The Weak Galerkin (WG) finite element method for the unsteady Stokes equations in the primary velocity-pressure formulation is introduced in this paper. Optimal-order error estimates are established for the corresponding numerical approximation in an H1 norm for the velocity, and L2 norm for both the velocity and the pressure by us e of the Stokes projection.
The finite element method for the unsteady Stokes equations developed over the last several decades is based on the weak formulation by constructing a pair of finite element spaces satisfying the inf-sup condition of Babuska [
In this paper, we study the initial-boundary value problems of the Stokes.
{ u t − Δ u + ∇ p = f ( x , t ) , t ∈ ( 0, T ) ∇ ⋅ u = 0, ( x , t ) ∈ Ω × ( 0, T ) u = 0, ( x , t ) ∈ ∂ Ω × ( 0, T ) u ( x ,0 ) = ψ ( x ) x ∈ Ω , t = 0 (1)
where u = ( u 1 , u 2 ) T is fluid velocity, p is pressure, f = ( f 1 , f 2 ) T is volumetric power density.
The solution of the Stokes equations forms an important aspect of both theoretical and computational fluid dynamics. A limited number of solutions of these non-linear partial differential equations mostly involving spatially one-dimensional problems are given in the literature. Solutions of practical interest have been obtained for cases where, with suitable approximations, the equations are reduced to linear partial differential equations.
Let W be a bounded domain in R2. We introduce function spaces
X = [ H 0 1 ( Ω ) ] 2 , V = { u ∈ X , d i v u = 0 } , M = { q ∈ L 2 ( Ω ) ; ∫ Ω q d x d y = 0 } , then the unsteady Stokes problem would take the following form: seek ( u , p ) ∈ X × M satisfying
{ ( u t , v ) + ( ∇ u , ∇ v ) − ( p , ∇ ⋅ v ) = ( f , v ) , ∀ v ∈ X ( q , ∇ ⋅ u ) = 0 , ∀ q ∈ W u ( x , 0 ) = ψ ( x ) (2)
We use ‖ ⋅ ‖ s , D and | ⋅ | s , D to be denote the norm and Semi-norm in the Sobolev space H s ( D ) for any s ≥ 0 , respectively. The inner product in H s ( D ) is denoted by ( ⋅ , ⋅ ) s , D . For example, for each s ≥ 0 , the Semi-norm | ⋅ | s , D is given by
| φ | s , D = ( ∑ | γ | = s ∫ D | ∂ γ φ | 2 d D ) 1 2
and ‖ ⋅ ‖ is said to be the norm of L 2 .
For w is [ 0, T ] to H s ( D ) , the definition is given by
‖ w ‖ L q ( 0 , T ; H s ( D ) ) = ( ∫ 0 T ‖ w ( ⋅ , t ) ‖ s , D q d t ) 1 q
for 1 ≤ q ≤ ∞ , we have
‖ w ‖ L ∞ ( 0 , T ; H s ( D ) ) = s u p 0 ≤ t ≤ T ‖ w ( ⋅ , t ) ‖ s , D
The space H s ( D ) and the norm defined in the H s ( D ) defined as
H ( d i v , Ω ) = { q : q ∈ [ L 2 ( Ω ) ] , ∇ ⋅ q ∈ L 2 ( Ω ) }
‖ q ‖ H ( d i v , Ω ) = ( ‖ q ‖ 2 + ‖ ∇ ⋅ q ‖ 2 ) 1 2
Let K be any polygonal or polyhedral domain with boundary ∂ K . A weak vector-valued function on the region K refers to a vector-valued function v = { v 0 , v b } such that v 0 ∈ [ L 2 ( K ) ] d and v b ∈ [ H 1 2 ( ∂ K ) ] d . The first component v 0 can be understood as the value of v in K, and the second component v b represents v on the boundary of K. Note that v b may not necessarily be related to the trace of v 0 on ∂ K should a trace be well-defined. Denote by ν ( K ) the space of weak functions on K;
ν ( K ) = { v = { v 0 , v b } : v 0 ∈ [ L 2 ( K ) ] d , v b ∈ [ H 1 2 ( ∂ K ) ] d } (3)
Definition 1. For any v ∈ ν ( K ) , the weak gradient of v is defined as a linear functional ∇ w v in the dual space of [ H ( d i v , K ) ] d , whose action on each q ∈ [ H ( d i v , K ) ] d is given by
( ∇ w v , q ) K = − ( v 0 , ∇ ⋅ q ) K + 〈 v b , q ⋅ n 〉 ∂ K (4)
where n is the outward normal direction to ∂ K , ( v 0 , ∇ ⋅ q ) K = ∫ K v 0 ( ∇ ⋅ q ) d K is the action of v 0 on ∇ ⋅ q , and 〈 v b , q ⋅ n 〉 ∂ K = ∫ ∂ K v b q ⋅ n d s is the action of q ⋅ n on v b ∈ [ H 1 2 ( ∂ K ) ] d .
