WEIGHTED ERROR ESTIMATES OF THE CONTINUOUS INTERIOR PENALTY METHOD FOR SINGULARLY PERTURBED PROBLEMS

Size: px
Start display at page:

Download "WEIGHTED ERROR ESTIMATES OF THE CONTINUOUS INTERIOR PENALTY METHOD FOR SINGULARLY PERTURBED PROBLEMS"

Transcription

1 WEIGHTED ERROR ESTIMATES OF THE CONTINUOUS INTERIOR PENALTY METHOD FOR SINGULARLY PERTURBED PROBLEMS ERIK BURMAN, JOHNNY GUZMÁN, AND DMITRIY LEYKEKHMAN Abstract. In tis paper we analyze local properties of te Continuous Interior Penalty CIP Metod for a model convection-doated singularly perturbed convection-diffusion problem. We sow weigted a priori error estimates, were te weigt function exponentially decays outside te subdomain of interest. Tis result sows tat locally, te CIP metod is comparable to te Streamline Diffusion SD or te Discontinuous Galerkin DG metods. 1. Introduction Te Continuous Interior Penalty CIP metod was originally proposed by Douglas and Dupont [9] for parabolic and elliptic equations. Te idea was to add a penalization term on te gradient jumps in order to increase robustness for elliptic problems wit a doating convection term. Te case of optimal convergence in te ig Péclet number regime was analyzed by Burman and Hansbo [2] and Burman [3] for first order conforg and non-conforg approximation and in te framework of p-finite elements by Burman and Ern [5]. In tis paper, we are interested in approximating te solution u of te following model problem 1.1 ε u + u x + u = f in, u = 0 on, were is a polygonal domain, 0 < ε 1, and f L 2. Let U denote te approximate solution and te quality of te mes. Typically te error is sown to satisfy u U L 2 C k+ 1 2 u H k+1 in te ig Péclet number regime, were k is te polynomial order and assug u as sufficient regularity. Optimal convergence in of te error in te streamline derivative can also be derived. Tese results are similar to te typical estimates for oter stabilized metods suc as Discontinuous Galerkin DG metod or te Streamline-Diffusion SD metod [13]. Te rigt-and side of tese estimates, owever, depends on a global Sobolev norm. Tis norm may be large in te presence of layers. Terefore te estimates can be considered to be of practical interest only in case te solution is smoot. Date: Feb 07, Matematics Subject Classification. 65N30,65N15. Key words and prases. singularly perturbed, convection-diffusion, local error analysis, continuous interior penalty metod. 1

2 2 BURMAN, GUZMÁN, AND LEYKEKHMAN However, many problems of interest do not ave smoot solutions. Very often tey exibit nonsmoot beavior, like socks, boundary or interior layers and interface discontinuities. Model problem 1.1 for example is known to exibit internal parabolic layer of order O ε log 1/ε and exponential outflow layers of order Oε log 1/ε. In designing a numerical metod it is important to know ow te metod will beave in te neigborood of suc a discontinuity, and weter or not te resulting effects are global or local. For example wen under-resolved layers are present it is well known tat te standard Galerkin metod suffers from oscillations tat pollute te wole solution. One approac to assess te propagation of perturbations for a given metod is to prove tat te local error away from te possibly under-resolved layer as optimal convergence. An appealing way to prove suc results is by using weigted a priori error estimates to prove local error estimates as was done for te Streamline- Diffusion metod by Jonson, Nävert and Pitkäranta in te case of local L 2 -norm estimates [13]. Te analysis was ten extended by Jonson, Walbin and Scatz to L -norm estimates [14]. Recently, similar results were proved in te case of te residual free bubble metod by Sangalli [16] and te Discontinuous Galerkin metod by Guzmán [12]. Te CIP metod is one instance of a class of symmetric stabilization metods tat as received increasing interest lately. Examples of oter members of tis group are te subgrid viscosity metod proposed by Guermond [11], te ortogonal subscale stabilization proposed by Codina and Blasco [8], and te local projection stabilization proposed by Becker and Braack [1]. In tis work we prove weigted a priori error estimates for te CIP metod. To our knowledge tis is te first time suc estimates ave been proved for a metod from te class of symmetric stabilizations. We sow tat te CIP metod as te same upwind and crosswind error propagation properties as te SD-metod. In particular te penalization of te jumps of te crosswind derivative allows an improved estimate of te error in te crosswind derivative in te case of piecewise linear approximation. Compared to te DG-metod or te SD-metod te proof of local estimates for te CIP-metod is more involved. Tis is due to te fact tat te stabilization only controls te part of te streamline derivative tat is not in te finite element space. Terefore te desired control of te streamline derivative is obtained in a more implicit fasion tan for te oter metods. In order to avoid unnecessary tecnicalities we consider te simple model problem 1.1 wit constant convection velocity and reaction terms. Also, for te sake of simplicity, we only give te detailed proof for te case of first order polynomial approximation. However te present analysis extends to te more general case of ig polynomial order and also allows to prove L -norm error estimates. We will comment on tis in te final section of te paper. In te remaining part of tis section we will introduce te CIP-metod for problem 1.1 and we state te main results Te CIP metod. Let {T } be a one-parameter family of face-to-face triangulations of, wit = sup T T T, were T = diamt. Te triangulations are assumed to be globally quasi-uniform, i.e. if necessary after a renormalization

3 WEIGHTED ESTIMATES OF CIP METHOD 3 of, diam T Cmeas T 1/2 T T. Troug out tis paper we assume 0 < < 1. By V we will denote a finite dimensional space of continuous piecewise linear polynomial functions. We will not pose any boundary conditions on V. We let E be te collection of boundary edges, E 0 be te collection of interior edges corresponding to T, and E = E 0 E. Te Continuous Interior Penalty approximation U V is defined as te unique solution to 1.2 BU, v = fv, v V, were Bu, v = εau, v + Mu, v + J u, v + J u, v, wit Au, v = Mu, v = u v u n v + v n u ds + γ bc u x v + uv + uv n x ds, [u x ][v x ] ds, J u, v = 2 e E 0 e J u, v = 2 [u y ][v y ] ds + 1/2 e E 0 e uv n y 2 ds. uv ds, Here n = n x, n y is te outward unit vector to. Te inflow part of te boundary is given by = {x, y : n x x, y < 0}. Te constant γ bc is a boundary penalty parameter tat as to be cosen large enoug in order to guarantee stability. Te coice of 2 in te definition of te stability terms Ju, v is not essential and can be replaced wit oter expressions cf. [3]. Te jump operator is given by [v]x, y = lim t 0 +vx, y tn e vx, y + tn e, were e is a mes interior edge, x, y e, and n e is a fixed unit vector normal to te edge e. Te last term of J does not appear in te original definition of te CIP metod appearing in [3]. We added tis term in order to improve te widt of te crosswind layer to order maxε 1/2, 3/4 log1/. Witout tis term we can only prove tat te widt of te crosswind layer is of order 1/2 log1/. By te regularity teory u H 3 2 +δ for some δ > 0 cf. [10]. Hence u satisfies 1.2 and we ave te usual Galerkin ortogonality property 1.3 Bu U, v = 0, v V.

4 4 BURMAN, GUZMÁN, AND LEYKEKHMAN Figure 1. Sketc of 0 and + s 1.2. Main Result. Te main goal of tis paper is to obtain weigted a priori error estimates, were te weigt ω is O1 on =, x 0 ] [y 1, y 2 ] and decays exponentially outside of a sligtly larger subdomain cf. Figure 1. More precisely, te weigt ω is a positive function wit te following properties: C 1 ωx, y C 2, for x, y 0, ωx, y Ce x x0/k, for x x 0 +, ωx, y Ce y y2/kσ, for y y 2 +, ωx, y Ce y1 y/kσ, for y y 1. Here C 1 and C 2 are two fixed positive constants, K > 1 is a sufficiently large number, and σ, te size of te crosswind layer, is 1.5 σ := maxε 1/2, 3/4. Teorem 1.1. For every u H 1 and U V tat satisfy 1.3 wit ε, tere exists a constant C independent of u and U, suc tat 1/2 ωu U x L2 + 3/4 ωu U y L2 + Qu U C χ V Lu χ, were Q 2 v :=ε ω v 2 L ω ω x 1/2 v 2 L 2 + ωv 2 L ω[v x ] 2 L 2 e + ω[v y] 2 L 2 e e E ωv n x 1/2 2 L 2 + 1/2 ωvn y 2 L 2 + γ bcε 1 ωv 2 L 2

5 WEIGHTED ESTIMATES OF CIP METHOD 5 and L 2 v := ω v 2 L ωv 2 L e E ω[v x ] 2 L 2 e + ω[v y] 2 L 2 e. We point out tat by penalizing te crosswind derivative i.e. by including te term J in te metod we reduce te size of te crosswind layer for piecewise linear elements to order σ = max 3/4, ε 1/2 instead of order σ = 1/2. To give an application of te above result, let 0 be as in 1.4 and define + s = {x x 0 + sk log, y 1 skσ log y y 2 + skσ log }. Corollary 1.2. Under te assumptions of Teorem 1.1 and assug u H t, tere exists a constant C independent of u and suc tat for any s > 0, wit r u = 2, t. 1/2 u U x L /4 u U y L u U L 2 0 C ru 1 2 u H ru + s + Cs+ru 1 2 u H ru, Remark 1. Note tat in te above estimate r u can be cosen to take different values in te different subdomains in te presence of singularities. Te rest of te paper is organized as follows. Te next tree sections are devoted to te proof of Teorem 1.1 for te piecewise linear case and constitute te main part of te paper. In Section 2 and Section 3, we collect some preliary results wic are necessary in order to carry out te proof of te Teorem 1.1. Te proofs of Teorem 1.1 and Corollary 1.2 are presented in Section 4. Te last section addresses te possible extensions, generalizations, and concluding remarks. Finally in te Appendix, we provide te proofs of te tecnical results stated in Section Preliary Results In tis section we collect some results we require in our analysis. First we recall te standard trace and inverse inequalities. Te proofs can be found in many textbooks on finite elements cf. [6] Trace, inverse and interpolation inequalities. For T T and v H 1 T we ave 2.1 v L 2 T C tr 1/2 v L 2 T + 1/2 v L 2 T, were C tr is independent of T and v. If v P 1 T, ten 2.2 v L2 T C inv 1 v L2 T, 2.3 v L 2 T C inv 1/2 v L 2 T, were C inv is independent of and v. Let T T and v H 2 T, ten te interpolation inequality reads 2.4 v L2 T C 1 v L2 T + D 2 v L2 T, for some constant C independent of T and v.

6 6 BURMAN, GUZMÁN, AND LEYKEKHMAN 2.2. Te weigt function. In addition to te properties ω described above, we assume tat ω satisfies, ω x x, y < 0, for all x, y, Dx α Dy β ω CK α α K β σ β ω, for α + β 2, Dx α Dy β ω CK 1 α 1 α K β σ β ω x, for α 1, α + β 2, ROS, ω + ROS, ω x C ω, for any ball S of radius K, were ROS, v = max x S vx / x S vx. Te explicit construction of suc a function is given in [14]. Te last property of ω enables us to apply te inverse inequalities 2.2 and 2.3 to functions of te form ωv, wit v P 1 T. For example, 2.9 and similarly ω v L 2 T max x T ωx v L 2 T max x T ωxc inv 1 v L 2 T C inv C ω x T ωx 1 v L2 T C inv C ω 1 ωv L2 T, 2.10 ωv L 2 T C inv C ω 1/2 ωv L 2 T. Finally, we would like to note tat in te subsequent analysis it is important tat ω 1 = 1 ω as te same properties as ω. For example, 2.11 D x ω 1 = ω 2 D x ω ω 2 CK 1 1 ω = CK 1 1 ω 1 and for any ball of radius K, x T ωx 1 max x T ωx 2.12 ROS, ω 1 = max x T [ωx] 1 1 x T [ωx] 1 = = ROS, ω C ω Quasi-interpolant operator. Next we introduce a quasi-interpolant operator wic is similar to te Clément operator [7]. Te properties of tis interpolant are essential to carry out te error analysis of te CIP metod; see [2, 3, 4]. In tis paper, we use te quasi-interpolant Π presented in [4]. For simplicity we only consider te piecewise linear case. Te value of Πv at a node x i of our finite element space V i.e. te nodes defined by te Lagrangian polynomial basis is given by 2.13 Πvx i := 1 N i {T :x i T } v T x i. Here N i is te number of triangles in T tat contain x i. One of te properties of Π we will use in te subsequent analysis is te following L 2 -stability result 2.14 Πv L2 τ C v L2 τ, v L 2 τ. 3. Tecnical results In tis section we state several results in te form of lemmas, wic we require in order te carry out te proof of te main result. Te proofs of tese lemmas are quite tecnical and are given in te Appendix.

