Sie sind auf Seite 1von 22

Tasa de la Pconvergencia de operadores resolventes discretizados via el m´etodo de elementos ﬁnitos

Rodiak Nicolai Figueroa L´opez y German Jesus Lozada Cruz

Departamento de Matem´aticas IBILCE/UNESP - S˜ao Jos´e do Rio Preto - Brasil

VII Fast Workshop on Applied and Computational Mathematics 8 de Enero del 2014

Rodiak Figueroa L´opez Tasa de la Pconvergencia de operadores resolventes

The Parabolic Problem

We consider the problem

  u t = Lu, u(t, x) = 0,

 

t > 0,

x

t > 0,

x

u(0, x) = a(x),

where a H

1 (Ω), Ω R n is a bounded domain with smooth

0

(1)

boundary Ω, n N, L is second order elliptic operator given by

Lu =

n

i,j=1

x i a ij (x)

u

x j +

n

j=1

b j (x) x u j + (c(x) + λ)u,

(2)

with coeﬃcients a ij , b j , c : Ω R smooth, λ R.

Rodiak Figueroa L´opez Tasa de la Pconvergencia de operadores resolventes

We assume that L is uniformly strongly elliptic operator, that is, there is a constant ϑ > 0 such that

n

i,j=1 a ij (x)ξ i ξ j ϑ

n

k=1

ξ

k 2 ,

for all x Ω, ξ = (ξ 1 ,

Let X = L 2 (Ω) be and deﬁne the sectorial linear operator

A : D (A) X X

We assume that λ is chosen such that Re σ(A) > 0. Then,

X 1 = D(A), X 0 = L 2 (Ω) and X 1/2 = H

ξ n ) R n with a ij = a ji , i, j = 1,

,

by Au = Lu and D (A) = H 2 (Ω) H

1

0

1 (Ω).

0

,

n.

(Ω).

With this notation, the problem (1) can be written in the abstract form

u˙ + Au = 0 u(0) = a X 1/2 .

(AP 0 )

Rodiak Figueroa L´opez Tasa de la Pconvergencia de operadores resolventes

We can associate a sesquilinear form σ : X 1/2 × X 1/2 C such that

σ(u, v ) = Au, v X ,

u D(A),

v X 1/2 ,

|σ(u, v)| Re σ(u, u)

c 1 u X 1/2 v X 1/2 ,

u, v X 1/2

2

c 2 u X 1/2 ,

u X 1/2 ,

(3)

(4)

(5)

where the constants c 1 , c 2 are positives. Also, there are positive constants θ 1 < π/2 and M 1 with the following properties:

S 0,θ 1 = {z C : θ 1 ≤ | arg(z)| ≤ π, z

= 0} ⊂ ρ(A)

(z A) 1 L (X) M |z| 1 ,

z ∈ S 0,θ 1 .

and

(6)

Furthermore, the solution of the initial value problem (AP 0 ) is given by u(t ) = e tA a, t 0, where {e tA : t 0} is the analytic semigroup in X 1/2 generated by A (see [1, Page 35]). Also, from Theorem 1.7.7 in , we have

where Γ 1
e −tA =
e −zt (z − A) −1 dz,
(7)
2πi Γ 1
Rodiak Figueroa L´opez
Tasa de la P−convergencia de operadores resolventes
−iθ
is the positively oriented boundary running from ∞e
1

Discretization of parabolic problems via FEM

Deﬁnition

Let R n be a given domain and let {T h } h(0,1] be a family of subdivisions such that

max{diam(T ) : T ∈ T h } ≤ h diam(Ω).

(8)

The family {T h } h(0,1] is said to be quasi-uniform if there exists ρ > 0 such that

min{diam(B T ) : T ∈ T h } ≥ ρh diam(Ω), h (0, 1],

(9)

where B T is the largest ball contained in T such that T is star-shaped with respect to B T .

Rodiak Figueroa L´opez Tasa de la Pconvergencia de operadores resolventes Figura : Example of a T ∈ T h star-shaped with respect to ball B T in R 2 .

