and it has also been combined with adaptive grid generation ë20ë and domain decomposition ë21ë. The functions a i : æærær 2!R, in è1.1è are assumed to

Size: px
Start display at page:

Download "and it has also been combined with adaptive grid generation ë20ë and domain decomposition ë21ë. The functions a i : æærær 2!R, in è1.1è are assumed to"

Transcription

1 Mixed Finite Element Method for onlinear Elliptic Problems: the h-p -Version Miyoung Lee and Fabio A. Milner Department of Mathematics, Purdue University West Lafayette, I , U.S.A. Abstract. Mixed ænite element methods for strongly nonlinear second order elliptic problems are proposed and analyzed. Existence and uniqueness of the approximate solution are demonstrated using a æxed point argument. Convergence and stability of the method are proved both with respect to mesh reænement and increase of the degree of the approximating polynomials. The analysis is carried out in detail using Raviart-Thomas-edelec spaces as an example. umerical results for minimal surface problems are obtained using Brezzi-Douglas-Marini spaces. Graphs of the approximate solutions are presented for two different problems. Key words. nonlinear elliptic problem, mixed method, minimal surfaces, B.D.M space. AMS subject classiæcation. 6512, 6515, 6522, 6530, 35J Introduction We shall consider the numerical solution of the following nonlinear Dirichlet problem, which appears in many engineering applications: è1.1è, P 2 i èx;u;ruè+a 0 èx;u;ruè = 0; x 2 æ; uèxè =,gèxè; x where æ is a rectangular domain in R 2 with The combination of the h and p versions of the ænite element method is in widespread use, for example, in structural mechanics and æuid dynamics, 1

2 and it has also been combined with adaptive grid generation ë20ë and domain decomposition ë21ë. The functions a i : æærær 2!R, in è1.1è are assumed to be as regular as needed for our arguments. We shall also assume that the quasilinear operator associated with è1.1è is elliptic in the following sense. Let èx;u;zè and æèx;u;zè denote, respectively, the minimum and maximum eigenvalues of the matrix A = ; which we shall assume to be j ii;j=1;2 Then, for all 2 R 2, 6= 0, and for all èx;u;zè 2 æ ærær 2, we have 0 éèx;u;zèkk 2 T A æèx;u;zèkk 2 : The variable x will normally be omitted in this notation. For 1 qé1 and k any nonnegative integer, let W k;q èæè = f 2 L q èæè : D æ 2 L q èæè if jæjkg denote the standard Sobolev space endowed with its usual norm kæk k;q;æ, with the obvious modiæcation in the case q = 1. Let H k èæè = W k;2 èæè denote L 2 -based Sobolev spaces with the norm kæk k = kæk k;2. In particular, the notation kæk 0 will mean kæk L 2 èæè or kæk L 2 èæè2. For 0 s é 1, let W s;q èæè, W s;q è@æè, H s èæè and H s è@æè denote the fractional order Sobolev spaces endowed with the norms kæk s;q;æ, kæk s;q;@æ, kæk s;æ and kæk s;@æ. The subscript æ in the norm will be omitted. Let be normed by and let V = Hèdiv;æè=fv 2 H 2 èæè 2 : div v 2 L 2 èæèg kvk V = kvk 0 + k div vk 0 ; W = L 2 èæè: The mixed ænite element method ë5ë approximates at the same time the solution of è1.1è, u, and the æux è1.2è =,aèu; ruè =,èa 1 èu; ruè;a 2 èu; ruèè; which is frequently the quantity that needs to be found in the problem at hand representing, for example, a velocity æeld or a stress tensor. By the implicit function theorem è1.2è can be inverted to obtain ru as a function of u and, say ru =,bèu; è: 2

3 We now set fèu; è =,a 0 èu; ruè =,a 0 èu;,bèu; èè in è1.1è. Then, the mixed weak form of è1.1è we shall consider consists of ænding è;uè 2 V æ W such that è1.3è èbèu; è;vè, èdiv v; uè =é g;væ né 8v 2 V; èdiv ;wè =èfèu; è;wè 8w 2 W; where n denotes the unit outer normal vector Let us consider a quasi-uniform family ft g of decompositions of æ by rectangles E èwith boundary elements allowed to have one curved edgeè. For each decomposition, let p 2 0 be the degree of the approximating piecewise polynomials used for this decomposition. ext, let V æ W V æ W be the Raviart-Thomas-edelec space of index p 0 associated with this decomposition ë11, 19, 23ë. The mixed ænite element method is a discrete form of è1.3è and consists of ænding è ;u è 2 V æ W such that è1.4è èbèu ; è;vè, èdiv v; u è = ég;væné 8v 2 V ; èdiv ;wè = èfèu ; è;wè 8w 2 W : We shall use an L 2 -projection onto W, P : L 2! W, given by è1.5è èp w, w; è =0; 2 W ;w 2 W: We shall also use the R-T projection of V onto V, : V! V ë23ë, which is locally deæned èon every element Eè èsee ë23ëè. Lemma 1.1 Let : V! V be the Raviart-Thomas projection. Let s 2 and réf1=2; 3=2, 3=sg. Then, for v 2 èh r èæèè 2, è1.6è k v, vk 0;s Ch minfp +1;rg,1+2=s p 5=2,r,4=s kvk r ; where C is a constant independent of h, p, and s but depends on r. Proof The result follows from a simple argument that combines known estimates on each element which are summed over all elements and combined with Corollary 2.1 in ë18ë and Lemma 3.2 in ë24ë. 3

4 Lemma 1.2 Let P be L 2 -projection onto W deæned by è1.5è and let = min fp +1;rg, s 2, m 3=2, 3=s. Then, for w 2 H r èæè, è1.7è kw, P wk 0;s Ch 2=s,1+ p,r+3=2,3=s kwk; where C is a constant independent of h, p, and w. Proof The proof is similar to that of Lemma 1.1, using è1.7è in ë18ë and Lemma 4 in ë3ë. We will also use the following inverse-type inequalities. Lemma 1.3 Let 2 L s èæè T W èor 2 L s èæè 2 T V è, 1 r s 1. Then, è1.8è Proof in ë22ë. kk 0;s Qh 2=s,2=r p 4=r,4=s kk 0;r : The proof follows easily, using inequalities è1.9è in ë18ë and è4.6è The theoretical results in this paper are essentially a combination of those in ë15ë and ë22ë. However, it is possible to choose h asymptotically bounded by some power of p in such a way that the regularity required of the solution of è1.1è is less than needed for the p-version of the method. We shall indicate this reduction along the way. The second half of the paper will be devoted to addressing some of the programming problems associated with the implementation of the method and presenting results from simulations for minimal surfaces. In section 2 we shall linearize the system è1.1è and analize the linearization. In section 3 we will show that è1.4è is uniquely solvable by using a æxed point argument, and that its solution è ;u è converges to è;uè in V T L 4 èæè 2 æ L 4 èæè. Then in section 4 we derive error estimates in L 2 for the approximation. The last four sections will contain, respectively, the description of a ewton iteration algorithm which may be used to handle the nonlinearities of the problem, programming techniques for BDM mixed ænite elements, numerical results obtained for minimal surface problems, and some concluding remarks. 4

5 2. Solvability of the Linearized Problem Following ë15ë and ë22ë, for 2 W, 2 V weintroduce ærst and second order Taylor expansions, è2.1è fè; è, fèu; è =,f u èu; èèu, è, f èu; èè, è+ ~ Qf èu, ;, è =, ~ f u è; èèu, è, ~ f è; èè, è: Similarly, for b =èb 1 ;b 2 è, bè; è, bèu; è =,b u èu; èèu, è, b èu; èè, è+ Qb ~ èu, ;, è =, ~ b u è; èèu, è, ~ b è; èèu, èè, è: We obtain our ærst error equations by subtracting è1.4è from è1.3è: è2.2è èbèu; è, bèu ; è;vè, èdiv v; u, u è=0 8v 2 V ; èdiv è, è;wè=èfèu; è, fèu ; è;wè 8w 2 W : Recall, ë11ë, that div æ = P æ div : H 1 èæè 2! W : Combining è2.1è-è2.2è with = u and =, we obtain the following form of the error equations, which we will need for our æxed point argument: èbèu; èë, ë;vè, èdiv v; P u, u è+è, 1 ëp u, u ë;vè =èbèu; èë, ë+, 1 ëp u, uë+ ~ Qb èu, u ;, è;vè 8v 2 V ; èdiv è, è;wè, è, 2 ë, ë;wè, èæëp u, u ë;wè =è,, 2 ë, ë, æëp u, uë, ~ Qf èu, u ;, è;wè 8w 2 W : Here we have set, just as in ë15ë and in ë22ë, Bèu; è =b èu; è =A,1 èu; è,, 1 = b u èu; è,, 2 = f èu; è, and æ = f u èu; è. ext we deæne, just as in ë15ë and ë22ë, M : H 2 èæè! L 2 èæè by è2.3è Mw =,div èaèu; èrw + Aèu; è, 1 wè+aèu; è, 2 ærw,èæ,, T 2 Aèu; è, 1èw and its formal adjoint M æ by M æ =,div èaèu; èr + Aèu; è, 2 è+aèu; è, 1 ær,èæ,, T 2 Aèu; è, 1è: 5

6 From ë11, 15ë we know that the restrictions of the operator M and M æ to H 2 èæè T H 1 0 èæè have bounded inverses, if we assume that èu; è can be extended to a pair è~u; ~è deæned on a domain æ 0 with a C 2 -boundary, such that æ æ 0 and measèæ 0, æè is arbitrary small ë1, 14ë. Then, for any è 2 L 2 èæè there is a unique 2 H 2 èæè T H 1 0èæè such that M = è èrespectively, M æ = èè and kk 2 Ckèk 0 if we assume that, for example, the zero order term of M æ is nonnegative, that is è2.4è æ, T 2 Aèu; è, 1, div èaèu; è, 2 è; where u 2 C 0;1 èæè and 2 C 0;1 èæè 2 ë12, 22ë. In order to be able to employ our duality arguments, we shall assume the structure condition è2.4è. ote that this condition is reduced to æ 0if, 2 =0. Let æ : V æ W! V æ W be given by æè; è =ès; qè where ès; qè is the solution of the following linear system: èbèu; èë, së;vè, èdiv v; P u, qè+è, 1 ëp u, që;vè =èbèu; èë, ë+, 1 ëp u, uë+ ~ Qb èu, ;, è;vè 8v 2 V ; èdiv è, sè;wè, è, 2 ë, së;wè, èæëp u, që;wè =è,, 2 ë, ë, æëp u, uë, ~ Qf èu, ;, è;wè 8w 2 W : ote that the left hand side of this system corresponds to the mixed method for the operator M given by è2.3è. We will show next that this system is uniquely solvable, so that the map æ is well-deæned. Thus, the problem we want to study is that of ænding èy; qè 2 W æ V such that, for a pair of given l 2 L 2 èæè 2 and m 2 L 2 èæè, è2.5è èbq;vè, èdiv v; yè+è, 1 y; vè = èl; vè 8v 2 V ; èdiv q; wè, è, 2 q; wè, èæy;wè = èm; wè 8w 2 W : Lemma 2.1 Let q 2 V, l 2 L 2 èæè 2, and m 2 L 2 èæè. If y 2 W satisæes the relation è2.5è, then for suæciently large p or small h, kyk 0 Q hh p,1=2 kqk 0 + h 2 p,2 kdiv qk 0 + klk 0 + kmk 0 i : Proof The proof follows exactly as in ë15, 22ë using Lemmas Lemma 2.2 If q 2 V satisæes è2.5è, then kqk 0 + kdiv qk 0 C ëkyk 0 + klk 0 + kmk 0 ë : 6

