Stabilization of a strong damped wave equation with acoustic boundary conditions. Nicolas Fourrier
U NIVERSITY
OF
V IRGINIA
[email protected]
September 11, 2012
Nicolas Fourrier (U.V.A.)
P.D.E. Seminar
September 11, 2012
1 / 42
Overview
1
Introduction
2
Theorems
3
Proof of the theorems
4
Numerical Calculations
Nicolas Fourrier (U.V.A.)
P.D.E. Seminar
September 11, 2012
2 / 42
Introduction
Overview
1
Introduction Presentation of the problem Mathematical model Preliminary definitions
2
Theorems
3
Proof of the theorems
4
Numerical Calculations
Nicolas Fourrier (U.V.A.)
P.D.E. Seminar
September 11, 2012
3 / 42
Introduction
Presentation of the problem
Physical Model
3-dimension bounded domain Ω. Boundary δΩ = Γ0 ∪ Γ1 Gas undergoes small irrotationnal perturbations away from Ω which are concentrated on a portion of the boundary Γ1 . Each point of the surface Γ1 acts like a spring in response to the excess pressure in the gas. There is no transverse tension between neighboring points of Γ1 , i.e., the springs are independent of each other. Dirichlet boundary conditions on Γ0 .
Nicolas Fourrier (U.V.A.)
P.D.E. Seminar
September 11, 2012
4 / 42
Introduction
Presentation of the problem
Previous results
Wave equation Viscous-frictional dampers lead to analytic and exponentially stable semigroup in the presence of classical boundary conditions (Dirichlet, Robin) No longer true in the case of acoustic boundary conditions.
Nicolas Fourrier (U.V.A.)
P.D.E. Seminar
September 11, 2012
5 / 42
Introduction
Mathematical model
utt − ∆u − cΩ ∆ut = 0 utt − δη (u + cΩ ut ) − kΓ ∆Γ (cΩ αut + u) = 0 u=0 Initial conditions u(0, x) = u0 , ut (0, x) = u1
in Ω in Γ1 in Γ0
(1)
on Ω
cΩ is the coefficient associated with the viscous interior damper. kΓ is the boundary coefficient. α = 0 or 1, allows to add a viscous boundary damper. If α = 0, kΓ is only related to speed on the boundary. If α = 1, kΓ also determines the speed, but it also determines with cΩ the intensity of the damping.
Nicolas Fourrier (U.V.A.)
P.D.E. Seminar
September 11, 2012
6 / 42
Introduction
Mathematical model
utt − ∆u − cΩ ∆ut = 0 utt − δη (u + cΩ ut ) − kΓ ∆Γ (cΩ αut + u) = 0 u=0 Initial conditions u(0, x) = u0 , ut (0, x) = u1
in Ω in Γ1 in Γ0
(1)
on Ω
cΩ is the coefficient associated with the viscous interior damper. kΓ is the boundary coefficient. α = 0 or 1, allows to add a viscous boundary damper. If α = 0, kΓ is only related to speed on the boundary. If α = 1, kΓ also determines the speed, but it also determines with cΩ the intensity of the damping.
Nicolas Fourrier (U.V.A.)
P.D.E. Seminar
September 11, 2012
6 / 42
Introduction
Mathematical model
utt − ∆u − cΩ ∆ut = 0 utt − δη (u + cΩ ut ) − kΓ ∆Γ (cΩ αut + u) = 0 u=0 Initial conditions u(0, x) = u0 , ut (0, x) = u1
in Ω in Γ1 in Γ0
(1)
on Ω
cΩ is the coefficient associated with the viscous interior damper. kΓ is the boundary coefficient. α = 0 or 1, allows to add a viscous boundary damper. If α = 0, kΓ is only related to speed on the boundary. If α = 1, kΓ also determines the speed, but it also determines with cΩ the intensity of the damping.
Nicolas Fourrier (U.V.A.)
