11 B E,11 ( 1 + ) 7/15 C E,11. Optimal multilevel preconditioning of strongly anisotropic problems.part II: non-conforming FEM. p. 25/36

26 Proof. Consider the generalized eigenvalue problem B E,11 v = λc E,11 v and the corresponding characteristic equation for λ, det (B E,11 λc E,11 ) = 0. The determinant is found to be (1 λ)(3 q + 2(1 + β + β 2 )) (1 λ)(q + 2) q 2 β 2 (1 λ)(q + 2) (1 λ)(3 q + 2(1 + α + α 2 )) q 2 α 2. q 2 β 2 q 2 α 2 (1 λ)(3 q + 2 (α 2 + αβ + β 2 )) Straightforward computation shows that µ i = 1 λ i, i = 1, 2, 3 satisfy µ 1 = 0, i.e., λ 1 = 1 µ 2 2,3 = (α + β)(α + αβ + β + 2(2α2 αβ) + β 2 ) (α + β + 2)[2(α 2 + αβ + β 2 ) + 3(α + αβ + β)]. Optimal multilevel preconditioning of strongly anisotropic problems.part II: non-conforming FEM. p. 26/36

27 We show below that which expanded form becomes µ 2 2,3 7 15, E(α, β) 16α 3 +16β 3 34α 2 34β 2 34α 2 β 34αβ 2 82αβ 42α 42β 0 (α, β) D = { (α, β) : 1 2 < α 1, 0 < β 1, α + β > 0, α β}. Case 1: Let α = 0. In this case E(0, β) 16β 2 34β 2 42β = 18β 2 42β 0. Case 2: Let α > 0. Then 16α β 3 34β 2 16β β 2 34β 2 0 and the remaining terms in E are negative, so E 0. Optimal multilevel preconditioning of strongly anisotropic problems.part II: non-conforming FEM. p. 27/36

28 Case 3: Let 1 2 < α < 0. We use the fact that β 1, i.e., β2 β and that αβ α + β. E = 16α β 3 34α 2 34β 2 34αβ(α + β) 82αβ 42(α + β) 16α β 3 14αβ 42(α + β) 16(α 3 + β 3 ) + 14(α + β) 42(α + β) = 16(α + β)(α 2 αβ + β 2 ) 28(α + β) It remains to prove that 16(α 2 αβ + β 2 ) 28 or α 2 αβ + β 2 7/4. The latter is true, since sup α,β (α 2 αβ + β 2 ) = sup (α 2 α + 1) = 7/4. α ( 1/2,0) E achieves its maximum value 0 for (α, β) equal to (0, 0) and ( 1/2, 1). Optimal multilevel preconditioning of strongly anisotropic problems.part II: non-conforming FEM. p. 28/36

29 The assembled global preconditioner C 11 inherits the properties of C E,11. We collect the results in the following theorem. Theorem 3.1. By deleting the two smallest off-diagonal elements in the local matrix B E,11 and assembling the modified element matrix, we construct a preconditioner C 11 to B 11 such that ( 1 ) 7/15 C 11 B 11 ( 1 + ) 7/15 C 11, κ(c 1 11 B 11) < 1 4 ( 11 + ) The spectral equivalence and the related condition number estimate hold independently of element size and shape, and problem anisotropy as well. Optimal multilevel preconditioning of strongly anisotropic problems.part II: non-conforming FEM. p. 29/36

30 4.2. Multiplicative preconditioner Following the construction first proposed by Margenov, Vassilevski (1994) (see Axelsson, Margenov (2003) for the complete analysis in the case of conforming FE), we partition the nodes in the block B 11 into two groups, where the first one contains the centers of parallelogram superelements Q which are weakly connected in the sense that the off-diagonal couplings are relatively small. With respect to this partitioning, B 11 admits the two-by-two block-factored form B 11 = D 11 F 11 T F 11 E 11 = D 11 0 T F 11 S 11 I D 11 1 F 11 0 I. Then, the multiplicative preconditioner C 11 is defined as C 11 = D 11 0 T F 11 E 11 I D 11 1 F 11 0 I. Optimal multilevel preconditioning of strongly anisotropic problems.part II: non-conforming FEM. p. 30/36

31 Theorem 3.2. The multiplicative preconditioner C 11 of B 11 has an optimal order convergence rate with a relative condition number uniformly bounded by Proof. κ ( C 11 1 B 11 ) < 15 8 = We again consider the generalized eigenvalue problem S 11:Q v Q = λ Q E 11:Q v Q. As it is seen from the analysis in the case of conforming FEM, λ (2) Q = λ(3) Q = λ(4) Q = 1, and λ(1) Q = 1 (µ (2,3)) 2. Here µ (2,3) = 1 λ (2,3) stand for the eigenvalues introduced in the analysis of the additive preconditioner to B 11. This immediately gives ( µ (2,3)) 2 7 < 13 λ (1) Q > 8 15 after what the proof of the theorem follows straightforwardly. Optimal multilevel preconditioning of strongly anisotropic problems.part II: non-conforming FEM. p. 31/36

32 4.3. Solvers for C 11 The computational complexity of solving systems with C 11 is determined by their connectivity pattern only. This means, that the optimal order direct solver considered in the case of conforming FEM are directly applicable here. For the additive algorithm, the matrix C 11 has a generalized tridiagonal structure, that is, it is tridiagonal under a proper ordering of the elements. For the multiplicative preconditioner of B 11, the optimal order direct solver incorporates the nested dissection (ND) algorithm for the reduced system. Optimal multilevel preconditioning of strongly anisotropic problems.part II: non-conforming FEM. p. 32/36

33 5. Preconditioner of Ã11 (DA) The block à 11 can be further decomposed by splitting the unknowns into two groups - associated with the interior (I), and associated with the sides (S) of the macroelements, à 11 = Ã11,II à 11,SI à 11,IS à 11,SS Elimination of the block Ã11,II corresponding to the inner nodes gives rise to the Schur complement. S = à 11,SS à 11,SI à 1 11,IIÃ11,IS S = B 11, where B 11 is the block appearing in the FR decomposition. Therefore, the optimal preconditioners C 11 of B 11 are directly applicable to construct C 11, namely C 11 = I 0 à 11,SI à 1 11,II I Ã11,II à 11,IS 0 C 11. Optimal multilevel preconditioning of strongly anisotropic problems.part II: non-conforming FEM. p. 33/36