The Sobolev space [ H 1 ( K ) ] d can be embedded into the space ν ( K ) by an inclusion map i ν : [ H 1 ( K ) ] d → ν ( K ) defined as follows
i ν ( ϕ ) = { ϕ | K , ϕ | ∂ K } , ϕ ∈ [ H 1 ( K ) ] d
With the help of the inclusion map i ν , the Sobolev space [ H 1 ( K ) ] d can be viewed as a subspace of ν ( K ) by identifying each ϕ ∈ [ H 1 ( K ) ] d with i ν ( ϕ ) .
Let P r ( K ) be the set of polynomials on K with degree no more than r.
Definition 2. The discrete weak gradient operator, denoted by ∇ w , r , K , is defined as the unique polynomial ( ∇ w , r , K v ) ∈ [ P r ( K ) ] d × d satisfying the following equation,
( ∇ w , r , K v , q ) K = − ( v 0 , ∇ ⋅ q ) K + 〈 v b , q ⋅ n 〉 ∂ K , (5)
for all q ∈ [ P r ( K ) ] d × d .
In what follows, we give the definition of weak divergence, first of all, we require weak function v = { v 0 , v b } such that v 0 ∈ [ L 2 ( K ) ] d an v b ⋅ n ∈ L 2 ( ∂ K ) Denote by V ( K ) the space of weak vector-valued functions on K;
V ( K ) = { v = { v 0 , v b } : v 0 ∈ [ L 2 ( K ) ] d , v b ⋅ n ∈ L 2 ( ∂ K ) } (6)
Definition 3. For any v ∈ V ( K ) , the weak divergence of v is defined as a linear functional ∇ w ⋅ v in the dual space of H 1 ( K ) whose action on each φ ∈ H 1 ( K ) is given by
( ∇ w ⋅ v , φ ) K = − ( v 0 , ∇ φ ) K + 〈 v b ⋅ n , φ 〉 ∂ K (7)
where n is the outward normal direction to ∂ K , ( v 0 , ∇ φ ) K is the action of v 0
on ∇ φ , and 〈 v b ⋅ n , φ 〉 ∂ K is the action of v b ⋅ n on φ ∈ H 1 2 ( ∂ K ) .
The Sobolev space [ H 1 ( K ) ] d can be embedded into the space V ( K ) by an inclusion map i v : [ H 1 ( K ) ] d → V ( K ) defined as follows
i ν ( ϕ ) = { ϕ | K , ϕ | ∂ K } , ϕ ∈ [ H 1 ( K ) ] d
Definition 4. A discrete weak divergence operator, denoted by ∇ w , r , K , is defined as the unique polynomial ( ∇ w , r , K ⋅ v ) ∈ P r ( K ) that satisfies the following equation.
( ∇ w , r , K ⋅ v , φ ) K = − ( v 0 , ∇ φ ) K + 〈 v b ⋅ n , φ 〉 ∂ K , (8)
for all φ ∈ P r ( K ) .