P.D.E. Seminar
September 11, 2012
6 / 42
Introduction
Mathematical model
utt − ∆u − cΩ ∆ut = 0 utt − δη (u + cΩ ut ) − kΓ ∆Γ (cΩ αut + u) = 0 u=0 Initial conditions u(0, x) = u0 , ut (0, x) = u1
in Ω in Γ1 in Γ0
(1)
on Ω
cΩ is the coefficient associated with the viscous interior damper. kΓ is the boundary coefficient. α = 0 or 1, allows to add a viscous boundary damper. If α = 0, kΓ is only related to speed on the boundary. If α = 1, kΓ also determines the speed, but it also determines with cΩ the intensity of the damping.
Nicolas Fourrier (U.V.A.)
P.D.E. Seminar
September 11, 2012
6 / 42
Introduction
Mathematical model
utt − ∆u − cΩ ∆ut = 0 utt − δη (u + cΩ ut ) − kΓ ∆Γ (u) = 0 u=0 Initial conditions u(0, x) = u0 , ut (0, x) = u1
in Ω in Γ1 in Γ0
(1)
on Ω
cΩ is the coefficient associated with the viscous interior damper. kΓ is the boundary coefficient. α = 0 or 1, allows to add a viscous boundary damper. If α = 0, kΓ is only related to speed on the boundary. If α = 1, kΓ also determines the speed, but it also determines with cΩ the intensity of the damping.
Nicolas Fourrier (U.V.A.)
P.D.E. Seminar
September 11, 2012
6 / 42
Introduction
Mathematical model
utt − ∆u − cΩ ∆ut = 0 utt − δη (u + cΩ ut ) − kΓ ∆Γ (cΩ αut + u) = 0 u=0 Initial conditions u(0, x) = u0 , ut (0, x) = u1
in Ω in Γ1 in Γ0
(1)
on Ω
cΩ is the coefficient associated with the viscous interior damper. kΓ is the boundary coefficient. α = 0 or 1, allows to add a viscous boundary damper. If α = 0, kΓ is only related to speed on the boundary. If α = 1, kΓ also determines the speed, but it also determines with cΩ the intensity of the damping.
Nicolas Fourrier (U.V.A.)
P.D.E. Seminar
September 11, 2012
6 / 42
Introduction
Preliminary definitions
Definition of the operators Neumann Map: N
z = Ng ⇔
∆z = 0 δη z|Γ = g 1 z|Γ = 0 0
3
3 −ǫ
By elliptic theory, N ∈ L(L2 (Γ1 ), H 2 (Ω) ⊂ D(A 4
on Ω on Γ on Γ0
(2)
)) ∀ǫ > 0
Laplacian in the interior A : L2 (Ω) ⊃ D(A) → L2 (Ω) Let the operator A : L2 (Ω) ⊃ D(A) → L2 (Ω) be defined by: 2
A(u − Nδη u) = −∆u, D(A) = {z ∈ H (Ω), z|Γ = 0, δη z|Γ = 0} 0
1
(3)
Laplace-Beltrami on the boundary B : L2 (Γ1 ) ⊃ D(B) → L2 (Γ1 )
2
2
Bz = −∆Γ z, D(B) = {z ∈ H (Γ1 ), z|δΓ = 0} = H0 (Γ1 )
(4)
1
1 R with the weighted norm kzk 1 = kΓ kB 2 zk 2 . = kΓ Γ |∇Γ z|2 = kΓ kzk 1 L (Γ1 ) H (Γ1 ) 1 0
B2
Nicolas Fourrier (U.V.A.)
P.D.E. Seminar
September 11, 2012
7 / 42
Introduction
Preliminary definitions
Energy space
Let (u, ut , u|Γ1 , u|Γ1 ) = (u1 , u2 , u3 , u4 ). The associated energy space is: 1
1
H = D(A 2 ) × L2 (Ω) × D(B 2 ) × L2 (Γ1 )
(5)
with associated norm: kukH = |∇u1 |2Ω + |u2 |2Ω + kΓ |∇Γ u3 |2Γ1 + |u4 |2Γ1 .
Nicolas Fourrier (U.V.A.)