34 Theorem 3.3. For any element size and shape and any problem anisotropy it holds that Proof. κ ( ) C 11 1 Ã11 < 1 4 ( ), if C 11 is the preconditioner to B 11, based on a modified element matrix; κ ( C 1 11 Ã11 form; ) < 15 8, if C 11 is the preconditioner to B 11 of factorized the cost of the application of preconditioners in both cases is proportional to the number of unknowns. The spectral equivalence is readily seen from the following expressions C 11 = Ã11,II Ã 11,SI Ã 11,IS, Ã 11 = Ã 11,SI Ã 1 11,IIÃ11,IS+C 11 Ã11,II Ã 11,SI Ã 11,IS. Ã 11,SI Ã 1 11,IIÃ11,IS+B 11 Optimal multilevel preconditioning of strongly anisotropic problems.part II: non-conforming FEM. p. 34/36

35 6. Concluding remarks To be able to construct multilevel hierarchical preconditioners for the non-conforming discretization we have to answer the question if the two-level splitting of the finite element spaces is recursively applicable, or in other words, how V (k) 2 relates to V (k 1). In the FR case, it turns out that the block B 22 approximates A H. The spectral relations of B 22 and A H can be studied locally. For the reference macroelement, we find numerically that c 1 v T B 22 v v T A H v c 2 v T B 22 v for all v, where c 1 = 0.5, c 2 = 0.75 for isotropic problems, 0.25 c 1 and c 2 1 for all considered anisotropies. In the DA case, Theorem 3.2. shows that Ã22 = 4A H, which enables the recursive extension of the two-level hierarchical construction to the multilevel version straightforwardly. Optimal multilevel preconditioning of strongly anisotropic problems.part II: non-conforming FEM. p. 35/36

36 References 1. Blaheta R, Margenov S, Neytcheva M. Uniform estimate of the constant in the strengthened CBS inequality for anisotropic non-conforming FEM systems, Numerical Linear Algebra with Applications 2004; 11: Blaheta R, Margenov S, Neytcheva M. Robust optimal multilevel preconditioners for non-conforming finite element systems, Numerical Linear Algebra with Applications 2005; 12: Blaheta R, Margenov S, M. Neytcheva. Multilevel methods and preconditioners: An overview. Technical Report, Center of Excellence BIS-21 grant ICA , June Margenov S, Vassilevski PS. Two-level preconditioning of non-conforming FEM systems, Griebel, Iliev, Margenov, Vassilevski, eds., Large-Scale Scientific Computations of Engineering and Environmental Problems, Notes on Numerical Fluid Mechanics VIEWEG 1998: V 62: Optimal multilevel preconditioning of strongly anisotropic problems.part II: non-conforming FEM. p. 36/36

Multilevel Preconditioning of Graph-Laplacians: Polynomial Approximation of the Pivot Blocks Inverses

Multilevel Preconditioning of Graph-Laplacians: Polynomial Approximation of the Pivot Blocks Inverses Multilevel Preconditioning of Graph-Laplacians: Polynomial Approximation of the Pivot Blocks Inverses P. Boyanova 1, I. Georgiev 34, S. Margenov, L. Zikatanov 5 1 Uppsala University, Box 337, 751 05 Uppsala,

More information

Uniform estimate of the constant in the strengthened CBS inequality for anisotropic non-conforming FEM systems

Uniform estimate of the constant in the strengthened CBS inequality for anisotropic non-conforming FEM systems Uniform estimate of te constant in te strengtened CBS inequality for anisotropic non-conforming FEM systems R. Blaeta S. Margenov M. Neytceva Version of November 0, 00 Abstract Preconditioners based on

More information

Comparative Analysis of Mesh Generators and MIC(0) Preconditioning of FEM Elasticity Systems

Comparative Analysis of Mesh Generators and MIC(0) Preconditioning of FEM Elasticity Systems Comparative Analysis of Mesh Generators and MIC(0) Preconditioning of FEM Elasticity Systems Nikola Kosturski and Svetozar Margenov Institute for Parallel Processing, Bulgarian Academy of Sciences Abstract.

More information

Lecture 9 Approximations of Laplace s Equation, Finite Element Method. Mathématiques appliquées (MATH0504-1) B. Dewals, C.

Lecture 9 Approximations of Laplace s Equation, Finite Element Method. Mathématiques appliquées (MATH0504-1) B. Dewals, C. Lecture 9 Approximations of Laplace s Equation, Finite Element Method Mathématiques appliquées (MATH54-1) B. Dewals, C. Geuzaine V1.2 23/11/218 1 Learning objectives of this lecture Apply the finite difference

More information

Schur Complement Matrix And Its (Elementwise) Approximation: A Spectral Analysis Based On GLT Sequences

Schur Complement Matrix And Its (Elementwise) Approximation: A Spectral Analysis Based On GLT Sequences Schur Complement Matrix And Its (Elementwise) Approximation: A Spectral Analysis Based On GLT Sequences Ali Dorostkar, Maya Neytcheva, and Stefano Serra-Capizzano 2 Department of Information Technology,

More information

INTRODUCTION TO MULTIGRID METHODS

INTRODUCTION TO MULTIGRID METHODS INTRODUCTION TO MULTIGRID METHODS LONG CHEN 1. ALGEBRAIC EQUATION OF TWO POINT BOUNDARY VALUE PROBLEM We consider the discretization of Poisson equation in one dimension: (1) u = f, x (0, 1) u(0) = u(1)

More information

Spectral element agglomerate AMGe

Spectral element agglomerate AMGe Spectral element agglomerate AMGe T. Chartier 1, R. Falgout 2, V. E. Henson 2, J. E. Jones 4, T. A. Manteuffel 3, S. F. McCormick 3, J. W. Ruge 3, and P. S. Vassilevski 2 1 Department of Mathematics, Davidson

More information

Solution methods for the Cahn-Hilliard equation discretized by conforming and non-conforming finite elements