Let T h be a partition of the domain W with mesh size h that consists of arbitrary polygons/polyhedra. In this paper, we assume that the partition T h is WG shape regular-defined by a set of conditions as detailed in references. Denote by ε h the set of all edges/flat faces in T h , and let ε h 0 = ε h ∂ Ω be the set of all interior edges/faces. For any integer k ≥ 1 , we define a weak Galerkin finite element space for the velocity variable as follows,
V h = { v = { v 0 , v b } : { v 0 , v b } | T ∈ [ P k ( T ) ] d × [ P k − 1 ( e ) ] d , e ⊂ ∂ T }
We would like to emphasize that there is only a single value v b defined on each edge e ∈ ε h . For the pressure variable, we have the following finite element space
W h = { q : q ∈ L 0 2 ( Ω ) , q | T ∈ P k − 1 ( T ) }
Denote by V h 0 the subspace of V h consisting of discrete weak functions with vanishing boundary value;
V h 0 = { v = { v 0 , v b } ∈ V h , v b = 0 on ∂ Ω }
The discrete weak gradient ∇ w , k − 1 and the discrete weak divergence ( ∇ w , k − 1 ) on the finite element space V h can be computed by using (5) and (8) on each element T, respectively. More precisely, they are given by
( ∇ w , k − 1 v ) | T = ∇ w , k − 1 , T ( v | T ) , ∀ v ∈ V h
( ∇ w , k − 1 ⋅ v ) | T = ∇ w , k − 1, T ⋅ ( v | T ) , ∀ v ∈ V h
For simplicity of notation, from now on we shall drop the subscript k − 1 in the notation ∇ w , k − 1 and ( ∇ w , k − 1 ) for the discrete weak gradient and the discrete weak divergence. The usual L 2 inner product can be written locally on each element as follows
( ∇ w v , ∇ w w ) = ∑ T ∈ T h ( ∇ w v , ∇ w w ) T
( ∇ w ⋅ v , q ) = ∑ T ∈ T h ( ∇ w ⋅ v , q ) T
Denote by Q 0 the L2 projection operator from [ L 2 ( T ) ] d onto [ P k ( T ) ] d . For each edge/face e ∈ ε h , denote by Q b the L2 projection from [ L 2 ( e ) ] d onto [ P k − 1 ( e ) ] d . We shall combine Q 0 with Q b by writing Q h = { Q 0 Q b } .
We are now in a position to describe a weak Galerkin finite element scheme for the Stokes Equations (1). To this end, we first introduce three bilinear forms as follows
s ( v , w ) = ∑ T ∈ T h h T − 1 〈 Q b v 0 − v b , Q b w 0 − w b 〉 ∂ T
a ( v , w ) = ( ∇ w v , ∇ w w ) + s ( v , w )
b ( v , q ) = ( ∇ w ⋅ v , q )
WG Algorithm. Seek ( u h = { u 0 , u b } , p h ) ∈ V h × W h satisfying
{ ( u h , t , v ) + a ( u h , v ) − b ( v , p h ) = ( f , v ) , ∀ v ∈ V h 0 b ( u h , q ) = 0, ∀ q ∈ W h u h ( x ,0 ) = ψ 0 ( x ) (9)
In the following, the proof process of Lemma 1-6 refers to reference [
Lemma 1. For any v ∈ V h 0 , the following equation hold true,
‖ | v | ‖ 2 = ∑ T ∈ T h ‖ ∇ w v ‖ T 2 + ∑ T ∈ T h h T − 1 ‖ Q b v 0 − v b ‖ ∂ T 2
Lemma 2. For any v , w ∈ V h 0 we have
| a ( v , w ) | ≤ ‖ | v | ‖ ‖ | w | ‖
a ( v , v ) = ‖ | v | ‖ 2 .
In addition to the projection Q h = { Q 0 , Q b } defined in the previous section, let Q h and Q h be two local L2 projections onto P k − 1 ( T ) and [ P k − 1 ( T ) ] d × d , respectively.