P.D.E. Seminar
September 11, 2012
8 / 42
Introduction
Preliminary definitions
Operator A With the definitions of A, B and N, we set: 0 I 0 0 −A(I − Nδη ) −cΩ A(I − Nδη ) 0 0 A= 0 0 0 I −δη −cΩ δη −kΓ B −qB
1
1
1
(6)
1
D(A) ={[u1 , u2 , u3 , u4 ]T ∈ D(A 2 ) × D(A 2 ) × D(B 2 ) × D(B 2 ), such that (I − Nδη )u1 + cΩ (I + Nδη )u2 ∈ D(A), kΓ u3 + qu4 ∈ D(B), u1 |Γ1 = N ∗ Au1 = u3 } Nicolas Fourrier (U.V.A.)
P.D.E. Seminar
September 11, 2012
9 / 42
Theorems
Overview
1
Introduction
2
Theorems Strong damping on the boundary No damping on the boundary Intermediate theorem
3
Proof of the theorems
4
Numerical Calculations
Nicolas Fourrier (U.V.A.)
P.D.E. Seminar
September 11, 2012
10 / 42
Theorems
Strong damping on the boundary
Having a strong damping on the boundary is equivalent to α = 1 in (1)
Theorem I With A defined in (6), generates an analytic C0 -semigroup {eAt }t≥0 on H. Also, {eAt }t≥0 is exponentially stable, i.e., ∃C, ω ≥ 0 such that keAt uk2H ≤ Ce−ωtkukH
Nicolas Fourrier (U.V.A.)
P.D.E. Seminar
September 11, 2012
11 / 42
Theorems
No damping on the boundary
No damping on the boundary is equivalent to α = 0 in (1)
Theorem II Aα=0 given by A in (6), generates a C0 -semigroup of contractions {eAα=0 t }t≥0 on H. Also, {eAα=0 t }t≥0 is strongly stable, i.e., ∀u ∈ H lim , eAα=0 t u → 0 t→∞
Nicolas Fourrier (U.V.A.)
P.D.E. Seminar
September 11, 2012
12 / 42
Theorems
Intermediate theorem
The following theorem will provide the necessary tool to obtain the stability results from Theorem I and Theorem II.
Theorem III With A defined in (6), σ(A) ∩ iR = ∅
Nicolas Fourrier (U.V.A.)
P.D.E. Seminar
September 11, 2012
13 / 42
Proof of the theorems
Overview
1
Introduction
2
Theorems
3
Proof of the theorems Theorem III Theorem I Theorem II
4
Numerical Calculations
Nicolas Fourrier (U.V.A.)
P.D.E. Seminar
September 11, 2012
14 / 42
Proof of the theorems
Theorem III
Statement
Theorem III With A defined in (6), σ(A) ∩ iR = ∅ Strategy This problem will give the stability results for both theorems. We will proceed in 3 steps: σc (A) ∩ iR = ∅ σp (A) ∩ iR = ∅ σr (A) ∩ iR = ∅
Nicolas Fourrier (U.V.A.)
P.D.E. Seminar
September 11, 2012
15 / 42
Proof of the theorems
Theorem III
Adjoint of A
0 −I 0 0 A(I − Nδη ) −cΩ A(I − Nδη ) 0 0 A∗ = 0 0 0 −I δη −cΩ δη kΓ B −qB 1
1
1
(7)
1
D(A∗ ) ={[u1 , u2 , u3 , u4 ]T ∈ D(A 2 ) × D(A 2 ) × D(B 2 ) × D(B 2 ), such that (I − Nδη )u1 − cΩ (I + Nδη )u2 ∈ D(A), u3 − qu4 ∈ D(B), u1 |Γ1 = N ∗ Au1 = u3 and u2 |Γ1 = N ∗ Au2 = u4 }
Nicolas Fourrier (U.V.A.)