7 Proof To bound kqk 0, choose v = q, w = y in è2.5è and add the resulting equations. The choice w = div q in è2.5è gives the bound for kdiv qk 0. Lemma 2.3 There exists one and only one solution of the system è2.5è. Proof Existence follows from uniqueness since the system is linear. Assume l = 0;m = 0. Then Lemma 2.1 implies kyk 0 Qh p,1=2 kqk V ; where kqk V = kqk 0 + kdiv qk 0. By Lemma 2.2, we have kqk V ckyk 0 èkyk 0 Qh p,1=2 kqk V Qh p,1=2 kyk 0 ; which implies y = 0 for large p, or small h. Then, q = 0 as needed. ow it is clear from Lemma 2.3 that the functional æ is well-deæned. 3. Existence and Uniqueness The solvability of è1.4è is now equivalenttoshowing that æ has a æxed point. We state this result in the following theorem. Theorem 3.1 For p suæciently large or h suæciently small, æ has a æxed point. In order to prove this, we shall need the following duality lemma. Lemma 3.1 Let! 2 V, l 2 L 2 èæè 2, and m 2 L 2 èæè. If 2 W satisæes the relation è3.1è èbèu; è!; vè, èdiv v; è+è, 1 ;vè =èl; vè 8v 2 V ; èdiv!; wè, è, 2!; wè, èæ;wè =èm; wè 8w 2 W ; then, for any, 2 4, ", there exists a positive constant C = Cè; u; ;, 1 ;, 2 ;æ;æ;"è, independent of p and h, such that kk 0: Cëh 2= p1=2,2= k!k 0 + h 2=+1 p,1,2= kdiv!k 0 + klk 0 + kmk 0 : ow let V = V with the stronger norm kvk V = kvk 0;4 + kdiv vk 0, and let W = W with the stronger norm kwk W = kwk 0;4. It follows from the Brouwer æxed point theorem that Theorem 3.1 is true if we can show the following. 7

8 Theorem 3.2 For æ é 0 suæciently small èdependent on p and h è, æ maps the ball of radius æ of V æw ; centered at è ;P uè, into itself. Proof Let k,qk V æ and kp u,yk W æ. We apply Lemma 3.1 with = P u, y,! =, q, and l = Bèu; èë, ë+, 1 ëp u, uë+ ~ Qb èu, ;, è m =,, 2 ë, ë, æëp u, uë, ~ Qf èu, ;, è: Using Lemmas and the relation 2ab a 2 + b 2, it follows that for suæciently large p or suæciently small h, è3.2è kp u, yk 0 ëh p,1=2 k, qk 0 +h 2 p,2 kdiv è, qèk 0 + klk 0 + kmk 0 ë Cëæ 2 + p 1=2,r ë where C = Cèkk r ; kuk m ;, 1 ; ~ Qb ; ~ Qf è. ext, by applying Lemma 2.2 to è3.1è, we have k, qk 0 + kdiv è, qèk CëkP u, yk 0 + klk 0 + kmk 0 ë Cëh r p 1 2,r + æ2 ë: è3.3è Also, using è1.8è we see that k, qk V Ch,1=2 p ëh r p1=2,r while è3.2è and è3.3è imply that kp u, yk W Ch,1=2 Hence, + æ 2 ë, p ëæ 2 + h r p1=2,r ë. kp u, yk W + k, qk V 2Ch,1=2 p ëæ 2 + h r p1=2,r ë; where C = Cèkk r ; kuk m ;, 1 ; Qb ~ ; Qf ~ è. Finally, we want to choose p ;h, and æ so that I =ë4ch r,1=2 p 3=2,r, ë is not empty. This requires that r é 5=2 èor r = 5=2 and h 1 4C h1=2 p,1 suæciently smallè, a severe regularity constraint. Then, for any æ 2 I, we have kp u, yk W æ and k, qk V æ as needed. However, given "é0 small, if we let k =3=è2"è, 1+" and impose the constraint h ép,k on the mesh sizes of the decompositions of æ, then for p large enough that p èk+1èr,k, C 2 and for r 1+", the interval I is not empty. Thus we see that existence of solutions is really guaranteed as 8

9 long as ré1. ext we shall prove a uniqueness result which holds provided that the coeæcients a i, i =0; 1; 2 of è1.1è are three times continuously diæerentiable. Theorem 3.3 If p is suæciently large or h suæciently small, there isa unique solution of è1.4è near the solution fu; g of è1.3è. Proof Let è i ;u i è 2V æw, i =1; 2 be solutions of è1.4è and let U = u 1, u 2 ; æ = 1, 2 ; i =, i ; i = u, u i ; i =1; 2: Rewrite è2.2è as èbèu; è, bèu 2 ; 2 è;vè, èdiv v; Uè =èbèu; è, bèu 1 ; 1 è;vè; 8v 2V ; Then, using è2.1è we obtain èdivæ;wè=èfèu 1 ; 1 è, fèu 2 ; 2 è;wè; 8w 2W : èbèu; èæ;vè,èdiv v; Uè+è, 1 U; vè =è~ Qb è 2 ; 2 è, ~ Qb è 1 ; 1 è;vè; 8v 2V ; èdiv æ;wè,è, 2 æ;wè,èæu;wè =è~ Qf è 2 ; 2 è, ~ Qf è 1 ; 1 è;wè; 8w 2W ; and then it follows from Lemma 2.2 that kæk 0 + kdiv æk 0 CëkUk 0 + k ~ Qb è 1 ; 1 è, ~ Qb è 2 ; 2 èk 0 +k ~ Qf è 1 ; 1 è, ~ Qf è 2 ; 2 èk 0 ë: Also, by Lemma 2.1, we see that kuk 0 Qfh p,1=2 kæk 0 + h 2 p,2 kdiv æk 0 + k ~ Qb è 1 ; 1 è, ~ Qb è 2 ; 2 èk 0 +k ~ Qf è 1 ; 1 è, ~ Qf è 1 ; 1 èk 0 g: ext we want to show that k Qb ~ è 1 ; 1 è, Qb ~ è 2 ; 2 èk 0 Ch 2r,1 p5,2rèkæk 0 + kuk 0 è; k Qf ~ è 1 ; 1 è, Qf ~ è 2 ; 2 èk 0 Ch 2r,1 p5,2rèkæk 0 + kuk 0 è; so that, if ré5=2, or r =5=2 and h is suæciently small, or r 1+" with the additional condition h ép 1,3=2"," given in the proof of Theorem 3.2, we will have è3.4è k ~ Qb è 1 ; 1 è, ~ Qb è 2 ; 2 èk 0 Cp,"2 èkæk 0 + kuk 0 è; k ~ Qf è 1 ; 1 è, ~ Qf è 2 ; 2 èk 0 Cp,"2 èkæk 0 + kuk 0 è: 9

10 To show è3.4è we use the mean value theorem on the quadratic forms ~ Qb and ~ Qf, just as in ë15, 22ë. It follows that, for example, k Qf ~ è 2 ; 2 è, Qf ~ è 1 ; 1 èk 0 Cëk 1 k 0;1 + k 2 k 0;1 + k 1 k 2 0;1 +k 1 k 0;1 + k 1 k 0;1 k 1 k 0;1 + k 2 k 0;1 + k 2 k 2 0;1ëëkæk 0 + kuk 0 ë: è3.5è ote that, from the inverse estimates è1.8è, it follows that kp u, u k 0;1 p kp u, u k 0;4 p h,1=2 k, k 0;1 p k, k 0;4 p h,1=2 So, we have æ Khr,1 p1=2,r ; æ Khr,1 p1=2,r : k i k 0;1 = ku i, u i k 0;1 ku i, P u i k 0;1 + kp u i, u i k 0;1 K h p,m+3=2 kuk m + p,"2 ; è3.6è k i k 0;1 = k i, i k 0;1 k i, i k 0;1 + k i, i k 0;1 K h p 5=2,r kk r + p,"2 ; and using è3.6è in è3.5è we obtain è3.4è: k ~ Qb è 1 ; 1 è, ~ Qb è 2 ; 2 èk 0 Kh " p3=2," ëkæk 0 + kuk 0 ë; k ~ Qf è 1 ; 1 è, ~ Qf è 2 ; 2 èk 0 Kh " p 3=2," ëkæk 0 + kuk 0 ë: By Lemmas 2.1 and 2.2, this concludes the proof of the theorem, provided ré5=2, or r =5=2 and h is suæciently small, or ré1 and h = Oèp,k è as in Theorem L 2 -Error Estimates In this section we shall bound the error of the solution of the h-píversion for problem è1.1è. In order to simplify the notation, from now on we shall omit the subindex on the parameters h and p. Theorem 4.1 Assume that the coeæcients of è1.1è are smooth enough that u 2 H r+1 èæè and 2 H r èæè 2, where r é 5=2, or r = 5=2 and h is sufæciently small, or r é 1 and h = Oèp,k è as in Theorem 3.2. Then, for suæciently large p or suæciently small h, the following estimates hold. 10

