intTypePromotion=1
zunia.vn Tuyển sinh 2024 dành cho Gen-Z zunia.vn zunia.vn
ADSENSE

FEM for Elliptic problems

Chia sẻ: Quang Huy | Ngày: | Loại File: PDF | Số trang:8

32
lượt xem
1
download
 
  Download Vui lòng tải xuống để xem tài liệu đầy đủ

FEM for Elliptic problems presented Introduction; variational formulation; existence of solutions: lax-milgram theorem; FEM problem,... Invite you to consult the documentation

Chủ đề:
Lưu

Nội dung Text: FEM for Elliptic problems

FEM for Elliptic Problems<br /> Sebastian Gonzalez Pintor<br /> November, 2016<br /> Proof. Multiply equation (D) by v ∈ V and integrate<br /> over the whole domain<br /> Z 1<br /> Z 1<br /> 00<br /> f v dx,<br /> u v dx =<br /> −<br /> <br /> We follow the results from [Joh12] and [And15].<br /> <br /> 1<br /> <br /> Introduction<br /> <br /> then we integrate by parts and use the boundary conditions on the left side to obtain<br /> Z 1<br /> Z 1<br /> −<br /> u00 v dx = − [u0 v]10 +<br /> u0 v 0 dx<br /> <br /> Outline: Variational form. and Minimization prob.<br /> •<br /> •<br /> •<br /> •<br /> <br /> 0<br /> <br /> 0<br /> <br /> Definition of (D), (V) and (M)<br /> Equivalence (D) ⇒ (V ) ⇔ (M )<br /> If u ∈ C 2 then (D) ⇐ (V )<br /> Uniqueness of (V ).<br /> <br /> 0<br /> <br /> 0<br /> <br /> = −u0 (1)v(1) + u0 (0)v(0) +(u0 , v 0 )<br /> {z<br /> }<br /> |<br /> v(1)=v(0)=0<br /> <br /> 0<br /> <br /> 0<br /> <br /> = (u , v ).<br /> We notice that u ∈ C 2 and u(0) = u(1) = 0 implies that<br /> u ∈ V , and because the choice of v ∈ V is arbitrary, the<br /> function u satisfies the equation<br /> <br /> Lets consider the following two-point boundary value<br /> problem<br /> − u00 (x) = f (x)<br /> <br /> for x ∈ (0, 1)<br /> <br /> (D)<br /> <br /> (u0 , v 0 ) = (f, v) ∀v ∈ V,<br /> <br /> u(0) = u(1) = 0<br /> where u0 = ux and f is a given continuous function. We<br /> introduce the notation<br /> Z 1<br /> (u, v) =<br /> v(x)w(x) dx<br /> <br /> what is the same as problem (V).<br /> Proposition 2 (V ⇒ M ). If u ∈ V is a solution of<br /> the variational problem (V), then u is a solution of the<br /> minimization problem (M) .<br /> <br /> 0<br /> <br /> for real valued piecewise continuous bounded functions,<br /> and the linear space<br /> Z 1<br /> V = {v : v ∈ C 0 ([0, 1]),<br /> |v 0 |2 dx < ∞,<br /> <br /> Proof. Let us consider an arbitrary function w ∈ V . We<br /> define v = w − u ∈ V , so that using the definition of the<br /> functional F we have<br /> 1 0 0<br /> (w , w ) − (w, f )<br /> 2<br /> 1<br /> = (v 0 + u0 , v 0 + u0 ) − (v + u, f )<br /> 2<br /> 1<br /> 1<br /> = (v 0 , v 0 ) + (u0 , v 0 ) + (u0 , u0 ) − (v, f ) − (u, f )<br /> 2<br /> 2<br /> 1<br /> 1<br /> = (v 0 , v 0 ) + (u0 , v 0 ) − (v, f ) + (u0 , u0 ) − (u, f )<br /> |<br /> {z<br /> } |2<br /> 2<br /> {z<br /> }<br /> <br /> 0<br /> <br /> F (w) =<br /> <br /> and v(0) = v(1) = 0}<br /> and the linear functional<br /> F (v) =<br /> <br /> 1 0 0<br /> (v , v ) − (f, v).