Solution methods for the Cahn-Hilliard equation discretized by conforming and non-conforming finite elements Solution methods for the Cahn-Hilliard equation discretized by conforming and non-conforming finite elements Petia Boyanova 2, Minh Do-Quang 3, Maya Neytcheva Uppsala University, Box 337, 75 5 Uppsala,

More information

A Robust Preconditioner for the Hessian System in Elliptic Optimal Control Problems

A Robust Preconditioner for the Hessian System in Elliptic Optimal Control Problems A Robust Preconditioner for the Hessian System in Elliptic Optimal Control Problems Etereldes Gonçalves 1, Tarek P. Mathew 1, Markus Sarkis 1,2, and Christian E. Schaerer 1 1 Instituto de Matemática Pura

More information

DISCRETE MAXIMUM PRINCIPLES in THE FINITE ELEMENT SIMULATIONS

DISCRETE MAXIMUM PRINCIPLES in THE FINITE ELEMENT SIMULATIONS DISCRETE MAXIMUM PRINCIPLES in THE FINITE ELEMENT SIMULATIONS Sergey Korotov BCAM Basque Center for Applied Mathematics http://www.bcamath.org 1 The presentation is based on my collaboration with several

More information

Block-tridiagonal matrices

Block-tridiagonal matrices Block-tridiagonal matrices. p.1/31 Block-tridiagonal matrices - where do these arise? - as a result of a particular mesh-point ordering - as a part of a factorization procedure, for example when we compute

More information

INTRODUCTION TO FINITE ELEMENT METHODS

INTRODUCTION TO FINITE ELEMENT METHODS INTRODUCTION TO FINITE ELEMENT METHODS LONG CHEN Finite element methods are based on the variational formulation of partial differential equations which only need to compute the gradient of a function.

More information

Auxiliary space multigrid method for elliptic problems with highly varying coefficients

Auxiliary space multigrid method for elliptic problems with highly varying coefficients Auxiliary space multigrid method for elliptic problems with highly varying coefficients Johannes Kraus 1 and Maria Lymbery 2 1 Introduction The robust preconditioning of linear systems of algebraic equations

More information

PARTITION OF UNITY FOR THE STOKES PROBLEM ON NONMATCHING GRIDS

PARTITION OF UNITY FOR THE STOKES PROBLEM ON NONMATCHING GRIDS PARTITION OF UNITY FOR THE STOES PROBLEM ON NONMATCHING GRIDS CONSTANTIN BACUTA AND JINCHAO XU Abstract. We consider the Stokes Problem on a plane polygonal domain Ω R 2. We propose a finite element method

More information

Domain Decomposition Preconditioners for Spectral Nédélec Elements in Two and Three Dimensions

Domain Decomposition Preconditioners for Spectral Nédélec Elements in Two and Three Dimensions Domain Decomposition Preconditioners for Spectral Nédélec Elements in Two and Three Dimensions Bernhard Hientzsch Courant Institute of Mathematical Sciences, New York University, 51 Mercer Street, New

More information

arxiv: v1 [math.na] 11 Jul 2011

arxiv: v1 [math.na] 11 Jul 2011 Multigrid Preconditioner for Nonconforming Discretization of Elliptic Problems with Jump Coefficients arxiv:07.260v [math.na] Jul 20 Blanca Ayuso De Dios, Michael Holst 2, Yunrong Zhu 2, and Ludmil Zikatanov

More information

Multispace and Multilevel BDDC. Jan Mandel University of Colorado at Denver and Health Sciences Center

Multispace and Multilevel BDDC. Jan Mandel University of Colorado at Denver and Health Sciences Center Multispace and Multilevel BDDC Jan Mandel University of Colorado at Denver and Health Sciences Center Based on joint work with Bedřich Sousedík, UCDHSC and Czech Technical University, and Clark R. Dohrmann,

More information

Assignment on iterative solution methods and preconditioning

Assignment on iterative solution methods and preconditioning Division of Scientific Computing, Department of Information Technology, Uppsala University Numerical Linear Algebra October-November, 2018 Assignment on iterative solution methods and preconditioning 1.

More information

MULTIGRID PRECONDITIONING IN H(div) ON NON-CONVEX POLYGONS* Dedicated to Professor Jim Douglas, Jr. on the occasion of his seventieth birthday.

MULTIGRID PRECONDITIONING IN H(div) ON NON-CONVEX POLYGONS* Dedicated to Professor Jim Douglas, Jr. on the occasion of his seventieth birthday. MULTIGRID PRECONDITIONING IN H(div) ON NON-CONVEX POLYGONS* DOUGLAS N ARNOLD, RICHARD S FALK, and RAGNAR WINTHER Dedicated to Professor Jim Douglas, Jr on the occasion of his seventieth birthday Abstract

More information

Finite difference method for elliptic problems: I

Finite difference method for elliptic problems: I Finite difference method for elliptic problems: I Praveen. C praveen@math.tifrbng.res.in Tata Institute of Fundamental Research Center for Applicable Mathematics Bangalore 560065 http://math.tifrbng.res.in/~praveen

More information

Geometric Multigrid Methods

Geometric Multigrid Methods Geometric Multigrid Methods Susanne C. Brenner Department of Mathematics and Center for Computation & Technology Louisiana State University IMA Tutorial: Fast Solution Techniques November 28, 2010 Ideas

More information

Parallel Discontinuous Galerkin Method

Parallel Discontinuous Galerkin Method Parallel Discontinuous Galerkin Method Yin Ki, NG The Chinese University of Hong Kong Aug 5, 2015 Mentors: Dr. Ohannes Karakashian, Dr. Kwai Wong Overview Project Goal Implement parallelization on Discontinuous

More information

A Subspace Correction Method for Discontinuous Galerkin Discretizations of Linear Elasticity Equations

A Subspace Correction Method for Discontinuous Galerkin Discretizations of Linear Elasticity Equations www.oeaw.ac.at A Subspace Correction Method for Discontinuous Galerkin Discretizations of Linear lasticity quations B. Ayuso de Dios, I. Georgiev, J. Kraus, L. Zikatanov RICAM-Report 2011-29 www.ricam.oeaw.ac.at

