[go: up one dir, main page]
More Web Proxy on the site http://driver.im/ Skip to content
Licensed Unlicensed Requires Authentication Published by De Gruyter April 20, 2021

Error Analysis of an Unconditionally Energy Stable Local Discontinuous Galerkin Scheme for the Cahn–Hilliard Equation with Concentration-Dependent Mobility

  • Fengna Yan and Yan Xu EMAIL logo

Abstract

In this paper, we mainly study the error analysis of an unconditionally energy stable local discontinuous Galerkin (LDG) scheme for the Cahn–Hilliard equation with concentration-dependent mobility. The time discretization is based on the invariant energy quadratization (IEQ) method. The fully discrete scheme leads to a linear algebraic system to solve at each time step. The main difficulty in the error estimates is the lack of control on some jump terms at cell boundaries in the LDG discretization. Special treatments are needed for the initial condition and the non-constant mobility term of the Cahn–Hilliard equation. For the analysis of the non-constant mobility term, we take full advantage of the semi-implicit time-discrete method and bound some numerical variables in L -norm by the mathematical induction method. The optimal error results are obtained for the fully discrete scheme.

MSC 2010: 65M12; 65M60; 35K55

Award Identifier / Grant number: NNW2019ZT4-B08

Award Identifier / Grant number: 11722112

Award Identifier / Grant number: 12071455

Funding statement: Research supported by the National Numerical Windtunnel Project NNW2019ZT4-B08 and NSFC grants 11722112, 12071455.

A Proof of Some Lemmas in Section 3.3

A.1 Proof of Lemma 3.3

To prove the existence of u h 0 , w h 0 in (3.9), we use the following lemma (see [25, Lemma 1.4 of Chapter 2]).

Lemma A.1

Let 𝕏 be a finite-dimensional Hilbert space with inner product ( , ) and norm . Let ℙ be a continuous mapping from 𝕏 to itself such that, for a sufficiently large constant q > 0 , ( P ( ζ ) , ζ ) > 0 for all ζ X such that ζ = q . Then there exists ζ X , ζ q , such that P ( ζ ) = 0 .

In our case, X = S h k := V h k × γ 1 2 Σ h k . Multiplying the last equation in (3.9) by 𝛾 and adding it to the second equation in (3.9), we have ( G h ( X ) , χ ) = 0 for all χ = ( χ 1 , γ 1 2 χ 2 ) S h k , where X := ( u h 0 , γ 1 2 w h 0 ) and for any ζ = ( ζ 1 , γ 1 2 ζ 2 ) S h k , G h ( ζ ) S h k is defined using Riesz’s representation theorem by

( G h ( ζ ) , χ ) = ( f L ( ζ 1 ) , χ 1 ) + γ ( ζ 2 , χ 1 ) - γ K ( ζ ^ 2 ν , χ 1 ) K - ( q h 0 , χ 1 ) + γ ( ζ 2 , χ 2 ) + γ ( ζ 1 , χ 2 ) - γ K ( ζ ^ 1 , ν χ 2 ) K for all χ = ( χ 1 , γ 1 2 χ 2 ) S h k .

It is obvious that G h is continuous.

Due to Lemma A.1, G h ( ϖ ) = 0 has a solution

ϖ B q = { χ = ( χ 1 , γ 1 2 χ 2 ) S h k : χ 2 = χ 1 2 + γ χ 2 2 q 2 }

if ( G h ( χ ) , χ ) > 0 for χ = q . For more details, we refer readers to see [13] and [26, Chapter 13].

Recalling the choices of the fluxes in (2.3) and the boundary conditions in (2.4), there holds

(A.1) K ( - ( χ 1 , χ 2 ) K + ( χ ^ 1 , ν χ 2 ) K ) = K ( - ( χ 1 , ν χ 2 ) K + ( χ 1 , χ 2 ) K + ( χ ^ 1 , ν χ 2 ) K ) = - K ( ( χ ^ 2 ν , χ 1 ) K - ( χ 2 , χ 1 ) K ) .