<br /> 2<br /> <br /> We can define the problems (M) and (V) as follows:<br /> <br /> (I)<br /> <br /> Find u ∈ V such that F (u) ≤ F (v) ∀v ∈ V,<br /> <br /> (M)<br /> <br /> Find u ∈ V such that (u0 , v 0 ) = (f, v) ∀v ∈ V.<br /> <br /> (II)<br /> <br /> (V) but because u is solution of (V) we have that (I) = 0,<br /> and we can rewrite (II) = F (u), so using that (v 0 , v 0 ) ≥ 0<br /> Now we will show that if u is a solution of the two- we obtain<br /> 1<br /> point boundary value problem (D), then u is a solution<br /> F (w) = (v 0 , v 0 ) + F (u) ≥ F (u).<br /> of the variational problem (V) (D ⇒ V ), and that u is<br /> 2<br /> a solution of the variational problem (V) if, and only<br /> Because w ∈ V was chosen arbitrarily, it means that u<br /> if, u is also a solution of the minimization problem (M)<br /> is a solution of (M).<br /> (V ⇔ M ).<br /> Proposition 1 (D ⇒ V ). If u ∈ C 2 is a solution of Proposition 3 (M ⇒ V ). If u ∈ V is a solution of the<br /> the two-point boundary value problem (D), then u is a minimization problem (M), then u is a solution of the<br /> solution of the variational problem (V) .<br /> variational problem (V) .<br /> 1<br /> <br /> and integrating by parts the first term, using that v(0) =<br /> v(1) = 0 and that u00 exists we get,<br /> <br /> Proof. If u is a solution of (M) we have that<br /> F (u) ≤ F (u + εv),<br /> <br /> (u0 , v 0 ) = (f, v) ∀v ∈ V,<br /> <br /> for all ε ∈ R, because u + εv ∈ V for every v ∈ V . Thus,<br /> the differentiable function<br /> <br /> − (u00 , v) = (f, v) ∀v ∈ V,<br /> (f + u00 , v) = 0 ∀v ∈ V,<br /> <br /> g(ε) ≡ F (u + εv)<br /> <br /> but with the assumption that f + u00 is continuous, this<br /> can only hold if f (x) + u00 (x) = 0 ∀x ∈ (0, 1) .<br /> <br /> 2<br /> <br /> 1 0 0<br /> ε<br /> (u , u ) + ε(u0 , v 0 ) + (v 0 , v 0 ) − (u, f ) − ε(v, f )<br /> 2<br /> 2<br /> ε2<br /> 1 0 0<br /> = (u , u ) − (u, f ) + ε[(u0 , v 0 ) − (v, f )] + (v 0 , v 0 ),<br /> 2<br /> 2<br /> <br /> =<br /> <br /> 2<br /> <br /> Variational Formulation<br /> <br /> 0<br /> <br /> has a minimum at ε = 0 and hence g (0) = 0. But<br /> Outline: Hilbert Spaces<br /> <br /> g 0 (ε) = (u0 , v 0 ) − (v, f ) + ε(v 0 , v 0 ),<br /> <br /> • Hilbert Spaces (L2 , H1 and H10 )<br /> – Scalar product, Cauchy-Schwarz inequality and Cauchy sequence<br /> • Natural and Essential boundary conditions<br /> <br /> and evaluating at ε = 0 we get<br /> g 0 (0) = (u0 , v 0 ) − (v, f ) = 0,<br /> hence u is a solution of (V).<br /> We can also show that<br /> Proposition 4 (Uniqueness). A solution to (V) is<br /> uniquely determined.