More information

Multilevel and Adaptive Iterative Substructuring Methods. Jan Mandel University of Colorado Denver

Multilevel and Adaptive Iterative Substructuring Methods. Jan Mandel University of Colorado Denver Multilevel and Adaptive Iterative Substructuring Methods Jan Mandel University of Colorado Denver The multilevel BDDC method is joint work with Bedřich Sousedík, Czech Technical University, and Clark Dohrmann,

More information

MPI parallel implementation of CBF preconditioning for 3D elasticity problems 1

MPI parallel implementation of CBF preconditioning for 3D elasticity problems 1 Mathematics and Computers in Simulation 50 (1999) 247±254 MPI parallel implementation of CBF preconditioning for 3D elasticity problems 1 Ivan Lirkov *, Svetozar Margenov Central Laboratory for Parallel

More information

Stabilization and Acceleration of Algebraic Multigrid Method

Stabilization and Acceleration of Algebraic Multigrid Method Stabilization and Acceleration of Algebraic Multigrid Method Recursive Projection Algorithm A. Jemcov J.P. Maruszewski Fluent Inc. October 24, 2006 Outline 1 Need for Algorithm Stabilization and Acceleration

More information

2. Linear algebra. matrices and vectors. linear equations. range and nullspace of matrices. function of vectors, gradient and Hessian

2. Linear algebra. matrices and vectors. linear equations. range and nullspace of matrices. function of vectors, gradient and Hessian FE661 - Statistical Methods for Financial Engineering 2. Linear algebra Jitkomut Songsiri matrices and vectors linear equations range and nullspace of matrices function of vectors, gradient and Hessian

More information

Scientific Computing WS 2018/2019. Lecture 15. Jürgen Fuhrmann Lecture 15 Slide 1

Scientific Computing WS 2018/2019. Lecture 15. Jürgen Fuhrmann Lecture 15 Slide 1 Scientific Computing WS 2018/2019 Lecture 15 Jürgen Fuhrmann juergen.fuhrmann@wias-berlin.de Lecture 15 Slide 1 Lecture 15 Slide 2 Problems with strong formulation Writing the PDE with divergence and gradient

More information

Algebraic Multigrid as Solvers and as Preconditioner

Algebraic Multigrid as Solvers and as Preconditioner Ò Algebraic Multigrid as Solvers and as Preconditioner Domenico Lahaye domenico.lahaye@cs.kuleuven.ac.be http://www.cs.kuleuven.ac.be/ domenico/ Department of Computer Science Katholieke Universiteit Leuven

More information

Adaptive algebraic multigrid methods in lattice computations

Adaptive algebraic multigrid methods in lattice computations Adaptive algebraic multigrid methods in lattice computations Karsten Kahl Bergische Universität Wuppertal January 8, 2009 Acknowledgements Matthias Bolten, University of Wuppertal Achi Brandt, Weizmann

More information

Overlapping Schwarz preconditioners for Fekete spectral elements

Overlapping Schwarz preconditioners for Fekete spectral elements Overlapping Schwarz preconditioners for Fekete spectral elements R. Pasquetti 1, L. F. Pavarino 2, F. Rapetti 1, and E. Zampieri 2 1 Laboratoire J.-A. Dieudonné, CNRS & Université de Nice et Sophia-Antipolis,

More information

Math 5520 Homework 2 Solutions

Math 5520 Homework 2 Solutions Math 552 Homework 2 Solutions March, 26. Consider the function fx) = 2x ) 3 if x, 3x ) 2 if < x 2. Determine for which k there holds f H k, 2). Find D α f for α k. Solution. We show that k = 2. The formulas

More information

Kasetsart University Workshop. Multigrid methods: An introduction

Kasetsart University Workshop. Multigrid methods: An introduction Kasetsart University Workshop Multigrid methods: An introduction Dr. Anand Pardhanani Mathematics Department Earlham College Richmond, Indiana USA pardhan@earlham.edu A copy of these slides is available

More information

LECTURE # 0 BASIC NOTATIONS AND CONCEPTS IN THE THEORY OF PARTIAL DIFFERENTIAL EQUATIONS (PDES)

LECTURE # 0 BASIC NOTATIONS AND CONCEPTS IN THE THEORY OF PARTIAL DIFFERENTIAL EQUATIONS (PDES) LECTURE # 0 BASIC NOTATIONS AND CONCEPTS IN THE THEORY OF PARTIAL DIFFERENTIAL EQUATIONS (PDES) RAYTCHO LAZAROV 1 Notations and Basic Functional Spaces Scalar function in R d, d 1 will be denoted by u,

More information

Chapter 5 A priori error estimates for nonconforming finite element approximations 5.1 Strang s first lemma

Chapter 5 A priori error estimates for nonconforming finite element approximations 5.1 Strang s first lemma Chapter 5 A priori error estimates for nonconforming finite element approximations 51 Strang s first lemma We consider the variational equation (51 a(u, v = l(v, v V H 1 (Ω, and assume that the conditions

More information

An Iterative Substructuring Method for Mortar Nonconforming Discretization of a Fourth-Order Elliptic Problem in two dimensions

An Iterative Substructuring Method for Mortar Nonconforming Discretization of a Fourth-Order Elliptic Problem in two dimensions An Iterative Substructuring Method for Mortar Nonconforming Discretization of a Fourth-Order Elliptic Problem in two dimensions Leszek Marcinkowski Department of Mathematics, Warsaw University, Banacha

More information

Schwarz Preconditioner for the Stochastic Finite Element Method

Schwarz Preconditioner for the Stochastic Finite Element Method Schwarz Preconditioner for the Stochastic Finite Element Method Waad Subber 1 and Sébastien Loisel 2 Preprint submitted to DD22 conference 1 Introduction The intrusive polynomial chaos approach for uncertainty

More information

Using an Auction Algorithm in AMG based on Maximum Weighted Matching in Matrix Graphs

Using an Auction Algorithm in AMG based on Maximum Weighted Matching in Matrix Graphs Using an Auction Algorithm in AMG based on Maximum Weighted Matching in Matrix Graphs Pasqua D Ambra Institute for Applied Computing (IAC) National Research Council of Italy (CNR) pasqua.dambra@cnr.it