11 ku, u k 0 Qfh r p 1=2,r kk r èkk r +1è+h r+1 p,r,1 kuk r+1 èkuk r+1 +1èg; k, k 0 Qfh r p 1=2,r kk r èkk r +1è+h r+1 p,r,1 kuk r+1 èkuk r+1 +1èg; kdiv è, èk 0 Qfh r p 1,r kk r èkk r +1è+h r+1 p,r,1 kuk r+1 èkuk r+1 +1èg; Proof Let =,, = u,u, =,, and = P u,u. ext, rewrite è2.2è in the form èbèu; è; vè, èdiv v; è+è, 1 ;vè =èbèu; èë, ë+, 1 ëp u, uë+ ~ Qb è;è;vè 8v 2 V ; èdiv ; wè, è, 2 ; wè, èæ;wè =è,, 2 ë, ë, æëp u, uë, ~ Qf è;è;wè 8w 2 W : Then, just as in Theorem 5.1 in ë15ë, we can see that kk 0 + kdiv k 0 Cëkk 0 + kk 2 0;4 + kk 2 0;4 + k, k 2 0;4 +k, k 0 + kp u, uk 2 0;4 + kp u, uk 0 ë Cëkk 0 + h,1=2 pkk 0;4 kk 0;2 + kk 0;4 h,1=2 pkk 0 +k, k 2 0;4 + k, k 0 + kp u, uk 2 0;4 + kp u, uk 0 ë Cëkk 0 + h r,1 p 5=2,r kk 0;2 + h r,1 p 5=2,r kk 0 +k, k 2 0;4 + k, k 0 + kp u, uk 2 0;4 + kp u, uk 0 ë Cëkk 0 + h r,1 p 5=2,r kk 0 + h r,1 p 5=2,r kk 0 +h r p 1 2,r kk r èkk r +1è+h r+1 p,r,1 kuk r+1 èkuk r+1 + 1èë: For p large enough or h small enough, we obtain the bound è4.1è kk 0 + kdiv k 0 Cëkk 0 + h r p 1=2,r kk r èkk r +1è +h r+1 p,r,1 kuk r+1 èkuk r+1 + 1èë: Likewise, by Lemma 2.1, kk 0 Qëhp,1=2 kk 0 + h 2 p,2 kdiv k 0 + klk 0 + kmk 0 ë Qëhp,1=2 kk 0 + h 2 p,2 kdiv k 0 + h r,1 p 5=2,r kk 0 + h r,1 p 5=2,r kk 0 +h r p 1=2,r kk r èkk r +1è+h r+1 p,r,1 kuk r+1 èkuk r+1 + 1èë; 11

12 and if p is suæciently large, it follows that è4.2è kk 0 Qëhp,1=2 kk 0 + h 2 p,2 kdiv k 0 + h r,1 p 5=2,r èkk 0 + kk 0 è +h r p 1=2,r kk r èkk r +1è+h r+1 p,r,1 kuk r+1 èkuk r+1 +1èg: Substituting è4.2è into è4.1è, we ænd that è4.3è kk V Qëh r p 1=2,r kk r èkk r +1è+h r+1 p,r,1 kuk r+1 èkuk r+1 +1èg; and substituting è4.3è into è4.2è we arrive at the estimate è4.4è kk 0 Qfh r p 1=2,r kk r èkk r +1è+h r+1 p,r,1 kuk r+1 èkuk r+1 +1èg: By combining è4.3è and è4.4è with Lemmas 1.1 and 1.2 we have ku, u k 0 Qfh r p1=2,r kk r èkk r +1è+h m p,m kuk m èkuk m +1èg; k, k 0 Qfh r p 1=2,r kk r èkk r +1è+h m p,m kuk m èkuk m +1èg; kdiv è, èk 0 Qfh r p 1,r kk r èkk r +1è+h m p,m kuk m èkuk m +1èg; as needed. 5. ewton's Method Following ë22ë, we compute a ewton approximation of f ;u g using a sequence f m ;u m g 1 m=0 in V æ W satisfying the following relations: For v 2 V and w 2 W, èbèu m ; m èè m+1, m è;vè, èdiv v; u m+1 è +è, 1 èu m ; m èèu m+1, u m è;vè=è,bèu m ; m è;vè+ ég;væné; è5.1è èdiv m+1 ;wè, è, 2 èu m ; m èè m+1, m è;wè,èæèu m ; m èèu m+1, u m è;wè=èfèu m ; m è;wè; where Bèu m ; m è=b èu m ; m è,, 1 èu m ; m è=b u èu m ; m è,, 2 èu m ; m è= f èu m ; m è, and æèu m ; m è=f u èu m ; m è. We shall show that the algorithm è5.1è is well deæned and that it converges quadratically. To this end, we shall need a technical lemma. Let æ = sup kvk0;1 + kwk 0;1 kvk 0 + kwk 0 ote æ =Oèh,1 p 2 èby the inverse inequality è1.8è. : fv; wg 2V æ W,f0; 0g : 12

13 Lemma 5.1 Given 0 é 0, there exist positive constants p 0, h 0, æ 0, and C such that the following holds. If p 0 p, h h 0, and 2 L 1 èæè and 2 L 1 èæè 2 satisfy the following relations, kk 0;1 + kk 0;1 0 and èk, k 0 + k, uk V èæ æ 0 ; and if l æ 2 L 2 èæè 2 and m æ 2 L 2 èæè, then there exists a unique f æ ;u æ g 2 V æ W such that èbè; è æ ;vè, èdiv v; u æ è+è, 1 è; èu æ ;vè=èl æ ;vè 8v 2 V ; èdiv æ ;wè, è, 2 è; è æ æ ;wè, èæè; èu æ ;wè=èm æ ;wè 82W : è5.2è Furthermore, f æ ;u æ g satisæes the bound è5.3è ku æ k 0 + k æ k V Cëkl æ k 0 + km æ k 0 ë: Proof It suæces to show that è5.3è holds, since this will imply that the solution è5.2è is unique and hence it exists. First, rewrite è5.2è in the following form: For v 2 V and w 2 W, è5.4è èbèu; è æ ;vè, èdiv v; u æ è+è, 1 è; èu æ ;vè=èl æ ;vè +èèbèu; è, èbè; èè æ ;vè + èè, 1 èu; è,, 1 è; èèu æ ;vè; èdiv æ ;wè, è, 2 èu; è æ æ ;wè, èæèu; èu æ ;wè=èm æ ;wè +èè, 2 è; è,, 2 èu; èè æ æ ;wè+èèæè; è, æèu; èèu æ ;wè: Recall, Lemma 2.2, that è5.5è k æ k V Cëku æ k 0 + kl æ k 0 + km æ k 0 ë: Also, it follows from Lemma 2.1 that ku æ k 0 Cëhp,1=2 k æ k V + kbèu; è, Bè; èk 0 k æ k 0;1 +k, 1 èu; è,, 1 è; èk 0 ku æ k 0;1 + k, 2 èu; è,, 2 è; èk 0 k æ k 0;1 +kæèu; è, æè; èk 0 ku æ k 0;1 + kl æ k 0 + km æ k 0 ë C 1 ëhp,1=2 k æ k V +èk, k 0 + ku, k 0 è æ èku æ k 0 + k æ k 0 è+kl æ k 0 + km æ k 0 ë; which implies that ku æ k 0 C 2 ëèh p,1=2 + æ 0 èk æ k V + æ 0 ku æ k 0 + kl æ k 0 + km æ k 0 ë: 13

14 Let now æ 0 é 1 2C 2 è5.6è so that ku æ k 0 C 3 ëèh p,1=2 + æ 0 èk æ k V + kl æ k 0 + km æ k 0 ë; Then, substituting è5.6è into è5.5è, we see that, k æ k V C 4 ëèhp,1=2 + æ 0 èk æ k V + kl æ k 0 + km æ k 0 ë: ow, for hp,1=2 é 1 4C 4, and æ 0 é minf 1 4C 4 ; 1 2C 2 g; we have which implies, from è5.6è, k æ k V C 5 ëkl æ k 0 + km æ k 0 ë; ku æ k 0 C 6 ëkl æ k 0 + km æ k 0 ë; and the result follows. We can state now our result concerning the existence and convergence of the ewton iterates, which follows immediately from Lemma 5.1. Theorem 5.1 There exist positive constants p 0, h 0, æ 0, and ~ C, such that, if p 0 ép, héh 0 and æ ëku 0, u k V + k 0, k 0 ë æ 0 ; then f m ;u m g 1 m=0 is well deæned, and m = ku m, u k 0 + k m, k V a decreasing sequence satisfying is m+1 ~ Cæ 2 m : 6. Programming techniques with BDM Spaces We shall apply now the numerical methods described in Section 5 to approximate the solution of the minimal surface problem è6.1è,div è ru è1+jruj 2 è 1=2 è = 0 8x 2 æ; u =,g 8x The approximation we ænd is based on the BDM èbrezzi-douglas-mariniè spaces ë6ë which are based on polynomials of some æxed total degree, rather than on tensor products. Consequently, the local dimension of these spaces 14

15 is much smaller than those of the corresponding Raviart-Thomas-edelec spaces. We give now a brief description of the BDM spaces. Let p 1 and let T h = fe i g be such that the length of each side of E i is h. We base the ænite element space for the approximation of the scalar function u on polynomials of total degree not exceeding p, 1. Let W p,1 èe i è=p p,1 = 8 é é: X i+jp,1 9 é= c i;j x i y j ; c i;j 2R é; : The number of degrees of freedom for W p,1 èe i èispèp +1è=2. The ænite element space for the approximation of the vector function is based on polynomials of total degree p augmented by a space of polynomials of degree p + 1 of dimension two. Let V p èe i è=p p èe i è M Spanècurl x p+1 y;curl xy p+1 è; where P p èe i è=p p æp p. The number of degrees of freedom for V p èe i èis èp + 1èèp + 2è + 2. Hence, the total local number of degrees of freedom for the BDM space M p is 1:5p 2 +3:5p +4which is about half the size of the local number of degrees of freedom for the Raviart-Thomas-edelec space of the same index, 3p 2 +2p. ext, we deæne V = V p h = fv 2 Hèdiv ; æè : vj E i 2 V p èe i è;e i 2T h g W = W p,1 h = fw : wj Ei 2 W p,1 èe i è;e i 2T h g M p h = V p h æw p,1 h ; and we seek è ;u è 2M p h, the solution of è1.4è. The analysis of this resulting mixed method is facilitated by the existence of the L 2 -orthogonal projection è1.5è, locally deæned on each element E i of the decomposition T h by the relations è6.2è èw, P p,1 h w; zè Ei =0; z 2P p,1 èe i è; E i 2T h : Q p Let h be the projection analogous to the Raviart-Thomas projection, locally deæned on each element E i of the decomposition T h by the following 15

