mathematical properties of pml in time domain
play

Mathematical Properties of PML in time domain Oliver Pfeifer by P. - PowerPoint PPT Presentation

Mathematical Properties of PML in time domain Oliver Pfeifer by P. Joly Overview Some properties of hyperbolic systems of first order PDE Well-posedness of the PML model Stability of the PML-method Cauchy problem in R 2 Lets have a


  1. Mathematical Properties of PML in time domain Oliver Pfeifer by P. Joly

  2. Overview • Some properties of hyperbolic systems of first order PDE • Well-posedness of the PML model • Stability of the PML-method

  3. Cauchy problem in R 2 Lets have a look at a Cauchy problem in R 2  ∂V ∂t + A x∂V ∂x + A y ∂V ∂y + BV = 0 , ( x, y ) ∈ R 2 , t < 0    V ( x, y, 0) = V 0 ( x, y ) , ( x, y ) ∈ R 2   V ∈ R d , ( A x , A y , B ) ∈ L ( R d ) 3  Observe that the PML-System in two dimensions: ∂U x ∂t + σU x + A x ∂ ∂x ( U x + U y ) = 0 ∂U y ∂y ( U x + U y ) = 0 ∂t + A y ∂ is a problem of this type if we set: � U x � 0 � σI � A x � � � � A x 0 0 V = , A x = , A y = , B = U y 0 0 0 0 A y A y

  4. We say that the Cauchy problem is well-posed if for every Definition 1. V 0 in H s , there exists a unique solution V ∈ C 0 ( R + , L 2 ) that satisfies a estimation on the type: ∀ t > 0 , � V ( t ) � L 2 � C ( t ) � V 0 � H s It is called strongly well-posed if s = 0 , otherwise it is called weakly well-posed. When is the above Cauchy problem a well-posed problem? → ˆ Applying the Fourier transform in space: V ( x, y, t ) − V ( k x , k y , t ) ∂ ˆ ) + B ) ˆ V ∂t + ( i ( k x A x + k y A y V = 0 � �� � A ( k ) V ( k x , k y , 0) = ˆ ˆ V 0 ( k x , k y )

  5. The solution to this problem is: V ( k x , k y , t ) = e i ( A ( k ) − iB ) t ˆ ˆ V 0 If we want an estimation like in the definition, we have to estimate the exponential term. In practice, this is the same as finding particular solution in forms of plane waves: V ( x, y, t ) = V ( k ) e i ( k x x + k y y ) e iω ( k ) t , k = ( k x , k y ) ∈ R 2 , ω ( k ) ∈ C This leads us to the dispersion relation: det( A ( k ) − iB − ωI ) = 0 We denote by { ω ( k ) } the set of all branch of solutions.

  6. For the problem to be well-posed it is necessary that: Im ω ( k ) is bounded below, ∀ k ∈ R 2 This is because of: e iω ( k ) t = e iRe ( ω ( k )) t e − Im ( ω ( k )) t Geometrically this can be interpreted as: k All the complex curves | k | → ω ( | k | · K ) , K = | k | stay above a half plane. Because the functions ω ( k ) are continuous, we only have to look at what happens if | k | goes to ∞ . The original problem can be interpreted as a perturbation of a homogenous system, where B = 0 . So we first have a look at that.

  7. The unperturbed system Definition 2. The unperturbed system is called hyperbolic if ∀ k ∈ R 2 , the eigenvalues of A ( k ) are real. It is called strongly hyperbolic (otherwise weakly hyperbolic) if ∀ k ∈ R 2 , A ( k ) can be diagonalised. The dispersion relation det( A ( k ) − ωI ) = 0 becomes the characteristic equation of the Matrix A ( k ) : ω has to be an eigenvalue of A ( k ) . The solutions ω ( k ) are homogenous functions of order 1 in k, so in this case the curves are straight lines. A ( k ) is real and therefore the eigenvalues are paired by complex conjugation. So the curves are also paired by symmetry to the reel axis.

  8. We have a well-posed problem, if all curves are on the real axes, which means that the system is hyperbolic. In fact there is a more precise result: (Kreiss) In the case B = 0 , the problem is well-posed if and Theorem 1. only if the system is hyperbolic, and: • strongly hyperbolic ⇒ strongly well-posed and ∀ t > 0 , � V ( t ) � L 2 ≤ C � V 0 � L 2 • weakly hyperbolic ⇒ weakly well-posed and ∀ t > 0 , � V ( t ) � L 2 ≤ C (1 + t ) s � V 0 � H s , s ≥ 1

  9. The perturbed system ( B � = 0) If the unperturbed system is strongly hyperbolic, then the Theorem 2. problem is strongly well-posed and it exists a constant K > 0 such that: ∀ t > 0 , � V ( t ) � L 2 ≤ Ce Kt � V 0 � L 2 If the unperturbed system is only weakly hyperbolic, then for certain matrices B the problem is ill-posed. The notation of well-posedness guarantees a unique solution, but it does not exclude exponential growth in time. Therefore this isn’t sufficient for a PML, witch should be absorbing.

  10. We suppose that the Cauchy problem is well-posed. Definition 3. Then the system is called strongly stable if the solution holds: ∀ t > 0 , � V ( t ) � L 2 ≤ C � V 0 � L 2 or weakly stable if it holds: � V ( t ) � L 2 ≤ C (1 + t ) s � V 0 � H s , s ≥ 1 ∀ t > 0 , The Cauchy problem is stable if and only if ∀ k ∈ R 2 the solutions ω ( k ) satisfy Imω ( k ) ≥ 0 . The existence of solutions ω with negative imaginary parts would correspond to plane wave solutions with exponential growth in time. A stable system does not admit such solutions.

