Partial Differential Equations - Modelling and ... - ResearchGate
Partial Differential Equations - Modelling and ... - ResearchGate Partial Differential Equations - Modelling and ... - ResearchGate
238 P. Neittaanmäki and D. Tiba Proof. The existence and the uniqueness of the optimal pair [u ε ,y ε ] ∈ L 2 (Ω 0 ) × H 1 (Ω 0 ) of the control problem (8), (9) is obvious. The pair [0,0] is clearly admissible and, for any ε>0, we obtain 1 2 |y ε − w| 2 H 1/2 (Γ ) + ε 2 |u ε| 2 L 2 (Ω 0) ≤ 1 2 |w|2 H 1/2 (Γ ) . Therefore, {y ε } and {ε 1/2 u ε } are bounded respectively in H 1/2 (Γ ), L 2 (Ω 0 ). We denote by l ∈ H 1/2 (Γ ) the weak limit (on a subsequence) of {y ε − w}. Let us define the adjoint system by: ⎡ ⎤ ∫ d∑ ∫ ⎣ ∂z ∂p ε a ij + a 0 zp ε ⎦ dx = (y ε − w)zdσ ∀z ∈ H 1 (Ω 0 ), (11) Ω 0 ∂x i ∂x j Γ i,j=1 which is a non-homogeneous Neumann problem and p ε ∈ H 1 (Ω 0 ). We also introduce the equation in variations ⎡ ⎤ ∫ d∑ ∫ ⎣ ∂µ ∂z a ij + a 0 µz⎦ dx = νz dx ∀z ∈ H 1 (Ω 0 ), (12) Ω 0 ∂x i ∂x j Ω 0 i,j=1 which defines the variations y ε + λµ, u ε + λν for any ν ∈ L 2 (Ω 0 )andλ ∈ R. A standard computation using (11), (12) and the optimality of [u ε ,y ε ] gives 0=ε(u ε ,ν) L 2 (Ω 0) +(y ε − w, µ) H 1/2 (Γ ) ⎡ ⎤ ∫ d∑ = ε(u ε ,ν) L2 (Ω 0) + ⎣ ∂µ ∂p ε a ij + a 0 µp ε ⎦ dx Ω 0 ∂x i,j=1 i ∂x j = ε(u ε ,ν) L 2 (Ω 0) +(p ε ,ν) L 2 (Ω 0). (13) Due to the convergence properties of the right-hand side in (11), {p ε } is bounded in H 1 (Ω 0 ) and we can pass to the limit (on a subsequence) p ε → p weakly in H 1 (Ω 0 ), to obtain ⎤ ∫ d∑ ∫ ⎣ ∂z ∂p a ij + a 0 zp⎦ dx = lz dσ ∀z ∈ H 1 (Ω 0 ). (14) ∂x i ∂x j Ω 0 ⎡ i,j=1 The passage to the limit in (13), as {ε 1/2 u ε } is bounded, gives that p ≡ 0in Ω 0 and (14) shows that l =0inΓ . We have proved (10) in the weak topology of H 1/2 (Γ ). The strong convergence is a consequence of the Mazur theorem [Yos80] and of a variational argument. Γ
Shape Optimization Problems with Neumann Boundary Condition 239 Remark 1. The Mazur theorem alone and the linearity of (9) produces a sequence ũ ε (of convex combinations of u ε ) such that the corresponding sequence of states ỹ ε satisfies (10). Theorem 1 gives a constructive answer to the approximate controllability property. If Ω 0 is smooth enough and w ∈ H 3/2 (Γ ), then the trace theorem ensures the existence of ŷ ∈ H 2 ∂ŷ (Ω 0 ) such that ∂n A = 0 (null conormal derivative) and ŷ| Γ = w. That is, the control û = − d∑ i,j=1 ( ) ∂ ∂ŷ a ij + a 0 ŷ ∂x j ∂x i ensures the exact controllability property. Notice that û is not unique since any element in H 2 0 (Ω 0 ) may be added to ŷ with all the properties being preserved. 