Lemma 3. The projection operators Q h , R h , and S h satisfy the following commutative properties
∇ w ( Q h v ) = R h ( ∇ v ) , ∀ v ∈ [ H 1 ( Ω ) ] d
∇ w ⋅ ( Q h v ) = S h ( ∇ ⋅ v ) , ∀ v ∈ H ( d i v Ω )
Lemma 4. There exists a positive constant b independent of h such that
s u p v ∈ V h 0 b ( v , ρ ) ‖ | v | ‖ ≥ β ‖ ρ ‖
for all ρ ∈ W h .
Lemma 5. Poincare inequality of Weak gradient operator: If v ∈ V h 0 , then exists a constant c satisfying
‖ v ‖ 2 ≤ c ‖ ∇ w v ‖ 2 ≤ c ‖ | v | ‖ 2
First of all, we study the existence and uniqueness of the solution for (9). The space defined as follows
T h = { v h ∈ V h ( j ) ; ( q , ∇ w ⋅ v ) = 0, ∀ q ∈ W h } .
Then we need to seek u h ( x , t ) : ( 0, T ) → T h satisfying
{ ( u h , t , v ) + a ( u h , v ) = ( f , v ) , v ∈ T h u h ( x , 0 ) = ψ 0 ( x ) (10)
Let u h ( x , t ) be the solution of (10) and which is unique, the linear bounded functional l = l ( u h ) on V h defined as follows.
〈 l , v 〉 = ( u h , t , v ) + a ( u h , v ) − ( f , v ) (11)
Then problem (9) is equivalent to seek P h ∈ W h satisfying
( P h , ∇ w ⋅ v ) = 〈 l , v 〉 , ∀ v ∈ V h (12)
Using LBB condition and Lax-Milgram Lemma, we know that the solution P h ∈ W h of (12) is unique.
Combing (11) and (12), it is concluded that if initial approximation u h ( x , 0 ) = ψ 0 ( x ) ∈ T h , the solution ( u h , P h ) ∈ V h × M h of (9) is unique.
In what follows, we introduce Stokes projection, which is the important approximation of projection.
Lemma 6. First of all, we introduce Stokes projection of ( u , p ) ∈ X × W , which is ( Q h 1 u , S h 1 p ) ∈ V h × W h need satisfying
{ a ( u h − Q h 1 u , v ) − b ( v , p h − S h 1 p ) = − φ u , p ( v ) , v ∈ V h b ( u h − Q h 1 u , q ) = 0 , q ∈ W h (13)
If let f ∗ = − Δ u + Δ p , easy to know that ( Q h 1 u , S h 1 p ) ∈ V h × W h satisfying
{ a ( Q h 1 u , v ) − b ( v , S h 1 p ) = ( f ∗ , v 0 ) , v ∈ V h b ( Q h 1 u , q ) = 0, q ∈ W h (14)
Then ( Q h 1 u , S h 1 p ) ∈ V h × W h is the finite element approximation of ( u , p ) ∈ X × W , so we have
{ ‖ u h − Q h 1 ‖ + h ‖ | u h − Q h 1 | ‖ ≤ C h k + 1 ( ‖ u ‖ k + 1 + ‖ p ‖ k ) ‖ p h − S h 1 p ‖ ≤ C h k ( ‖ u ‖ k + 1 + ‖ p ‖ k ) ‖ u h − Q h 1 ‖ + h ‖ | ( u h − Q h 1 ) t | ‖ ≤ C h k + 1 ( ‖ u ‖ k + 1 + ‖ p ‖ k + ‖ u t ‖ k + 1 + ‖ p t ‖ k ) (15)
In what follows, we list Lemma 7 to prove the error estimation of approximate solution for Semi-discrete scheme.