<br /> <br /> Definition 1. A map a(·, ·) : V × V → R that is symmetric and bilinear, i.e., ∀u, v, w ∈ V , ∀α, β ∈ R<br /> a(v, w) = (w, v),<br /> <br /> Proof. Suppose that u1 and u2 are solutions of (V), i.e.,<br /> u1 , u2 ∈ V and<br /> (u01 , v 0 ) = (f, v) ∀v ∈ V,<br /> (u0w , v 0 )<br /> <br /> (symmetric)<br /> <br /> a(αu + βv, w) = αa(u, w) + βa(v, w)<br /> <br /> (a(·, w) is linear)<br /> <br /> a(u, αv + βw) = αa(u, v) + βa(u, w)<br /> <br /> (a(u, ·) is linear)<br /> <br /> is called a scalar product on V if<br /> <br /> = (f, v) ∀v ∈ V.<br /> <br /> a(v, v) > 0, ∀v ∈ V, v 6= 0,<br /> <br /> Subtracting these two equations and v = u1 − u2 ∈ V<br /> gives<br /> Z 1<br /> (u01 − u02 )2 dx = 0<br /> <br /> and the norm associated with the scalar product is defined<br /> by<br /> ||v||a = (a(v, v))1/2 , ∀v ∈ V.<br /> <br /> 0<br /> <br /> Moreover, if < ·, · > is a scalar product with corresponding norm || · ||, we have that the following CauchySchwarz’s inequality is satisfied<br /> <br /> which shows that<br /> u01 (x) − u02 (x) = 0 ∀x ∈ [0, 1],<br /> <br /> | < v, w > | ≤ ||v|| ||w||, ∀v, w ∈ V.<br /> <br /> so u1 − u2 is constant in [0, 1], which together with the<br /> boundary conditions u1 (0) = u2 (0) = 0 gives u1 (x) =<br /> u2 (x), ∀x ∈ [0, 1], and the uniqueness follows.<br /> <br /> Definition 2. A linear space V with a scalar product<br /> < ·, · > and the corresponding norm || · || is said to be<br /> a Hilbert space if V is complete, i.e., if every Cauchy<br /> sequence with respect to || · || is convergent, where<br /> <br /> Also, because every solution of (D) or (M) is also solution<br /> of (V), but the solution of (V) is unique, it follows the<br /> uniqueness of the solutions of (D) and (M). We notice<br /> here that we have not proved yet that the solution exists,<br /> just uniqueness in case of existence. Existence will be<br /> shown later for a general variational problem, given that<br /> it satisfies some conditions.<br /> <br /> • a sequence is said to be Cauchy if<br /> ∀ε > 0 ∃N ∈ N such that ||vi − vj || < ε if i, j ≥ N,<br /> • a sequence is convergent in V if ∃v ∈ V such that<br /> <br /> ∀ε > 0 ∃N ∈ N such that ||vi − v|| < ε if i ≥ N.<br /> Last, we show that<br /> Proposition 5. If u00 (x) exists and it is continuous for<br /> a solution of (V), then u is also a solution of (D).<br /> Example 2.1. We can see that the space V :=<br /> C 0 ([−1, 1]) is a linear space, and that (·, ·)V : V × V → R<br /> Proof. For this, assuming that u is a solution of (V), we defined by<br /> have<br /> Z 1<br /> (v,<br /> w)<br /> =<br /> v(x)w(x) dx,<br /> V<br /> (u0 , v 0 ) = (f, v) ∀v ∈ V,<br /> −1<br /> 2<br /> <br /> 1.2<br /> <br /> together with the scalar product and its associated<br /> norm as follows<br /> Z<br /> (v, w)L2 (Ω) =<br /> vw dx,<br /> ZΩ<br /> ||v||L2 (Ω) =<br /> v 2 dx.