Then we have

( G h ( χ ) , χ ) = ( f L ( χ 1 ) , χ 1 ) + γ χ 2 2 - ( q h 0 , χ 1 ) = ( χ 1 3 - χ 1 , χ 1 ) | χ 1 | L + ( a χ 1 + a , χ 1 ) χ 1 < - L + ( a χ 1 - a , χ 1 ) χ 1 > L + γ χ 2 2 - ( q h 0 , χ 1 ) χ 1 2 | χ 1 | L 2 - 1 2 χ 1 2 | χ 1 | L 2 - | Ω | 2 + a 2 χ 1 | χ 1 | > L 2 - a 2 a | Ω | + γ χ 2 2 - C q h 0 2 - ε 0 χ 1 2 1 2 | Ω | χ 1 | χ 1 | L 4 + a 2 χ 1 | χ 1 | > L 2 - | Ω | 2 - a 2 a | Ω | + γ χ 2 2 - C q h 0 2 - ε 0 χ 1 2 = 1 2 | Ω | χ 1 | χ 1 | L 4 - ε 0 χ 1 | χ 1 | L 2 + ( a 2 - ε 0 ) χ 1 | χ 1 | > L 2 - | Ω | 2 - a 2 a | Ω | + γ χ 2 2 - C q h 0 2 = 1 2 | Ω | χ 1 | χ 1 | L 4 - ε 0 χ 1 | χ 1 | L 2 + ( a 2 - ε 0 ) χ 1 | χ 1 | > L 2 + γ χ 2 2 - ( | Ω | 2 + a 2 a | Ω | + C q h 0 2 ) ,

where ε 0 is an arbitrarily small positive constant generated by Young’s inequality.

Since | Ω | , a , a , q h 0 are bounded and

χ 2 = χ 1 | χ 1 | L 2 + χ 1 | χ 1 | > L 2 + γ χ 2 2 = q 2 ,

it is easy to get that ( G h ( χ ) , χ ) is positive if q is large enough, provided by

ε 0 < min { 1 2 | Ω | , a 2 } .

A.2 Proof of Lemma 3.4

We consider the following auxiliary equations:

(A.2) { θ * = δ * in Ω , ρ * = θ * in Ω , θ * ν = 0 on Ω ,

with the regularity estimates

(A.3) ρ * H 2 ( Ω ) + θ * H 1 ( Ω ) δ * L 2 ( Ω ) .

If u h 1 0 , w h 1 0 and u h 2 0 , w h 2 0 all satisfy the equations in (3.9), then we have (A.4)

(A.4a) ( f L ( u h 1 0 ) - f L ( u h 2 0 ) , ξ ) K = - γ ( w h 1 0 - w h 2 0 , ξ ) K + γ ( ( w ^ h 1 0 - w ^ h 2 0 ) ν , ξ ) K ,
(A.4b) ( w h 1 0 - w h 2 0 , ϕ ) K = - ( u h 1 0 - u h 2 0 , ϕ ) K + ( u ^ h 1 0 - u ^ h 2 0 , ν ϕ ) K .
Letting δ * = u h 1 0 - u h 2 0 , we get

(A.5) ( u h 1 0 - u h 2 0 , u h 1 0 - u h 2 0 ) K = ( u h 1 0 - u h 2 0 , θ * ) K = ( u h 1 0 - u h 2 0 , ( θ * - Π θ * ) ) K + ( u h 1 0 - u h 2 0 , Π θ * ) K = ( u h 1 0 - u h 2 0 , ( θ * - Π θ * ) ) K + ( u ^ h 1 0 - u ^ h 2 0 , ν Π θ * ) K - ( w h 1 0 - w h 2 0 , Π θ * ) K = ( u h 1 0 - u h 2 0 , ν ( θ * - Π θ * ) ) K - ( u ^ h 1 0 - u ^ h 2 0 , ν ( θ * - Π θ * ) ) K + ( u ^ h 1 0 - u ^ h 2 0 , ν θ * ) K - ( w h 1 0 - w h 2 0 , Π θ * - θ * ) K - ( w h 1 0 - w h 2 0 , θ * ) K ,

where the third equality depends on equation (A.4b), and the last equality depends on the property of the interpolation operator Π in (3.3).