P.D.E. Seminar
September 11, 2012
16 / 42
Proof of the theorems
Theorem III
Step 1: σc (A) ∩ iR = ∅ Theorem If λ ∈ σc (A), then A − λ does not have a closed range. Assumption With A as given in (6), assume that λ = ir ∈ σc (A) with r 6= 0. ∀f = [f1 , f2 , f3 , f4 ]T ∈ H, suppose that u = [u1, u2, u3, u4] ∈ D(A) such that: (ir − A)u = f
(8)
Recall that: ∗
u1 |Γ = N Au1 = u3 1 Then, this leads to a single equation: 2
2
∗
∗
∗
− r u1 + (1 + cΩ ir )Au1 − AN(r N Au1 − kΓ BN Au1 − qirBN Au1 ) = irf1 + cΩ Af1 + f2 + AN(f4 + irf3 + qBf3 )
Nicolas Fourrier (U.V.A.)
P.D.E. Seminar
September 11, 2012
(9)
17 / 42
Proof of the theorems
1
Theorem III
1
Let V = D(A 2 ) ∩ D(B 2 ) with associated norm: 1
1
2
2
kukV = |A 2 u|Ω + |B 2 u|Γ
(10)
1
Consider the left hand side of (9) and define the operators T , M, K : V → V ∗ , by T M K 1
1
1
=M+K = A + kΓ ANBN ∗ A + ir (cΩ A + qANBN ∗ A) = −r 2 (I + ANN ∗ A)
(11)
1
Define F : D(A 2 ) × D(A 2 ) × D(B 2 ) × D(B 2 ) → V ∗ by: F = (ir + cΩ A)
Nicolas Fourrier (U.V.A.)
I
AN(ir + qB)
P.D.E. Seminar
AN
(12)
September 11, 2012
18 / 42
Proof of the theorems
Theorem III
Compactness of K 1
By definition of A, N and N ∗ A, K is compact from D(A 2 ) into its dual.
M is boundedly invertible Given M, u ∈ V and v ∈ V ∗ , define its bilinear form M(., .), by:
M(u, v ) = (1 + cΩ ir )
1 1 A 2 u, A 2 v
Ω
+ (kΓ + qir )
1
1
B 2 u, B 2 v
Γ1
(13)
M(., .) is a coercive and bounded bilinear form. Therefore, by Lax-Milgram M is bounded invertible.
Nicolas Fourrier (U.V.A.)
P.D.E. Seminar
September 11, 2012
19 / 42
Proof of the theorems
Theorem III
T is invertible By the Fredholm’s alternative, since we can deduce the desired invertibility of T provided that T is injective. Suppose that Tu1 = 0, assume that r 6= 0 and take the duality pairing with respect to u1 ∈ V ∗ , then: 0 = ((1 + cΩ ir )Au1 , v )
2
1 1 [D(A 2 )]∗ ×D(A 2 )
∗
− r (I + ANN Au1 , v ∗
1 1 [D(A 2 )]∗ ×D(A 2 )
+ (kΓ + qir )ANBN Au1 , v
(14)
1 1 [D(A 2 )]∗ ×D(A 2 )
1 1 ∗ 1 1 ∗ 2 2 2 2 2 = (1 − r )|A 2 u1 |Ω + kΓ |B 2 N Au1 |Γ + ir cΩ |A 2 u1 |Ω + q|B 2 N Au1 |Γ 1 1 Taking the imaginary part in (14) implies u1 = 0. Thus T is injective.
Nicolas Fourrier (U.V.A.)
P.D.E. Seminar
September 11, 2012
20 / 42
Proof of the theorems
Theorem III
End of the proof σp (A) ∩ iR = ∅ With A as given in (6), if for r ∈ R and r 6= 0, there exists u = [u1, u2, u3, u4]T ∈ D(A) such that: u1 u1 u 2 u 2 A u3 = ir u3 u4 u4
(15)
This turns to be equivalent to solve: Tu1 = 0. σr (A) ∩ iR = ∅ First, recall the theorem: If the eigenvalue λ ∈ C is in the residual spectrum of A, then λ ∈ σp (A∗ ) With A∗ as given in (7), if for r ∈ R and r 6= 0, there exists u = [u1, u2, u3, u4]T ∈ D(A∗ ) such that: u1 u1 u 2 u2 ∗ A = ir u 3 u3 u4 u4
(16)
This is also equivalent to solve: Tu1 = 0.