3 A Variational Fixed Domain Formulation We assume that Ω = Ω g ,whereg ∈ C( ¯D), is as in (5). Motivated by the result in the previous section, we consider the following homogeneous Neumann problem in D: d∑ ( ) ∂ ∂ỹ − a ij + a 0 ỹ = f +(1− H(g))u in D, (15) ∂x i,j=1 j ∂x i ∂y =0 on∂D. (16) ∂n A Here H(·) is the Heaviside function in R and H(g) is, consequently, the characteristic function of Ω g . Under conditions of Theorem 1, the restriction y =ỹ| Ωg is the solution of (2) in Ω = Ω g . Moreover, since g =0on∂Ω g , under smoothness conditions, ∇g is parallel to ¯n, the normal to ∂Ω g . Then, we can rewrite (4) as d∑ ∂y a ij ∇g · e i =0 on∂Ω g , (17) ∂x i,j=1 j where we use that cos(¯n, x i ) = cos(∇g, x i )ande i is the vector of the axis x i . If the elliptic operator is the Laplace operator, then (17) becomes simply ∇g ·∇y =0 on∂Ω g . In order to fix a unique u ∈ L 2 (D) satisfying to (15), (16), (17), we define the following optimal control problem with state constraints: { ∫ } 1 Min u 2 dx , (18) u∈L 2 (D) 2 governed by the state system (15), (16) and subject to the state constraint (17). D
- Page 185 and 186: Numerical Analysis of a Finite Elem
- Page 187 and 188: Numerical Analysis of a Finite Elem
- Page 189 and 190: 188 A. Bonito et al. of the model a
- Page 191 and 192: 190 A. Bonito et al. Fig. 1. The sp
- Page 193 and 194: 192 A. Bonito et al. 3 1 16 4 1 1 1
- Page 195 and 196: 194 A. Bonito et al. ∫ v n+1 h
- Page 197 and 198: 196 A. Bonito et al. −pn +2µD(v)
- Page 199 and 200: 198 A. Bonito et al. each of its pa
- Page 201 and 202: 200 A. Bonito et al. The normal vec
- Page 203 and 204: 202 A. Bonito et al. with initial c
- Page 205 and 206: 204 A. Bonito et al. Fig. 8. Jet bu
- Page 207 and 208: 206 A. Bonito et al. References [AM
- Page 209 and 210: 208 A. Bonito et al. [Set96] J. A.
- Page 211 and 212: 210 J. Hao et al. due to shear flow
- Page 213 and 214: 212 J. Hao et al. The backward reac
- Page 215 and 216: 214 J. Hao et al. where u and p den
- Page 217 and 218: 216 J. Hao et al. * * * * * * * * *
- Page 219 and 220: 218 J. Hao et al. and solve for V n
- Page 221 and 222: 220 J. Hao et al. Table 2. The calc
- Page 223 and 224: 222 J. Hao et al. References [ASS80
- Page 225 and 226: Computing the Eigenvalues of the La
- Page 227 and 228: Eigenvalues of the Laplace-Beltrami
- Page 229 and 230: Eigenvalues of the Laplace-Beltrami
- Page 231 and 232: Eigenvalues of the Laplace-Beltrami
- Page 233 and 234: A Fixed Domain Approach in Shape Op
- Page 235: Shape Optimization Problems with Ne
- Page 239 and 240: Shape Optimization Problems with Ne
- Page 241 and 242: Shape Optimization Problems with Ne
- Page 243 and 244: Reduced-Order Modelling of Dispersi
- Page 245 and 246: Reduced-Order Modelling of Dispersi
- Page 247 and 248: Reduced-Order Modelling of Dispersi
- Page 249 and 250: Reduced-Order Modelling of Dispersi
- Page 251 and 252: Reduced-Order Modelling of Dispersi
- Page 253 and 254: Reduced-Order Modelling of Dispersi
- Page 255 and 256: Calibration of Lévy Processes with
- Page 257 and 258: Calibration of Lévy Processes with
- Page 259 and 260: Calibration of Lévy Processes with
- Page 261 and 262: Calibration of Lévy Processes with