Due to θ * = ρ * and equation (A.4a), we obtain

(A.6) ( w h 1 0 - w h 2 0 , θ * ) K = ( w h 1 0 - w h 2 0 , ρ * ) K = ( w h 1 0 - w h 2 0 , ( ρ * - P ρ * ) ) K + ( w h 1 0 - w h 2 0 , P ρ * ) K = ( w h 1 0 - w h 2 0 , ( ρ * - P ρ * ) ) K + ( w ^ h 1 0 - w ^ h 2 0 ν , P ρ * ) K - 1 γ ( f L ( u h 1 0 ) - f L ( u h 2 0 ) , P ρ * ) K = ( w h 1 0 - w h 2 0 , ( ρ * - P ρ * ) ) K + ( w ^ h 1 0 - w ^ h 2 0 ν , P ρ * - ρ * + ρ * ) K - 1 γ ( f L ( u h 1 0 ) - f L ( u h 2 0 ) , P ρ * - ρ * + ρ * ) K .

From (A.5) and (A.6), we have

u h 1 0 - u h 2 0 K 2 = ( u h 1 0 - u h 2 0 , ν ( θ * - Π θ * ) ) K - ( u ^ h 1 0 - u ^ h 2 0 , ν ( θ * - Π θ * ) ) K + ( u ^ h 1 0 - u ^ h 2 0 , ν θ * ) K
- ( w h 1 0 - w h 2 0 , Π θ * - θ * ) K - ( w h 1 0 - w h 2 0 , ( ρ * - P ρ * ) ) K
- ( w ^ h 1 0 - w ^ h 2 0 ν , P ρ * - ρ * + ρ * ) K + 1 γ ( f L ( u h 1 0 ) - f L ( u h 2 0 ) , P ρ * - ρ * + ρ * ) K
= ( u h 1 0 - u h 2 0 , ν ( θ * - Π θ * ) ) K - ( u ^ h 1 0 - u ^ h 2 0 , ν ( θ * - Π θ * ) ) K + ( u ^ h 1 0 - u ^ h 2 0 , ν θ * ) K
- ( w h 1 0 - w h 2 0 , Π θ * - θ * ) K - ( w h 1 0 - w h 2 0 , ( ρ * - P ρ * ) ) K
- ( w ^ h 1 0 - w ^ h 2 0 ν , P ρ * - ρ * + ρ * ) K
(A.7) + 1 γ ( f L ( u h 1 0 ) - f L ( u h 2 0 ) , P ρ * - ρ * ) K + 1 γ ( f L ( u h 1 0 ) - f L ( u h 2 0 ) , ρ * - ρ ¯ * + ρ ¯ * ) K ,
where ρ ¯ * = 1 | Ω | Ω ρ * ( x ) d x .

Due to the definition of u ^ h 0 in (2.3), we have

(A.8) K ( ( u h 1 0 - u h 2 0 , ν ( θ * - Π θ * ) ) K - ( u ^ h 1 0 - u ^ h 2 0 , ν ( θ * - Π θ * ) ) K ) = K ( u h 1 0 - u h 2 0 , ν ( θ ^ * - Π θ ^ * ) ) K = 0 ,

where θ ^ * = θ * | R and the last equation depends on the property of the projection Π.

Because of the choice of u ^ h 0 in (2.3), the boundary condition of θ * in (A.2) and the continuity of θ * , we have

(A.9) K ( u ^ h 1 0 - u ^ h 2 0 , ν θ * ) K = 0 .