7 WEIGHTED ESTIMATES OF CIP METHOD Weigted estimates for a quasi-interpolant operator. Te first result is a weigted version of Lemma 3.1 in [4]. Tis result is important in our presentation and will be used often. Tis type of estimate is also used in te a posteriori error estimation for non-conforg finite element metods and can be considered to be of independent interest. Lemma 3.1. Let U V and te operator Π be given by Tere exists a constant C independent of U,, and K suc tat, 3.1 ωu x ΠU x 2 L 2 C e E ω 1 ω 2 U x Πω 2 U x 2 L 2 C e E 0 and ω 1 ω 2 U x Πω 2 U x 2 L 2 T C e E 0 ω[u x ] 2 L 2 e, ω[u x ] 2 L 2 e + CK 2 ωu x 2 L 2, ω[u x ] 2 L 2 e + CK 2 ωu x 2 L Estimates for te crosswind derivative. Te next result is special for te piecewise linear case. Tis lemma allows us to prove tat in piecewise linear case te size of te crosswind layer is of order σ = max 3/4, ε 1/2. We would like to point out tat te result below olds true for any function in V and does not use te fact tat U is te approximate solution. Lemma 3.2. Let U V. Tere exists a constant C independent of U and, suc tat 3/2 ωu y 2 L 2 C 2 ω[u y ] 2 L 2 e + ωu 2 L 2 + 1/2 ωun y 2 L 2. e E Estimates for te streamline derivative. In te next result we will give an estimate for te upwind term. In contrast to te above result, tis lemma makes a strong use of te fact tat U is an approximate solution to 1.1. One of te main tecnical tools used to prove te following lemma is te weigted quasi-interpolant results in Lemma 3.1. Lemma 3.3. Let U V and u H 1 satisfy 1.3. Tere is a constant C independent of U, u, and, suc tat ωu x 2 L 2 C Q 2 U + L 2 u Superapproximation. Te following superapproximation result is similar to te superapproximation result in [14] and [12]. Te difference from te above mentioned superapproximation results is tat instead of a local interpolant operator we ere use a global L 2 -projection. Because of tis, te lemma as an independent interest. Te ortogonal properties of te L 2 -projection will be essential in our proof of te teorem. Indeed, tis is wat allows us to control te convective term by te gradient jump operator.

8 8 BURMAN, GUZMÁN, AND LEYKEKHMAN Lemma 3.4. Let U V and u H 1 satisfy 1.3. Let P : L 2 V denote te L 2 -projection defined by 3.4 P v χ dx = v χ dx, χ V. Set E = ω 2 U P ω 2 U. Tere exists a constant C independent of U, u,, and K, suc tat ω 1 E 2 L 2 + ω 1 E 2 L 2 CK 1 Q 2 U + L 2 u. Remark 2. By using te trace inequality 2.1 and property of ω 2.8, we may include ω 1 E L2 on te left and side of Lemma 3.4, since, ω 1 E 2 L 2 x T [ωx] 2 E 2 L 2 T Ctr 2 x T [ωx] 2 1 E 2 L 2 T + E 2 L 2 T CtrC 2 ω 2 1 ω 1 E 2 L 2 + ω 1 E 2 L 2. Te last result states an important inequality wic we will use in te second part of te proof of Teorem 1.1. Te main tecnical tool used to prove tis lemma is te above superappoximation result, Lemma 3.4. Lemma 3.5. Let U V and u H 1 satisfy 1.3. Tere exist constants C and δ, te latter may be cosen arbitrary small, independent of U, u,, and K, and a constant C δ, wic depends on δ, but not on U, u,, and K, suc tat BU, ω 2 U CK 1 Q 2 U + CδQ 2 U + ωu x 2 L 2 + C δl 2 u. 4. Proof of Teorem 1.1 and Corollary 1.2 After collecting all te necessary tecnical results in te previous section, we are ready to present a proof of Teorem 1.1. We prove te teorem in several steps. Proof. Step 1: Reduction to weigted stability. In tis step we will prove tat it is sufficient to sow tat 4.1 QU C δ Lu + Cδ 1/2 ωu x L 2, for all U V and u H 1 tat satisfy 1.3. Here δ is a constant tat can be made as small as required. By te triangle inequality, we ave 1/2 ωu u x L2 + 3/4 ωu u y L2 + QU u 1/2 ωu χ x L2 + 3/4 ωu χ y L2 + QU χ + CLu χ. We ave used tat 1/2 ωu χ x L2 + 3/4 ωu χ y L2 + Qu χ CLu χ, wic follows from te properties of ω, te assumption ε, and te trace inequality Hence, it is enoug to sow tat for any χ V, 1/2 ωu χ x L 2 + 3/4 ωu χ y L 2 + QU χ CLu χ.

9 WEIGHTED ESTIMATES OF CIP METHOD 9 We note tat if χ V, ten Ũ := U χ V, ũ := u χ H 1 and Terefore, it is enoug to sow tat B Ũ ũ, v = 0, v V. 1/2 ωũx L2 + 3/4 ωũy L2 + QŨ CLũ. In view of Lemma 3.2, 3/4 ωũy L2 CQŨ. Moreover, using Lemma 3.3, it is sufficient to sow 4.2 QŨ C δlũ + Cδ 1/2 ωũx L 2, were δ is a constant tat can be made as small as desired. Using U and u instead of Ũ and ũ we see tat we need to sow 4.1 for all U V and u H 1 tat satisfy 1.3. Step 2: Relating Q 2 U to BU, ω 2 U. In tis step we will sow tat 4.3 Q 2 U = BU, ω 2 U 2ε Uω ω U 2ε U n ω2 U + U 2 ω ω n ds. Recalling te definition 1.2, we ave BU, ω 2 U = εau, ω 2 U + MU, ω 2 U + J U, ω 2 U + J U, ω 2 U = ε ω U 2 L 2 + 2ε Uω ω U U ε n ω2 U + ω2 U n U ds + εγ bc ωu 2 L 2 + U x ω 2 U + ωu 2 L 2 + ω 2 U 2 n x ds [Ux ][ω 2 U x ] + [U y ][ω 2 U y ] ds + 1/2 ωu n y 2 L e E 0 e Let us first treat te first term in MU, ω 2 U, namely U xω 2 U. By integration by parts and using tat ω x < 0, we ave U x ω 2 U = U 2 ωω x + 1 ω 2 U 2 n x ds 2 = ω ω x 1/2 U 2 L ω 2 U 2 n x ds + 1 ω 2 U 2 n x ds. 2 2 \ Because n x < 0 on, we ave ω 2 U 2 n x ds + 1 ω 2 U 2 n x ds + 1 ω 2 U 2 n x ds 2 2 \ = 1 ω 2 U 2 n x ds + 1 ω 2 U 2 n x ds 2 2 \ = 1 2 ωu n x 1/2 2 L 2.

10 10 BURMAN, GUZMÁN, AND LEYKEKHMAN Since, ω is smoot and U is continuous, te jump terms [ω 2 x U] and [ω 2 y U] vanis and we ave [Ux ][ω 2 U x ] + [U y ][ω 2 U y ] ds e = [Ux ][ω 2 x U + ω 2 U x ] + [U y ][ω 2 y U + ω 2 U y ] ds e = ω[u x ] 2 L 2 e + ω[u y] 2 L 2 e. Taking into account te above arguments, we ave sown 4.3. Step 3: Initial Estimate for Q 2 U. In tis step we will bound te last two terms on te rigt and side of 4.3 to sow 4.4 Q 2 U CBU, ω 2 U. For te second term appearing in te rigt and side of 4.3 we note tat Uω ω U = Uωω x U x + ω y U y, ence by te Caucy-Scwarz inequality, 2.6, and using tat ε, ε Uωω x U x ε ωu x L 2 ω x U L ε ωu x L 2 C 1/2 K 1/2 ω ω x 1/2 U L 2 CK 1/2 ε ω U 2 L 2 + ε 1 ω ω x 1/2 U 2 L 2 CK 1/2 Q 2 U. Similarly, by 2.6 and using tat by definition 1.5 σ = max ε 1/2, 3/4 and terefore ε σ 2, 4.6 ε ωuω y U y ε ωu y L2 ω y U L2 ε ωu y L2 Cσ 1 K 1 ωu L2 CK 1 ε ω U 2 L 2 + εσ 2 ωu 2 L 2 CK 1 Q 2 U. Combining, 4.5 and 4.6, we obtain 4.7 ε ωu ω U CK 1/2 Q 2 U. Next we will estimate te boundary term. By te Caucy-Scwarz inequality, te property 2.8 of ω, and te trace inequality 2.10, U ε n ω2 U ds ε ω U n L 2 ωu L2 εc ω C inv 1/2 ω U L2 ωu L2 ε 4 ω U 2 L 2 + ωc 2 εc2 inv ωu 2 L 2. Also, using 2.6 and 2.8, we see tat ε U 2 ω ω n ds ε CK 1 ωu 2 L 2.

11 WEIGHTED ESTIMATES OF CIP METHOD 11 Terefore, if γ bc > 2C 2 inv C2 ω and by taking K large enoug, we ave U 4.8 2ε n ω2 U + U 2 ω ω ds 1 n 2 Q2 U. If we substitute 4.8 and 4.7 into 4.3 and coose K sufficienlty large we ave 4.4. Step 4: Final estimate for Q 2 U. Applying Lemma 3.5 to te rigt and side of 4.4, we obtain Q 2 U CK 1 Q 2 U + CδQ 2 U + ωu x 2 L 2 + C δl 2 u. Coosing K large and δ small, sows 4.1. Tis proves te teorem. Now we will prove Corollary 1.2 Proof. By Teorem 1.1 and te triangle inequality, for any χ V we ave, 1/2 u U x L /4 u U y L2 0 + u U L2 0 C 1/2 ωu U x L 2 + 3/4 ωu U y L 2 + ωu U L 2 C Lu χ C 1/2 u χ L 2 + s + 1/2 u χ L 2 + s + + 1/2 ω u χ L 2 \ + s + 1/2 ωu χ L 2 \ + s 1/2 + ω u χ 2 L 2 e. e E \ + s e E + s u χ 2 L 2 e Taking χ = I u to be te Lagrange interpolant of u and using te approximation teory togeter wit te trace inequality 2.1, we obtain 1 2 1/2 u I u L 2 + s + 1/2 u I u L 2 + s + u I u 2 L 2 e 1 2 C ru 1 2 u H ru + s. e E + s For te oter terms using tat ω = O s outside of + s, we ave 1/2 ω u I u L2 \ + s + 1/2 ωu I u L2 \ + s 1/2 + ω u I u 2 L e 2 e E \ + s 1/2 C s 1/2 u I u L2 + 1/2 u I u L2 + u I u 2 L 2 e. e E

12 12 BURMAN, GUZMÁN, AND LEYKEKHMAN Similarly, by te trace inequality 2.1 and te approximation teory, we ave 1/2 u I u L 2 + 1/2 u I u L 2 + u I u 2 L 2 e e E C 1/2 u I u L 2 + 1/2 u I u L 2 + ru 1 2 D r u u L C ru 1 2 u H ru. Tis proves te corollary. 5. Extensions and concluding remarks Tere are several extensions of te above analysis tat are straigtforward altoug tecnical. Below we will comment on some of te more important ones Hig-order elements. Te extension of te present analysis to ig-order elements is straigtforward. In our analysis, only Lemma 3.2 makes strong use of te vanising second derivative in an essential way. It seems tat te crosswind jump term i.e. J is useful to control crosswind smearing only in te piecewise linear case. It remains unclear weter te crosswind stabilization for ig-order elements as any effect. However, we can still prove a sligtly weaker result tan Teorem 1.1 of te form 5.1 1/2 ωu U x L 2 + Qu U C χ V Lu χ, were we need to take σ = 1/2. Tis means tat we can only prove tat te crosswind layer is of size 1/2 rater tan max 3/4, ε 1/2. It is not clear if tis result is sarp for ig-order elements. Since te Lemma 3.2 does not old for k 2, in order to establis 5.1 one will need a sligtly different superapproximation result of te form ω 1 E 2 L 2 + ω 1 E 2 L 2 CK 1 ωu 2 L 2 + ωω x 1/2 U 2 L L -norm estimates. In tis paper we ave presented weigted error estimates in L 2 -norm. Using similar weigted stability estimates one can prove suboptimal L -norm estimates in regions away from boundary layers. More precisely, if we use te tecnique in [15], we can prove u U L 0 C 11/8, were 0 and te distance from 0 to te outflow part of te boundary of is at least K log1/ for a sufficiently large constant K No reaction term and variable convection. In our presentation, for simplicity, we assumed te reaction term in 1.1 is just u. However, if instead we ave used te weigt function presented in [12, 16], we could ave sown te results for a more general problem 5.3 ε u + β u + cu = f in, u = 0 were β [L ] 2 and c L, cx 0. on,