16 degrees of freedom: With fe i jg 4 j=1 being the sides of the element E i and n i j the exterior unit normal vector to E i along e i j, for any q 2 V, Z w q, Y è p è p X è6.3è q æ n i e i h j dx =0; w 2 a s x s ; a s 2R ; 1 j 4; j s=0 q, ZEi Y p è6.4è q æ v dx dy =0; v 2 P p,2 èe i è: h Here we shall consider æ = ë0; 1ë 2. For the minimal surface problem we consider, the mixed ænite element formulation è1.4è is simply èbèu ; è;vè, èdiv v; u è = ég;væ n é 8v 2 V ; èdiv v; wè = 0 8w 2 W : We use the iterative algorithm which was described in Section 5 in order to solve this nonlinear algebraic system. Since in this case, 1 =0,, 2 = 0, and æ = 0 in è5.1è, then we have the following simpler iteration algorithm: è6.5è èbèu n ; n èè n+1, n è;vè, èdiv v; u n+1 è =,èbèu n ; n è;vè+ ég;væ n é; v 2 V ; èdiv n+1 ;wè=0; w 2 W ; where èsee ë17ëè and Then, it follows that è6.6è Bèu n ; n è= bèu; è =,ru = Bèu; è = p 1,jj è! 1 1, è n 2 è 2 1 nn 2 è1,j n j 2 è 3=2 1 nn 2 1, è1 n è 2 : ow, let f i g m 1 i=1 and f jg m 2 j=1 be, respectively, bases of V and W. Assume that i and j are supported in E 2T h. We then have the following iteration algorithm: For any initial guess f 0 ;u 0 g2v æ W, è : è6.7è èbèu m ; m è m+1 ; i è, èdiv i ;u m+1 è=èbèu m ; m è m ; i è,èbèu m ; m è; i è+ ég; i æ n E é; i 2 V ; èdiv m+1 ; i è=0; i 2 W 16

17 We obtain è m ;u m è recursively and, in each iteration, use it to compute the coeæcients for the next iteration. In order to compute è m ;u m èwe need to solve the following linear system of equations, using any direct or iterative method ë13ë: where Sx + Yy = f 1 ; Y T x = 0; ësë i;j =èbèu n ; n è j ; i è; Y i;j = èdiv i ; j è; u = Xm 2 j=1 y j j ; = x =èx 1 ;x 2 ; :::; x m1 è T ; and y =èy 1 ;y 2 ; :::; y m2 è T : ow we write this system in the form Xm 1 i=1 x i i ; è6.8è where M = " S Y Y T 0 è ; b = Mz = b " f1 0 è ; and z = " x y è : We shall show next how to select a local basis of V to avoid diæculties with the necessary continuity in the normal direction across inter-element boundaries when we extend the local basis to a global basis. At the same time, we shall obtain a sparse matrix associated with the system of equations è6.8è at each iteration step. Recall that p is the degree of the approximating piecewise polynomials in V. We nowchoose a basis on the reference element R =ë,1; 1ë 2 based on the following obvious decomposition. V èrè = Span èl i èxèl j èyè; 0è : 0 i + j p M Span è0;l i èxèl j èyèè : 0 i + j p M Z x Z y Span,curl y L p èè d; curl x,1,1 L p èè d ; where L s is the Legendre polynomial of degree s. Then, we see that a basis for V èrè can be chosen as B x 0 ë B x 1 ëæææëb x p ë B y 0 ëæææëb y p ; 17

18 where, for 0 k p, 2, Bk x = L i èxèl k èyè; 0 ; 0 i + k p, 2 ë è,1è p,k+1 L p,k èxèl k èyè+è,1è p,k L p,k,1 èxèl k èyè; 0 B x p,1 = B x p = ë L p,k èxèl k èyè+l p,k,1 èxèl k èyè; 0 ; L 1 èxèl p,1 èyè, L p,1 èyè; 0 èx, 1èL p èyè; ë 1 2p +1 ël p,1èyè, L p+1 èyèë ë èx +1èL p èyè; L 1 èxèl p,1 èyè+l p,1 èyè; 0 1 2p +1 ël p,1èyè, L p+1 èyèë We obtain B y k in an analogous way. Also, it is obvious that a local basis for W can be chosen as fl i èxèl j èyè; 0 i + j p, 1g: We introduce now the Lagrange interpolation polynomials based on Gauss-Lobatto points ë7ë. For 1 i k + 1, let where fx j g k+1 j=1 `ki èxè = k+1 Y j=1;j6=i èx, x j è èx i, x j è ; is the set of Gauss-Lobatto points of degree k on ë,1; 1ë, ordered as,1 =x 1 éx 2 é æææ éx k+1 =1: We know that, for 0 k p,1, SpanfBkg x = Span ; 1 j p, k +1 ; and `p,k j èxèl k èyè; 0 SpanfBpg x 1 = Span `1 1èxèL p èyè;, 2è2p +1è ël p,1èyè, L p+1 èyèë 1 `12èxèL p èyè; 2è2p +1è ël p,1èyè, L p+1 èyèë : Also, note that there exists a nonsingular èp, k +1èæ èp, k +1è matrix M k è2æ2 if k=pè such that M k l k =g k : ; ; 18

19 where, for 0 k p, 1, while l k =èl 1 ;l 2 ; æææ;l p,k+1 è T where B x k consists of fl 1 ;l 2 ; æææ;l p,k+1 g and, for 0 k p, 1, g k = l p =èl 1 ;l 2 è T where B x p consists of fl 1 ;l 2 g; ë`p,k 1 èxèl k èyè; 0ë; ë`p,k 2 èxèl k èyè; 0ë; æææ; ë`p,k k+1 èxèl kèyè; 0ë T while g p = `11 èxèl 1 pèyè;, 2è2p +1è ël p,1èyè, L p+1 èyèë ; 1 `12èxèL p èyè; 2è2p +1è ël p,1èyè, L p+1 èyèë T : We proceed analogously for B y k, 0 k p, and we can ænd matrices ^Mk such that ^M k^lk = ^h k ; where, for 0 k p, 1, while ^lk =èl 0 1;l 0 2; æææ;l 0 p,k+1è T where B y k consists of fl0 1;l 0 2; æææ;l 0 p,k+1g; and, for 0 k p, 1, ^h k = ^lp =èl 0 1;l 0 2è T where B x p consists of fl 0 1;l 0 2g; ë0;`p,k 1 èyèl k èxèë; ë0;`p,k 2 èyèl k èxèë; æææ; ë0;`p,k k+1 èyèl kèxèë T ; while ^h p =, 1 2è2p +1è ël p,1èxè, L p+1 èxèë;`1 1èyèL p èxè 1 2è2p +1è ël p,1èxè, L p+1 èxèë;`12èyèl p èxè 19 ; T :

20 Then we replace the system è6.8è by the following equivalent one: M 0... M p ^M0... ^Mp 30 7B 5@ l 0 æ l p ^l 0 æ ^l p 1 = C A 0 g 0 æ g p ^h 0 æ ^h p 1 : C A To compute ësë i;j =èbèu m ; m è j ; i è for the system è6.7è we use the Gauss- Lobatto quadrature of degree 5p: S i;j = 5p+1 X i=1 5p+1 X j=1 w i w j èbèu m èx i ;y j è; m èx i ;y j èè j èx i ;y j è; i èx i ;y j èè where w k, 1 k 5p +1, is the k-th Gauss-Lobatto weight of degree 5p. We also use the Gauss-Lobatto quadrature of degree 5p to calculate èbèu m ; m è;vè and ég;væn é. Finally,we use the Gauss-Lobatto quadrature of degree p for èdiv v; u m+1 è since èdiv vèu m+1 is a polynomial of degree at most 2p, 1 and the Gauss-Lobatto quadrature of degree p is exact up to this degree. 7. umerical Results ; In this section we shall present some results from numerical simulations of minimal surfaces modeled by è6.1è. The approximations were obtained both by reæning the mesh and by increasing the degree of the approximating polynomials. The boundary data was chosen as è7.1è gèx; yè = 8 é é: logëcosèy, :5èë, logëcosè,:5èë; 0 y 1; x =0; logëcosèy, :5èë, logëcosè:5èë; 0 y 1; x =1; logëcosè:5èë, logëcosèx, :5èë; 0 x 1; y =1; logëcosè,:5èë, logëcosèx, :5èë; 0 x 1; y =0; so that the exact solution for for this problem is known and it is given by uèx; yè = log cosèy, 0:5è cosèx, 0:5è ; 20

21 which belongs to C 1 èæè. The knowledge of the exact solution allows us to compute the actual errors of each approximation we ænd, and thus we are able to compare the eæective rates of convergence in h èfor a æxed pè and in p èfor a æxed hè, with the theoretically predicted ones, Theorem 4.1. We present in Figures 7.1 and 7.2 approximations obtained by mesh reænement using p = 2. Figure 7.1 depicts the approximations obtained for mesh sizes h = 1=2 ètopè and mesh size h = 1=4 èbottomè, while Figure 7.2 depicts the approximation obtained for mesh size h =1=8 ètopè and the exact solution èbottomè. 21

22 Figure 7.1: h=1è2 and h=1è4 22

23 Figure 7.2: h=1è8 and exact solution 23

24 We also presentintables 7.1 and 7.2 the errors in the approximation of u and both in L 2 and in L 1, in correspondence with the numerical solutions depicted in Figures 7.1 and 7.2 using the boundary condition given by è7.1è. These give an idea of the eæective performance of the method with respect to mesh reænement è"h-version"è. The error in L 2 was computed using numerical integration. The convergence rate indicated on the tables was calculated under the standard assumption that the errors = u, u and =, are of the form Kh!. Then! is found from two approximations computed for diæerent values of h. In this case wehave chosen h =2, and we show the convergence rates computed, respectively, as! = lnè0 = 2 è ;! = lnè0 = 2 è : 2ln2 2ln2 mesh size h lnècosèy, :5è= cosèx, :5èè ku, uk 1 ku, uk E-01 1è E E-01 1è E E-02 1è E E-03 convergence rate h 2 h 2 Table 7.1: Error in the scalar function, u, u èp =2è mesh size h lnècosèy, :5è= cosèx, :5èè k, k 1 k, k è E-01 1è E-01 1è E E-02 convergence rate h 1:2 h 1:6 Table 7.2: Error in the vector function,, èp =2è Tables 7.3 and 7.4 give, respectively, the errors in the numerical solutions obtained for a æxed mesh of size h = 1=2, and for diæerent degrees of the approximating polynomials. These give an idea of the eæective performance of the method with respect to the increase of the degree of the approximating polynomials è"p-version"è. 24

25 degree of polynomial lnècosèy, :5è= cosèx, :5èè ku, uk 1 ku, uk E E E E E E E E E E E E-13 convergence rate p,14 p,14 Table 7.3: Error in the scalar function, u, u èh =1=2è degree of polynomial lnècosèy, :5è= cosèx, :5èè k, k 1 k, k E E E E E E E E E E E-12 convergence rate p,14 p,14 Table 7.4: Error in the vector function,, èh =1=2è The convergence rate indicated on these tables was calculated under the standard assumption that the errors = u, u and =, are of the form Kp,è. Then è is found from two approximations computed for diæerent values of p. In this case we have chosen p 2 = 3 and p 6 = 15, and we show the convergence rates computed, respectively, as è = lnè2 = 6 è ; è = ln 5 lnè 2 = 6 è : ln 5 Finally, we present in Figure 7.3 the minimal surfaces corresponding to the boundary condition gèx; yè = 8 é é:,èsin 2yè=20; 0 y 1; x =0; èsin 2yè=20; 0 y 1; x =1;,èsin 2xè=20; 0 x 1; y =1; èsin 2xè=20; 0 x 1; y =0: 25