  11. Example in one dimension � u � u � u � � � ∂ + A ∂ v v + B v = 0 ∂t ∂x w w w � 0 � 0 0 where the matrix A is given by: A = 0 2 1 0 − 1 0 The eigenvalues of A are 0 (simple) and 1 (double), and it is not diagonalisable. Consider successively ( a ∈ R ) : � 0 � 0 � � − 1 0 − 1 0 B = B 1 = 0 0 a , B = B 2 = a 0 0 0 0 0 0 0 0

  12. With B = B 1 we obtain the following equation for u if we eliminate v and w � ∂ 2 u ∂t 2 + 2 ∂ 2 u ∂x∂t + ∂ 2 u � ∂ ∂x 2 + a∂u = 0 ∂t ∂x iω ( − ω 2 − 2 kω − k 2 − iak ) = 0 , The dispersion relation is and the solutions are: 1 2 1+ i ω = 0 , ω = − k ± | ak | √ 2 One of them has a imaginary part that goes to −∞ when | k | goes to + ∞ . So in this case the problem is ill-posed.

  13. With B = B 2 we obtain a slightly different equation for u � ∂ 2 u ∂t 2 + 2 ∂ 2 u ∂x∂t + ∂ 2 u ∂ � ∂x 2 + au = 0 ∂t iω ( − ω 2 − 2 kω − k 2 + a ) = 0 , Now the dispersion relation is and the solutions are: ω = 0 , ω = − k ± √ a There imaginary parts are uniformly bounded and therefore the problem is well-posed. But we see also, that the system is stable if a ≥ 0 and instable if a < 0 .

  14. Well-posedness of the PML model First, have a look at the PML system for acoustic waves:  ρ ( ∂u x ∂t + σu x ) − ∂v x ∂x = 0      ∂x ( u x + u y ) = 0  µ − 1 ( ∂v x ∂t + σv x ) − ∂   ρ ∂u y ∂t − ∂v y ∂y = 0       ∂y ( u x + u y ) = 0 µ − 1 ∂v y  ∂t − ∂  If we set ρ = µ = 1 and σ = const. , we can simplify: � 2 ( ∂ 2 u � ∂ ∂t 2 − ∂ 2 u ∂ 4 u ∂t + σ ∂y 2 ) − ∂x 2 ∂t 2 = 0

  15. From this we get the dispersion relation: det( A ( k ) − iB − ωI ) = ( iω + σ ) 2 ( ω 2 − k 2 y ) + ω 2 k 2 x = 0 ω 2 ( ω 2 − k 2 y − k 2 And the one for the unperturbed system: x ) = 0 ω = 0 is a double eigenvalue of A ( k ) and it is not diagonalisable. ⇒ not strongly hyperbolic To show that the system stays well-posed anyway, we have to look at the development of the four solutions ω ( k ) for great values of | k | :  x / | k | 2 + O ( | k | − 1 ) , ω ( k ) = ± | k | + iσk 2   ± k x k y + ik 2 + O ( | k | − 1 ) , y  ω ( k ) = σ  | k | 2 The imaginary parts are bounded regardless of the sign of σ . One can show that: σ > 0 ⇒ 0 ≤ Im ω ( k ) ≤ σ, ∀ k ∈ R 2

  16. The Cauchy problem associated to the PML system for Theorem 3. acoustic waves with constant coefficients is weakly well-posed. More 0 , u y 0 , v x, 0 , v y, 0 ) in L 2 ( R 2 ) there precisely, for every initial data V 0 = ( u x exists a unique solution: ( u = u x + u y , v x , v y ) ∈ C 0 ( R + , L 2 ( R 2 )) , e = u x − u y ∈ C 0 ( R + , H − 1 ( R 2 )) , and if σ > 0 , we have the estimations: � � u ( ., t ) � L 2 + � v x ( ., t ) � L 2 + � v y ( ., t ) � L 2 ≤ C � V 0 � L 2 � e ( ., t ) � H − 1 ≤ C (1 + t ) � V 0 � L 2 If σ < 0 there are similar estimations with Ce | σ | t in place of C . The loss of regularity concerns only the ”not physical” value e.

  17. Energy estimation For the unperturbed system:  ∂x − ∂v y ρ ∂u ∂t − ∂v x ∂y = 0     µ − 1 ∂v x ∂t − ∂u ∂t = 0    µ − 1 ∂v y ∂t − ∂u  ∂t = 0 we have conservation of the energy: � � ρ | u | 2 + µ − 1 | v | 2 dx � 1 d = 0 2 dt Is there also a energy estimation for the PML system?

  18. The formulation of Zhao-Cangellaris We have to make a change of the unknown functions: ( u x , u y , v x , v y ) − → ( u, v x , v y , v ∗ y ) ∂v ∗ where the new functions are given by: u = u x + u y , ∂t = ∂v y y ∂t + σv y From the first and the third equations we get: � � ∂t + σu x � � ∂u x ∂ − ∂v x ρ = 0 ∂t ∂x � � � � ∂ ρ ∂u y ∂t − ∂v y ∂t + σ = 0 ∂y After summation, it becomes � ∂ � ∂ � � � � ∂ u − ∂v x − ∂ ρ ∂t + σ ∂t + σ v y = 0 ∂t ∂x ∂y

  19. Finally we get: ∂v ∗ � ∂ � u − ∂v x y ρ ∂t + σ ∂x − ∂y = 0 together with the second and fourth equation and the one for v ∗ y , we have the new system:  ∂v ∗ � ∂u � − ∂v x y ρ ∂t + σu ∂x − ∂y = 0       µ − 1 ( ∂v x ∂t + σv x ) − ∂u  ∂x = 0  µ − 1 ∂v y ∂t − ∂u  ∂y = 0      ∂v ∗  ∂t = ∂v y  y ∂t + σv y

Recommend


More recommend