<br /> <br /> 1<br /> vi<br /> <br /> 0.8<br /> <br /> vj<br /> <br /> 0.6<br /> <br /> Ω<br /> 0.4<br /> <br /> • We also introduce the linear space of square integrable derivatives up to order 1,<br /> <br /> 0.2<br /> <br /> H1 (Ω) = {v ∈ L2 (Ω), vxi ∈ L2 , i = 1, . . . , d},<br /> <br /> 0<br /> 1/N<br /> <br /> together with the scalar product and the associated<br /> norm as follows<br /> Z<br /> (v, w)H1 (Ω) =<br /> vw + ∇v · ∇w dx,<br /> Z Ω<br /> ||v||H1 (Ω) = v 2 +|∇v|2 dx = ||v||L2 (Ω) + ||∇v||L2 (Ω) .<br /> <br /> -0.2<br /> -1<br /> <br /> -0.5<br /> <br /> 0<br /> <br /> 0.5<br /> <br /> 1<br /> <br /> Figure 1: For i, j > N > 1/ε, the square of the area<br /> between vi and vj is smaller than ε.<br /> is a scalar product, with the corresponding associated<br /> 1/2<br /> norm ||v||V = (v, v)V . But V with this norm is not<br /> complete, thus it is not a Hilbert space.<br /> <br /> Ω<br /> <br /> • And the space<br /> H10 (Ω) := {v ∈ H1 (Ω) : v = 0 on Γ},<br /> <br /> Proof. To see this we can build a Cauchy sequence that<br /> is not convergent inside V . For example, consider the<br /> sequence<br /> <br /> <br /> if x ∈ [−1, 0],<br /> 0<br /> vn = nx if x ∈ (0, 1/n),<br /> <br /> <br /> 1<br /> if x ∈ [1/n, 1].<br /> <br /> together with the same scalar product and the associated norm as for H1 (Ω), where Γ := ∂Ω is the<br /> boundary of Ω.<br /> <br /> 2.1<br /> <br /> Boundary Conditions<br /> <br /> Take ε > 0, and 1/ε < N ∈ N, we can see (Figure 1 that<br /> The boundary conditions for the differential equation are<br /> the sequence is Cauchy<br /> usually classified as Dirichlet (when they impose a value<br /> Z 1<br /> for the function on the boundary), Neumann (when they<br /> 2<br /> 2<br /> ||vi − vj ||V =<br /> (vi (x) − vj (x)) dx<br /> force a value for the derivative of the function on the<br /> −1<br /> boundary), or mixed (mixing Dirichlet and Newman),<br /> 0≤·≤1<br /> Z 1/N z<br /> }|<br /> {<br /> =<br /> (vi (x) − vj (x))2 dx<br /> u=g<br /> on ∂Ω<br /> (Dirichlet)<br /> 0<br /> Z 1/N<br /> ~n∇u = h<br /> on ∂Ω<br /> (Neumann)<br /> ≤<br /> dx = 1/N < ε,<br /> γu + ~n∇u = g<br /> on ∂Ω<br /> (Mixed)<br /> 0<br /> <br /> where g, h, γ are given functions. Building the variational formulation, when integrating by parts the original equation multiplied by a test function, different<br /> boundary conditions will appear in different ways in<br /> the variational formulation. In particular, we focus on<br /> Dirichlet and Neumann boundary conditions, that for<br /> second order differential equations will become essential<br /> and natural boundary conditions, meaning that<br /> <br /> but the sequence converges to the function<br /> (<br /> 0 if x ∈ [−1, 0],<br /> H(x) =<br /> 1 if x ∈ (0, 1],<br /> which is not inside V . It means that the space V with<br /> this norm is not complete, so it is not a Hilbert space.