Similarly,

(A.10) K ( w ^ h 1 0 - w ^ h 2 0 ν , ρ * ) K = 0 .

Taking ξ = 1 in (A.4a), it is easy to obtain

(A.11) ( f L ( u h 1 0 ) - f L ( u h 2 0 ) , 1 ) = 0 .

Inserting (A.8)–(A.11) into (A.7) and summing up equation (A.7) over all elements 𝐾, the following estimate holds:

(A.12) u h 1 0 - u h 2 0 2 = - ( w h 1 0 - w h 2 0 , Π θ * - θ * ) - ( w h 1 0 - w h 2 0 , ( ρ * - P ρ * ) ) - K ( w ^ h 1 0 - w ^ h 2 0 ν , P ρ * - ρ * ) K + 1 γ ( f L ( u h 1 0 ) - f L ( u h 2 0 ) , P ρ * - ρ * ) + 1 γ ( f L ( u h 1 0 ) - f L ( u h 2 0 ) , ρ * - ρ ¯ * ) C h θ * H 1 ( Ω ) w h 1 0 - w h 2 0 + C h ρ * H 2 ( Ω ) w h 1 0 - w h 2 0 + C h ρ * H 1 ( Ω ) u h 1 0 - u h 2 0 ,

where the above inequality depends on the trace inequalities for the polynomials and the Sobolev functions, the interpolation properties of 𝑃 and Π, the global Lipschitz continuity of f L and the mean-value technique.

By choosing ξ = u h 1 0 - u h 2 0 , ϕ = γ ( w h 1 0 - w h 2 0 ) in (A.4a), (A.4b), respectively, together with property (A.1), we have

w h 1 0 - w h 2 0 2 + ( f L ( u h 1 0 ) - f L ( u h 2 0 ) , u h 1 0 - u h 2 0 ) = 0 ,

which implies

(A.13) w h 1 0 - w h 2 0 2 C u h 1 0 - u h 2 0 2 .

Inserting (A.13) into (A.12), and using the regularity estimates in (A.3), we deduce that

u h 1 0 - u h 2 0 2 C h u h 1 0 - u h 2 0 2 ,

i.e. u h 1 0 = u h 2 0 . From (A.13), it is obvious to see that w h 1 0 = w h 2 0 . Then we prove the uniqueness of u h 0 and w h 0 .

A.3 Proof of Lemma 3.5

From (3.9) and (3.13), we have the error equations (A.14)

(A.14a) ( e q 0 , ξ ) K = ( f L ( u 0 ) - f L ( u h 0 ) , ξ ) K + γ ( e w 0 , ξ ) K - γ ( e w ^ 0 ν , ξ ) K ,
(A.14b) ( e w 0 , ϕ ) K = - ( e u 0 , ϕ ) K + ( e u ^ 0 , ν ϕ ) K .
We use a method similar to the proof for the uniqueness of u h 0 and w h 0 . Taking δ * = u 0 - u h 0 in (A.2), with the property of Π and (A.14b), we have

(A.15) ( e u 0 , e u 0 ) K = ( e u 0 , θ * ) K = ( e u 0 , ( θ * - Π θ * ) ) K + ( e u 0 , Π θ * ) K = ( e u 0 , ( θ * - Π θ * ) ) K + ( e u ^ 0 , ν Π θ * ) K - ( e w 0 , Π θ * ) K = - ( ( u 0 - P u 0 ) , θ * - Π θ * ) K + ( e u 0 , ν ( θ * - Π θ * ) ) K + ( e u ^ 0 , ν ( Π θ * - θ * ) ) K + ( e u ^ 0 , ν θ * ) K - ( e w 0 , Π θ * - θ * ) K - ( e w 0 , θ * ) K = - ( ( u 0 - P u 0 ) , θ * - Π θ * ) K - ( u h 0 - u ^ h 0 , ν ( θ * - Π θ * ) ) K + ( e u ^ 0 , ν θ * ) K - ( e w 0 , Π θ * - θ * ) K - ( e w 0 , θ * ) K .