Nicolas Fourrier (U.V.A.)
P.D.E. Seminar
September 11, 2012
21 / 42
Proof of the theorems
Theorem I
Statement
Theorem I With A defined in (6), generates an analytic C0 -semigroup {eAt }t ≥0 on H. Also, {eAt }t ≥0 is exponentially stable, i.e., ∃C, ω ≥ 0 such that ke
At
2
ukH ≤ Ce−ωtkukH
Strategy Introducing a new variable z = u + cΩ u t
(17)
we will to transform the system into the heat equation with General Wentzell Boundary Conditions (GWBC) with a perturbation.
Nicolas Fourrier (U.V.A.)
P.D.E. Seminar
September 11, 2012
22 / 42
Proof of the theorems
Theorem I
Heat Equation with GWBC Using z = u + cΩ ut , we obtain:
(
zt ∆z + δη z − kΓ ∆Γ z
= cΩ ∆z + cz − cu Ω Ω
(18)
=0
Theorem by Favini Consider the heat equation with GWBC: zt = ∆z in Ω
(19)
∆z + βδη z + γz − ζβ∆Γ = 0 on δΩ
(20)
Then the closure G2 of ∆ in X2 with domain 2
¯ z|δΩ satisfies (20)} D(G2 ) = {Z = (z|Ω , z|δΩ ) ∈ H (Ω), The operator G2 is self-adjoint and generates a cosine function and a quasicontraction (contraction, if γ ≥ 0) semigroup on X2 which is analytic in the right half plane.
Nicolas Fourrier (U.V.A.)
P.D.E. Seminar
September 11, 2012
23 / 42
Proof of the theorems
Theorem I
Perturbation Theory
Let U = (z, u|Ω , u|Γ )T and set β = cΩ , ζ = kΓ , then one can rewrite (17) and (18) as follows: 1 u| z|Ω − 2Ω G 2 cΩ z c2 c zt u|Ω Ω Ω − c z|Ω U t = u t |Ω = Ω + c Ω u|Γ u t |Γ z|Γ 1 − c 1 1
Ω
(21)
cΩ
= (B + P)U
Note that the second equation is simply an ordinary differential equation. We can let u to be defined on H 1 (Ω). ¯ × H 1 (Ω)} is analytic on X2 × H 1 (Ω). Then A with domain D(A) = {H 2 (Ω) It is enough to show that B is relatively A-bounded. z|
u|
But | 2Ω − 2Ω |B ≤ C(kzkH 1 (Ω) + kukH 1 (Ω) ) ≤ C(kzkX + kukH 1 (Ω) ) = CkU kA 2 c c Ω Ω Therefore B is relatively A-bounded and thus analytic.
Nicolas Fourrier (U.V.A.)
P.D.E. Seminar
September 11, 2012
24 / 42
Proof of the theorems
Theorem I
Perturbation Theorem Let the operator (A, D(A)) generate an analytic semigroup T (z) on a Banach space X. Then there exists a constant α such that (A + B, D(A + B)) generates an analytic semigroup for every A-bounded operator B having A-bound α0 < α. Conclusion Then A generates an analytic semigroup {eAt }t ≥0 on H. Note that the proof also holds for kΓ = 0. Since σ(A) ∩ iR = ∅ by Theorem III, then {eAt }t ≥0 is exponentially stable by the stability theorem which achieves the proof theorem Theorem I.
Nicolas Fourrier (U.V.A.)
P.D.E. Seminar
September 11, 2012
25 / 42
Proof of the theorems
Theorem II
Statement
Theorem II Aα=0 given by A in (6), generates a C0 -semigroup of contractions {eAα=0 t }t ≥0 on H. Also, {eAα=0 t }t ≥0 is strongly stable, i.e., A t ∀u ∈ H lim , e α=0 u → 0 t →∞
Strategy Using Lumer-Phillips theorem, we need to show that Aα=0 is maximal dissipative.
Nicolas Fourrier (U.V.A.)