We know that ( u , p ) ∈ X × M and ( u h , p h ) ∈ X h × M h be solution of (1) and Galerkin finite element solution of (9), respectively. The L2 projection of u in the finite element space V h is given by Q h u = { Q 0 u , Q b u } . Similarly, the pressure p is projected into W h as S h p . Denote by e h and ε h the corresponding error given by
{ e h = { e 0 , e b } = { Q 0 u − u 0 , Q b u − u b } ε h = S h p − p h (16)
Lemma 7. Let ( w , p ) ∈ [ H 1 ( Ω ) ] d × L 2 ( Ω ) be sufficiently smooth and satisfy the following equation
w t − Δ w + ∇ ρ = η (17)
in the domain W. Let Q h w = { Q 0 w , Q b w } and S h ρ be the L2 projection of ( w , p ) into the finite element space V h × W h . Then, the following equation holds true
( Q h w t , v 0 ) + ( ∇ w ( Q h w ) , ∇ w v ) − ( ∇ w ⋅ v , S h ρ ) = ( η , v 0 ) + l w ( v ) − θ ρ ( v ) , (18)
for all v ∈ V h 0 . Where l w and θ ρ are two linear functionals on V h 0 defined by
l w ( v ) = ∑ T ∈ T h 〈 v 0 − v b , ∇ w ⋅ n − R h ( ∇ w ) ⋅ n 〉 ∂ T
θ ρ ( v ) = ∑ T ∈ T h 〈 v 0 − v b , ( ρ − S h ρ ) n 〉 ∂ T
Proof. Together Lemma 3, Equation (5) and integration by parts. we obtain
( ∇ w ( Q h w ) , ∇ w v ) T = ( R h ( ∇ w ) , ∇ w v ) T = − ( v 0 , ∇ ⋅ R h ( ∇ w ) ) T + 〈 v b , ∇ ⋅ R h ( ∇ w ) ⋅ n 〉 ∂ T = ( ∇ v 0 , R h ( ∇ w ) ) T − 〈 v 0 − v b , ∇ ⋅ R h ( ∇ w ) ⋅ n 〉 ∂ T = ( ∇ w , ∇ v ) T − 〈 v 0 − v b , ∇ ⋅ R h ( ∇ w ) ⋅ n 〉 ∂ T (19)
Next, Combing Lemma 3 and Equation (8), the fact that ∑ T ∈ T h 〈 v b , p n 〉 ∂ T = 0 ,
then using integration by parts, we obtain
( ∇ w ⋅ v , S h ρ ) = − ∑ T ∈ T h 〈 v 0 , ∇ ( S h ρ ) 〉 T + ∑ T ∈ T h 〈 v b , ( S h ρ ) n 〉 ∂ T = ∑ T ∈ T h 〈 ∇ ⋅ v 0 , S h ρ 〉 T − ∑ T ∈ T h 〈 v 0 − v b , ( S h ρ ) n 〉 ∂ T = ∑ T ∈ T h 〈 ∇ ⋅ v 0 , ρ 〉 T − ∑ T ∈ T h 〈 v 0 − v b , ( S h ρ ) n 〉 ∂ T
= − ∑ T ∈ T h 〈 v 0 , ∇ ρ 〉 T + ∑ T ∈ T h 〈 v 0 , ρ n 〉 ∂ T − ∑ T ∈ T h 〈 v 0 − v b , ( S h ρ ) n 〉 ∂ T = − ∑ T ∈ T h 〈 v 0 , ∇ ρ 〉 T + ∑ T ∈ T h 〈 v 0 − v b , ρ n 〉 ∂ T − ∑ T ∈ T h 〈 v 0 − v b , ( S h ρ ) n 〉 ∂ T = − ( v 0 , ∇ ρ ) + ∑ T ∈ T h 〈 v 0 − v b , ( ρ − S h ρ ) n 〉 ∂ T
We can imply that
( v 0 , ∇ p ) = − ( ∇ w ⋅ v , S h ρ ) + ∑ T ∈ T h 〈 v 0 − v b , ( ρ − S h ρ ) n 〉 ∂ T (20)
Next, we test (17) by using v 0 in v = { v 0 , v b } ∈ V h 0 to obtain, we can obtain
( w t , v 0 ) − ( Δ w , v 0 ) + ( ∇ ρ , v 0 ) = ( η , v 0 ) (21)
It follows from the usual integration by parts that
− ( Δ w , v 0 ) = ∑ T ∈ T h ( ∇ w , ∇ v 0 ) T − ∑ T ∈ T h 〈 v 0 − v b , ∇ w ⋅ n 〉 ∂ T
Where we have used the fact that ∑ T ∈ T h 〈 v b , ∇ w ⋅ n 〉 ∂ T = 0 . using Equations (19) and (20), we have
− ( Δ w , v 0 ) = ( ∇ w ( Q h w ) , ∇ w v ) − ∑ T ∈ T h 〈 v 0 − v b , ∇ w ⋅ n − R h ( ∇ w ) ⋅ n 〉 ∂ T (22)
Substituting (20), (22) and ( Q h w t , v 0 ) = ( w t , v 0 ) into (21) yields
( Q h w t , v 0 ) + ( ∇ w ( Q h w ) , ∇ w v ) − ( ∇ w ⋅ v , S h ρ ) = ( η , v 0 ) + l w ( v ) − θ ρ (v)
which completes the proof of the lemma.