More information

ASM-BDDC Preconditioners with variable polynomial degree for CG- and DG-SEM

ASM-BDDC Preconditioners with variable polynomial degree for CG- and DG-SEM ASM-BDDC Preconditioners with variable polynomial degree for CG- and DG-SEM C. Canuto 1, L. F. Pavarino 2, and A. B. Pieri 3 1 Introduction Discontinuous Galerkin (DG) methods for partial differential

More information

A greedy strategy for coarse-grid selection

A greedy strategy for coarse-grid selection A greedy strategy for coarse-grid selection S. MacLachlan Yousef Saad August 3, 2006 Abstract Efficient solution of the very large linear systems that arise in numerical modelling of real-world applications

More information

Iterative Methods for Linear Systems

Iterative Methods for Linear Systems Iterative Methods for Linear Systems 1. Introduction: Direct solvers versus iterative solvers In many applications we have to solve a linear system Ax = b with A R n n and b R n given. If n is large the

More information

Robust Domain Decomposition Preconditioners for Abstract Symmetric Positive Definite Bilinear Forms

Robust Domain Decomposition Preconditioners for Abstract Symmetric Positive Definite Bilinear Forms www.oeaw.ac.at Robust Domain Decomposition Preconditioners for Abstract Symmetric Positive Definite Bilinear Forms Y. Efendiev, J. Galvis, R. Lazarov, J. Willems RICAM-Report 2011-05 www.ricam.oeaw.ac.at

More information

Additive Average Schwarz Method for a Crouzeix-Raviart Finite Volume Element Discretization of Elliptic Problems

Additive Average Schwarz Method for a Crouzeix-Raviart Finite Volume Element Discretization of Elliptic Problems Additive Average Schwarz Method for a Crouzeix-Raviart Finite Volume Element Discretization of Elliptic Problems Atle Loneland 1, Leszek Marcinkowski 2, and Talal Rahman 3 1 Introduction In this paper

More information

A MULTIGRID ALGORITHM FOR. Richard E. Ewing and Jian Shen. Institute for Scientic Computation. Texas A&M University. College Station, Texas SUMMARY

A MULTIGRID ALGORITHM FOR. Richard E. Ewing and Jian Shen. Institute for Scientic Computation. Texas A&M University. College Station, Texas SUMMARY A MULTIGRID ALGORITHM FOR THE CELL-CENTERED FINITE DIFFERENCE SCHEME Richard E. Ewing and Jian Shen Institute for Scientic Computation Texas A&M University College Station, Texas SUMMARY In this article,

More information

Comparison of V-cycle Multigrid Method for Cell-centered Finite Difference on Triangular Meshes

Comparison of V-cycle Multigrid Method for Cell-centered Finite Difference on Triangular Meshes Comparison of V-cycle Multigrid Method for Cell-centered Finite Difference on Triangular Meshes Do Y. Kwak, 1 JunS.Lee 1 Department of Mathematics, KAIST, Taejon 305-701, Korea Department of Mathematics,

More information

1. Introduction. We consider the model problem that seeks an unknown function u = u(x) satisfying

1. Introduction. We consider the model problem that seeks an unknown function u = u(x) satisfying A SIMPLE FINITE ELEMENT METHOD FOR LINEAR HYPERBOLIC PROBLEMS LIN MU AND XIU YE Abstract. In this paper, we introduce a simple finite element method for solving first order hyperbolic equations with easy

More information

Weak Galerkin Finite Element Scheme and Its Applications

Weak Galerkin Finite Element Scheme and Its Applications Weak Galerkin Finite Element Scheme and Its Applications Ran Zhang Department of Mathematics Jilin University, China IMS, Singapore February 6, 2015 Talk Outline Motivation WG FEMs: Weak Operators + Stabilizer

More information

Algebraic multigrid for moderate order finite elements

Algebraic multigrid for moderate order finite elements Algebraic multigrid for moderate order finite elements Artem Napov and Yvan Notay Service de Métrologie Nucléaire Université Libre de Bruxelles (C.P. 165/84) 50, Av. F.D. Roosevelt, B-1050 Brussels, Belgium.

More information

VARIATIONAL AND NON-VARIATIONAL MULTIGRID ALGORITHMS FOR THE LAPLACE-BELTRAMI OPERATOR.

VARIATIONAL AND NON-VARIATIONAL MULTIGRID ALGORITHMS FOR THE LAPLACE-BELTRAMI OPERATOR. VARIATIONAL AND NON-VARIATIONAL MULTIGRID ALGORITHMS FOR THE LAPLACE-BELTRAMI OPERATOR. ANDREA BONITO AND JOSEPH E. PASCIAK Abstract. We design and analyze variational and non-variational multigrid algorithms

More information

Numerical Analysis of Differential Equations Numerical Solution of Elliptic Boundary Value

Numerical Analysis of Differential Equations Numerical Solution of Elliptic Boundary Value Numerical Analysis of Differential Equations 188 5 Numerical Solution of Elliptic Boundary Value Problems 5 Numerical Solution of Elliptic Boundary Value Problems TU Bergakademie Freiberg, SS 2012 Numerical

More information

Multigrid Methods and their application in CFD

Multigrid Methods and their application in CFD Multigrid Methods and their application in CFD Michael Wurst TU München 16.06.2009 1 Multigrid Methods Definition Multigrid (MG) methods in numerical analysis are a group of algorithms for solving differential

More information

1. Fast Iterative Solvers of SLE

1. Fast Iterative Solvers of SLE 1. Fast Iterative Solvers of crucial drawback of solvers discussed so far: they become slower if we discretize more accurate! now: look for possible remedies relaxation: explicit application of the multigrid

More information

Math Linear Algebra II. 1. Inner Products and Norms

Math Linear Algebra II. 1. Inner Products and Norms Math 342 - Linear Algebra II Notes 1. Inner Products and Norms One knows from a basic introduction to vectors in R n Math 254 at OSU) that the length of a vector x = x 1 x 2... x n ) T R n, denoted x,

More information

2 Two-Point Boundary Value Problems

2 Two-Point Boundary Value Problems 2 Two-Point Boundary Value Problems Another fundamental equation, in addition to the heat eq. and the wave eq., is Poisson s equation: n j=1 2 u x 2 j The unknown is the function u = u(x 1, x 2,..., x