P.D.E. Seminar
September 11, 2012
26 / 42
Proof of the theorems
Theorem II
Dissipativity
∀u ∈ D(Aα=0 ), take the inner product:
< Aα=0 U , V > = (u2 , u1 )H 1 (Ω) + (∆u1 + cΩ ∆u2 , u2 )Ω + (u4 , u3 )H 1 (Γ ,k dS) + −δη (u1 + cΩ u2 ) + kΓ ∆Γ u3 , u4 Γ 1 1 Ω
= (∇u2 , ∇u1 )Ω − (∇u1 + cΩ ∇u2 , ∇u2 )Ω + δη (u1 + cΩ u2 ), u2 |δΩ Γ 1
+ kΓ h∇Γ u4 , ∇Γ u3 iΓ − δη (u1 + cΩ u2 ), u4 Γ − hkΓ ∇Γ u3 , ∇Γ u4 iΓ 1
1
1
2
= −cΩ |∇u2 |Ω < 0
(22)
Nicolas Fourrier (U.V.A.)
P.D.E. Seminar
September 11, 2012
27 / 42
Proof of the theorems
Theorem II
Range Condition I
if λ ∈ C, Reλ > 0 and F = f1
f2
f3
f4 T ∈ H is given, then the elliptic equation: f1 u1 u2 = f2 (λ − Aα=0 ) f3 u3 f4 u4
is satisfied for some u1
u2
u3
(23)
u4 T ∈ D(Aα=0 )
Then the strategy is to create a bilinear form in order to eventually use Lax-Milgram. After multiplying by V = (v1 , v2 , v3 , v4 ) ∈ H, integrating over Ω and simplifying, we get: D E 2 2 + (∇u1 , ∇v1 )Ω + (cΩ λ∇u1 , ∇v1 )Ω + λ u3 , v3 λ u 1 , v1 Ω
Γ1
+ hkΓ ∇Γ u3 , ∇Γ v3 iΓ
1
(24)
= (f2 + λf1 , v1 )Ω + (cΩ λ∇f1 , ∇v1 )Ω + hf4 + λf3 , v3 iΓ 1
Let L(U , V ) be the left hand side and F (V ) be the right hand side.
Nicolas Fourrier (U.V.A.)
P.D.E. Seminar
September 11, 2012
28 / 42
Proof of the theorems
Theorem II
Range Condition II
Lax-Milgram conditions
2
|L(U , V )| ≤ max {λ + cΩ λ + kΓ , 1}kU kH kV kH |F (V )| ≤ max {λ, 1}kF kH kV kH ReL(U , U ) =
D E 2 2 + (∇u1 , ∇u1 )Ω + (cΩ λ∇u1 , ∇u1 )Ω + λ u3 , u3 λ u1 , u1 Ω
(25)
Γ1
+ hkΓ ∇Γ u3 , ∇Γ u3 iΓ
1
(26)
2
2 2 2 2 2 ≥ Reλ |u1 |Ω + |∇u1 |Ω + Reλ |u3 |Γ + |kΓ ∇Γ u3 |Γ 1 1
Conclusion L(U , V ) defined a bounded (25), coercive (26) bilinear form. So Aα=0 generates a semigroup of contraction {eAt }t ≥0 . By the stability theorem from Arendt-Batty, Theorem III shows the strong stability for {eAt }t ≥0 .
Nicolas Fourrier (U.V.A.)
P.D.E. Seminar
September 11, 2012
29 / 42
Numerical Calculations
Overview
1
Introduction
2
Theorems
3
Proof of the theorems
4
Numerical Calculations Spectrum of a classical strong damped wave equation Spectrums of the strong damped equation with Acoustic Boundary Conditi Solution’s behavior
Nicolas Fourrier (U.V.A.)
P.D.E. Seminar
September 11, 2012
30 / 42
Numerical Calculations
Spectrum of a classical strong damped wave equation
Spectral result In the paper, Proof of extensions of two conjecteures o structural damping for elastic systems, Pacific Journal of Mathematics - S. Chen and R. Triggiani, the eigenvalues for a strong damped wave equation are given: q +,− −1 α 2 λn = µn −cΩ ± cΩ − µn (27) ∞ where {µ}n=1 are the eigenvalues of ∆in 2D. The following picture shows the first eigenvalues λ+,− n .