<br /> In particular, for a bounded domain Ω ∈ Rd , we define<br /> the following function spaces, with appropriate scalar<br /> products and associated norms, which are Hilbert spaces:<br /> <br /> • essential boundary conditions will appear in the<br /> function space V , while<br /> <br /> • natural boundary conditions will naturally appear<br /> • The linear space of square-integrable functions in a<br /> in the variational form.<br /> bounded domain Ω ∈ Rd ,<br /> In general, for an equation of order 2k, the variational<br /> Z<br /> 2<br /> 2<br /> form needs an space with functions differentiable up to<br /> L (Ω) = {v :<br /> |v| dx < ∞},<br /> Ω<br /> order k, so essential boundary conditions will be inposed<br /> 3<br /> <br /> in the space for orders 0, . . . , k − 1, and natural boundary conditions will be built in the variational form for<br /> orders k, . . . , 2k − 1. It is also important to notice that<br /> essential boundary conditions will generate two spaces,<br /> not necessarily the same, one for which the condition is<br /> imposed to the solution, and other one for the test function where the functions would be zero. For instance,<br /> consider the boundary, Γ, split in two parts, Γ1 and Γ2 ,<br /> then the two spaces are<br /> <br /> In addition, we can see that the minimization problem<br /> associated to this variational problem can be written by<br /> Find u ∈ V1 such that<br /> F (u) ≤ F (v)∀v ∈ V1<br /> where<br /> 1<br /> a(v, v) − L(v),<br /> 2<br /> and it is equivalent to the variational problem above.<br /> F (v) =<br /> <br /> • u ∈ V1 = {v : v ∈ H1 , and v = g on Γ1 },<br /> • v ∈ V2 = {v : v ∈ H1 , and v = 0 on Γ1 },<br /> <br /> 3<br /> <br /> and they are the same only when g = 0. We notice<br /> that the Neumann boundary conditions do not have any<br /> effect into the function spaces; instead, they will appear<br /> in the variational formulation.<br /> <br /> Existence of solutions:<br /> Lax-Milgram theorem<br /> Outline: Lax-Milgram<br /> <br /> • Parallelogram law<br /> • Lax-Milgram Theorem<br /> Example 2.2. Consider a bounded domain Ω ∈ Rn ,<br /> • (V ) ⇔ (M )<br /> with boundary Γ = ∂Ω ∈ C 1 piecewise, and the bound• Poincar´e Inequalities for H1 and H10<br /> ary is split in two parts Γ1 and Γ2 . The problem reads<br /> • Trace Inequality<br /> <br /> n<br /> <br /> −∆u<br /> =<br /> f<br /> in<br /> Ω<br /> ⊂<br /> R<br /> ,<br /> <br /> u=g<br /> on Γ1<br /> <br /> <br /> ~n · ∇u = h on Γ2<br /> The complete proofs of the following results are provided<br /> in [And15].<br /> Proof. We use the spaces<br /> Parallelogram Law: Let ||·|| be a norm associated to a<br /> V1 = {v : v ∈ H1 , and v = g on Γ1 },<br /> scalar product < ·, · >. The following equivalence holds<br /> V2 = {v : v ∈ H1 , and v = 0 on Γ1 },<br /> <br /> ||x + y||2 + ||x − y||2 = 2||x||2 + 2||y||2 .<br /> <br /> where the solution must be in V1 , and V2 is the space of<br /> test functions. Now, multiply the original equation by<br /> a test function v and integrate over the whole domain,<br /> obtaining<br /> Z<br /> Z<br /> −<br /> ∆uv dx =<br /> f v dx, ∀v ∈ V2 ,<br /> <br /> Theorem 1 (Lax-Milgram). Let V be a Hilbert space<br /> with scalar product < ·, · >V and associated norm || · ||V .