More information

FEniCS Course. Lecture 0: Introduction to FEM. Contributors Anders Logg, Kent-Andre Mardal

FEniCS Course. Lecture 0: Introduction to FEM. Contributors Anders Logg, Kent-Andre Mardal FEniCS Course Lecture 0: Introduction to FEM Contributors Anders Logg, Kent-Andre Mardal 1 / 46 What is FEM? The finite element method is a framework and a recipe for discretization of mathematical problems

More information

Non-Conforming Finite Element Methods for Nonmatching Grids in Three Dimensions

Non-Conforming Finite Element Methods for Nonmatching Grids in Three Dimensions Non-Conforming Finite Element Methods for Nonmatching Grids in Three Dimensions Wayne McGee and Padmanabhan Seshaiyer Texas Tech University, Mathematics and Statistics (padhu@math.ttu.edu) Summary. In

More information

Weak Galerkin Finite Element Methods and Applications

Weak Galerkin Finite Element Methods and Applications Weak Galerkin Finite Element Methods and Applications Lin Mu mul1@ornl.gov Computational and Applied Mathematics Computationa Science and Mathematics Division Oak Ridge National Laboratory Georgia Institute

More information

Friedrich symmetric systems

Friedrich symmetric systems viii CHAPTER 8 Friedrich symmetric systems In this chapter, we describe a theory due to Friedrich [13] for positive symmetric systems, which gives the existence and uniqueness of weak solutions of boundary

More information

Algorithms for Scientific Computing

Algorithms for Scientific Computing Algorithms for Scientific Computing Finite Element Methods Michael Bader Technical University of Munich Summer 2016 Part I Looking Back: Discrete Models for Heat Transfer and the Poisson Equation Modelling

More information

multigrid, algebraic multigrid, AMG, convergence analysis, preconditioning, ag- gregation Ax = b (1.1)

multigrid, algebraic multigrid, AMG, convergence analysis, preconditioning, ag- gregation Ax = b (1.1) ALGEBRAIC MULTIGRID FOR MODERATE ORDER FINITE ELEMENTS ARTEM NAPOV AND YVAN NOTAY Abstract. We investigate the use of algebraic multigrid (AMG) methods for the solution of large sparse linear systems arising

More information

ON THE TWO-LEVEL PRECONDITIONING IN LEAST SQUARES METHOD

ON THE TWO-LEVEL PRECONDITIONING IN LEAST SQUARES METHOD PROCEEDINGS OF THE YEREVAN STATE UNIVERSITY Physical and Matheatical Sciences 04,, p. 7 5 ON THE TWO-LEVEL PRECONDITIONING IN LEAST SQUARES METHOD M a t h e a t i c s Yu. A. HAKOPIAN, R. Z. HOVHANNISYAN

More information

A posteriori error estimation for elliptic problems

A posteriori error estimation for elliptic problems A posteriori error estimation for elliptic problems Praveen. C praveen@math.tifrbng.res.in Tata Institute of Fundamental Research Center for Applicable Mathematics Bangalore 560065 http://math.tifrbng.res.in

More information

Chapter Two: Numerical Methods for Elliptic PDEs. 1 Finite Difference Methods for Elliptic PDEs

Chapter Two: Numerical Methods for Elliptic PDEs. 1 Finite Difference Methods for Elliptic PDEs Chapter Two: Numerical Methods for Elliptic PDEs Finite Difference Methods for Elliptic PDEs.. Finite difference scheme. We consider a simple example u := subject to Dirichlet boundary conditions ( ) u

More information

Local Mesh Refinement with the PCD Method

Local Mesh Refinement with the PCD Method Advances in Dynamical Systems and Applications ISSN 0973-5321, Volume 8, Number 1, pp. 125 136 (2013) http://campus.mst.edu/adsa Local Mesh Refinement with the PCD Method Ahmed Tahiri Université Med Premier

More information

An Introduction to Algebraic Multigrid (AMG) Algorithms Derrick Cerwinsky and Craig C. Douglas 1/84

An Introduction to Algebraic Multigrid (AMG) Algorithms Derrick Cerwinsky and Craig C. Douglas 1/84 An Introduction to Algebraic Multigrid (AMG) Algorithms Derrick Cerwinsky and Craig C. Douglas 1/84 Introduction Almost all numerical methods for solving PDEs will at some point be reduced to solving A

More information

arxiv: v2 [math.na] 17 Jun 2010

arxiv: v2 [math.na] 17 Jun 2010 Numerische Mathematik manuscript No. (will be inserted by the editor) Local Multilevel Preconditioners for Elliptic Equations with Jump Coefficients on Bisection Grids Long Chen 1, Michael Holst 2, Jinchao

More information

Applied/Numerical Analysis Qualifying Exam

Applied/Numerical Analysis Qualifying Exam Applied/Numerical Analysis Qualifying Exam August 9, 212 Cover Sheet Applied Analysis Part Policy on misprints: The qualifying exam committee tries to proofread exams as carefully as possible. Nevertheless,

More information

A DELTA-REGULARIZATION FINITE ELEMENT METHOD FOR A DOUBLE CURL PROBLEM WITH DIVERGENCE-FREE CONSTRAINT

A DELTA-REGULARIZATION FINITE ELEMENT METHOD FOR A DOUBLE CURL PROBLEM WITH DIVERGENCE-FREE CONSTRAINT A DELTA-REGULARIZATION FINITE ELEMENT METHOD FOR A DOUBLE CURL PROBLEM WITH DIVERGENCE-FREE CONSTRAINT HUOYUAN DUAN, SHA LI, ROGER C. E. TAN, AND WEIYING ZHENG Abstract. To deal with the divergence-free

More information

Numerical Programming I (for CSE)

Numerical Programming I (for CSE) Technische Universität München WT 1/13 Fakultät für Mathematik Prof. Dr. M. Mehl B. Gatzhammer January 1, 13 Numerical Programming I (for CSE) Tutorial 1: Iterative Methods 1) Relaxation Methods a) Let