26 Figure 7.3: very æne mesh, h=1è20, and high polynomial degree, p=15 26

27 The exact solution for this problem is not known. The ærst picture in Figure 7.3 depicts the approximation computed using a very æne mesh, h = 1=20, and p = 2. The second picture depicts the approximation computed on a coarse mesh, h =1=2, with polynomials of high degree, p = Conclusions and Future Directions We have shown that the h and the p versions of the mixed ænite element method can be combined for the numerical solution of quite general nonlinear second order elliptic problems in divergence form. The resulting methods have much better convergence than the p version without mesh reænement, both in the regularity required of the exact solution and although we have not indicated them in the CPU times needed for the computations. The theoretical results have some serious restrictions. First, they require either a very regular solution, u 2 H 7=2+", or extremely æne meshes, h é p,k. Secondly, there is a requirement that the domain be rectangular in order to be able to use approximation estimates for the Raviart-Thomas projection in the p version of the mixed method. Such estimates only exist on domains which are unions of rectangles and there does not seem to exist an easy way to extend them to domains with curved boundaries. As it frequently happens, numerical simulations seem to indicate that these restrictions may be artiæcial, that is, due to the nature of the proof of these results. In fact, using fairly coarse meshes and polynomials of low degree, one can still see the numerical approximations converge. The computational results are quite satisfactory. As indicated by the ægures, the numerical approximations are quite good even with relatively coarse meshes and polynomials of low degree. Several extensions of this work should be pursued. It would be highly desirable to have error estimates available on domains with curved boundary, as well as on non-convex domains. It would also be very useful to reduce the regularity required of the solution of the diæerential problem without having to require extremely æne meshes. References ë1ë R. A. Adams, Sobolev Spaces. 1975, Academic Press, ew York- San Francisco-London. 27

28 ë2ë I. Babuska and B. Guo, The h-p version of the ænite element method for problems with nonhomogeneous essential boundary condition, Comput. Meth. Appl. Mech. Engin., 74 è1989è, pp ë3ë I. Babuska and M. Suri, The h, p version of the ænite element method with quasiuniform meshes. RAIRO Model. Math. Anal. umer., 21 è1987è pp ë4ë J. Bergh and J. Líofstríom, Interpolation Spaces: An Introduction. 1976, Springer-Verlag, Berlin and ew York. ë5ë F. Brezzi, On the existence, uniqueness, and approximation of saddle point problems arising from Lagrangian multipliers, RAIRO, Anal. umer., 2 è1974è, pp ë6ë F. Brezzi,J. Douglas, Jr. and L. D. Marini, Two families of mixed ænite elements for second order elliptic problems, umer. Math, 47 è1985è, pp ë7ë C. Canuto, M. Y. Hussaini, A. Quateroni and T. A. Zang, Spectral Methods in Fluid Dynamics, Springer-Verlag, ew York, ë8ë C. Canuto and A. Quateroni, Approximation results for orthogonal polynomials in Sobolev spaces Math. Comp., 38 è1982è, pp ë9ë P.G. Ciarlet, The ænite element method for elliptic problems. 1978, orth-holland, Amsterdam. ë10ë J. Douglas, Jr., H 1 -Galerkin methods for a nonlinear Dirichlet problem, in Mathematical Aspects of Finite Element MethodsèHeidelbergè, vol. 606 of Lecture otes in Mathematics, Springer-Verlag, Heidelberg, ë11ë J. Douglas, Jr. and J. E. Roberts, Global estimates for mixed methods for second order elliptic equations. Math. Comp., 44 è1985è, pp ë12ë D. Gilbarg and. S. Trudinger, Elliptic Partial Diæerential Equations of Second Order, 2nd ed., Springer-Verlag, Berlin, ë13ë G. Golub and C. V. Loan, Matrix Computation 2nd ed., The Johns Hopkins University Press, Baltimore and London,

29 ë14ë J. Kim and D. Sheen, An elliptic regularity of a Helmholtz- Type problem with an absorbing boundary condition, RIM-GARC Preprint Series 95-29, July 1995, Department of Mathematics, Seoul ational University, Seoul , Korea. ë15ë M. Lee and F. A. Milner, Mixed ænite element method for nonlinear elliptic problems: The p-version, to appear in umer. Meth. Part. Diæ, Eq. ë16ë F. A. Milner, Mixed ænite element methods for quasilinear secondorder elliptic problems. Math. Comp., 44 è1985è, pp ë17ë F. A. Milner and E. J. Park, Mixed ænite element method for a strongly nonlinear second order elliptic problem. Math. Comp., 65 è1995è, pp ë18ë F. A. Milner and M. Suri, Mixed ænite element methods for Quasilinear second order elliptic problems: the p-version. RAIRO Model. Math. Anal. umer., 26 è1992è, pp ë19ë J. C. edelec, A new family of mixed ænite elements in R 3. umer. Math., 50 è1986è, pp ë20ë J. T. Oden and L. Demkowicz, h-p adaptive ænite element methods in computational æuid dynamics. Comput. Meth. Appl. Mech. Engin., 89 è1991è, pp ë21ë J. T. Oden, A. Patra, and Y. Feng, Parallel domain decomposition solver for adaptive hp ænite element methods., TICAM Report è1994è. ë22ë E. J. Park, Mixed ænite element methods for nonlinear second order elliptic problems, SIAM um. Anal., 32 è1995è, pp ë23ë P. A. Raviart and J. M. Thomas, mixed ænite element method for 2nd order elliptic problems, Proceed. Conf. on Mathematical Aspects of Finite Element Methods. volume 606 of Lecture otes in Mathematics, G. F. Hewitt, J. M. Delhaye, and. Zuber,eds., Springer-Verlag, Berlin, 1987, pp ë24ë M. Suri, On the stability and convergence of higher order mixed ænite element methods for second order elliptic problems. Math. Comp, 54 è1990è pp

30 ë25ë A. F. Timan, The Theory of Approximation of Functions of Real Variable. Pergamon Press, Oxford, ë26ë H. Triebel, Interpolation Theory, Function Spaces, Diæerential Operators., 1978, orth-holland, Amsterdam. 30

LECTURE Review. In this lecture we shall study the errors and stability properties for numerical solutions of initial value.

LECTURE Review. In this lecture we shall study the errors and stability properties for numerical solutions of initial value. LECTURE 24 Error Analysis for Multi-step Methods 1. Review In this lecture we shall study the errors and stability properties for numerical solutions of initial value problems of the form è24.1è dx = fèt;

More information

W 1 æw 2 G + 0 e? u K y Figure 5.1: Control of uncertain system. For MIMO systems, the normbounded uncertainty description is generalized by assuming

W 1 æw 2 G + 0 e? u K y Figure 5.1: Control of uncertain system. For MIMO systems, the normbounded uncertainty description is generalized by assuming Chapter 5 Robust stability and the H1 norm An important application of the H1 control problem arises when studying robustness against model uncertainties. It turns out that the condition that a control

More information

Lecture Note III: Least-Squares Method

Lecture Note III: Least-Squares Method Lecture Note III: Least-Squares Method Zhiqiang Cai October 4, 004 In this chapter, we shall present least-squares methods for second-order scalar partial differential equations, elastic equations of solids,

More information

A Mixed Nonconforming Finite Element for Linear Elasticity

A Mixed Nonconforming Finite Element for Linear Elasticity A Mixed Nonconforming Finite Element for Linear Elasticity Zhiqiang Cai, 1 Xiu Ye 2 1 Department of Mathematics, Purdue University, West Lafayette, Indiana 47907-1395 2 Department of Mathematics and Statistics,

More information

b i (x) u + c(x)u = f in Ω,

b i (x) u + c(x)u = f in Ω, SIAM J. NUMER. ANAL. Vol. 39, No. 6, pp. 1938 1953 c 2002 Society for Industrial and Applied Mathematics SUBOPTIMAL AND OPTIMAL CONVERGENCE IN MIXED FINITE ELEMENT METHODS ALAN DEMLOW Abstract. An elliptic

More information

where Sènè stands for the set of n æ n real symmetric matrices, and æ is a bounded open set in IR n,typically with a suæciently regular boundary, mean

where Sènè stands for the set of n æ n real symmetric matrices, and æ is a bounded open set in IR n,typically with a suæciently regular boundary, mean GOOD AND VISCOSITY SOLUTIONS OF FULLY NONLINEAR ELLIPTIC EQUATIONS 1 ROBERT JENSEN 2 Department of Mathematical and Computer Sciences Loyola University Chicago, IL 60626, U.S.A. E-mail: rrj@math.luc.edu

More information

322 HENDRA GUNAWAN AND MASHADI èivè kx; y + zk çkx; yk + kx; zk: The pair èx; kæ; ækè is then called a 2-normed space. A standard example of a 2-norme

322 HENDRA GUNAWAN AND MASHADI èivè kx; y + zk çkx; yk + kx; zk: The pair èx; kæ; ækè is then called a 2-normed space. A standard example of a 2-norme SOOCHOW JOURNAL OF MATHEMATICS Volume 27, No. 3, pp. 321-329, July 2001 ON FINITE DIMENSIONAL 2-NORMED SPACES BY HENDRA GUNAWAN AND MASHADI Abstract. In this note, we shall study ænite dimensional 2-normed

More information

arxiv: v1 [math.na] 27 Jan 2016

arxiv: v1 [math.na] 27 Jan 2016 Virtual Element Method for fourth order problems: L 2 estimates Claudia Chinosi a, L. Donatella Marini b arxiv:1601.07484v1 [math.na] 27 Jan 2016 a Dipartimento di Scienze e Innovazione Tecnologica, Università

More information

arxiv: v1 [math.na] 27 Jan 2016

arxiv: v1 [math.na] 27 Jan 2016 Virtual Element Method for fourth order problems: L 2 estimates Claudia Chinosi a, L. Donatella Marini b arxiv:1601.07484v1 [math.na] 27 Jan 2016 a Dipartimento di Scienze e Innovazione Tecnologica, Università