Nicolas Fourrier (U.V.A.)
P.D.E. Seminar
September 11, 2012
31 / 42
Numerical Calculations
Spectrum of a classical strong damped wave equation
Imaginary Axis
3i
Real Axis −24
Nicolas Fourrier (U.V.A.)
P.D.E. Seminar
0
September 11, 2012
32 / 42
Numerical Calculations
21
22
11 16
23
12 17
7 11
13
8
14
9
1
10 15
6
8
2 2
20
14
Γ0 7
25
19
13
5
1
24
18
12
6
Spectrums of the strong damped equation with Acoustic Boundary
9
3 3
10
4 4
5
Γ
Nicolas Fourrier (U.V.A.)
P.D.E. Seminar
September 11, 2012
33 / 42
Numerical Calculations
Spectrums of the strong damped equation with Acoustic Boundary
In the next two pictures, we will observe that the equation (1) with α = 1, i.e., with a damping on the boundary presents a very similar spectrum to the previous case. However, when there is no more damping on the boundary (α = 0), this leads to a new configuration.
Nicolas Fourrier (U.V.A.)
P.D.E. Seminar
September 11, 2012
34 / 42
Numerical Calculations
Spectrums of the strong damped equation with Acoustic Boundary
Imaginary Axis
cΩ =0.15 cΓ =1 kΩ =1 kΓ =1
4i
α =1
Real Axis −31
utt − cΩ ∆ut − ∆u = 0 ut +u) − kΓ ∆Γ (cΩ αut + u) = 0 u + δ(cΩδη tt u=0
Nicolas Fourrier (U.V.A.)
0
on Ω on Γ1 on Γ0
P.D.E. Seminar
September 11, 2012
35 / 42
Numerical Calculations
Spectrums of the strong damped equation with Acoustic Boundary
Imaginary Axis
cΩ =0.15 cΓ =1 kΩ =1 kΓ =1
11i
α =0
Real Axis −30
utt − cΩ ∆ut − ∆u = 0 ut +u) − kΓ ∆Γ (cΩ αut + u) = 0 u + δ(cΩδη tt u=0
Nicolas Fourrier (U.V.A.)
0
on Ω on Γ1 on Γ0
P.D.E. Seminar
September 11, 2012
36 / 42
Numerical Calculations
Solution’s behavior
Model utt − ∆u − cΩ ∆ut = 0 utt − δη (u + cΩ ut ) − kΓ ∆Γ (cΩ αut + u) = 0 u=0 Initial conditions u(0, x) = u0 , ut (0, x) = u1
in Ω in Γ1 in Γ0
(28)
on Ω
Solution with α = 1 Solution with α = 0 Comparaison between these two cases. Comparaison of the L2 and H 1 norm.
Nicolas Fourrier (U.V.A.)
P.D.E. Seminar
September 11, 2012
37 / 42
Numerical Calculations
Nicolas Fourrier (U.V.A.)
Solution’s behavior
P.D.E. Seminar
September 11, 2012
38 / 42
Numerical Calculations
Nicolas Fourrier (U.V.A.)
Solution’s behavior
P.D.E. Seminar
September 11, 2012
39 / 42
Numerical Calculations
Nicolas Fourrier (U.V.A.)
Solution’s behavior
P.D.E. Seminar
September 11, 2012
40 / 42
Numerical Calculations
Nicolas Fourrier (U.V.A.)
Solution’s behavior
P.D.E. Seminar
−−−− |uα = 1|L2(Ω)
−−−− |uα = 0|L2(Ω)
..... |uα = 1|H1(Ω)
..... |uα = 0|H1(Ω)
September 11, 2012
41 / 42
Numerical Calculations
Solution’s behavior
Lili Rose
Nicolas Fourrier (U.V.A.)
P.D.E. Seminar
September 11, 2012
42 / 42