Similarly to (A.6), we get

(A.16) ( e w 0 , θ * ) K = ( e w 0 , ρ * ) K = ( e w 0 , ( ρ * - P ρ * ) ) K + ( e w 0 , P ρ * ) K = ( e w 0 , ( ρ * - P ρ * ) ) K + ( e w ^ 0 ν , P ρ * ) K - 1 γ ( f L ( u 0 ) - f L ( u h 0 ) , P ρ * ) K + 1 γ ( e q 0 , P ρ * ) K = ( e w 0 , ( ρ * - P ρ * ) ) K + ( e w ^ 0 ν , P ρ * - ρ * + ρ * ) K - 1 γ ( f L ( u 0 ) - f L ( u h 0 ) , P ρ * - ρ * ) K - 1 γ ( f L ( u 0 ) - f L ( u h 0 ) , ρ * - ρ ¯ * + ρ ¯ * ) K + 1 γ ( e q 0 , P ρ * - ρ * + ρ * - ρ ¯ * + ρ ¯ * ) K .

By taking ξ = - P e u 0 , ϕ = γ Π e w 0 in (A.14a) and (A.14b), respectively, we have

γ Π e w 0 2 = - γ ( w 0 - Π w 0 , Π e w 0 ) - ( u 0 - P u 0 , Π e w 0 ) + K ( u ^ 0 - P u ^ 0 , ν Π e w 0 ) K - ( e q 0 , P e u 0 ) + ( f L ( u 0 ) - f L ( u h 0 ) , P e u 0 ) .

Due to the error estimate of e q 0 in (3.14), the superconvergent property of 𝑃 in (3.6) and the interpolation properties (3.2) and (3.4)–(3.5), we deduce that

(A.17) Π e w 0 2 C h 2 k + 2 + C P e u 0 2 .

By a method similar to (A.8)–(A.11), we obtain

K ( u h 0 - u ^ h 0 , ν ( θ * - Π θ * ) ) K = K ( u h 0 , ν ( θ ^ * - Π θ ^ * ) ) K = 0 , K ( e u ^ 0 , ν θ * ) K = 0 , K ( e w ^ 0 ν , ρ * ) K = 0 , ( f L ( u 0 ) - f L ( u h 0 ) , 1 ) = 0 .

Inserting the above equations, the first equation in (3.10) and (A.16) into (A.15) and summing up over 𝐾, we have

( e u 0 , e u 0 ) = - ( ( u 0 - P u 0 ) , θ * - Π θ * ) - ( e w 0 , Π θ * - θ * ) - ( e w 0 , ( ρ * - P ρ * ) ) - K ( w ^ 0 - w ^ h 0 ν , P ρ * - ρ * ) K + 1 γ ( f L ( u 0 ) - f L ( u h 0 ) , P ρ * - ρ * + ρ * - ρ ¯ * ) - 1 γ ( e q 0 , P ρ * - ρ * + ρ * - ρ ¯ * ) .

By the trace inequalities for the polynomials and the Sobolev functions, the interpolation properties of 𝑃 and Π, the global Lipschitz continuity of f L and the mean-value technique, we get

e u 0 2 C h k + 1 ( θ * H 1 + ρ * H 2 ( Ω ) ) + C h ( θ * H 1 ( Ω ) + ρ * H 2 ( Ω ) ) Π e w 0 + C h ρ * H 2 ( Ω ) u 0 - u h 0 C h k + 1 u 0 - u h 0 + C h u 0 - u h 0 2 + C h u 0 - u h 0 P e u 0 C h k + 1 u 0 - u h 0 + C h u 0 - u h 0 2 ,

where the second inequality depends on (A.17) and the regularity estimates in (A.3). Then there hold the following estimates:

P e u 0 C h k + 1 , Π e w 0 C h k + 1 .