<br /> Assume that a(·, ·) is a bilinear functional and L a linear<br /> functional satisfying:<br /> <br /> Ω<br /> <br /> (1) a(·, ·) is symmetric, i.e.,<br /> <br /> Ω<br /> <br /> but integrating by parts (Green’s formula) the left part<br /> Z<br /> −<br /> ∆uv dx<br /> Ω<br /> Z<br /> Z<br /> =<br /> ∇u · ∇v dx − ~n∇uv ds<br /> Z<br /> ZΩ<br /> ZΓ<br /> ~n∇uv ds<br /> =<br /> ∇u · ∇v dx −<br /> ~n∇uv ds −<br /> Γ<br /> Ω<br /> Γ<br /> } | 2 {z<br /> }<br /> | 1 {z<br /> v=0 on Γ1<br /> <br /> Z<br /> <br /> (2) a(·, ·) is V-elliptic, i.e.,<br /> ∃α > 0 s.t. a(v, v) ≥ α||v||2V , ∀v ∈ V ;<br /> (3) a(·, ·) is continuous, i.e.,<br /> ∃C ∈ R s.t. |a(v, w)| ≤ C||v||V ||w||V , ∀v, w ∈ V ;<br /> <br /> ~<br /> n∇u=h on Γ2<br /> <br /> Z<br /> ∇u · ∇v dx −<br /> <br /> =<br /> <br /> a(v, w) = a(w, v), ∀v, w ∈ V ;<br /> <br /> Ω<br /> <br /> (4) L(·) is continuous, i.e.,<br /> <br /> hv ds,<br /> Γ2<br /> <br /> ∃Λ ∈ R s.t. |L(v)| ≥ Λ||v||V , ∀v ∈ V ;<br /> <br /> so the variational formulation can be rewritten as<br /> Find u ∈ V1 such that<br /> Z<br /> Z<br /> Z<br /> f v dx +<br /> hv ds, ∀v ∈ V2 .<br /> ∇u · ∇v dx =<br /> Γ2<br /> | Ω {z<br /> } |Ω<br /> {z<br /> }<br /> a(u,v)<br /> <br /> Then there exists a unique function u ∈ V such that<br /> a(u, v) = L(v), ∀v ∈ V,<br /> and the following estimate holds<br /> <br /> L(v)<br /> <br /> ||u||V ≤<br /> 4<br /> <br /> Λ<br /> .<br /> α<br /> <br /> Proof. A sketch of the proof (complete proof in [And15])<br /> <br /> It is sometimes useful to be able to bound the norm of<br /> the function with the bound of the derivative in order to<br /> verify conditions (2) − (4) of the Lax-Milgram theorem.<br /> It is easy to show that this can not be done with constant<br /> functions, so that we can use the followings Poincar´e<br /> inequalities, always having a mechanism that identify<br /> constant functions with the function zero.<br /> Theorem 2 (Poincar´e inequality for H10 (Ω)). Let Ω be a<br /> bounded domain in Rn with boundary ∂Ω ∈ C 1 piecewise.<br /> Let u ∈ H10 (Ω). Then there exists a constant C > 0 such<br /> that<br /> <br /> • Build a solution of (M): ∃u ∈ V solving (M)<br /> – Norm ||v|| = a(v, v) is equivalent to || · ||V<br /> – We see that β = inf F (v) > −∞<br /> v∈V<br /> <br /> – {vi } ⊂ V , with F (vi ) → β, is Cauchy.<br /> – Completeness implies that ∃u ∈ V s.t. vi → u<br /> – Show that F (u) = β. Thus u solves (M).<br /> <br /> ||u||L2 (Ω) ≤ C||∇u||L2 (Ω) .<br /> <br /> • (M)⇒(V): Because of proposition 6.<br /> <br /> • Uniqueness: If u1 and u2 solve (V), subtracting the Proof. A sketch of the proof (complete proof in [And15])<br /> P<br /> equations we obtain a(u1 − u2 , v) = 0, and using (2)<br /> • We use the function φ(x) = 1/(2n) x2i , with ∆φ =<br /> for v = u1 − u2 we get the result.