In what follows, we give the derivation of the error equation of (9).
Lemma 8. Let e h and ε h be the error of the weak Galerkin finite element solution arising from (9), as defined by (16). Then, we have
{ ( e h , t , v ) + a ( e h , v ) − b ( v , ε h ) = φ u , p ( v ) b ( e h , q ) = 0 , (23)
for all v ∈ V h 0 and q ∈ W h , where φ u , p ( v ) = l u ( v ) − θ p ( v ) + s ( Q h u , v ) is a linear functional defined on V h 0 .
Proof. Since ( u , p ) satisfies the Equation (17) with η = f , then from Lemma 6 we have
( Q h u t , v 0 ) + ( ∇ w ( Q h u ) , ∇ w v ) − ( ∇ w ⋅ v , S h p ) = ( f , v 0 ) + l u ( v ) − θ ρ (v)
Adding s ( Q h u , v ) to both side of the above equation give
( Q h u t , v 0 ) + a ( Q h u , v ) − b ( v , S h p ) = ( f , v 0 ) + l u ( v ) − θ ρ ( v ) + s ( Q h u , v ) (24)
The difference of (24) and (9) yields the following equation,
( e h , t , v 0 ) + a ( e h , v ) − b ( v , ε h ) = l u ( v ) − θ ρ ( v ) + s ( Q h u , v )
for all v ∈ V h 0 , where e h = { e 0 , e b } = { Q 0 u − u 0 , Q b u − u b } . This completes the derivation of (23).
As to (24), we test Equation (1) by q ∈ W h and use (9) to obtain
0 = ( ∇ ⋅ u , q ) = ( ∇ w ⋅ Q h u , q ) = b ( Q h u , q ) (25)
The difference of (25) and (9) yields the following equation
b ( e h , q ) = 0
for all q ∈ W h .
Which completes the proof of the lemma.
In the following, the proof process of Lemma 9 refers to reference [
Lemma 9. If ( w , ρ , r ) ∈ [ H r + 1 ( Ω ) ] d × H r ( Ω ) × V h and 1 ≤ r ≤ k , with the
precondition of regular-shape T h , we have the following estimation.
| s ( Q h w , v ) | ≤ C h r ‖ w ‖ r + 1 ‖ | v | ‖
| l w ( v ) | ≤ C h r ‖ w ‖ r + 1 ‖ | v | ‖
| θ ρ ( v ) | ≤ C h r ‖ w ‖ r + 1 ‖ | v | ‖
The following theorem is the main result of this paper.
Theorem 1. Let ( u , p ) ∈ [ H 0 1 ( Ω ) ∩ H k + 1 ( Ω ) ] d × [ L 0 2 ( Ω ) ∩ H k ( Ω ) ] and ( u h , p h ) ∈ V h × W h be the solution of (1) and (9), respectively. the following error estimates is true.