- Page 263 and 264: We have proved Calibration of Lévy
- Page 265 and 266: Calibration of Lévy Processes with
- Page 267 and 268: Calibration of Lévy Processes with
- Page 269 and 270: Calibration of Lévy Processes with
- Page 271 and 272: Note that p ∗ satisfies Calibrati
- Page 273 and 274: Calibration of Lévy Processes with
- Page 275 and 276: 280 S. Ikonen and J. Toivanen the p
- Page 277 and 278: 282 S. Ikonen and J. Toivanen Merto
- Page 279 and 280: 284 S. Ikonen and J. Toivanen For H
- Page 281 and 282: 286 S. Ikonen and J. Toivanen and {
- Page 283 and 284: 288 S. Ikonen and J. Toivanen 1.6 1
- Page 285 and 286: 290 S. Ikonen and J. Toivanen 8 Con
238 P. Neittaanmäki <strong>and</strong> D. Tiba<br />
Proof. The existence <strong>and</strong> the uniqueness of the optimal pair [u ε ,y ε ] ∈<br />
L 2 (Ω 0 ) × H 1 (Ω 0 ) of the control problem (8), (9) is obvious. The pair [0,0]<br />
is clearly admissible <strong>and</strong>, for any ε>0, we obtain<br />
1<br />
2 |y ε − w| 2 H 1/2 (Γ ) + ε 2 |u ε| 2 L 2 (Ω 0) ≤ 1 2 |w|2 H 1/2 (Γ ) .<br />
Therefore, {y ε } <strong>and</strong> {ε 1/2 u ε } are bounded respectively in H 1/2 (Γ ), L 2 (Ω 0 ).<br />
We denote by l ∈ H 1/2 (Γ ) the weak limit (on a subsequence) of {y ε − w}.<br />
Let us define the adjoint system by:<br />
⎡<br />
⎤<br />
∫ d∑<br />
∫<br />
⎣<br />
∂z ∂p ε<br />
a ij + a 0 zp ε<br />
⎦ dx = (y ε − w)zdσ ∀z ∈ H 1 (Ω 0 ), (11)<br />
Ω 0<br />
∂x i ∂x j Γ<br />
i,j=1<br />
which is a non-homogeneous Neumann problem <strong>and</strong> p ε ∈ H 1 (Ω 0 ). We also<br />
introduce the equation in variations<br />
⎡<br />
⎤<br />
∫ d∑<br />
∫<br />
⎣<br />
∂µ ∂z<br />
a ij + a 0 µz⎦ dx = νz dx ∀z ∈ H 1 (Ω 0 ), (12)<br />
Ω 0<br />
∂x i ∂x j Ω 0<br />
i,j=1<br />
which defines the variations y ε + λµ, u ε + λν for any ν ∈ L 2 (Ω 0 )<strong>and</strong>λ ∈ R.<br />
A st<strong>and</strong>ard computation using (11), (12) <strong>and</strong> the optimality of [u ε ,y ε ]<br />
gives<br />
0=ε(u ε ,ν) L 2 (Ω 0) +(y ε − w, µ) H 1/2 (Γ )<br />
⎡<br />
⎤<br />
∫ d∑<br />
= ε(u ε ,ν) L2 (Ω 0) + ⎣<br />
∂µ ∂p ε<br />
a ij + a 0 µp ε<br />
⎦ dx<br />
Ω 0<br />
∂x<br />
i,j=1 i ∂x j<br />
= ε(u ε ,ν) L 2 (Ω 0) +(p ε ,ν) L 2 (Ω 0). (13)<br />
Due to the convergence properties of the right-h<strong>and</strong> side in (11), {p ε } is<br />
bounded in H 1 (Ω 0 ) <strong>and</strong> we can pass to the limit (on a subsequence) p ε → p<br />
weakly in H 1 (Ω 0 ), to obtain<br />
⎤<br />
∫ d∑<br />
∫<br />
⎣<br />
∂z ∂p<br />
a ij + a 0 zp⎦ dx = lz dσ ∀z ∈ H 1 (Ω 0 ). (14)<br />
∂x i ∂x j<br />
Ω 0<br />
⎡<br />
i,j=1<br />
The passage to the limit in (13), as {ε 1/2 u ε } is bounded, gives that p ≡ 0in<br />
Ω 0 <strong>and</strong> (14) shows that l =0inΓ .<br />
We have proved (10) in the weak topology of H 1/2 (Γ ). The strong convergence<br />
is a consequence of the Mazur theorem [Yos80] <strong>and</strong> of a variational<br />
argument.<br />
Γ