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

ENERGY NORM A POSTERIORI ERROR ESTIMATES FOR MIXED FINITE ELEMENT METHODS

ENERGY NORM A POSTERIORI ERROR ESTIMATES FOR MIXED FINITE ELEMENT METHODS ENERGY NORM A POSTERIORI ERROR ESTIMATES FOR MIXED FINITE ELEMENT METHODS CARLO LOVADINA AND ROLF STENBERG Abstract The paper deals with the a-posteriori error analysis of mixed finite element methods

More information

APPENDIX E., where the boundary values on the sector are given by = 0. n=1. a 00 n + 1 r a0 n, n2

APPENDIX E., where the boundary values on the sector are given by = 0. n=1. a 00 n + 1 r a0 n, n2 APPENDIX E Solutions to Problem Set 5. èproblem 4.5.4 in textè èaè Use a series expansion to nd èèr;è satisfying èe.è è rr + r è r + r 2 è =, r 2 cosèè in the sector éré3, 0 éé 2, where the boundary values

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

Local flux mimetic finite difference methods

Local flux mimetic finite difference methods Local flux mimetic finite difference methods Konstantin Lipnikov Mikhail Shashkov Ivan Yotov November 4, 2005 Abstract We develop a local flux mimetic finite difference method for second order elliptic

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

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

A New Invariance Property of Lyapunov Characteristic Directions S. Bharadwaj and K.D. Mease Mechanical and Aerospace Engineering University of Califor

A New Invariance Property of Lyapunov Characteristic Directions S. Bharadwaj and K.D. Mease Mechanical and Aerospace Engineering University of Califor A New Invariance Property of Lyapunov Characteristic Directions S. Bharadwaj and K.D. Mease Mechanical and Aerospace Engineering University of California, Irvine, California, 92697-3975 Email: sanjay@eng.uci.edu,

More information

Analysis of Two-Grid Methods for Nonlinear Parabolic Equations by Expanded Mixed Finite Element Methods

Analysis of Two-Grid Methods for Nonlinear Parabolic Equations by Expanded Mixed Finite Element Methods Advances in Applied athematics and echanics Adv. Appl. ath. ech., Vol. 1, No. 6, pp. 830-844 DOI: 10.408/aamm.09-m09S09 December 009 Analysis of Two-Grid ethods for Nonlinear Parabolic Equations by Expanded

More information

Hamburger Beiträge zur Angewandten Mathematik

Hamburger Beiträge zur Angewandten Mathematik Hamburger Beiträge zur Angewandten Mathematik Numerical analysis of a control and state constrained elliptic control problem with piecewise constant control approximations Klaus Deckelnick and Michael

More information

Simple Examples on Rectangular Domains

Simple Examples on Rectangular Domains 84 Chapter 5 Simple Examples on Rectangular Domains In this chapter we consider simple elliptic boundary value problems in rectangular domains in R 2 or R 3 ; our prototype example is the Poisson equation

More information

A NOTE ON THE LADYŽENSKAJA-BABUŠKA-BREZZI CONDITION

A NOTE ON THE LADYŽENSKAJA-BABUŠKA-BREZZI CONDITION A NOTE ON THE LADYŽENSKAJA-BABUŠKA-BREZZI CONDITION JOHNNY GUZMÁN, ABNER J. SALGADO, AND FRANCISCO-JAVIER SAYAS Abstract. The analysis of finite-element-like Galerkin discretization techniques for the

More information

Continuity of Bçezier patches. Jana Pçlnikovça, Jaroslav Plaçcek, Juraj ç Sofranko. Faculty of Mathematics and Physics. Comenius University

Continuity of Bçezier patches. Jana Pçlnikovça, Jaroslav Plaçcek, Juraj ç Sofranko. Faculty of Mathematics and Physics. Comenius University Continuity of Bezier patches Jana Plnikova, Jaroslav Placek, Juraj Sofranko Faculty of Mathematics and Physics Comenius University Bratislava Abstract The paper is concerned about the question of smooth

More information

with Applications to Elasticity and Compressible Flow Daoqi Yang March 20, 1997 Abstract

with Applications to Elasticity and Compressible Flow Daoqi Yang March 20, 1997 Abstract Stabilized Schemes for Mixed Finite Element Methods with Applications to Elasticity and Compressible Flow Problems Daoqi Yang March 20, 1997 Abstract Stabilized iterative schemes for mixed nite element

More information

A Finite Element Method Using Singular Functions for Poisson Equations: Mixed Boundary Conditions

A Finite Element Method Using Singular Functions for Poisson Equations: Mixed Boundary Conditions A Finite Element Method Using Singular Functions for Poisson Equations: Mixed Boundary Conditions Zhiqiang Cai Seokchan Kim Sangdong Kim Sooryun Kong Abstract In [7], we proposed a new finite element method

More information

Yongdeok Kim and Seki Kim

Yongdeok Kim and Seki Kim J. Korean Math. Soc. 39 (00), No. 3, pp. 363 376 STABLE LOW ORDER NONCONFORMING QUADRILATERAL FINITE ELEMENTS FOR THE STOKES PROBLEM Yongdeok Kim and Seki Kim Abstract. Stability result is obtained for

More information

Economics Midterm Answer Key. Q1 èiè In this question we have a Marshallian demand function with arguments Cèp;mè =

Economics Midterm Answer Key. Q1 èiè In this question we have a Marshallian demand function with arguments Cèp;mè = Economics 7 997 Midterm Answer Key PART A Q èiè In this question we have a Marshallian demand function with arguments Cè;mè = Cè; w; w Lè. We can determine this function from the solution to max fc;lg

More information

An interpolation operator for H 1 functions on general quadrilateral and hexahedral meshes with hanging nodes

An interpolation operator for H 1 functions on general quadrilateral and hexahedral meshes with hanging nodes An interpolation operator for H 1 functions on general quadrilateral and hexahedral meshes with hanging nodes Vincent Heuveline Friedhelm Schieweck Abstract We propose a Scott-Zhang type interpolation

More information

also has x æ as a local imizer. Of course, æ is typically not known, but an algorithm can approximate æ as it approximates x æ èas the augmented Lagra

also has x æ as a local imizer. Of course, æ is typically not known, but an algorithm can approximate æ as it approximates x æ èas the augmented Lagra Introduction to sequential quadratic programg Mark S. Gockenbach Introduction Sequential quadratic programg èsqpè methods attempt to solve a nonlinear program directly rather than convert it to a sequence

More information

In the previous chapters we have presented synthesis methods for optimal H 2 and

In the previous chapters we have presented synthesis methods for optimal H 2 and Chapter 8 Robust performance problems In the previous chapters we have presented synthesis methods for optimal H 2 and H1 control problems, and studied the robust stabilization problem with respect to

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

Find (u,p;λ), with u 0 and λ R, such that u + p = λu in Ω, (2.1) div u = 0 in Ω, u = 0 on Γ.

Find (u,p;λ), with u 0 and λ R, such that u + p = λu in Ω, (2.1) div u = 0 in Ω, u = 0 on Γ. A POSTERIORI ESTIMATES FOR THE STOKES EIGENVALUE PROBLEM CARLO LOVADINA, MIKKO LYLY, AND ROLF STENBERG Abstract. We consider the Stokes eigenvalue problem. For the eigenvalues we derive both upper and

More information

Abstract. 1. Introduction

Abstract. 1. Introduction Journal of Computational Mathematics Vol.28, No.2, 2010, 273 288. http://www.global-sci.org/jcm doi:10.4208/jcm.2009.10-m2870 UNIFORM SUPERCONVERGENCE OF GALERKIN METHODS FOR SINGULARLY PERTURBED PROBLEMS

More information

ENERGY NORM A POSTERIORI ERROR ESTIMATES FOR MIXED FINITE ELEMENT METHODS

ENERGY NORM A POSTERIORI ERROR ESTIMATES FOR MIXED FINITE ELEMENT METHODS MATHEMATICS OF COMPUTATION Volume 75, Number 256, October 2006, Pages 1659 1674 S 0025-57180601872-2 Article electronically published on June 26, 2006 ENERGY NORM A POSTERIORI ERROR ESTIMATES FOR MIXED

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

Chapter 12. Partial di erential equations Di erential operators in R n. The gradient and Jacobian. Divergence and rotation

Chapter 12. Partial di erential equations Di erential operators in R n. The gradient and Jacobian. Divergence and rotation Chapter 12 Partial di erential equations 12.1 Di erential operators in R n The gradient and Jacobian We recall the definition of the gradient of a scalar function f : R n! R, as @f grad f = rf =,..., @f

More information

A Finite Element Method for an Ill-Posed Problem. Martin-Luther-Universitat, Fachbereich Mathematik/Informatik,Postfach 8, D Halle, Abstract

A Finite Element Method for an Ill-Posed Problem. Martin-Luther-Universitat, Fachbereich Mathematik/Informatik,Postfach 8, D Halle, Abstract A Finite Element Method for an Ill-Posed Problem W. Lucht Martin-Luther-Universitat, Fachbereich Mathematik/Informatik,Postfach 8, D-699 Halle, Germany Abstract For an ill-posed problem which has its origin

More information

Parabolic Layers, II

Parabolic Layers, II Discrete Approximations for Singularly Perturbed Boundary Value Problems with Parabolic Layers, II Paul A. Farrell, Pieter W. Hemker, and Grigori I. Shishkin Computer Science Program Technical Report Number

More information

Existence of minimizers for the pure displacement problem in nonlinear elasticity

Existence of minimizers for the pure displacement problem in nonlinear elasticity Existence of minimizers for the pure displacement problem in nonlinear elasticity Cristinel Mardare Université Pierre et Marie Curie - Paris 6, Laboratoire Jacques-Louis Lions, Paris, F-75005 France Abstract

More information

Adaptive methods for control problems with finite-dimensional control space

Adaptive methods for control problems with finite-dimensional control space Adaptive methods for control problems with finite-dimensional control space Saheed Akindeinde and Daniel Wachsmuth Johann Radon Institute for Computational and Applied Mathematics (RICAM) Austrian Academy

More information

An Accurate Fourier-Spectral Solver for Variable Coefficient Elliptic Equations

An Accurate Fourier-Spectral Solver for Variable Coefficient Elliptic Equations An Accurate Fourier-Spectral Solver for Variable Coefficient Elliptic Equations Moshe Israeli Computer Science Department, Technion-Israel Institute of Technology, Technion city, Haifa 32000, ISRAEL Alexander

More information

LECTURE 1: SOURCES OF ERRORS MATHEMATICAL TOOLS A PRIORI ERROR ESTIMATES. Sergey Korotov,

LECTURE 1: SOURCES OF ERRORS MATHEMATICAL TOOLS A PRIORI ERROR ESTIMATES. Sergey Korotov, LECTURE 1: SOURCES OF ERRORS MATHEMATICAL TOOLS A PRIORI ERROR ESTIMATES Sergey Korotov, Institute of Mathematics Helsinki University of Technology, Finland Academy of Finland 1 Main Problem in Mathematical

More information

arxiv: v2 [math.na] 23 Apr 2016

arxiv: v2 [math.na] 23 Apr 2016 Improved ZZ A Posteriori Error Estimators for Diffusion Problems: Conforming Linear Elements arxiv:508.009v2 [math.na] 23 Apr 206 Zhiqiang Cai Cuiyu He Shun Zhang May 2, 208 Abstract. In [8], we introduced

More information

P.B. Stark. January 29, 1998

P.B. Stark. January 29, 1998 Statistics 210B, Spring 1998 Class Notes P.B. Stark stark@stat.berkeley.edu www.stat.berkeley.eduèçstarkèindex.html January 29, 1998 Second Set of Notes 1 More on Testing and Conædence Sets See Lehmann,

More information

ETNA Kent State University

ETNA Kent State University Electronic Transactions on Numerical Analysis. Volume 37, pp. 166-172, 2010. Copyright 2010,. ISSN 1068-9613. ETNA A GRADIENT RECOVERY OPERATOR BASED ON AN OBLIQUE PROJECTION BISHNU P. LAMICHHANE Abstract.