‖ Q h u − u h ‖ ≤ C h k + 1 ( ‖ u ‖ k + 1 + ‖ p ‖ k + ∫ 0 t ( ‖ u ‖ k + 1 + ‖ p ‖ k + ‖ u t ‖ k + 1 + ‖ p t ‖ k ) d τ ) ‖ Q h u − u h ‖ ≤ C h k ( ‖ u ‖ k + 1 + ‖ p ‖ k + ‖ u t ‖ k + 1 + ‖ p t ‖ k ) ‖ S h p − p h ‖ ≤ C h k + 1 ( ‖ u ‖ k + 1 + ‖ p ‖ k + ‖ u t ‖ k + 1 + ‖ p t ‖ k + ∫ 0 t ( ‖ u ‖ k + 1 + ‖ p ‖ k + ‖ u t ‖ k + 1 + ‖ p t ‖ k + ‖ u t t ‖ k + 1 + ‖ p t t ‖ k ) d τ )
Proof. Let
e h = Q h u − u h = Q h u − Q h 1 u + Q h 1 u − u h = θ + η
e h ( ⋅ , 0 ) = θ ( ⋅ , 0 ) = η ( ⋅ , 0 ) = 0
By the error of Equation (23), we have
( θ t , v ) + a ( θ , v ) − b ( v , S h p − S h 1 p ) = φ u , p ( v ) − ( η t , v ) − a ( η , v ) + b ( v , S h 1 p − p h ) (26)
Substituting (13) into (26), we obtain
( θ t , v ) + a ( θ , v ) − b ( v , S h p − S h 1 p ) = − ( η t , v ) (27)
Let v = θ = Q h u − Q h 1 u , combing the Equation (25) and (14), we have
b ( θ , S h p − S h 1 p ) = 0
That is
( θ t , θ ) + a ( θ , θ ) = − ( η t , θ )
By Lemma 2 and Cauchy inequality, we have
1 2 d d t ‖ θ ‖ 2 + ‖ | θ | ‖ 2 ≤ ‖ η t ‖ ‖ θ ‖ ≤ ‖ η t ‖ 2 + ‖ θ ‖ 2 (28)
By Gronwall Lemma, we have
1 2 d d t ‖ θ ‖ 2 + ‖ | θ | ‖ 2 ≤ ( C h k + 1 ( ‖ u ‖ k + 1 + ‖ p ‖ k + ‖ u t ‖ k + 1 + ‖ p t ‖ k ) ) 2 (29)
By Cauchy inequality, we have
‖ | θ | ‖ ≤ C h k + 1 ( ‖ u ‖ k + 1 + ‖ p ‖ k + ‖ u t ‖ k + 1 + ‖ p t ‖ k ) (30)
Then take the integration about t of both side of Equation (28)
‖ θ ( ⋅ , t ) ‖ 2 + 2 ∫ 0 t ‖ | θ | ‖ 2 d τ ≤ ‖ θ ( ⋅ ,0 ) ‖ 2 + C ( ∫ 0 t ‖ η t ‖ d τ ) 2 + 1 4 s u p τ ≤ t ‖ θ ( τ ) ‖ 2
Since ‖ θ ( ⋅ , 0 ) ‖ = 0 , then
‖ θ ‖ ≤ s u p τ ≤ t ‖ θ ( τ ) ‖ ≤ C ∫ 0 t ‖ η t ‖ d τ ≤ C h k + 1 ∫ 0 t ( ‖ u ‖ k + 1 + ‖ p ‖ k + ‖ u t ‖ k + 1 + ‖ p t ‖ k ) d τ (31)
Combing the Equations (15), (29), (30) and triangle inequality, we have
‖ Q h u − u h ‖ ≤ C h k + 1 ( ‖ u ‖ k + 1 + ‖ p ‖ k + ∫ 0 t ( ‖ u ‖ k + 1 + ‖ p ‖ k + ‖ u t ‖ k + 1 + ‖ p t ‖ k ) d τ ) (32)
‖ Q h u − u h ‖ ≤ C h k ( ‖ u ‖ k + 1 + ‖ p ‖ k + ‖ u t ‖ k + 1 + ‖ p t ‖ k ) (33)
Next, we proof the error estimate of pressure approximation ‖ S h p − p h ‖ , by using error Equation (23), we have
b ( v , S h p − p h ) = ( e h , t , v ) + a ( e h , v ) − φ u , p (v)