13 WEIGHTED ESTIMATES OF CIP METHOD Summary and concluding remarks. We ave proved a weigted a priori error estimate for te CIP metod applied to a convection doated second order convection-diffusion equation. We also proved tat by including extra terms we can reduce te size of numerical layer in te crosswind direction. Typically te numerical layer in te upwind direction is of order O log 1/ wereas it is of order O 1/2 log 1/ in te crosswind direction. In our analysis of te piecewise linear case by adding an extra penalty term and using σ = max 3/4, ε 1/2, we are able to reduce te size of te crosswind numerical layer to order O 3/4 log 1/. Te reason we were able to accomplis tis is tat once Lemma 3.2 was establised, we could use te superapproximation result, Lemma 3.4. Te reason tat Lemma 3.2 olds is tat we ave included te term J in te definition of te CIP metod. In oter words, penalizing te jumps of te crosswind derivative allows us to reduce te size of te crosswind numerical layer to σ = maxε 1/2, 3/4 in te case of piecewise linear approximations. Tis argument is similar to te results of [14], were more crosswind diffusion was added in order to reduce te crosswind smear. Note owever tat in spite of te fact tat te CIP-crosswind diffusion is weakly consistent tis property does not seem to generalize to te case of iger order polynomial approximation. Tese types of results are expected to be of interest in problems in fluid mecanics. For example, if similar results eld for te CIP-metod of [4] applied to te Oseen s equation it would mean tat, away from boundary layers and singularities, solutions to te linearized equations of incompressible flow are accurately approximated by te CIP-metod also in te ig Reynolds number regime Proof of Lemma Appendix: Tecnical proofs Proof. Let T denote te closure of te set of triangles T T suc tat T T. We start by proving 3.1. By [4], for any piecewise polynomial function p 6.1 p Πp 2 L 2 T C [p] 2 L 2 e. e E 0,e T Using 2.8 and 6.1, we can estimate te left and side of 3.1 as ωu x ΠU x 2 L 2 max x T [ωx]2 U x ΠU x 2 L 2 T C max x T [ωx]2 [U x ] 2 L 2 e e E 0,e T CC 2 ω e E 0 ω[u x ] 2 L 2 e, wic proves 3.1. Next we will establis 3.2. By te triangle inequality, ω 1 ω 2 U x Πω 2 U x 2 L 2 C ω 1 Πω 2 U x Πω 2 U x 2 L 2 + ω 1 Πω 2 U x ω 2 U x 2 L 2 + ω 1 ω 2 U x ω 2 U x 2 L 2 = CI 1 + I 2 + I 3,

14 14 BURMAN, GUZMÁN, AND LEYKEKHMAN were ω 2 = 1 T T ω2 is te average of ω 2 over eac triangle. Using 2.12 and te stability of Π in te L 2 -norm, we ave 6.2 I 1 = ω 1 Πω 2 U x Πω 2 U x 2 L 2 T C x T [ωx] 2 ω 2 U x ω 2 U x 2 L 2 T CC 2 ω ω 1 ω 2 U x ω 2 U x 2 L 2 = CC2 ωi 3. Tus, we need only to estimate I 2 and I 3. To estimate I 3 we use te fact tat ω 2 ω 2 L T C ω 2 L T and te properties of ω 2.6 and Tus, 6.3 I 3 = ω 1 ω 2 U x ω 2 U x 2 L 2 T C x T [ωx] 2 2 ω 2 2 L T U x 2 L 2 T C x T [ωx] 2 K 2 ω 4 L T U x 2 L 2 T CCω 4 K 2 ωu x 2 L 2 T = CC4 ω K 2 ωu x 2 L 2. Tus, we are only left to estimate I 2. Again using 2.12, 6.1, and te triangle inequality, we ave 6.4 I 2 = ω 1 Πω 2 U x ω 2 U x 2 L 2 T x T [ωx] 2 Πω 2 U x ω 2 U x 2 L 2 T C x T [ωx] 2 C x T [ωx] 2 [ω 2 U x ] 2 L 2 e e E 0,e T e E 0,e T [ω 2 U x ] 2 L 2 e + [ω2 ω 2 U x ] 2 L 2 e. Since ω is smoot, [ω 2 U x ] = ω 2 [U x ], and using 2.12, we ave x T [ωx] 2 ω 2 [U x ] 2 L 2 e C2 ω ω[u x ] 2 L 2 e e E 0,e T e E 0,e T CCω 2 ω[u x ] 2 L 2 e. e E 0

15 WEIGHTED ESTIMATES OF CIP METHOD 15 To estimate te remaining term in I 2, we proceed similarly to 6.3 and use te inverse inequality 2.3 applied to [U x ], x T [ωx] 2 [ω 2 ω 2 U x ] 2 L 2 e e E 0,e T C x T [ωx] 2 2 ω 2 2 L T [U x ] 2 L 2 e e E 0,e T CCinv 2 x T [ωx] 2 K 2 ω 4 L T 1 U x 2 L 2 T T T CCinvC 2 ωk 4 2 ωu x 2 L 2 T CC 2 invc 4 ωk 2 T T ωu x 2 L 2 T = CC2 invc 4 ωk 2 ωu x 2 L 2. Combining te above two estimates wit estimates 6.2, 6.4, and 6.3, we prove 3.2. Finally, to obtain 3.3, we apply 2.12, triangle inequality, and te inverse inequality to ω 2 U x Πω 2 U x, we obtain, 2 ω 1 ω 2 U x Πω 2 U x 2 L 2 T C 2 x T [ωx] 2 ω 2 U x Πω 2 U x 2 L 2 T C 2 x T [ωx] 2 ω 2 U x ω 2 U x 2 L 2 T + ω2 U x Πω 2 U x 2 L 2 T C x T [ωx] 2 2 ω 2 U x ω 2 U x 2 L 2 T + C2 inv ω 2 U x Πω 2 U x 2 L 2 T CCω 2 2 ω 1 ω 2 ω 2 U x 2 L 2 T + CC2 invcω 2 ω 1 ω 2 U x Πω 2 U x 2 L 2 T CCω 2 2 ω 1 ω 2 ω 2 U x 2 L 2 T + CC2 invc 2 ωi 1 + I 2. Since we already sowed te desired estimate for I 1 and I 2, to finis te proof we use 2.6, 2.12, and noticing tat U x = 0, we obtain, 2 ω 1 ω 2 ω 2 U x 2 L 2 T = 2 2 ωu x 2 L 2 T CK 2 ωu x 2 L 2 T. Tis concludes te proof of te lemma Proof of Lemma 3.2. Proof. Using te integration by parts and te fact tat U y is piecewise constant, 6.5 ωu y 2 L 2 = T ω 2 U 2 y = 2ωω y UU y + ω 2 UU y n y ds. T T

16 16 BURMAN, GUZMÁN, AND LEYKEKHMAN Using te property of ω, namely 2.6, te Caucy-Scwarz and aritmetic-geometric mean inequalities, we can estimate te first term on te rigt and side by 6.6 3/2 T 2ωω y UU y C 3/4 K 1 ωu y L 2 ωu L 2 3/2 4 ωu y 2 L 2 + CK 2 ωu 2 L 2. By te Caucy-Scwarz and aritmetic-geometric mean inequalities, we can estimate te second term on te rigt and side by 3/2 Since by 2.10 we obtain 6.7 T ω 2 UU y n y ds 3/2 e E 0 ω[u y ] L2 e ωu L2 e + 3/2 ωun y L 2 ωu y L ω[u y ] 2 L 2 2 e + ωu 2 L 2 e e E 0 ωu y 2 L 2 = ωu y 2 L 2 e 3/2 e E C 2 invc 2 ω 1 T + 5/2 4Cinv 2 ωu y 2 C2 L 2 + C1/2 ωun y 2 L 2. ω ωu y 2 L 2 T ωu y 2 L 2 T = C2 invc 2 ω 1 ωu y 2 L 2, ω 2 UU y n y ds C 2 e E 0 ω[u y ] 2 L 2 e + C ωu 2 L 2 + 3/2 4 ωu y 2 L 2 + C1/2 ωun y 2 L 2. By absorbing te term 3/2 4 ωu y 2 L 2 appearing in 6.6 and 6.7 by te left and side of 6.5, we complete te proof of te lemma Proof of Lemma 3.3. Proof. We start te proof of te lemma by adding and subtracting te Clément interpolant of ω 2 U x, 6.8 ωu x 2 L 2 = U x Πω 2 U x + U x ω 2 U x Πω 2 U x.

17 WEIGHTED ESTIMATES OF CIP METHOD 17 Using te Caucy-Scwarz and geometric-aritmetic mean inequalities and te Lemma 3.1, we can estimate te last term on te rigt and side by U x ω 2 U x Πω 2 U x ωu x L 2 ω 1 ω 2 U x Πω 2 U x L ωu x 2 L 2 + ω 1 ω 2 U x Πω 2 U x 2 L ωu x 2 L 2 + CK 2 ωu x 2 L 2 + C e E 0 ω[u x ] 2 L 2 e. By coosing K large enoug we can kick back te first two terms on te rigt and side and from 6.8 we obtain ωu x 2 L 2 2 U x Πω 2 U x + C e E 0 ω[u x ] 2 L 2 e. Since 2 e E ω[u 0 x ] 2 L 2 e is one of te terms of Q2 U, to complete te proof, we sall sow 6.9 U x Πω 2 U x 4 ωu x 2 L 2 + C Q 2 U + L 2 u. To establis 6.9 we will use te ortogonality condition 1.3, 6.10 BU, Πω 2 U x = Bu, Πω 2 U x. Tus from 6.10, 6.11 U x Πω 2 U x = Bu, Πω 2 U x U Πω 2 U x n x ds U Πω 2 U x εau, Πω 2 U x J U, Πω 2 U x J U, Πω 2 U x. We bound eac term of te rigt and side separately. In details we will only demonstrate te estimates for εau, Πω 2 U x and J U, Πω 2 U x. Te estimates for te oter terms are very similar. We start wit 6.12 εau, Πω 2 U x = ε U Πω 2 U x U + ε n Πω2 U x + Πω2 U x U ds n ε γ bc 1 UΠω 2 U x ds = I 1 + I 2 + I 3. Adding and subtracting ω 2 U x we ave I 1 = ε T U ω 2 U x ε T U Πω 2 U x ω 2 U x.

18 18 BURMAN, GUZMÁN, AND LEYKEKHMAN Te first term on te rigt and side we can estimate by using te properties of te weigt function and te fact tat U x is piecewise constant, ε U ω 2 U x = ε U ω 2 U x 6.13 T T = 2ε U ωωu x CεK 1 1 ω U 2 L 2 T = CεK 1 1 ω U 2 L 2. To estimate te oter term, we use te Caucy-Scwarz and aritmetic-geometric mean inequalities and Lemma 3.1, ε U Πω 2 U x ω 2 U x 6.14 T ε 2 1 T ω U 2 L 2 T + ε 2 ε 2 1 ω U 2 L 2 + Cε e E 0 Cε 1 ω U 2 L 2 + C e E 0 ω 1 Πω 2 U x ω 2 U x 2 L 2 T ω[u x ] 2 L 2 e + Cε 1 K 2 ωu x 2 L 2 ω[u x ] 2 L 2 e. In te last step we used te assumption ε. Tus, from 6.13 and 6.14, we get I 1 C 1 Q 2 U. Next, we will bound te remaining terms of εau, Πω 2 U x, namely I 2 and I 3. By te Caucy-Scwarz inequality I 2 ε ω U L 2 ω 1 Πω 2 U x L 2 + ε ω 1 Πω 2 U x L 2 ωu L 2. By te aritmetic-geometric mean inequality ω U L2 ω 1 Πω 2 U x L2 1 2 ω U 2 L ω 1 Πω 2 U x 2 L 2. Using te properties of ω 2.8 and te inverse inequality 2.3, ω U 2 L 2 e E max x e [ωx]2 U 2 L 2 e Cinv 2 1 max x T [ωx]2 U 2 L 2 T CωC 2 inv 2 1 ω U 2 L 2 T = C2 ωc 2 inv 1 ω U 2 L 2.