More information

Adaptive C1 Macroelements for Fourth Order and Divergence-Free Problems

Adaptive C1 Macroelements for Fourth Order and Divergence-Free Problems Adaptive C1 Macroelements for Fourth Order and Divergence-Free Problems Roy Stogner Computational Fluid Dynamics Lab Institute for Computational Engineering and Sciences University of Texas at Austin March

More information

Numerical methods for PDEs FEM convergence, error estimates, piecewise polynomials

Numerical methods for PDEs FEM convergence, error estimates, piecewise polynomials Platzhalter für Bild, Bild auf Titelfolie hinter das Logo einsetzen Numerical methods for PDEs FEM convergence, error estimates, piecewise polynomials Dr. Noemi Friedman Contents of the course Fundamentals

More information

Overlapping Schwarz Preconditioners for Spectral. Problem in H(curl)

Overlapping Schwarz Preconditioners for Spectral. Problem in H(curl) Overlapping Schwarz Preconditioners for Spectral Nédélec Elements for a Model Problem in H(curl) Technical Report TR2002-83 November 22, 2002 Department of Computer Science Courant Institute of Mathematical

More information

AMS subject classifications. Primary, 65N15, 65N30, 76D07; Secondary, 35B45, 35J50

AMS subject classifications. Primary, 65N15, 65N30, 76D07; Secondary, 35B45, 35J50 A SIMPLE FINITE ELEMENT METHOD FOR THE STOKES EQUATIONS LIN MU AND XIU YE Abstract. The goal of this paper is to introduce a simple finite element method to solve the Stokes equations. This method is in

More information

Finite Elements. Colin Cotter. February 22, Colin Cotter FEM

Finite Elements. Colin Cotter. February 22, Colin Cotter FEM Finite Elements February 22, 2019 In the previous sections, we introduced the concept of finite element spaces, which contain certain functions defined on a domain. Finite element spaces are examples of

More information

On an Approximation Result for Piecewise Polynomial Functions. O. Karakashian

On an Approximation Result for Piecewise Polynomial Functions. O. Karakashian BULLETIN OF THE GREE MATHEMATICAL SOCIETY Volume 57, 010 (1 7) On an Approximation Result for Piecewise Polynomial Functions O. arakashian Abstract We provide a new approach for proving approximation results

More information

Multigrid Methods for Elliptic Obstacle Problems on 2D Bisection Grids

Multigrid Methods for Elliptic Obstacle Problems on 2D Bisection Grids Multigrid Methods for Elliptic Obstacle Problems on 2D Bisection Grids Long Chen 1, Ricardo H. Nochetto 2, and Chen-Song Zhang 3 1 Department of Mathematics, University of California at Irvine. chenlong@math.uci.edu

More information

Multigrid and Domain Decomposition Methods for Electrostatics Problems

Multigrid and Domain Decomposition Methods for Electrostatics Problems Multigrid and Domain Decomposition Methods for Electrostatics Problems Michael Holst and Faisal Saied Abstract. We consider multigrid and domain decomposition methods for the numerical solution of electrostatics

More information

AN INTRODUCTION TO DOMAIN DECOMPOSITION METHODS. Gérard MEURANT CEA

AN INTRODUCTION TO DOMAIN DECOMPOSITION METHODS. Gérard MEURANT CEA Marrakech Jan 2003 AN INTRODUCTION TO DOMAIN DECOMPOSITION METHODS Gérard MEURANT CEA Introduction Domain decomposition is a divide and conquer technique Natural framework to introduce parallelism in the

More information

Multigrid Method ZHONG-CI SHI. Institute of Computational Mathematics Chinese Academy of Sciences, Beijing, China. Joint work: Xuejun Xu

Multigrid Method ZHONG-CI SHI. Institute of Computational Mathematics Chinese Academy of Sciences, Beijing, China. Joint work: Xuejun Xu Multigrid Method ZHONG-CI SHI Institute of Computational Mathematics Chinese Academy of Sciences, Beijing, China Joint work: Xuejun Xu Outline Introduction Standard Cascadic Multigrid method(cmg) Economical

More information

Scientific Computing WS 2017/2018. Lecture 18. Jürgen Fuhrmann Lecture 18 Slide 1

Scientific Computing WS 2017/2018. Lecture 18. Jürgen Fuhrmann Lecture 18 Slide 1 Scientific Computing WS 2017/2018 Lecture 18 Jürgen Fuhrmann juergen.fuhrmann@wias-berlin.de Lecture 18 Slide 1 Lecture 18 Slide 2 Weak formulation of homogeneous Dirichlet problem Search u H0 1 (Ω) (here,

More information

Institut für Mathematik

Institut für Mathematik U n i v e r s i t ä t A u g s b u r g Institut für Mathematik Xuejun Xu, Huangxin Chen, Ronald H.W. Hoppe Local Multilevel Methods for Adaptive Nonconforming Finite Element Methods Preprint Nr. 21/2009

More information

The Discontinuous Galerkin Finite Element Method

The Discontinuous Galerkin Finite Element Method The Discontinuous Galerkin Finite Element Method Michael A. Saum msaum@math.utk.edu Department of Mathematics University of Tennessee, Knoxville The Discontinuous Galerkin Finite Element Method p.1/41

More information

Chapter 3 Conforming Finite Element Methods 3.1 Foundations Ritz-Galerkin Method

Chapter 3 Conforming Finite Element Methods 3.1 Foundations Ritz-Galerkin Method Chapter 3 Conforming Finite Element Methods 3.1 Foundations 3.1.1 Ritz-Galerkin Method Let V be a Hilbert space, a(, ) : V V lr a bounded, V-elliptic bilinear form and l : V lr a bounded linear functional.