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

The method of teepet decent i probably the bet known procedure for ænding aymptotic behavior of integral of the form Z è1è Ièè = gèzè e f èzè dz; C wh

The method of teepet decent i probably the bet known procedure for ænding aymptotic behavior of integral of the form Z è1è Ièè = gèzè e f èzè dz; C wh UNIFORM ASYMPTOTIC EXPANSIONS R. Wong Department of Mathematic City Univerity of Hong Kong Tat Chee Ave Kowloon, Hong Kong for NATOèASI Special Function 2000 1 The method of teepet decent i probably the

More information

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

SECOND ORDER TIME DISCONTINUOUS GALERKIN METHOD FOR NONLINEAR CONVECTION-DIFFUSION PROBLEMS

SECOND ORDER TIME DISCONTINUOUS GALERKIN METHOD FOR NONLINEAR CONVECTION-DIFFUSION PROBLEMS Proceedings of ALGORITMY 2009 pp. 1 10 SECOND ORDER TIME DISCONTINUOUS GALERKIN METHOD FOR NONLINEAR CONVECTION-DIFFUSION PROBLEMS MILOSLAV VLASÁK Abstract. We deal with a numerical solution of a scalar

More information

PSEUDO-COMPRESSIBILITY METHODS FOR THE UNSTEADY INCOMPRESSIBLE NAVIER-STOKES EQUATIONS

PSEUDO-COMPRESSIBILITY METHODS FOR THE UNSTEADY INCOMPRESSIBLE NAVIER-STOKES EQUATIONS PSEUDO-COMPRESSIBILITY METHODS FOR THE UNSTEADY INCOMPRESSIBLE NAVIER-STOKES EQUATIONS Jie Shen Department of Mathematics, Penn State University University Par, PA 1680, USA Abstract. We present in this

More information

LECTURE NOTES ELEMENTARY NUMERICAL METHODS. Eusebius Doedel

LECTURE NOTES ELEMENTARY NUMERICAL METHODS. Eusebius Doedel LECTURE NOTES on ELEMENTARY NUMERICAL METHODS Eusebius Doedel TABLE OF CONTENTS Vector and Matrix Norms 1 Banach Lemma 20 The Numerical Solution of Linear Systems 25 Gauss Elimination 25 Operation Count

More information

Priority Program 1253

Priority Program 1253 Deutsche Forschungsgemeinschaft Priority Program 1253 Optimization with Partial Differential Equations Klaus Deckelnick and Michael Hinze A note on the approximation of elliptic control problems with bang-bang

More information

Finite Element Superconvergence Approximation for One-Dimensional Singularly Perturbed Problems

Finite Element Superconvergence Approximation for One-Dimensional Singularly Perturbed Problems Finite Element Superconvergence Approximation for One-Dimensional Singularly Perturbed Problems Zhimin Zhang Department of Mathematics Wayne State University Detroit, MI 48202 Received 19 July 2000; accepted

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

A posteriori error estimates for non conforming approximation of eigenvalue problems

A posteriori error estimates for non conforming approximation of eigenvalue problems A posteriori error estimates for non conforming approximation of eigenvalue problems E. Dari a, R. G. Durán b and C. Padra c, a Centro Atómico Bariloche, Comisión Nacional de Energía Atómica and CONICE,

More information

problem of detection naturally arises in technical diagnostics, where one is interested in detecting cracks, corrosion, or any other defect in a sampl

problem of detection naturally arises in technical diagnostics, where one is interested in detecting cracks, corrosion, or any other defect in a sampl In: Structural and Multidisciplinary Optimization, N. Olhoæ and G. I. N. Rozvany eds, Pergamon, 1995, 543í548. BOUNDS FOR DETECTABILITY OF MATERIAL'S DAMAGE BY NOISY ELECTRICAL MEASUREMENTS Elena CHERKAEVA

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

QUADRILATERAL H(DIV) FINITE ELEMENTS

QUADRILATERAL H(DIV) FINITE ELEMENTS QUADRILATERAL H(DIV) FINITE ELEMENTS DOUGLAS N. ARNOLD, DANIELE BOFFI, AND RICHARD S. FALK Abstract. We consider the approximation properties of quadrilateral finite element spaces of vector fields defined

More information

on! 0, 1 and 2 In the Zienkiewicz-Zhu SPR p 1 and p 2 are obtained by solving the locally discrete least-squares p

on! 0, 1 and 2 In the Zienkiewicz-Zhu SPR p 1 and p 2 are obtained by solving the locally discrete least-squares p Analysis of a Class of Superconvergence Patch Recovery Techniques for Linear and Bilinear Finite Elements Bo Li Zhimin Zhang y Abstract Mathematical proofs are presented for the derivative superconvergence

More information

The Nearest Doubly Stochastic Matrix to a Real Matrix with the same First Moment

The Nearest Doubly Stochastic Matrix to a Real Matrix with the same First Moment he Nearest Doubly Stochastic Matrix to a Real Matrix with the same First Moment William Glunt 1, homas L. Hayden 2 and Robert Reams 2 1 Department of Mathematics and Computer Science, Austin Peay State

More information

ICES REPORT A Generalized Mimetic Finite Difference Method and Two-Point Flux Schemes over Voronoi Diagrams

ICES REPORT A Generalized Mimetic Finite Difference Method and Two-Point Flux Schemes over Voronoi Diagrams ICS RPORT 15-17 July 2015 A Generalized Mimetic Finite Difference Method and Two-Point Flux Schemes over Voronoi Diagrams by Omar Al-Hinai, Mary F. Wheeler, Ivan Yotov The Institute for Computational ngineering

More information

Mimetic Finite Difference methods

Mimetic Finite Difference methods Mimetic Finite Difference methods An introduction Andrea Cangiani Università di Roma La Sapienza Seminario di Modellistica Differenziale Numerica 2 dicembre 2008 Andrea Cangiani (IAC CNR) mimetic finite

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

WEAK GALERKIN FINITE ELEMENT METHODS ON POLYTOPAL MESHES

WEAK GALERKIN FINITE ELEMENT METHODS ON POLYTOPAL MESHES INERNAIONAL JOURNAL OF NUMERICAL ANALYSIS AND MODELING Volume 12, Number 1, Pages 31 53 c 2015 Institute for Scientific Computing and Information WEAK GALERKIN FINIE ELEMEN MEHODS ON POLYOPAL MESHES LIN

More information

A FAST SOLVER FOR ELLIPTIC EQUATIONS WITH HARMONIC COEFFICIENT APPROXIMATIONS

A FAST SOLVER FOR ELLIPTIC EQUATIONS WITH HARMONIC COEFFICIENT APPROXIMATIONS Proceedings of ALGORITMY 2005 pp. 222 229 A FAST SOLVER FOR ELLIPTIC EQUATIONS WITH HARMONIC COEFFICIENT APPROXIMATIONS ELENA BRAVERMAN, MOSHE ISRAELI, AND ALEXANDER SHERMAN Abstract. Based on a fast subtractional

More information

M.A. Botchev. September 5, 2014

M.A. Botchev. September 5, 2014 Rome-Moscow school of Matrix Methods and Applied Linear Algebra 2014 A short introduction to Krylov subspaces for linear systems, matrix functions and inexact Newton methods. Plan and exercises. M.A. Botchev

More information

INTERGRID OPERATORS FOR THE CELL CENTERED FINITE DIFFERENCE MULTIGRID ALGORITHM ON RECTANGULAR GRIDS. 1. Introduction

INTERGRID OPERATORS FOR THE CELL CENTERED FINITE DIFFERENCE MULTIGRID ALGORITHM ON RECTANGULAR GRIDS. 1. Introduction Trends in Mathematics Information Center for Mathematical Sciences Volume 9 Number 2 December 2006 Pages 0 INTERGRID OPERATORS FOR THE CELL CENTERED FINITE DIFFERENCE MULTIGRID ALGORITHM ON RECTANGULAR

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

A.V. SAVKIN AND I.R. PETERSEN uncertain systems in which the uncertainty satisæes a certain integral quadratic constraint; e.g., see ë5, 6, 7ë. The ad

A.V. SAVKIN AND I.R. PETERSEN uncertain systems in which the uncertainty satisæes a certain integral quadratic constraint; e.g., see ë5, 6, 7ë. The ad Journal of Mathematical Systems, Estimation, and Control Vol. 6, No. 3, 1996, pp. 1í14 cæ 1996 Birkhíauser-Boston Robust H 1 Control of Uncertain Systems with Structured Uncertainty æ Andrey V. Savkin

More information

A Remark on the Regularity of Solutions of Maxwell s Equations on Lipschitz Domains

A Remark on the Regularity of Solutions of Maxwell s Equations on Lipschitz Domains A Remark on the Regularity of Solutions of Maxwell s Equations on Lipschitz Domains Martin Costabel Abstract Let u be a vector field on a bounded Lipschitz domain in R 3, and let u together with its divergence

More information

A local-structure-preserving local discontinuous Galerkin method for the Laplace equation

A local-structure-preserving local discontinuous Galerkin method for the Laplace equation A local-structure-preserving local discontinuous Galerkin method for the Laplace equation Fengyan Li 1 and Chi-Wang Shu 2 Abstract In this paper, we present a local-structure-preserving local discontinuous

More information

Error estimates for the Raviart-Thomas interpolation under the maximum angle condition

Error estimates for the Raviart-Thomas interpolation under the maximum angle condition Error estimates for the Raviart-Thomas interpolation under the maximum angle condition Ricardo G. Durán and Ariel L. Lombardi Abstract. The classical error analysis for the Raviart-Thomas interpolation

More information

256 Facta Universitatis ser.: Elect. and Energ. vol. 11, No.2 è1998è primarily concerned with narrow range of frequencies near ærst resonance èwhere s

256 Facta Universitatis ser.: Elect. and Energ. vol. 11, No.2 è1998è primarily concerned with narrow range of frequencies near ærst resonance èwhere s FACTA UNIVERSITATIS èni ç Sè Series: Electronics and Energetics vol. 11, No.2 è1998è, 255-261 EFFICIENT CALCULATION OF RADAR CROSS SECTION FOR FINITE STRIP ARRAY ON DIELECTRIC SLAB Borislav Popovski and

More information

Pointwise convergence rate for nonlinear conservation. Eitan Tadmor and Tao Tang

Pointwise convergence rate for nonlinear conservation. Eitan Tadmor and Tao Tang Pointwise convergence rate for nonlinear conservation laws Eitan Tadmor and Tao Tang Abstract. We introduce a new method to obtain pointwise error estimates for vanishing viscosity and nite dierence approximations

More information

On the Optimal Insulation of Conductors 1

On the Optimal Insulation of Conductors 1 JOURNAL OF OPTIMIZATION THEORY AND APPLICATIONS: Vol. 100, No. 2, pp. 253-263, FEBRUARY 1999 On the Optimal Insulation of Conductors 1 S. J. COX, 2 B. KAWOHL, 3 AND P. X. UHLIG 4 Communicated by K. A.