19 WEIGHTED ESTIMATES OF CIP METHOD 19 Using 2.12, te inverse inequality 2.3, te triangle inequality, and Lemma 3.1, we ave ω 1 Πω 2 U x 2 L 2 x e [ωx] 2 Πω 2 U x 2 L 2 e 6.15 C 2 inv 1 e E x T [ωx] 2 Πω 2 U x 2 L 2 T Cinv 2 1 x T [ωx] 2 Πω 2 U x ω 2 U x 2 L 2 T + ω2 U x 2 L 2 T CωC 2 inv 2 1 ω 1 Πω 2 U x ω 2 U x 2 L 2 T + ωu x 2 L 2 T C e E 0 C e E 0 ω[u x ] 2 L 2 e + C 1 K 2 ωu x 2 L 2 + C2 ωc 2 inv 1 ωu x 2 L 2 ω[u x ] 2 L 2 e + C 1 ω U 2 L 2. Using te assumption ε, we ave sown, 6.16 ε ω U L2 ω 1 Πω 2 U x L2 C 1 Q 2 U. Similarly, by te aritmetic-geometric mean inequality ω 1 Πω 2 U x L 2 ωu L 2 2 ω 1 Πω 2 U x 2 L ωu 2 L 2. Proceeding exactly as in te estimate 6.15, we obtain 2 ω 1 Πω 2 U x 2 L 2 C e E 0 Again using te assumption ε, we ave ω[u x ] 2 L 2 e + C 1 ω U 2 L ε ω U L 2 ω 1 Πω 2 U x L 2 C 1 Q 2 U. Tus, from 6.16 and 6.17, we ave Similarly, we can sow Tus, we ave sown I 2 C 1 Q 2 U. I 3 C 1 Q 2 U εau, Πω 2 U x CQ 2 U. In a similar fasion, we can sow UΠω 2 U x n x ds + UΠω 2 U x + J U, Πω 2 U x J U, Πω 2 U x CQ2 U + CK 2 + δ ωu x 2 L 2. We will demonstrate tis for J U, Πω 2 U x = 2 [U y ][Πω 2 U x y ]ds + 1/2 e E 0 e UΠω 2 U x n y 2 ds.

20 20 BURMAN, GUZMÁN, AND LEYKEKHMAN We start wit te last term. By te Caucy-Scwarz and aritmetic-geometric mean inequalities, 1/2 UΠω 2 U x n y 2 ds C δ 1/2 ωun y 2 L 2 + δ3/2 ω 1 Πω 2 U x 2 L 2 Since 1/2 ωun y 2 L 2 is one of te terms of Q2 U, we only need to treat te last term. By te properties of ω, te inverse inequality 2.3, te triangle inequality, and Lemma 3.1, ω 1 Πω 2 U x 2 L 2 C2 invc 2 ω 1 ω 1 Πω 2 U x 2 L 2 Tus, 6.20 C 2 invc 2 ω 1 ω 1 Πω 2 U x ω 2 U x 2 L 2 + ωu x 2 L 2 C e E 0 1/2 ω[u x ] 2 L 2 e + 1 CK 2 + C 2 invc 2 ω ωu x 2 L 2. UΠω 2 U x n y 2 ds C δ 1/2 ωun y 2 L 2 + C 3/2 e E 0 ω[u x ] 2 L 2 e + δ1/2 CK 2 + C 2 invc 2 ω ωu x 2 L 2. Next we estimate te oter term of J. By te aritmetic-geometric mean inequality, 2 y ][Πω e[u 2 U x y ]ds C ω[u y ] 2 L 2 e + 3 ω 1 [Πω 2 U x y ] 2 L 2 e. e E 0 e E 0 Again, 2 e E ω[u 0 y ] 2 L 2 e is one of te terms of Q2 U, ence we only need to treat te last term. By te properties of ω, te inverse inequality 2.3, te triangle inequality, and Lemma 3.1, 3 ω 1 [Πω 2 U x y ] 2 L 2 e C2 x T [ωx] 2 Πω 2 U x 2 L 2 T e E 0 C 2 ω 1 Πω 2 U x ω 2 U x 2 L 2 T + ωu x 2 L 2 T Tus, 6.21 C e E 0 2 e E 0 ω[u x ] 2 L 2 e + CK 2 ωu x 2 L 2. e[u y ][Πω 2 U x y ]ds C e E 0 Combining 6.20 and 6.21 we obtain + CK 2 ωu x 2 L 2. ω[u x ] 2 L 2 e + ω[u y] 2 L 2 e J U, Πω 2 U x CQ 2 U + CK 2 + δ 3/2 C 2 invc 2 ω ωux 2 L 2. Terefore multiplying 6.11 by and using 6.18 and 6.19, 6.22 U x Πω 2 U x CQ 2 U + CK 2 + δ ωu x 2 L 2 + Bu, Πω2 U x.

21 WEIGHTED ESTIMATES OF CIP METHOD 21 It remains to bound Bu, Πω 2 U x = εau, Πω 2 U x +Mu, Πω 2 U x +J u, Πω 2 U x +J u, Πω 2 U x. Te bound of te first term εau, Πω 2 U x follows in te same fasion as Now, we bound Mu, Πω 2 U x = u x Πω 2 U x + uπω 2 U x n x ds + uπω 2 U x. Adding and subtracting ω 2 U x, we ave u x Πω 2 U x = u x ω 2 U x + u x Πω 2 U x ω 2 U x. By te aritmetic-geometric mean inequality u x ω 2 U x 1 16 ωu x 2 L 2 + C ωu x 2 L 2, and by Lemma 3.1, u x Πω 2 U x ω 2 U x C ωu x 2 L 2 + C ω 1 Πω 2 U x ω 2 U x 2 L 2 C ωu x 2 L 2 + C 1 Q 2 U + CK 2 ωu x 2 L 2. Terefore, u x Πω 2 U x 16 ωu x 2 L 2 + CL2 u + CQ 2 U + CK 2 ωu x 2 L 2. We can easily bound te remaining terms of Mu, Πω 2 U x to arrive at Mu, Πω 2 U x 8 ωu x 2 L 2 + CL2 u + CQ 2 U + CK 2 ωu x 2 L 2. Te estimates of J u, Πω 2 U x and J u, Πω 2 U x can be derived along te lines of te estimate for J U, Πω 2 U x. Assembling all te bounds on te terms of Bu, Πω 2 U x we obtain Bu, Πω 2 U x 6 ωu x 2 L 2 + CL2 u + CQ 2 U + CK 2 + δ ωu x 2 L 2. Using te above inequality, estimate 6.22, and taking K large enoug and δ small enoug, proves 6.9. Tis completes te proof of te lemma Proof of Lemma 3.4. As we ave already mentioned above, tis superapproximation result is similar to te superapproximation results of [13] and [10], but ere instead of a local interpolant operator we ave to deal wit a global L 2 - projection. Because of tis fact te proof is muc more involved. Proof. Recall tat E = ω 2 U P ω 2 U, were P is te L 2 -projection defined in 3.4. Using tat u P u is ortogonal to V, we ave ω 1 E 2 L 2 = ω 2 ω 2 U P ω 2 Uω 2 U P ω 2 U = ω 2 U P ω 2 UU ω 2 P ω 2 U = ω 2 U P ω 2 UI ω 2 P ω 2 U ω 2 P ω 2 U,

22 22 BURMAN, GUZMÁN, AND LEYKEKHMAN were I denotes te Lagrange interpolant. Tus, by te Caucy-Scwarz inequality, Hence, ω 1 E 2 L 2 ω 1 E L2 ωi ω 2 P ω 2 U ω 2 P ω 2 U L2. ω 1 E L2 ωi ω 2 P ω 2 U ω 2 P ω 2 U L2. By following te proof of Lemma 2.2 in [14] and using 2.11 and 2.12, we get 1 ωi ω 2 P ω 2 U ω 2 P ω 2 U L2 CK 1/2 1/2 1/2 ω 1 P ω 2 U x L 2 + 3/4 ω 1 P ω 2 U y L 2 + ω 1 P ω 2 U L2 + ω 1 ω 1 x 1/2 P ω 2 U L2. Terefore by te triangle inequality, ω 1 E L 2 = CK 1/2 1/2 S 1 + S 2, were and S 1 = 1/2 ω 1 ω 2 U x L 2 + 3/4 ω 1 ω 2 U y L 2 + ωu L 2 + ω 1 ω 1 x 1/2 ω 2 U L 2. S 2 = 1/2 ω 1 P ω 2 U ω 2 U x L 2 + 3/4 ω 1 P ω 2 U ω 2 U y L 2 + ω 1 P ω 2 U ω 2 U L2 + ω 1 ω 1 x 1/2 P ω 2 U ω 2 U L2. One can sow using te product rule and 2.6, 2.7, and 2.8 tat S 1 C 1/2 ωu x L 2 + C 3/4 ωu y L 2 + ωu L 2 + ωω x 1/2 U L 2. Terefore, by Lemma 3.2 and Lemma 3.3, we ave S 1 CQU + Lu. Now we bound S 2. It easily follows tat S2 2 C ω 1 P ω 2 U ω 2 U x 2 L 2 T + 3/2 ω 1 P ω 2 U ω 2 U y 2 L 2 T + ω 1 P ω 2 U ω 2 U 2 L 2 T + ω 1 ω 1 x 1/2 P ω 2 U ω 2 U 2 L 2 T We analyze te first term. By using 2.12 and te interpolation inequality 2.4, we obtain ω 1 P ω 2 U ω 2 U x 2 L 2 T = CCω 2. x T [ωx] 2 P ω 2 U ω 2 U x 2 L 2 T 2 ω 1 P ω 2 U ω 2 U 2 L 2 T + 2 ω 1 D 2 P ω 2 U ω 2 U 2 L 2 T Since P ω 2 U is piecewise linear, ω 1 D 2 P ω 2 U ω 2 U 2 L 2 T = ω 1 D 2 ω 2 U 2 L 2 T,.

23 WEIGHTED ESTIMATES OF CIP METHOD 23 and we can sow using te product rule along wit properties of ω tat 3 ωu x 2 L 2 + 3/2 ωu y 2 L ω 1 D 2 ω 2 U 2 L 2 T C + ωu 2 L 2 + ωω x 1/2 U 2 L 2 Togeter wit Lemma 3.3 and Lemma 3.2, we obtain 3 ω 1 D 2 P ω 2 U ω 2 U 2 L 2 T C Q2 U + L 2 u. Terefore, we ave sown ω 1 P ω 2 U ω 2 U x 2 L 2 T C 1 ω 1 E 2 L 2 + CQ2 U + L 2 u. In a similar manner we can bound te remaining terms of S 2 2 to get S 2 2 C 1 ω 1 E 2 L 2 + C Q2 U + L 2 u. By taking te square root of bot sides we get S 2 C 1/2 ω 1 E L2 + C QU + Lu. Terefore, if we use te bounds for S 1 and S 2, 6.23, we see tat for K large enoug By 2.8 and 2.4, we get 1 ω 1 E L2 C 1/2 K 1/2 QU + Lu. ω 1 E 2 L 2 C 2 ω 1 E 2 L 2 + C2 Te proof is complete once we use te estimate ω 1 D 2 ω 2 U L Proof of Lemma 3.5. Presenting te proof of tis lemma, we assume tat te reader is already familiar wit proofs of te previous lemmas. Hence, in te proof below, we skip some steps wic appeared already several times in te proofs of te previous lemmas. Proof. By adding and subtracting BU, P ω 2 U and using te ortogonality property 1.3, we ave 6.25 BU, ω 2 U = BU, ω 2 U P ω 2 U + BU, P ω 2 U = BU, E + Bu, P ω 2 U, wit E = ω 2 U P ω 2 U, were P is te L 2 -projection defined in 3.4. First we bound BU, E. Recall tat We start wit εau, E = ε BU, E = εau, E + MU, E + J U, E + J U, E.. U E ε U n E + E n U ds + γ bc ε UE ds.