<br /> 1, with the Green’s formula and bcs. to get<br /> Z<br /> • Estimate: Combining (2) and (4) with (V).<br /> 2u(∇u · ∇φ) dx<br /> ||u||2L2 (Ω) = −<br /> Ω<br /> <br /> Proposition 6 ((V)⇔(M)). Let a and L be such that • and we use Cauchy-Schwarz inequality and boundness<br /> they satisfy the conditions of Theorem 1. Then, the folof φ in Ω<br /> lowing problems are equivalent:<br /> <br /> Z<br /> <br /> <br /> ≤ C||u||L2 (Ω) ||∇u||L2 (Ω)<br /> −<br /> 2u(∇u<br /> ·<br /> ∇φ)<br /> dx<br /> <br /> <br /> • Find u ∈ V such that a(u, v) = L(v), ∀v ∈ V ; (V)<br /> Ω<br /> • Find u ∈ V such that F (u) ≤ F (v), ∀v ∈ V ;<br /> where F (v) =<br /> <br /> (M)<br /> <br /> 1<br /> a(v, v) − L(v).<br /> 2<br /> <br /> Theorem 3 (Poincar´e inequality for H1 (Ω)). Let Ω be a<br /> bounded domain in Rn with boundary ∂Ω ∈ C 1 piecewise.<br /> Let u ∈ H1 (Ω) and let<br /> Z<br /> 1<br /> u<br /> ¯=<br /> u dx, (¯<br /> u is the average over Ω).<br /> |Ω| Ω<br /> <br /> Proof. Similar to (V⇔M) before (complete proof<br /> in [And15]).<br /> <br /> Then there exists a constant C > 0 such that<br /> Example 3.1. Show, using Lax-Milgram theorem, that<br /> the following problem has a unique solution:<br /> Find u ∈ V := H1 (Ω), Ω ⊂ R2 , such that<br /> Z<br /> Z<br /> (∇u · ∇v + uv) dx =<br /> f v dx, ∀v ∈ H1 (Ω)<br /> Ω<br /> <br /> ||u − u<br /> ¯||L2 (Ω) ≤ C||∇u||L2 (Ω) .<br /> Proof. We skip the proof.<br /> The next result (Trace inequality) can be used to prove<br /> continuity for a or L when boundary integrals are involved, i.e., when Neumann boundary conditions are<br /> used.<br /> Theorem 4 (Trace inequality). Let Ω be a bounded domain in Rn with boundary ∂Ω ∈ C 1 piecewise. Then<br /> there exists a constant C > 0 such that<br /> <br /> Ω<br /> <br /> Proof. Using<br /> Z<br /> a(v, w) :=<br /> (∇v · ∇w + vw) dx,<br /> <br /> Z<br /> L(w) :=<br /> <br /> Ω<br /> <br /> f w dx,<br /> Ω<br /> <br /> we check for the different conditions:<br /> <br /> 1/2<br /> <br /> 1/2<br /> <br /> ||u||L2 (∂Ω) ≤ C||u||L2 (Ω) ||u||H1 (Ω) .<br /> <br /> 1. a is clearly symmetric.<br /> 2. a(v, v) = ||v||2H1 (Ω) proves V-ellipticity<br /> <br /> Proof. We skip the proof.<br /> <br /> 3. Cauchy-Schwarz in H1 (Ω) provides continuity for<br /> |a(v, w)| = ||v||H1 (Ω) ||w||H1 (Ω)<br /> <br /> Example 3.2. Prove, using Lax-Milgram theorem, that<br /> 4. Cauchy-Schwarz in L (Ω) provides continuity for there exists a unique solution for the variational formu|L(v)| = |(f, v)L2 (Ω) | = ||f ||L2 (Ω) ||v||L2 (Ω) ≤ Λ||v||H1 (Ω) lation associated to the problem<br /> (<br /> with Λ = ||f ||L2 (Ω) .<br /> −∆u = f in Ω,<br /> Thus, using theorem 1 there exists a unique solution.<br /> u=0<br /> on ∂Ω.<br /> 2<br /> <br /> 5<br /> <br />
ADSENSE

CÓ THỂ BẠN MUỐN DOWNLOAD

 

Đồng bộ tài khoản
3=>0