Rodiak Figueroa L´opez Tasa de la Pconvergencia de operadores resolventes

We describe the discretization of the domain Ω R n with n 2, using the ﬁnite element method. Thus, we will be able to discretize the problem (AP 0 ) and study the limitations given in  using the theory made in  for its generalization to n 2. We assume the following hypothesis about the domain Ω. Assumption
Let Ω ⊂ R n be a polyhedral domain with n ≥ 2 which has
{T h } h∈(0,1] a quasi-uniform family of subdivisions with positive
constant ρ (see Deﬁnition 0.1). Let (K , P, N ) be a reference
element of class C 0 , satisfying
(i)
K is star-shaped with respect to some ball,
(ii)
P 1 ⊆
P ⊆ W 2,∞ (K )
and
+
(iii) N ⊆ C (K) , ∈ Z
0 .

Rodiak Figueroa L´opez Tasa de la Pconvergencia de operadores resolventes

Theorem

Let {T h } h(0,1] be a quasi-uniform family of subdivisions of a

polyhedral domain R n and (K , P, N ) the reference element of

class C 0 , satisfying

(i) K is star-shaped with respect a some ball,

 (ii) P m−1 ⊆ P ⊆ W m,∞ (K ) and (iii) N ⊆ C (K) .

Suposse

m n 0 when p = 1. For all T ∈ T h , h (0, 1], let (T, P T , N T ) be the aﬁn-equivalence element. Then, there exist a positive constant C depending of the reference element, n, m, p and ρ in (9) such that for 0 m,

1 p ≤ ∞ and or m n/p > 0 when p > 1 or

υ − I h υ W ,p () Ch m |υ| W m,p () , υ W m,p (Ω).

(10)

Rodiak Figueroa L´opez Tasa de la Pconvergencia de operadores resolventes

Theorem Let {T h } h∈(0,1] be a quasi-uniform family of subdivisions of a
polyhedral domain Ω ⊆ R n . Let (K , P, N ) be a reference ﬁnite
element of class C 0 such that P ⊆ W l,p (K) ∩ W m,q (K ) where
1 ≤ p, q ≤ ∞ and 0 ≤ m ≤ l. For T ∈ T h , let (T , P T , N T ) be
the aﬃne-equivalent element, and
V h = {v : v is measurable and v |T ∈ P T , ∀T ∈ P T }. Then there
exists a constant C = C (l, p, q, ρ) such that
n n
v W l,p (Ω) ≤ Ch m−l+min(0,
p −
q
(11)
) v W m,q (Ω)
for all v ∈ V h .

Rodiak Figueroa L´opez Tasa de la Pconvergencia de operadores resolventes

With this assumption, we can deﬁne the space

X

1/2

h

:= {I h υ : υ C (Ω), υ = 0} ⊂ X 1/2 C (Ω),

which has ﬁnite dimension and I h is the interpolation operator. (See [1, Page 108])

We can see that in the Assumption 0.2 we still do not know the value of . This value, , comes from the inequality m n/p > 0 with p [1, ] for v W m,p (Ω) from Theorem 4.4.20 in . In our case, m = 2 = p and n 2, then < 1. Thus, = 0.

Rodiak Figueroa L´opez Tasa de la Pconvergencia de operadores resolventes

With the Assumption 0.2 the Theorems 4.4.20 and 4.5.11 in 

holds. Then, there exists positive constants C and

ˆ

C such that

v − I h v W 1,2 ()

v W 1,2 (Ω)

Ch|v | W 2,2 () , v X 1 and

Ch 1 v L 2 () , v X

h

ˆ

1/2

.

(12)

(13)

In this framework, our ﬁnite element approximation

A h : X

1/2

h

1/2

X

h

of A is deﬁned by

A h φ h , ψ h X = σ(φ h , ψ h ),

1/2

φ h , ψ h X

h

.

Thus, the discretization of problem (AP 0 ) can be written as

u˙ h + A h u h = 0, u h (0) = a h

(AP h )

with solution u h (t) = e tA h a h for all t 0.

Rodiak Figueroa L´opez Tasa de la Pconvergencia de operadores resolventes

Deﬁnition

1/2

The orthogonal projection P h : X X inner product of X is deﬁned by

h

with relation to the

P h g, v X = g, v X , v X

 1/2 (14) h .

The following result follows immediately from the above deﬁnition.

Lemma

The ortogonal projection P h satisﬁes

P h f X f X , h (0, 1].