24 24 BURMAN, GUZMÁN, AND LEYKEKHMAN Te first term can be bounded by using Lemma 3.4 and te assumption ε as follows: ε U E ε ω U L 2 ω 1 E L 2 C ε 1/2 K 1/2 ω U L 2 QU + Lu CK 1/2 ε ω U 2 L 2 + CK 1/2 Q 2 U + L 2 u CK 1/2 Q 2 U + L 2 u. Te remaining terms of εau, E can be bounded in a similar way. Tus we get εau, E CK 1/2 Q 2 U + L 2 u. Te next term we will treat is MU, E = U x E + UE n x ds + UE. By using tat E is ortogonal to V, te Caucy-Scwarz inequality, Lemma 3.1 and Lemma 3.4, U x E = U x ΠU x E C 1/2 ωu x ΠU x L2 1/2 ω 1 E L2 C e E 0 2 ω[u x ] 2 L 2 e + 1 ω 1 E 2 L 2 CK 1 Q 2 U + CL 2 u. Similarly, we can bound te last two terms of MU, E. Tus, we obtain MU, E CK 1/2 Q 2 U + CL 2 u. In a similar fasion we may bound te remaining terms of BU, E following te proof of Lemma 3.3 to get 6.26 BU, E CK 1/2 Q 2 U + CL 2 u. It remains to estimate Bu, P ω 2 U. We start wit u εau, P ω 2 U =ε u P ω 2 U ε n P ω2 U + P ω2 U u ds n + εγ bc 1 up ω 2 U ds. Using te Caucy-Scwarz, aritmetic-geometric mean, and te triangle inequalities, we can bound te first term on te rigt and side as follows: ε u P ω 2 U Cε ω u L 2 ω 1 P ω 2 U L 2 C δ ε ω u 2 L 2 ω + εδ 1 E 2 L 2 + ω 1 ω 2 U 2 L 2, were δ is some small number to be cosen later. By te superapproximation result, Lemma 3.4, we ave ω 1 E 2 L 2 C 1 K 1 Q 2 U + L 2 u,

25 WEIGHTED ESTIMATES OF CIP METHOD 25 and by te triangle inequality ω 1 ω 2 U 2 L 2 ω U 2 L ωu 2 L 2. By te properties of te weigt function 2.6 and 2.7, we ave 6.27 ωu 2 L 2 = ωxu ωyu 2 2 C 1 K 1 ω ω x U 2 + Cσ 2 K 2 ω 2 U 2. Using tat ε and ε σ 2 we ave ε u, P ω 2 U C δ ε ω u 2 L 2 + CδK 1 Q 2 U + L 2 u + CδK 1 ω ω x 1/2 U 2 L 2 + CδK 2 ωu 2 L 2 + Cδε ω U 2 L 2 C δ ε ω u 2 L 2 + CδK 1 Q 2 U + L 2 u. Now we will treat te εγ bc 1 up ω2 U ds term. By te aritmetic-geometric mean and triangle inequalities, we ave up ω 2 U ds C δ ωu 2 L 2 + δ ω 1 E 2 L 2 + δ ωu 2 L 2. By te trace inequality 2.1 ωu 2 L 2 C 1 ωu 2 L 2 + C ωu 2 L 2 C 1 ωu 2 L 2 ωu + C 2 L 2 + ω u 2 L 2. Using 2.7, we ave ωu 2 L 2 C 2 K 2 ωu 2 L 2. Tus, 6.28 ωu 2 L 2 C 1 ωu 2 L 2 + ω u 2 L 2 CL 2 u. By te Remark 2, Tus, εγ bc 1 ω 1 E 2 L 2 CK 1 Q 2 U + L 2 u. up ω 2 U ds C δ L 2 u + CK 1 Q 2 U + CδQ 2 U. Next we bound J u, P ω 2 U = 2 [u x ][P ω 2 U x ] ds. e E 0 e By te Caucy-Scwarz and te triangle inequalities 2 [u x ][P ω 2 U x ] ds C δ 2 ω[ u] 2 L 2 e + δ2 ω 1 [P ω 2 U x ] 2 L 2 e. e By 2.12, te inverse inequality 2.3, and te triangle inequality, we ave 2 ω 1 [P ω 2 U x ] 2 L 2 e C ω 1 P ω 2 U x 2 L 2 T e E 0 C ω 1 ω 2 U x 2 L 2 + ω 1 E x 2 L 2 C ω x U 2 L 2 + ωu x 2 L 2 + ω 1 E x 2 L 2

26 26 BURMAN, GUZMÁN, AND LEYKEKHMAN Since ω x U 2 L 2 CK 1 ω ω x 1/2 U 2 L 2, by Lemma 3.3 and Lemma 3.4, we ave 2 ω 1 [P ω 2 U x ] 2 L 2 e C K 1 Q 2 U + Q 2 U + L 2 u. Hence e E 0 J u, P ω 2 U C δq 2 U + L 2 u. Te estimate of J u, P ω 2 U is similar. It remains to bound Mu, P ω 2 U = Mu, E + Mu, ω 2 U Te first term can be controlled by using te Caucy-Scwarz inequality and te superapproximation result of Lemma 3.4. Te second term we integrate by parts and split te term in te following fasion Mu, ω 2 U = uω x ωu + uω 2 U x 1 2 ωu L 2 ω ω x 1 2 U L ωu L ωux L 2. Similarly to te analysis above, we obtain, Tus, Mu, ω 2 U δq 2 U + ωu x 2 L 2 + C δl 2 u Bu, P ω 2 U CK 1 Q 2 U + CδQ 2 U + ωu x 2 L 2 + C δl 2 u. Combining estimates 6.28 and 6.29 we conclude te proof of te lemma. Acknowledgments: Te autors would like to tank te anonymous reviewers for very insigtful comments and for elping to improve te presentation of te paper. References [1] R. Becker and M. Braack, Two-level stabilization sceme for te Navier-Stokes equations, Numerical Matematics and Advanced Applications, , Springer, Berlin, [2] E. Burman and P. Hansbo, Edge stabilization for Galerkin approximations of convectiondiffusion-reaction problems, Comput. Metods Appl. Mec. Engrg , [3] E. Burman, A unified analysis of conforg and non-conforg stabilized finite element metods using interior penalty, SIAM J. Num. Anal , [4] E. Burman, M. Fernández, and P. Hansbo, Continuous Interior Penalty for Oseen s Equations, SIAM J. Num. Anal , [5] E. Burman and A. Ern, Continuous interior penalty p-finite element metods for advection and advection-diffusion equations, Mat. Comp , [6] P.G. Ciarlet, Te Finite Element Metod for Elliptic Problems, Nort-Holland, Amsterdam, [7] P. Clément, Approximation by finite elements functions using local regularization, RAIRO Anal. Numer , [8] R. Codina and J. Blasco, Analysis of a stabilized finite element approximation of te transient convection-diffusion-reaction equation using ortogonal subscales. Comput. Vis. Sci , no. 3, [9] J. Douglas and T. Dupont, Interior penalty procedures for elliptic and parabolic Galerkin metods In Computing metods in applied sciences Second Internat. Sympos., Versailles, 1975, pages Lecture Notes in Pys., Vol. 58. Springer, Berlin, [10] P. Grisvard, Singularities in Boundary Value Problems, Recerces en Matmatiques Appliques [Researc in Applied Matematics], 22. Masson, Paris; Springer-Verlag, Berlin, 1992.

27 WEIGHTED ESTIMATES OF CIP METHOD 27 [11] J.-L. Guermond. Stabilization of Galerkin approximations of transport equations by subgrid modeling. M2AN Mat. Model. Numer. Anal , no. 6, [12] J. Guzmán, Local analysis of discontinuous Galerkin metods applied to singularly perturbed problems, J. Numer. Mat , [13] C. Jonson, U. Nävert, J. Pitkänta, Finite element metods for linear yperbolic problems, Comput. Metods Appl. Mec. Engrg , [14] C. Jonson, A.H. Scatz, and L.B. Walbin, Crosswind smear and pointwise errors in streamline diffusion finite element metods, Mat. Comp , [15] K. Niijima, Pointwise error estimates for a streamline diffusion finite element sceme, Numer. Mat , [16] G. Sangalli, Global and local error analysis for te residual-free bubbles metod applied to advection-doated problems, SIAM J. Numer. Anal , Department of Matematics, University of Sussex, Brigton, BN1 9RF UK, E.N.Burman@sussex.ac.uk. Scool of Matematics, University of Minnesota, Minneapolis, MN 55455, USA, guzma033@umn.edu. Department of Matematics, University of Connecticut, Storrs, CT 06269, USA, leykekman@mat.uconn.edu.

1. Introduction. We consider the model problem: seeking an unknown function u satisfying

1. Introduction. We consider the model problem: seeking an unknown function u satisfying A DISCONTINUOUS LEAST-SQUARES FINITE ELEMENT METHOD FOR SECOND ORDER ELLIPTIC EQUATIONS XIU YE AND SHANGYOU ZHANG Abstract In tis paper, a discontinuous least-squares (DLS) finite element metod is introduced

More information

Numerical Experiments Using MATLAB: Superconvergence of Nonconforming Finite Element Approximation for Second-Order Elliptic Problems

Numerical Experiments Using MATLAB: Superconvergence of Nonconforming Finite Element Approximation for Second-Order Elliptic Problems Applied Matematics, 06, 7, 74-8 ttp://wwwscirporg/journal/am ISSN Online: 5-7393 ISSN Print: 5-7385 Numerical Experiments Using MATLAB: Superconvergence of Nonconforming Finite Element Approximation for

More information

Different Approaches to a Posteriori Error Analysis of the Discontinuous Galerkin Method

Different Approaches to a Posteriori Error Analysis of the Discontinuous Galerkin Method WDS'10 Proceedings of Contributed Papers, Part I, 151 156, 2010. ISBN 978-80-7378-139-2 MATFYZPRESS Different Approaces to a Posteriori Error Analysis of te Discontinuous Galerkin Metod I. Šebestová Carles

More information

A Mixed-Hybrid-Discontinuous Galerkin Finite Element Method for Convection-Diffusion Problems

A Mixed-Hybrid-Discontinuous Galerkin Finite Element Method for Convection-Diffusion Problems A Mixed-Hybrid-Discontinuous Galerkin Finite Element Metod for Convection-Diffusion Problems Herbert Egger Joacim Scöberl We propose and analyse a new finite element metod for convection diffusion problems

More information

A Hybrid Mixed Discontinuous Galerkin Finite Element Method for Convection-Diffusion Problems

A Hybrid Mixed Discontinuous Galerkin Finite Element Method for Convection-Diffusion Problems A Hybrid Mixed Discontinuous Galerkin Finite Element Metod for Convection-Diffusion Problems Herbert Egger Joacim Scöberl We propose and analyse a new finite element metod for convection diffusion problems

More information

AN ANALYSIS OF THE EMBEDDED DISCONTINUOUS GALERKIN METHOD FOR SECOND ORDER ELLIPTIC PROBLEMS

AN ANALYSIS OF THE EMBEDDED DISCONTINUOUS GALERKIN METHOD FOR SECOND ORDER ELLIPTIC PROBLEMS AN ANALYSIS OF THE EMBEDDED DISCONTINUOUS GALERKIN METHOD FOR SECOND ORDER ELLIPTIC PROBLEMS BERNARDO COCKBURN, JOHNNY GUZMÁN, SEE-CHEW SOON, AND HENRYK K. STOLARSKI Abstract. Te embedded discontinuous

More information

LEAST-SQUARES FINITE ELEMENT APPROXIMATIONS TO SOLUTIONS OF INTERFACE PROBLEMS

LEAST-SQUARES FINITE ELEMENT APPROXIMATIONS TO SOLUTIONS OF INTERFACE PROBLEMS SIAM J. NUMER. ANAL. c 998 Society for Industrial Applied Matematics Vol. 35, No., pp. 393 405, February 998 00 LEAST-SQUARES FINITE ELEMENT APPROXIMATIONS TO SOLUTIONS OF INTERFACE PROBLEMS YANZHAO CAO

More information

arxiv: v1 [math.na] 27 Jan 2014

arxiv: v1 [math.na] 27 Jan 2014 L 2 -ERROR ESTIMATES FOR FINITE ELEMENT APPROXIMATIONS OF BOUNDARY FLUXES MATS G. LARSON AND ANDRÉ MASSING arxiv:1401.6994v1 [mat.na] 27 Jan 2014 Abstract. We prove quasi-optimal a priori error estimates