More information

Numerical Solutions of Laplacian Problems over L-Shaped Domains and Calculations of the Generalized Stress Intensity Factors

Numerical Solutions of Laplacian Problems over L-Shaped Domains and Calculations of the Generalized Stress Intensity Factors WCCM V Fifth World Congress on Computational Mechanics July 7-2, 2002, Vienna, Austria Eds.: H.A. Mang, F.G. Rammerstorfer, J. Eberhardsteiner Numerical Solutions of Laplacian Problems over L-Shaped Domains

More information

STABILIZED DISCONTINUOUS FINITE ELEMENT APPROXIMATIONS FOR STOKES EQUATIONS

STABILIZED DISCONTINUOUS FINITE ELEMENT APPROXIMATIONS FOR STOKES EQUATIONS STABILIZED DISCONTINUOUS FINITE ELEMENT APPROXIMATIONS FOR STOKES EQUATIONS RAYTCHO LAZAROV AND XIU YE Abstract. In this paper, we derive two stabilized discontinuous finite element formulations, symmetric

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 Nonconformity and the Consistency Error First Strang Lemma Abstract Error Estimate

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

A Two-Grid Stabilization Method for Solving the Steady-State Navier-Stokes Equations

A Two-Grid Stabilization Method for Solving the Steady-State Navier-Stokes Equations A Two-Grid Stabilization Method for Solving the Steady-State Navier-Stokes Equations Songul Kaya and Béatrice Rivière Abstract We formulate a subgrid eddy viscosity method for solving the steady-state

More information

IMPROVED LEAST-SQUARES ERROR ESTIMATES FOR SCALAR HYPERBOLIC PROBLEMS, 1

IMPROVED LEAST-SQUARES ERROR ESTIMATES FOR SCALAR HYPERBOLIC PROBLEMS, 1 Computational Methods in Applied Mathematics Vol. 1, No. 1(2001) 1 8 c Institute of Mathematics IMPROVED LEAST-SQUARES ERROR ESTIMATES FOR SCALAR HYPERBOLIC PROBLEMS, 1 P.B. BOCHEV E-mail: bochev@uta.edu

More information

Math 302 Outcome Statements Winter 2013

Math 302 Outcome Statements Winter 2013 Math 302 Outcome Statements Winter 2013 1 Rectangular Space Coordinates; Vectors in the Three-Dimensional Space (a) Cartesian coordinates of a point (b) sphere (c) symmetry about a point, a line, and a

More information

A Locking-Free MHM Method for Elasticity

A Locking-Free MHM Method for Elasticity Trabalho apresentado no CNMAC, Gramado - RS, 2016. Proceeding Series of the Brazilian Society of Computational and Applied Mathematics A Locking-Free MHM Method for Elasticity Weslley S. Pereira 1 Frédéric

More information

Lecture 2: Review of Prerequisites. Table of contents

Lecture 2: Review of Prerequisites. Table of contents Math 348 Fall 217 Lecture 2: Review of Prerequisites Disclaimer. As we have a textbook, this lecture note is for guidance and supplement only. It should not be relied on when preparing for exams. In this

More information

A Spectral Method for Elliptic Equations: The Neumann Problem

A Spectral Method for Elliptic Equations: The Neumann Problem A Spectral Method for Elliptic Equations: The Neumann Problem Kendall Atkinson Departments of Mathematics & Computer Science The University of Iowa David Chien, Olaf Hansen Department of Mathematics California

More information

A Least-Squares Finite Element Approximation for the Compressible Stokes Equations

A Least-Squares Finite Element Approximation for the Compressible Stokes Equations A Least-Squares Finite Element Approximation for the Compressible Stokes Equations Zhiqiang Cai, 1 Xiu Ye 1 Department of Mathematics, Purdue University, 1395 Mathematical Science Building, West Lafayette,

More information

LECTURE 16 GAUSS QUADRATURE In general for Newton-Cotes (equispaced interpolation points/ data points/ integration points/ nodes).

LECTURE 16 GAUSS QUADRATURE In general for Newton-Cotes (equispaced interpolation points/ data points/ integration points/ nodes). CE 025 - Lecture 6 LECTURE 6 GAUSS QUADRATURE In general for ewton-cotes (equispaced interpolation points/ data points/ integration points/ nodes). x E x S fx dx hw' o f o + w' f + + w' f + E 84 f 0 f

More information

Projected Surface Finite Elements for Elliptic Equations

Projected Surface Finite Elements for Elliptic Equations Available at http://pvamu.edu/aam Appl. Appl. Math. IN: 1932-9466 Vol. 8, Issue 1 (June 2013), pp. 16 33 Applications and Applied Mathematics: An International Journal (AAM) Projected urface Finite Elements

More information

Improved Newton s method with exact line searches to solve quadratic matrix equation

Improved Newton s method with exact line searches to solve quadratic matrix equation Journal of Computational and Applied Mathematics 222 (2008) 645 654 wwwelseviercom/locate/cam Improved Newton s method with exact line searches to solve quadratic matrix equation Jian-hui Long, Xi-yan

More information

Course Summary Math 211

Course Summary Math 211 Course Summary Math 211 table of contents I. Functions of several variables. II. R n. III. Derivatives. IV. Taylor s Theorem. V. Differential Geometry. VI. Applications. 1. Best affine approximations.

More information

Local discontinuous Galerkin methods for elliptic problems

Local discontinuous Galerkin methods for elliptic problems COMMUNICATIONS IN NUMERICAL METHODS IN ENGINEERING Commun. Numer. Meth. Engng 2002; 18:69 75 [Version: 2000/03/22 v1.0] Local discontinuous Galerkin methods for elliptic problems P. Castillo 1 B. Cockburn

More information

EQUADIFF 6. Jean-Claude Nédélec Mixed finite element in 3D in H(div) and H(curl) Terms of use:

EQUADIFF 6. Jean-Claude Nédélec Mixed finite element in 3D in H(div) and H(curl) Terms of use: EQUADIFF 6 Jean-Claude Nédélec Mixed finite element in 3D in H(div) and H(curl) In: Jaromír Vosmanský and Miloš Zlámal (eds.): Equadiff 6, Proceedings of the International Conference on Differential Equations

More information

Journal of Universal Computer Science, vol. 3, no. 11 (1997), submitted: 8/8/97, accepted: 16/10/97, appeared: 28/11/97 Springer Pub. Co.

Journal of Universal Computer Science, vol. 3, no. 11 (1997), submitted: 8/8/97, accepted: 16/10/97, appeared: 28/11/97 Springer Pub. Co. Journal of Universal Computer Science, vol. 3, no. 11 (1997), 1250-1254 submitted: 8/8/97, accepted: 16/10/97, appeared: 28/11/97 Springer Pub. Co. Sequential Continuity of Linear Mappings in Constructive

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

A mixed finite element approximation of the Stokes equations with the boundary condition of type (D+N)

A mixed finite element approximation of the Stokes equations with the boundary condition of type (D+N) wwwijmercom Vol2, Issue1, Jan-Feb 2012 pp-464-472 ISSN: 2249-6645 A mixed finite element approximation of the Stokes equations with the boundary condition of type (D+N) Jaouad El-Mekkaoui 1, Abdeslam Elakkad

More information

The Kalman filter is arguably one of the most notable algorithms

The Kalman filter is arguably one of the most notable algorithms LECTURE E NOTES «Kalman Filtering with Newton s Method JEFFREY HUMPHERYS and JEREMY WEST The Kalman filter is arguably one of the most notable algorithms of the 0th century [1]. In this article, we derive

More information

Preliminary Examination in Numerical Analysis

Preliminary Examination in Numerical Analysis Department of Applied Mathematics Preliminary Examination in Numerical Analysis August 7, 06, 0 am pm. Submit solutions to four (and no more) of the following six problems. Show all your work, and justify

More information

Recurrence Relations and Fast Algorithms

Recurrence Relations and Fast Algorithms Recurrence Relations and Fast Algorithms Mark Tygert Research Report YALEU/DCS/RR-343 December 29, 2005 Abstract We construct fast algorithms for decomposing into and reconstructing from linear combinations

More information

Numerische Mathematik

Numerische Mathematik Numer. Math. (2003) 94: 195 202 Digital Object Identifier (DOI) 10.1007/s002110100308 Numerische Mathematik Some observations on Babuška and Brezzi theories Jinchao Xu, Ludmil Zikatanov Department of Mathematics,

More information

Remarks on the analysis of finite element methods on a Shishkin mesh: are Scott-Zhang interpolants applicable?

Remarks on the analysis of finite element methods on a Shishkin mesh: are Scott-Zhang interpolants applicable? Remarks on the analysis of finite element methods on a Shishkin mesh: are Scott-Zhang interpolants applicable? Thomas Apel, Hans-G. Roos 22.7.2008 Abstract In the first part of the paper we discuss minimal

More information

On Pressure Stabilization Method and Projection Method for Unsteady Navier-Stokes Equations 1

On Pressure Stabilization Method and Projection Method for Unsteady Navier-Stokes Equations 1 On Pressure Stabilization Method and Projection Method for Unsteady Navier-Stokes Equations 1 Jie Shen Department of Mathematics, Penn State University University Park, PA 1682 Abstract. We present some

More information