Lemma

 If (4) and (5) holds, then there is a positive constant C such that P h v − v X s ≤ Ch 2−2s |v| H 2 (Ω) , ∀v ∈ X and s = 0, 1/2.  (15) Rodiak Figueroa L´opez

Tasa de la Pconvergencia de operadores resolventes

 Deﬁnition ˜ 1/2 h The map P h : X 1/2 → X deﬁned by ˜ σ( P h v, χ) = σ(v, χ), ∀χ ∈ X 1/2 h . is called elliptic projection in X 1/2 .

Lemma

˜

If (4) and (5) holds, then the map P h is well-deﬁned, that is, for

all v X 1/2 , there is a unique v h X

1/2

h

such that

σ(v v h , χ) = 0, χ X

 1/2 (16) h .

Moreover, there is a constant C > 0 such that

P h v X 1/2 C v X 1/2 , v X 1/2 .

˜

(17)

Rodiak Figueroa L´opez Tasa de la Pconvergencia de operadores resolventes

Theorem

If the assumption 0.2 holds, then there exist a constant C > 0 such that

˜

u P h u X Ch u X 1/2 ,

˜

u P h u X 1/2 Ch u X 1,

˜

u P h u X Ch 2 u X 1,

 u ∈ X 1/2 , (18) u ∈ X 1 , (19) u ∈ X 1 , . (20)

Proof:

using (16), we have

˜

We denote by u˜ h := P h u, uˆ h := I h u and e := u u˜ h , and

(e, ϕ h ) X 1/2 = 0,

1/2

ϕ h X

h

.

(21)

Hence, taking ϕ h = u˜ h uˆ h , using (12) and (21), we obtain

2

e X 1/2 = (e,

e + ϕ h ) X 1/2 = (e, u uˆ h ) X 1/2 Ch e X 1/2 u H 2 () ,

which implies (19). We recall that

e X =

0

sup

=φX |(e, φ) X |
.
(22)
φ X

Rodiak Figueroa L´opez

Tasa de la Pconvergencia de operadores resolventes

From theory of elliptic operators (see ) we kwon that given φ X , there is a ψ X 1/2 , such that

Re σ(v , ψ) = (v , φ) X ,

Moreover, ψ H 2 (Ω) and

v X 1/2 .

ψ H 2 () C φ X .

(23)

(24)

Now, substitute (23) and (24) in (22), we have

e X C

0

sup

=ψX 1

| Re σ(e, ψ)|

ψ H 2 (Ω)

.

(25)

But, again using the Galerkin orthogonality, we obtain

1/2

Re σ(e, ψ) = Re σ(e, ψ)Re σ(e, ψ h ) = Re σ(e, ψψ h ), ψ h X

h

(26)

.

From continuity of σ and (26), it follows

| Re σ(e, ψ)| = | Re σ(e, ψ ψ h )| ≤ c 1 e X 1/2 ψ ψ h X 1/2 .

(27) Rodiak Figueroa L´opez

Tasa de la Pconvergencia de operadores resolventes

Choosing ψ h satisfying the inequality (19), we obtain

ψ ψ h X 1/2

˜

Ch ψ H 2 () .

(28)

Then, from (25), (27) and (28), we have

e X Cc 1

0

sup

=ψX 1

˜

e X 1/2 Ch ψ H 2 ()

ψ H 2 (Ω)

.

Thus,

(29)

Since e X 1/2 u X 1/2 by the deﬁnition of u˜ h and (29) we obtain (18). Finally, replacing (19) in (29) it follows (20).

 ¯ e X ≤ Ch e X 1/2 .

Lemma If a subdivision {T h } h∈(0,1] is quasi-uniform family, then there exist
a positive constant C such that
P h v X 1/2 ≤ C v X 1/2 , ∀ h ∈ (0, 1] and v ∈ X 1/2 .
(30)
Rodiak Figueroa L´opez
Tasa de la P−convergencia de operadores resolventes

Theorem (Theorem 3.1 in )

There exist a positive constant C and an acute angle θ 1 such that

for any f

X and z ∈ S 0,θ 1 we have

 (z − A) −1 f − (z − A h ) −1 P h f X 1/2 ≤ Ch f X , (31) (z − A) −1 f − (z − A h ) −1 P h f X ≤ Ch 2 f X , (32) (z − A) −1 f − (z − A h ) −1 P h f X ≤ Ch|z| −1/2 f X , (33)

where (z A) 1 and (z A h ) 1 are the resolvent operators of A and A h , respectively.