More information

Superconvergence of energy-conserving discontinuous Galerkin methods for. linear hyperbolic equations. Abstract

Superconvergence of energy-conserving discontinuous Galerkin methods for. linear hyperbolic equations. Abstract Superconvergence of energy-conserving discontinuous Galerkin metods for linear yperbolic equations Yong Liu, Ci-Wang Su and Mengping Zang 3 Abstract In tis paper, we study superconvergence properties of

More information

Poisson Equation in Sobolev Spaces

Poisson Equation in Sobolev Spaces Poisson Equation in Sobolev Spaces OcMountain Dayligt Time. 6, 011 Today we discuss te Poisson equation in Sobolev spaces. It s existence, uniqueness, and regularity. Weak Solution. u = f in, u = g on

More information

arxiv: v1 [math.na] 28 Apr 2017

arxiv: v1 [math.na] 28 Apr 2017 THE SCOTT-VOGELIUS FINITE ELEMENTS REVISITED JOHNNY GUZMÁN AND L RIDGWAY SCOTT arxiv:170500020v1 [matna] 28 Apr 2017 Abstract We prove tat te Scott-Vogelius finite elements are inf-sup stable on sape-regular

More information

ERROR BOUNDS FOR THE METHODS OF GLIMM, GODUNOV AND LEVEQUE BRADLEY J. LUCIER*

ERROR BOUNDS FOR THE METHODS OF GLIMM, GODUNOV AND LEVEQUE BRADLEY J. LUCIER* EO BOUNDS FO THE METHODS OF GLIMM, GODUNOV AND LEVEQUE BADLEY J. LUCIE* Abstract. Te expected error in L ) attimet for Glimm s sceme wen applied to a scalar conservation law is bounded by + 2 ) ) /2 T

More information

Part VIII, Chapter 39. Fluctuation-based stabilization Model problem

Part VIII, Chapter 39. Fluctuation-based stabilization Model problem Part VIII, Capter 39 Fluctuation-based stabilization Tis capter presents a unified analysis of recent stabilization tecniques for te standard Galerkin approximation of first-order PDEs using H 1 - conforming

More information

Error estimates for a semi-implicit fully discrete finite element scheme for the mean curvature flow of graphs

Error estimates for a semi-implicit fully discrete finite element scheme for the mean curvature flow of graphs Interfaces and Free Boundaries 2, 2000 34 359 Error estimates for a semi-implicit fully discrete finite element sceme for te mean curvature flow of graps KLAUS DECKELNICK Scool of Matematical Sciences,

More information

MIXED DISCONTINUOUS GALERKIN APPROXIMATION OF THE MAXWELL OPERATOR. SIAM J. Numer. Anal., Vol. 42 (2004), pp

MIXED DISCONTINUOUS GALERKIN APPROXIMATION OF THE MAXWELL OPERATOR. SIAM J. Numer. Anal., Vol. 42 (2004), pp MIXED DISCONTINUOUS GALERIN APPROXIMATION OF THE MAXWELL OPERATOR PAUL HOUSTON, ILARIA PERUGIA, AND DOMINI SCHÖTZAU SIAM J. Numer. Anal., Vol. 4 (004), pp. 434 459 Abstract. We introduce and analyze a

More information

Variational Localizations of the Dual Weighted Residual Estimator

Variational Localizations of the Dual Weighted Residual Estimator Publised in Journal for Computational and Applied Matematics, pp. 192-208, 2015 Variational Localizations of te Dual Weigted Residual Estimator Tomas Ricter Tomas Wick Te dual weigted residual metod (DWR)

More information

A Finite Element Primer

A Finite Element Primer A Finite Element Primer David J. Silvester Scool of Matematics, University of Mancester d.silvester@mancester.ac.uk. Version.3 updated 4 October Contents A Model Diffusion Problem.................... x.

More information

arxiv: v1 [math.na] 12 Mar 2018

arxiv: v1 [math.na] 12 Mar 2018 ON PRESSURE ESTIMATES FOR THE NAVIER-STOKES EQUATIONS J A FIORDILINO arxiv:180304366v1 [matna 12 Mar 2018 Abstract Tis paper presents a simple, general tecnique to prove finite element metod (FEM) pressure

More information

Volume 29, Issue 3. Existence of competitive equilibrium in economies with multi-member households

Volume 29, Issue 3. Existence of competitive equilibrium in economies with multi-member households Volume 29, Issue 3 Existence of competitive equilibrium in economies wit multi-member ouseolds Noriisa Sato Graduate Scool of Economics, Waseda University Abstract Tis paper focuses on te existence of

More information

A Weak Galerkin Method with an Over-Relaxed Stabilization for Low Regularity Elliptic Problems

A Weak Galerkin Method with an Over-Relaxed Stabilization for Low Regularity Elliptic Problems J Sci Comput (07 7:95 8 DOI 0.007/s095-06-096-4 A Weak Galerkin Metod wit an Over-Relaxed Stabilization for Low Regularity Elliptic Problems Lunji Song, Kaifang Liu San Zao Received: April 06 / Revised:

More information

Mass Lumping for Constant Density Acoustics

Mass Lumping for Constant Density Acoustics Lumping 1 Mass Lumping for Constant Density Acoustics William W. Symes ABSTRACT Mass lumping provides an avenue for efficient time-stepping of time-dependent problems wit conforming finite element spatial

More information

Differentiation in higher dimensions

Differentiation in higher dimensions Capter 2 Differentiation in iger dimensions 2.1 Te Total Derivative Recall tat if f : R R is a 1-variable function, and a R, we say tat f is differentiable at x = a if and only if te ratio f(a+) f(a) tends

More information

CLEMSON U N I V E R S I T Y

CLEMSON U N I V E R S I T Y A Fractional Step θ-metod for Convection-Diffusion Equations Jon Crispell December, 006 Advisors: Dr. Lea Jenkins and Dr. Vincent Ervin Fractional Step θ-metod Outline Crispell,Ervin,Jenkins Motivation

More information

Journal of Computational and Applied Mathematics

Journal of Computational and Applied Mathematics Journal of Computational and Applied Matematics 94 (6) 75 96 Contents lists available at ScienceDirect Journal of Computational and Applied Matematics journal omepage: www.elsevier.com/locate/cam Smootness-Increasing

More information

ERROR ESTIMATES FOR THE DISCONTINUOUS GALERKIN METHODS FOR PARABOLIC EQUATIONS. 1. Introduction. We consider the parabolic PDE of the form,

ERROR ESTIMATES FOR THE DISCONTINUOUS GALERKIN METHODS FOR PARABOLIC EQUATIONS. 1. Introduction. We consider the parabolic PDE of the form, ERROR ESTIMATES FOR THE DISCONTINUOUS GALERKIN METHODS FOR PARABOLIC EQUATIONS K. CHRYSAFINOS AND NOEL. J. WALKINGTON Abstract. We analyze te classical discontinuous Galerkin metod for a general parabolic

More information

MATH745 Fall MATH745 Fall

MATH745 Fall MATH745 Fall MATH745 Fall 5 MATH745 Fall 5 INTRODUCTION WELCOME TO MATH 745 TOPICS IN NUMERICAL ANALYSIS Instructor: Dr Bartosz Protas Department of Matematics & Statistics Email: bprotas@mcmasterca Office HH 36, Ext

More information

A SYMMETRIC NODAL CONSERVATIVE FINITE ELEMENT METHOD FOR THE DARCY EQUATION

A SYMMETRIC NODAL CONSERVATIVE FINITE ELEMENT METHOD FOR THE DARCY EQUATION A SYMMETRIC NODAL CONSERVATIVE FINITE ELEMENT METHOD FOR THE DARCY EQUATION GABRIEL R. BARRENECHEA, LEOPOLDO P. FRANCA 1 2, AND FRÉDÉRIC VALENTIN Abstract. Tis work introduces and analyzes novel stable

More information

Order of Accuracy. ũ h u Ch p, (1)

Order of Accuracy. ũ h u Ch p, (1) Order of Accuracy 1 Terminology We consider a numerical approximation of an exact value u. Te approximation depends on a small parameter, wic can be for instance te grid size or time step in a numerical

More information

Weierstraß-Institut. im Forschungsverbund Berlin e.v. Preprint ISSN

Weierstraß-Institut. im Forschungsverbund Berlin e.v. Preprint ISSN Weierstraß-Institut für Angewandte Analysis und Stocastik im Forscungsverbund Berlin e.v. Preprint ISSN 0946 8633 Stability of infinite dimensional control problems wit pointwise state constraints Micael

More information

H(div) conforming and DG methods for incompressible Euler s equations

H(div) conforming and DG methods for incompressible Euler s equations H(div) conforming and DG metods for incompressible Euler s equations Jonny Guzmán Filánder A. Sequeira Ci-Wang Su Abstract H(div) conforming and discontinuous Galerkin (DG) metods are designed for incompressible

More information

A method of Lagrange Galerkin of second order in time. Une méthode de Lagrange Galerkin d ordre deux en temps

A method of Lagrange Galerkin of second order in time. Une méthode de Lagrange Galerkin d ordre deux en temps A metod of Lagrange Galerkin of second order in time Une métode de Lagrange Galerkin d ordre deux en temps Jocelyn Étienne a a DAMTP, University of Cambridge, Wilberforce Road, Cambridge CB3 0WA, Great-Britain.

More information

arxiv: v1 [math.na] 20 Jul 2009

arxiv: v1 [math.na] 20 Jul 2009 STABILITY OF LAGRANGE ELEMENTS FOR THE MIXED LAPLACIAN DOUGLAS N. ARNOLD AND MARIE E. ROGNES arxiv:0907.3438v1 [mat.na] 20 Jul 2009 Abstract. Te stability properties of simple element coices for te mixed

More information

arxiv: v1 [math.na] 19 Mar 2018

arxiv: v1 [math.na] 19 Mar 2018 A primal discontinuous Galerkin metod wit static condensation on very general meses arxiv:1803.06846v1 [mat.na] 19 Mar 018 Alexei Lozinski Laboratoire de Matématiques de Besançon, CNRS UMR 663, Univ. Bourgogne

More information

Adaptive Finite Element Method

Adaptive Finite Element Method 39 Capter 3 Adaptive inite Element Metod 31 Introduction As already pointed out in capter 2, singularities occur in interface problems Wen discretizing te problem (221) wit inite Elements, te singularities

More information

Computers and Mathematics with Applications. A nonlinear weighted least-squares finite element method for Stokes equations

Computers and Mathematics with Applications. A nonlinear weighted least-squares finite element method for Stokes equations Computers Matematics wit Applications 59 () 5 4 Contents lists available at ScienceDirect Computers Matematics wit Applications journal omepage: www.elsevier.com/locate/camwa A nonlinear weigted least-squares

More information

Inf sup testing of upwind methods

Inf sup testing of upwind methods INTERNATIONAL JOURNAL FOR NUMERICAL METHODS IN ENGINEERING Int. J. Numer. Met. Engng 000; 48:745 760 Inf sup testing of upwind metods Klaus-Jurgen Bate 1; ;, Dena Hendriana 1, Franco Brezzi and Giancarlo

More information

Analysis of A Continuous Finite Element Method for H(curl, div)-elliptic Interface Problem

Analysis of A Continuous Finite Element Method for H(curl, div)-elliptic Interface Problem Analysis of A Continuous inite Element Metod for Hcurl, div)-elliptic Interface Problem Huoyuan Duan, Ping Lin, and Roger C. E. Tan Abstract In tis paper, we develop a continuous finite element metod for

More information

Polynomial Interpolation

Polynomial Interpolation Capter 4 Polynomial Interpolation In tis capter, we consider te important problem of approximatinga function fx, wose values at a set of distinct points x, x, x,, x n are known, by a polynomial P x suc

More information

A SPLITTING LEAST-SQUARES MIXED FINITE ELEMENT METHOD FOR ELLIPTIC OPTIMAL CONTROL PROBLEMS

A SPLITTING LEAST-SQUARES MIXED FINITE ELEMENT METHOD FOR ELLIPTIC OPTIMAL CONTROL PROBLEMS INTERNATIONAL JOURNAL OF NUMERICAL ANALSIS AND MODELING Volume 3, Number 4, Pages 6 626 c 26 Institute for Scientific Computing and Information A SPLITTING LEAST-SQUARES MIED FINITE ELEMENT METHOD FOR