More information

Algebraic Multigrid Preconditioners for Computing Stationary Distributions of Markov Processes

Algebraic Multigrid Preconditioners for Computing Stationary Distributions of Markov Processes Algebraic Multigrid Preconditioners for Computing Stationary Distributions of Markov Processes Elena Virnik, TU BERLIN Algebraic Multigrid Preconditioners for Computing Stationary Distributions of Markov

More information

Numerical Solutions to Partial Differential Equations

Numerical Solutions to Partial Differential Equations Numerical Solutions to Partial Differential Equations Zhiping Li LMAM and School of Mathematical Sciences Peking University Variational Problems of the Dirichlet BVP of the Poisson Equation 1 For the homogeneous

More information

Sophomoric Matrix Multiplication

Sophomoric Matrix Multiplication Sophomoric Matrix Multiplication Carl C. Cowen IUPUI (Indiana University Purdue University Indianapolis) Universidad de Zaragoza, 3 julio 2009 Linear algebra students learn, for m n matrices A, B, y C,

More information

PANM 17. Marta Čertíková; Jakub Šístek; Pavel Burda Different approaches to interface weights in the BDDC method in 3D

PANM 17.  Marta Čertíková; Jakub Šístek; Pavel Burda Different approaches to interface weights in the BDDC method in 3D PANM 17 Marta Čertíková; Jakub Šístek; Pavel Burda Different approaches to interface weights in the BDDC method in 3D In: Jan Chleboun and Petr Přikryl and Karel Segeth and Jakub Šístek and Tomáš Vejchodský

More information

EXACT DE RHAM SEQUENCES OF SPACES DEFINED ON MACRO-ELEMENTS IN TWO AND THREE SPATIAL DIMENSIONS

EXACT DE RHAM SEQUENCES OF SPACES DEFINED ON MACRO-ELEMENTS IN TWO AND THREE SPATIAL DIMENSIONS EXACT DE RHAM SEQUENCES OF SPACES DEFINED ON MACRO-ELEMENTS IN TWO AND THREE SPATIAL DIMENSIONS JOSEPH E. PASCIAK AND PANAYOT S. VASSILEVSKI Abstract. This paper proposes new finite element spaces that

More information

A SHORT NOTE COMPARING MULTIGRID AND DOMAIN DECOMPOSITION FOR PROTEIN MODELING EQUATIONS

A SHORT NOTE COMPARING MULTIGRID AND DOMAIN DECOMPOSITION FOR PROTEIN MODELING EQUATIONS A SHORT NOTE COMPARING MULTIGRID AND DOMAIN DECOMPOSITION FOR PROTEIN MODELING EQUATIONS MICHAEL HOLST AND FAISAL SAIED Abstract. We consider multigrid and domain decomposition methods for the numerical

More information

Chapter 7 Iterative Techniques in Matrix Algebra

Chapter 7 Iterative Techniques in Matrix Algebra Chapter 7 Iterative Techniques in Matrix Algebra Per-Olof Persson persson@berkeley.edu Department of Mathematics University of California, Berkeley Math 128B Numerical Analysis Vector Norms Definition

More information

Boundary Value Problems and Iterative Methods for Linear Systems

Boundary Value Problems and Iterative Methods for Linear Systems Boundary Value Problems and Iterative Methods for Linear Systems 1. Equilibrium Problems 1.1. Abstract setting We want to find a displacement u V. Here V is a complete vector space with a norm v V. In

More information

An additive average Schwarz method for the plate bending problem

An additive average Schwarz method for the plate bending problem J. Numer. Math., Vol. 10, No. 2, pp. 109 125 (2002) c VSP 2002 Prepared using jnm.sty [Version: 02.02.2002 v1.2] An additive average Schwarz method for the plate bending problem X. Feng and T. Rahman Abstract

More information

18. Balancing Neumann-Neumann for (In)Compressible Linear Elasticity and (Generalized) Stokes Parallel Implementation

18. Balancing Neumann-Neumann for (In)Compressible Linear Elasticity and (Generalized) Stokes Parallel Implementation Fourteenth nternational Conference on Domain Decomposition Methods Editors: smael Herrera, David E Keyes, Olof B Widlund, Robert Yates c 23 DDMorg 18 Balancing Neumann-Neumann for (n)compressible Linear

More information

Implicitly Defined High-Order Operator Splittings for Parabolic and Hyperbolic Variable-Coefficient PDE Using Modified Moments

Implicitly Defined High-Order Operator Splittings for Parabolic and Hyperbolic Variable-Coefficient PDE Using Modified Moments Implicitly Defined High-Order Operator Splittings for Parabolic and Hyperbolic Variable-Coefficient PDE Using Modified Moments James V. Lambers September 24, 2008 Abstract This paper presents a reformulation

More information

Oversampling for the partition of unity parallel finite element algorithm

Oversampling for the partition of unity parallel finite element algorithm International conference of Computational Mathematics and its application Urumqi, China, July 25-27, 203 Oversampling for the partition of unity parallel finite element algorithm HAIBIAO ZHENG School of

More information

A Balancing Algorithm for Mortar Methods

A Balancing Algorithm for Mortar Methods A Balancing Algorithm for Mortar Methods Dan Stefanica Baruch College, City University of New York, NY 11, USA Dan Stefanica@baruch.cuny.edu Summary. The balancing methods are hybrid nonoverlapping Schwarz

More information

On Surface Meshes Induced by Level Set Functions

On Surface Meshes Induced by Level Set Functions On Surface Meshes Induced by Level Set Functions Maxim A. Olshanskii, Arnold Reusken, and Xianmin Xu Bericht Nr. 347 Oktober 01 Key words: surface finite elements, level set function, surface triangulation,

More information

Multigrid Methods for Saddle Point Problems

Multigrid Methods for Saddle Point Problems Multigrid Methods for Saddle Point Problems Susanne C. Brenner Department of Mathematics and Center for Computation & Technology Louisiana State University Advances in Mathematics of Finite Elements (In

More information

Homework 2. Solutions T =

Homework 2. Solutions T = Homework. s Let {e x, e y, e z } be an orthonormal basis in E. Consider the following ordered triples: a) {e x, e x + e y, 5e z }, b) {e y, e x, 5e z }, c) {e y, e x, e z }, d) {e y, e x, 5e z }, e) {

More information

1. Introduction. The Stokes problem seeks unknown functions u and p satisfying

1. Introduction. The Stokes problem seeks unknown functions u and p satisfying A DISCRETE DIVERGENCE FREE WEAK GALERKIN FINITE ELEMENT METHOD FOR THE STOKES EQUATIONS LIN MU, JUNPING WANG, AND XIU YE Abstract. A discrete divergence free weak Galerkin finite element method is developed

More information