Lemma From Theorem 0.10 and Lemma 0.6, we obtain
(z − A h ) −1 P h f −P h (z − A) −1 f X s ≤ Ch 2−2s f X ,
(34)
for all z ∈ S 0,θ 1 , f
∈ X
and s = 0, 1/2.
Rodiak Figueroa L´opez
Tasa de la P−convergencia de operadores resolventes

Gracias !!!!!!!!!!!!!

Rodiak Figueroa L´opez Tasa de la Pconvergencia de operadores resolventes

BRENNER, S. C.; SCOTT, L. R. The mathematical theory of ﬁnite element methods. New York: Springer-Verlag, 1996.

Brezis, H., Analyse Fonctionnelle: Th´eorie et applications. Collection Math´ematiques appliqu´ees pour la maˆıtrise, (2)

(1987).

Bruschi, Simone M.; Carvalho, Alexandre N.; Ruas-Filho, Jos´e G. The dynamics of a one-dimensional parabolic problem versus the dynamics of its discretization. J. Diﬀerential Equations 168 (2000), n. 1, p. 67–92.

Burenkov, V. I. Sobolev Spaces on Domains. Leipzig:

Teubner-Texte zur Mathematik, volume 137, 1998.

Carvalho, A.N., Piskarev, S., A general approximations scheme for attractors of abstract parabolic problems, Numer. Funct. Anal. Optim. 27 (2006), n. 7-8, p. 785–829.

Rodiak Figueroa L´opez Tasa de la Pconvergencia de operadores resolventes

Cholewa, J. W.; Dlotko, T. Global Attractors in Abstract Parabolic Problems. Cambridge: University Press, 2000.

Friedman, A. Partial Diﬀerential Equations. New York:

Holt, Rinehart and Winston, 1969.

FUJITA, H.; MIZUTANI, A. On the ﬁnite element method for parabolic equations I: approximation of holomorphic semi-groups. Journal of the Mathematical Society of Japan, Tokyo, v. 28, n. 4, p. 749–771, 1976.

Hale, J. K. Asymptotic behavior of dissipative systems, Providence: American Mathematical Society, 1988.

Henry, D. Geometric theory of semilinear parabolic equations, Berlin: Springer-Verlag, 1981.

Kato, T., Perturbation theory for linear operators, Berlin:

Springer-Verlag, 1980.

Rodiak Figueroa L´opez Tasa de la Pconvergencia de operadores resolventes

Ladyzhenskaya, O., Attractors for semigroups and evolution equations. [Lincei Lectures] Cambridge University Press, Cambridge, 1991.

Nolasco de Carvalho, A. Analise Funcional II, Notas de aula, ICMC-USP, S˜ao Carlos, 2011.

Pazy, A., Semigroups of linear operators and applications to partial diﬀerential equations, Applied Mathematical Sciences, Springer-Verlag, New york, 2nd edition, 1992.

Stummel, F., Diskrete Konvergenz linearer Operatoren I. Mathematische Annalen, v. 190, 1970, p. 45–92.

Vainikko, G., Approximative methods for nonlinear equations (Two approaches to the convergence problem). Nonlinear Analysis, Theory, Melhods and Applications. v. 2, n. 6, p. 647–687.

Rodiak Figueroa L´opez Tasa de la Pconvergencia de operadores resolventes

Vainikko G., Multidimensional weakly singular integral equations, Berlin: Springer-Verlag, 1993.

Vainikko G. Funktionalanalysis der Diskretisierungsmethoden, Teubner-Texte zur Mathematik, Verlagsgesellschaft, Leipzig, 1976.

Vainikko, G., Regular convergence of operators and approximate solution of equations. Translated from Itogi Nauki i Tekhniki, Seriya Matematicheskii Analiz, Vol. 16, 1979, pp. 5-53.

Wolf-J¨urgen, B.; Piskarev, S., Shadowing for discrete approximations of abstract parabolic equations. Discrete Contin. Dyn. Syst. Ser. B 10 (2008), n. 1, pp. 19-42.

Zheng, S. Nonlinear evolution equations. Florida: Chapman and Hall/CRC, 2004.

Rodiak Figueroa L´opez Tasa de la Pconvergencia de operadores resolventes