More information

ERROR ESTIMATES FOR A FULLY DISCRETIZED SCHEME TO A CAHN-HILLIARD PHASE-FIELD MODEL FOR TWO-PHASE INCOMPRESSIBLE FLOWS

ERROR ESTIMATES FOR A FULLY DISCRETIZED SCHEME TO A CAHN-HILLIARD PHASE-FIELD MODEL FOR TWO-PHASE INCOMPRESSIBLE FLOWS ERROR ESTIMATES FOR A FULLY DISCRETIZED SCHEME TO A CAHN-HILLIARD PHASE-FIELD MODEL FOR TWO-PHASE INCOMPRESSIBLE FLOWS YONGYONG CAI, AND JIE SHEN Abstract. We carry out in tis paper a rigorous error analysis

More information

NUMERICAL DIFFERENTIATION. James T. Smith San Francisco State University. In calculus classes, you compute derivatives algebraically: for example,

NUMERICAL DIFFERENTIATION. James T. Smith San Francisco State University. In calculus classes, you compute derivatives algebraically: for example, NUMERICAL DIFFERENTIATION James T Smit San Francisco State University In calculus classes, you compute derivatives algebraically: for example, f( x) = x + x f ( x) = x x Tis tecnique requires your knowing

More information

Function Composition and Chain Rules

Function Composition and Chain Rules Function Composition and s James K. Peterson Department of Biological Sciences and Department of Matematical Sciences Clemson University Marc 8, 2017 Outline 1 Function Composition and Continuity 2 Function

More information

arxiv: v1 [math.na] 17 Jul 2014

arxiv: v1 [math.na] 17 Jul 2014 Div First-Order System LL* FOSLL* for Second-Order Elliptic Partial Differential Equations Ziqiang Cai Rob Falgout Sun Zang arxiv:1407.4558v1 [mat.na] 17 Jul 2014 February 13, 2018 Abstract. Te first-order

More information

Jian-Guo Liu 1 and Chi-Wang Shu 2

Jian-Guo Liu 1 and Chi-Wang Shu 2 Journal of Computational Pysics 60, 577 596 (000) doi:0.006/jcp.000.6475, available online at ttp://www.idealibrary.com on Jian-Guo Liu and Ci-Wang Su Institute for Pysical Science and Tecnology and Department

More information

lecture 26: Richardson extrapolation

lecture 26: Richardson extrapolation 43 lecture 26: Ricardson extrapolation 35 Ricardson extrapolation, Romberg integration Trougout numerical analysis, one encounters procedures tat apply some simple approximation (eg, linear interpolation)

More information

Consider a function f we ll specify which assumptions we need to make about it in a minute. Let us reformulate the integral. 1 f(x) dx.

Consider a function f we ll specify which assumptions we need to make about it in a minute. Let us reformulate the integral. 1 f(x) dx. Capter 2 Integrals as sums and derivatives as differences We now switc to te simplest metods for integrating or differentiating a function from its function samples. A careful study of Taylor expansions

More information

CONVERGENCE ANALYSIS OF FINITE ELEMENT SOLUTION OF ONE-DIMENSIONAL SINGULARLY PERTURBED DIFFERENTIAL EQUATIONS ON EQUIDISTRIBUTING MESHES

CONVERGENCE ANALYSIS OF FINITE ELEMENT SOLUTION OF ONE-DIMENSIONAL SINGULARLY PERTURBED DIFFERENTIAL EQUATIONS ON EQUIDISTRIBUTING MESHES INTERNATIONAL JOURNAL OF NUMERICAL ANALYSIS AND MODELING Volume, Number, Pages 57 74 c 5 Institute for Scientific Computing and Information CONVERGENCE ANALYSIS OF FINITE ELEMENT SOLUTION OF ONE-DIMENSIONAL

More information

Parameter Fitted Scheme for Singularly Perturbed Delay Differential Equations

Parameter Fitted Scheme for Singularly Perturbed Delay Differential Equations International Journal of Applied Science and Engineering 2013. 11, 4: 361-373 Parameter Fitted Sceme for Singularly Perturbed Delay Differential Equations Awoke Andargiea* and Y. N. Reddyb a b Department

More information

Analysis of the grad-div stabilization for the time-dependent Navier Stokes equations with inf-sup stable finite elements

Analysis of the grad-div stabilization for the time-dependent Navier Stokes equations with inf-sup stable finite elements arxiv:161.517v3 [mat.na] 2 May 217 Analysis of te grad-div stabilization for te time-dependent Navier Stokes equations wit inf-sup stable finite elements Javier de Frutos Bosco García-Arcilla Volker Jon

More information

Numerical performance of discontinuous and stabilized continuous Galerkin methods for convection diffusion problems

Numerical performance of discontinuous and stabilized continuous Galerkin methods for convection diffusion problems Numerical performance of discontinuous and stabilized continuous Galerkin metods for convection diffusion problems Paola F. Antonietti MOX, Dipartimento di Matematica, Politecnico di Milano, Piazza Leonardo

More information

A First-Order System Approach for Diffusion Equation. I. Second-Order Residual-Distribution Schemes

A First-Order System Approach for Diffusion Equation. I. Second-Order Residual-Distribution Schemes A First-Order System Approac for Diffusion Equation. I. Second-Order Residual-Distribution Scemes Hiroaki Nisikawa W. M. Keck Foundation Laboratory for Computational Fluid Dynamics, Department of Aerospace

More information

EXTENSION OF A POSTPROCESSING TECHNIQUE FOR THE DISCONTINUOUS GALERKIN METHOD FOR HYPERBOLIC EQUATIONS WITH APPLICATION TO AN AEROACOUSTIC PROBLEM

EXTENSION OF A POSTPROCESSING TECHNIQUE FOR THE DISCONTINUOUS GALERKIN METHOD FOR HYPERBOLIC EQUATIONS WITH APPLICATION TO AN AEROACOUSTIC PROBLEM SIAM J. SCI. COMPUT. Vol. 26, No. 3, pp. 821 843 c 2005 Society for Industrial and Applied Matematics ETENSION OF A POSTPROCESSING TECHNIQUE FOR THE DISCONTINUOUS GALERKIN METHOD FOR HYPERBOLIC EQUATIONS

More information

arxiv: v1 [math.na] 9 Sep 2015

arxiv: v1 [math.na] 9 Sep 2015 arxiv:509.02595v [mat.na] 9 Sep 205 An Expandable Local and Parallel Two-Grid Finite Element Sceme Yanren ou, GuangZi Du Abstract An expandable local and parallel two-grid finite element sceme based on

More information

Numerical Solution to Parabolic PDE Using Implicit Finite Difference Approach

Numerical Solution to Parabolic PDE Using Implicit Finite Difference Approach Numerical Solution to arabolic DE Using Implicit Finite Difference Approac Jon Amoa-Mensa, Francis Oene Boateng, Kwame Bonsu Department of Matematics and Statistics, Sunyani Tecnical University, Sunyani,

More information

Overlapping domain decomposition methods for elliptic quasi-variational inequalities related to impulse control problem with mixed boundary conditions

Overlapping domain decomposition methods for elliptic quasi-variational inequalities related to impulse control problem with mixed boundary conditions Proc. Indian Acad. Sci. (Mat. Sci.) Vol. 121, No. 4, November 2011, pp. 481 493. c Indian Academy of Sciences Overlapping domain decomposition metods for elliptic quasi-variational inequalities related

More information

Decay of solutions of wave equations with memory

Decay of solutions of wave equations with memory Proceedings of te 14t International Conference on Computational and Matematical Metods in Science and Engineering, CMMSE 14 3 7July, 14. Decay of solutions of wave equations wit memory J. A. Ferreira 1,

More information

Key words. Finite element method; convection-diffusion-reaction; nonnegativity; boundedness

Key words. Finite element method; convection-diffusion-reaction; nonnegativity; boundedness PRESERVING NONNEGATIVITY OF AN AFFINE FINITE ELEMENT APPROXIMATION FOR A CONVECTION-DIFFUSION-REACTION PROBLEM JAVIER RUIZ-RAMÍREZ Abstract. An affine finite element sceme approximation of a time dependent

More information

arxiv: v1 [math.na] 11 May 2018

arxiv: v1 [math.na] 11 May 2018 Nitsce s metod for unilateral contact problems arxiv:1805.04283v1 [mat.na] 11 May 2018 Tom Gustafsson, Rolf Stenberg and Jua Videman May 14, 2018 Abstract We derive optimal a priori and a posteriori error

More information

Fourier Type Super Convergence Study on DDGIC and Symmetric DDG Methods

Fourier Type Super Convergence Study on DDGIC and Symmetric DDG Methods DOI 0.007/s095-07-048- Fourier Type Super Convergence Study on DDGIC and Symmetric DDG Metods Mengping Zang Jue Yan Received: 7 December 06 / Revised: 7 April 07 / Accepted: April 07 Springer Science+Business

More information

Higher order unfitted isoparametric space-time FEM on moving domains

Higher order unfitted isoparametric space-time FEM on moving domains Higer order unfitted isoparametric space-time FEM on moving domains Master s tesis by: Janosc Preuß Supervisor: Jun.-Prof. Dr. Cristop Lerenfeld Second Assessor: Prof. Dr. Gert Lube prepared at te Institute

More information

Polynomial Interpolation

Polynomial Interpolation Capter 4 Polynomial Interpolation In tis capter, we consider te important problem of approximating a function f(x, wose values at a set of distinct points x, x, x 2,,x n are known, by a polynomial P (x

More information

One-Sided Position-Dependent Smoothness-Increasing Accuracy-Conserving (SIAC) Filtering Over Uniform and Non-uniform Meshes

One-Sided Position-Dependent Smoothness-Increasing Accuracy-Conserving (SIAC) Filtering Over Uniform and Non-uniform Meshes DOI 10.1007/s10915-014-9946-6 One-Sided Position-Dependent Smootness-Increasing Accuracy-Conserving (SIAC) Filtering Over Uniform and Non-uniform Meses JenniferK.Ryan Xiaozou Li Robert M. Kirby Kees Vuik

More information

A posteriori error estimation for unilateral contact with matching and non-matching meshes

A posteriori error estimation for unilateral contact with matching and non-matching meshes Comput. Metods Appl. Mec. Engrg. 186 (2000) 65±83 www.elsevier.com/locate/cma A posteriori error estimation for unilateral contact wit matcing and non-matcing meses Patrice Coorevits a, *, Patrick Hild

More information

arxiv: v1 [math.na] 6 Dec 2010

arxiv: v1 [math.na] 6 Dec 2010 MULTILEVEL PRECONDITIONERS FOR DISCONTINUOUS GALERKIN APPROXIMATIONS OF ELLIPTIC PROBLEMS WITH JUMP COEFFICIENTS BLANCA AYUSO DE DIOS, MICHAEL HOLST, YUNRONG ZHU, AND LUDMIL ZIKATANOV arxiv:1012.1287v1

More information

Numerical Differentiation

Numerical Differentiation Numerical Differentiation Finite Difference Formulas for te first derivative (Using Taylor Expansion tecnique) (section 8.3.) Suppose tat f() = g() is a function of te variable, and tat as 0 te function

More information

A trace finite element method for a class of coupled bulk-interface transport problems

A trace finite element method for a class of coupled bulk-interface transport problems Numerical Analysis and Scientific Computing Preprint Seria A trace finite element metod for a class of coupled bulk-interface transport problems S. Gross M.A. Olsanskii A. Reusken Preprint #28 Department

More information

Department of Mathematical Sciences University of South Carolina Aiken Aiken, SC 29801

Department of Mathematical Sciences University of South Carolina Aiken Aiken, SC 29801 RESEARCH SUMMARY AND PERSPECTIVES KOFFI B. FADIMBA Department of Matematical Sciences University of Sout Carolina Aiken Aiken, SC 29801 Email: KoffiF@usca.edu 1. Introduction My researc program as focused

More information

An approximation method using approximate approximations

An approximation method using approximate approximations Applicable Analysis: An International Journal Vol. 00, No. 00, September 2005, 1 13 An approximation metod using approximate approximations FRANK MÜLLER and WERNER VARNHORN, University of Kassel, Germany,

More information

arxiv: v2 [math.na] 5 Jul 2017

arxiv: v2 [math.na] 5 Jul 2017 Trace Finite Element Metods for PDEs on Surfaces Maxim A. Olsanskii and Arnold Reusken arxiv:1612.00054v2 [mat.na] 5 Jul 2017 Abstract In tis paper we consider a class of unfitted finite element metods