By using Lemma 2, Lemma 5 and Lemma 9, we obtain
b ( v , S h p − p h ) ≤ ‖ e h , t ‖ ‖ v ‖ + ‖ | e h | ‖ ‖ | v | ‖ + C h k + 1 ( ‖ u ‖ k + 1 + ‖ p ‖ k ) ‖ | v | ‖ ≤ C ‖ e h , t ‖ ‖ v ‖ + ‖ | e h | ‖ ‖ | v | ‖ + C h k + 1 ( ‖ u ‖ k + 1 + ‖ p ‖ k ) ‖ | v | ‖
By Lemma 4, we have
‖ S h p − p h ‖ ≤ C ‖ e h , t ‖ + ‖ | e h | ‖ + C h k + 1 ( ‖ u ‖ k + 1 + ‖ p ‖ k ) (34)
Next we seek error estimate ‖ e h , t ‖ , then take the derivation about t of both sides of Equation (27)
( θ t t , v ) + a ( θ t , v ) − b ( v , S h p t − S h 1 p t ) = − ( η t t , v )
Let v = θ t , take the derivation about t of both side of Equations (14) and (25), we obtain
b ( θ t , S h p t − S h 1 p t ) = 0
That is
( θ t t , θ t ) + a ( θ t , θ t ) = − ( η t t , θ t )
By Lemma 2 and Cauchy inequality, we have
1 2 d d t ‖ θ t ‖ 2 + ‖ | θ t | ‖ 2 ≤ ‖ η t t ‖ ‖ θ t ‖
That is
‖ θ t ( ⋅ , t ) ‖ 2 ≤ ‖ θ t ( ⋅ ,0 ) ‖ 2 + C ( ∫ 0 t ‖ η t t ‖ d τ ) 2 + 1 4 s u p τ ≤ t ‖ θ t ( τ ) ‖ 2 (35)
Since ‖ θ t ( ⋅ , 0 ) ‖ = 0 , that is
‖ θ t ‖ ≤ s u p τ ≤ t ‖ θ t ( ⋅ , t ) ‖ ≤ C h k + 1 ∫ 0 t ( ‖ u ‖ k + 1 + ‖ p ‖ k + ‖ u t ‖ k + 1 + ‖ p t ‖ k + ‖ u t t ‖ k + 1 + ‖ p t t ‖ k ) d τ (36)
Combing the Equations (15) and triangle inequality, we have
‖ e h , t ‖ ≤ C h k + 1 ( ‖ u ‖ k + 1 + ‖ p ‖ k + ‖ u t ‖ k + 1 + ‖ p t ‖ k + ∫ 0 t ( ‖ u ‖ k + 1 + ‖ p ‖ k + ‖ u t ‖ k + 1 + ‖ p t ‖ k + ‖ u t t ‖ k + 1 + ‖ p t t ‖ k ) d τ ) (37)
Substituting (33) and (36) into (34), we have
‖ S h p − p h ‖ ≤ C h k + 1 ( ‖ u ‖ k + 1 + ‖ p ‖ k + ‖ u t ‖ k + 1 + ‖ p t ‖ k + ∫ 0 t ( ‖ u ‖ k + 1 + ‖ p ‖ k + ‖ u t ‖ k + 1 + ‖ p t ‖ k + ‖ u t t ‖ k + 1 + ‖ p t t ‖ k ) d τ )
This completes the proof. Thus, the error estimates of Theorem 1 hold. Optimal-order error estimates are established for the corresponding numerical approximation in an H1 norm for the velocity, and L2 norm for both the velocity and the pressure by use of the Stokes projection.
Ning, C. and Gu, H.M. (2018) Weak Galerkin Finite Element Method for the Unsteady Navier-Stokes Equation. American Journal of Computational Mathematics, 8, 108-119. https://doi.org/10.4236/ajcm.2018.81009