More information

Numerical analysis of a free piston problem

Numerical analysis of a free piston problem MATHEMATICAL COMMUNICATIONS 573 Mat. Commun., Vol. 15, No. 2, pp. 573-585 (2010) Numerical analysis of a free piston problem Boris Mua 1 and Zvonimir Tutek 1, 1 Department of Matematics, University of

More information

FINITE ELEMENT EXTERIOR CALCULUS FOR PARABOLIC EVOLUTION PROBLEMS ON RIEMANNIAN HYPERSURFACES

FINITE ELEMENT EXTERIOR CALCULUS FOR PARABOLIC EVOLUTION PROBLEMS ON RIEMANNIAN HYPERSURFACES FINITE ELEMENT EXTERIOR CALCULUS FOR PARABOLIC EVOLUTION PROBLEMS ON RIEMANNIAN HYPERSURFACES MICHAEL HOLST AND CHRIS TIEE ABSTRACT. Over te last ten years, te Finite Element Exterior Calculus (FEEC) as

More information

arxiv: v1 [math.na] 20 Nov 2018

arxiv: v1 [math.na] 20 Nov 2018 An HDG Metod for Tangential Boundary Control of Stokes Equations I: Hig Regularity Wei Gong Weiwei Hu Mariano Mateos Jon R. Singler Yangwen Zang arxiv:1811.08522v1 [mat.na] 20 Nov 2018 November 22, 2018

More information

A New Class of Zienkiewicz-Type Nonconforming Element in Any Dimensions

A New Class of Zienkiewicz-Type Nonconforming Element in Any Dimensions Numerisce Matematik manuscript No. will be inserted by te editor A New Class of Zienkiewicz-Type Nonconforming Element in Any Dimensions Wang Ming 1, Zong-ci Si 2, Jincao Xu1,3 1 LMAM, Scool of Matematical

More information

2009 Elsevier Science. Reprinted with permission from Elsevier.

2009 Elsevier Science. Reprinted with permission from Elsevier. P. Hansbo and M. Juntunen. 009. Weakly imposed Diriclet boundary conditions for te Brinkman model of porous media flow. Applied Numerical Matematics, volume 59, number 6, pages 174 189. doi:10.1016/j.apnum.008.07.003.

More information

arxiv: v3 [math.na] 15 Dec 2009

arxiv: v3 [math.na] 15 Dec 2009 THE NAVIER-STOKES-VOIGHT MODEL FOR IMAGE INPAINTING M.A. EBRAHIMI, MICHAEL HOLST, AND EVELYN LUNASIN arxiv:91.4548v3 [mat.na] 15 Dec 9 ABSTRACT. In tis paper we investigate te use of te D Navier-Stokes-Voigt

More information

Some Error Estimates for the Finite Volume Element Method for a Parabolic Problem

Some Error Estimates for the Finite Volume Element Method for a Parabolic Problem Computational Metods in Applied Matematics Vol. 13 (213), No. 3, pp. 251 279 c 213 Institute of Matematics, NAS of Belarus Doi: 1.1515/cmam-212-6 Some Error Estimates for te Finite Volume Element Metod

More information

AN ANALYSIS OF NEW FINITE ELEMENT SPACES FOR MAXWELL S EQUATIONS

AN ANALYSIS OF NEW FINITE ELEMENT SPACES FOR MAXWELL S EQUATIONS Journal of Matematical Sciences: Advances and Applications Volume 5 8 Pages -9 Available at ttp://scientificadvances.co.in DOI: ttp://d.doi.org/.864/jmsaa_7975 AN ANALYSIS OF NEW FINITE ELEMENT SPACES

More information

Discontinuous Galerkin Methods for Relativistic Vlasov-Maxwell System

Discontinuous Galerkin Methods for Relativistic Vlasov-Maxwell System Discontinuous Galerkin Metods for Relativistic Vlasov-Maxwell System He Yang and Fengyan Li December 1, 16 Abstract e relativistic Vlasov-Maxwell (RVM) system is a kinetic model tat describes te dynamics

More information

Preconditioning in H(div) and Applications

Preconditioning in H(div) and Applications 1 Preconditioning in H(div) and Applications Douglas N. Arnold 1, Ricard S. Falk 2 and Ragnar Winter 3 4 Abstract. Summarizing te work of [AFW97], we sow ow to construct preconditioners using domain decomposition

More information

On convergence of the immersed boundary method for elliptic interface problems

On convergence of the immersed boundary method for elliptic interface problems On convergence of te immersed boundary metod for elliptic interface problems Zilin Li January 26, 2012 Abstract Peskin s Immersed Boundary (IB) metod is one of te most popular numerical metods for many

More information

Grad-div stabilization for the evolutionary Oseen problem with inf-sup stable finite elements

Grad-div stabilization for the evolutionary Oseen problem with inf-sup stable finite elements Noname manuscript No. will be inserted by te editor Grad-div stabilization for te evolutionary Oseen problem wit inf-sup stable finite elements Javier de Frutos Bosco García-Arcilla Volker Jon Julia Novo

More information

A MULTILEVEL PRECONDITIONER FOR THE INTERIOR PENALTY DISCONTINUOUS GALERKIN METHOD

A MULTILEVEL PRECONDITIONER FOR THE INTERIOR PENALTY DISCONTINUOUS GALERKIN METHOD A MULTILEVEL PRECONDITIONER FOR THE INTERIOR PENALTY DISCONTINUOUS GALERKIN METHOD KOLJA BRIX, MARTIN CAMPOS PINTO, AND WOLFGANG DAHMEN Abstract. In tis article we present a multilevel preconditioner for

More information

FEM solution of the ψ-ω equations with explicit viscous diffusion 1

FEM solution of the ψ-ω equations with explicit viscous diffusion 1 FEM solution of te ψ-ω equations wit explicit viscous diffusion J.-L. Guermond and L. Quartapelle 3 Abstract. Tis paper describes a variational formulation for solving te D time-dependent incompressible

More information

MANY scientific and engineering problems can be

MANY scientific and engineering problems can be A Domain Decomposition Metod using Elliptical Arc Artificial Boundary for Exterior Problems Yajun Cen, and Qikui Du Abstract In tis paper, a Diriclet-Neumann alternating metod using elliptical arc artificial

More information

LIMITS AND DERIVATIVES CONDITIONS FOR THE EXISTENCE OF A LIMIT

LIMITS AND DERIVATIVES CONDITIONS FOR THE EXISTENCE OF A LIMIT LIMITS AND DERIVATIVES Te limit of a function is defined as te value of y tat te curve approaces, as x approaces a particular value. Te limit of f (x) as x approaces a is written as f (x) approaces, as

More information

A Local Projection Stabilization/Continuous Galerkin Petrov Method for Incompressible Flow Problems

A Local Projection Stabilization/Continuous Galerkin Petrov Method for Incompressible Flow Problems A Local Projection Stabilization/Continuous Galerkin Petrov Metod for Incompressible Flow Problems Naveed Amed, Volker Jon, Gunar Matties, Julia Novo Marc 22, 218 Abstract A local projection stabilization

More information

Finite Difference Method

Finite Difference Method Capter 8 Finite Difference Metod 81 2nd order linear pde in two variables General 2nd order linear pde in two variables is given in te following form: L[u] = Au xx +2Bu xy +Cu yy +Du x +Eu y +Fu = G According

More information

A C 0 interior penalty discontinuous Galerkin method for fourth order total variation flow. II: Existence and uniqueness

A C 0 interior penalty discontinuous Galerkin method for fourth order total variation flow. II: Existence and uniqueness Numerical nalysis and Scientific Computing Preprint Seria C 0 interior penalty discontinuous Galerkin metod for fourt order total variation flow. II: xistence and uniqueness C. andari R. H. W. Hoppe R.

More information

The Laplace equation, cylindrically or spherically symmetric case

The Laplace equation, cylindrically or spherically symmetric case Numerisce Metoden II, 7 4, und Übungen, 7 5 Course Notes, Summer Term 7 Some material and exercises Te Laplace equation, cylindrically or sperically symmetric case Electric and gravitational potential,

More information

The derivative function

The derivative function Roberto s Notes on Differential Calculus Capter : Definition of derivative Section Te derivative function Wat you need to know already: f is at a point on its grap and ow to compute it. Wat te derivative

More information

Analysis of time-dependent Navier-Stokes flow coupled with Darcy

Analysis of time-dependent Navier-Stokes flow coupled with Darcy Analysis of time-dependent Navier-Stokes flow coupled wit Darcy flow Ayçıl Çeşmelioğlu and Béatrice Rivière Abstract Tis paper formulates and analyzes a weak solution to te coupling of time-dependent Navier-Stokes

More information

How to Find the Derivative of a Function: Calculus 1

How to Find the Derivative of a Function: Calculus 1 Introduction How to Find te Derivative of a Function: Calculus 1 Calculus is not an easy matematics course Te fact tat you ave enrolled in suc a difficult subject indicates tat you are interested in te

More information

1. Introduction. Consider a semilinear parabolic equation in the form

1. Introduction. Consider a semilinear parabolic equation in the form A POSTERIORI ERROR ESTIMATION FOR PARABOLIC PROBLEMS USING ELLIPTIC RECONSTRUCTIONS. I: BACKWARD-EULER AND CRANK-NICOLSON METHODS NATALIA KOPTEVA AND TORSTEN LINSS Abstract. A semilinear second-order parabolic

More information

Posteriori Analysis of a Finite Element Discretization for a Penalized Naghdi Shell

Posteriori Analysis of a Finite Element Discretization for a Penalized Naghdi Shell International Journal of Difference Equations ISSN 0973-6069, Volume 8, Number 1, pp. 111 124 (2013) ttp://campus.mst.edu/ijde Posteriori Analysis of a Finite Element Discretization for a Penalized Nagdi

More information

ERROR BOUNDS FOR FINITE-DIFFERENCE METHODS FOR RUDIN OSHER FATEMI IMAGE SMOOTHING

ERROR BOUNDS FOR FINITE-DIFFERENCE METHODS FOR RUDIN OSHER FATEMI IMAGE SMOOTHING ERROR BOUNDS FOR FINITE-DIFFERENCE METHODS FOR RUDIN OSHER FATEMI IMAGE SMOOTHING JINGYUE WANG AND BRADLEY J. LUCIER Abstract. We bound te difference between te solution to te continuous Rudin Oser Fatemi

More information

A UNIFORM INF SUP CONDITION WITH APPLICATIONS TO PRECONDITIONING

A UNIFORM INF SUP CONDITION WITH APPLICATIONS TO PRECONDITIONING A UNIFORM INF SUP CONDIION WIH APPLICAIONS O PRECONDIIONING KEN ANDRE MARDAL, JOACHIM SCHÖBERL, AND RAGNAR WINHER Abstract. A uniform inf sup condition related to a parameter dependent Stokes problem is

More information

Une méthode de pénalisation par face pour l approximation des équations de Navier-Stokes à nombre de Reynolds élevé

Une méthode de pénalisation par face pour l approximation des équations de Navier-Stokes à nombre de Reynolds élevé Une méthode de pénalisation par face pour l approximation des équations de Navier-Stokes à nombre de Reynolds élevé CMCS/IACS Ecole Polytechnique Federale de Lausanne Erik.Burman@epfl.ch Méthodes Numériques

More information

arxiv: v1 [math.na] 18 Jul 2015

arxiv: v1 [math.na] 18 Jul 2015 ENERGY NORM ERROR ESTIMATES FOR FINITE ELEMENT DISCRETIZATION OF PARABOLIC PROBLEMS HERBERT EGGER arxiv:150705183v1 [matna] 18 Jul 2015 Department of Matematics, TU Darmstadt, Germany Abstract We consider

More information

Lecture 15. Interpolation II. 2 Piecewise polynomial interpolation Hermite splines

Lecture 15. Interpolation II. 2 Piecewise polynomial interpolation Hermite splines Lecture 5 Interpolation II Introduction In te previous lecture we focused primarily on polynomial interpolation of a set of n points. A difficulty we observed is tat wen n is large, our polynomial as to

More information

Finite Difference Methods Assignments

Finite Difference Methods Assignments Finite Difference Metods Assignments Anders Söberg and Aay Saxena, Micael Tuné, and Maria Westermarck Revised: Jarmo Rantakokko June 6, 1999 Teknisk databeandling Assignment 1: A one-dimensional eat equation

More information