If ℎ is sufficiently small and k + 1 > d 2 , then it is easy to get that

u h 0 P e u 0 + u 0 - P u 0 + u 0 C h - d 2 h k + 1 + u 0 L .

The proof of Theorem 3.1 is complete.

B Proof of Some Lemmas in Section 4.3

B.1 Proof of Lemma 4.3

Taking θ = Π e s n in equations (4.2b) and (4.3b), we have

( e s n , Π e s n ) K = ( b ( u n ) p n - b ( u h n - 1 ) p h n , Π e s n ) K .

Adding and subtracting b ( u h n - 1 ) p n , we obtain

( Π e s n , Π e s n ) K = - ( s n - Π s n , Π e s n ) K + ( ( b ( u n ) - b ( u h n - 1 ) ) p n , Π e s n ) K + ( ( p n - p h n ) b ( u h n - 1 ) , Π e s n ) K = - ( s n - Π s n , Π e s n ) K + ( ( b ( u n ) - b ( u n - 1 ) ) p n , Π e s n ) K + ( ( b ( u n - 1 ) - b ( u h n - 1 ) ) p n , Π e s n ) K + ( ( p n - p h n ) b ( u h n - 1 ) , Π e s n ) K .

Collecting the above equation over all elements 𝐾, with the interpolation properties of the projections 𝑃, Π and the property of 𝑏 in (1.2), there holds

Π e s n 2 C h 2 k + 2 + C τ 2 + C ( P e u n - 1 2 + Π e p n 2 ) ,

where 𝐶 depends on s n H k + 1 ( Ω ) , u n - 1 H k + 1 ( Ω ) , p n H k + 1 ( Ω ) , p n L ( Ω ) and u t L ( ( 0 , T ) ; L 2 ( Ω ) ) .

B.2 Proof of Lemma 4.4

Subtracting (2.2e) from (4.1e), we have

(B.1) ( e w n , ϕ ) K = - ( e u n , ϕ ) K + ( e u ^ n , ν ϕ ) K .

Choosing ρ = P e u n , θ = - Π e w n , ϕ = Π e s n in equations (4.2a), (4.2b), (4.3a), (4.3b), (B.1), we have the error equations

( δ t e u n , P e u n ) K = - ( e s n , P e u n ) K + ( e s ^ n ν , P e u n ) K + ( δ t u n - u t n , P e u n ) K ,
- ( e s n , Π e w n ) K = - ( b ( u n ) p n - b ( u h n - 1 ) p h n , Π e w n ) K ,
( e w n , Π e s n ) K = - ( e u n , Π e s n ) K + ( e u ^ n , ν Π e s n ) K .
According to the choices of the fluxes s ^ h n , u ^ h n in (2.3) and the boundary conditions in (2.4), we have

K [ - ( Π e s n , P e u n ) K + ( Π e s ^ n ν , P e u n ) K - ( P e u n , Π e s n ) K + ( P e u ^ n , ν Π e s n ) K ] = 0 .

Then, by the definition of the projection Π, we obtain

( δ t P e u n , P e u n ) = - ( δ t ( u n - P u n ) , P e u n ) + ( δ t u n - u t n , P e u n ) + ( s n - Π s n , Π e w n ) - ( ( b ( u n ) p n - b ( u h n - 1 ) p h n ) , Π e w n ) - ( w n - Π w n , Π e s n ) - ( u n - P u n , Π e s n ) + K ( u ^ n - P u ^ n , Π e s n ν ) K .

By the interpolation properties of the projections 𝑃 and Π in (3.2), (3.4)–(3.6), we get

P e u n 2 - P e u n - 1 2 2 τ C h 2 k + 2 + C τ 2 + ε ( Π e s n 2 + Π e w n 2 + P e u n 2 ) + C ( P e u n - 1 2 + Π e p n 2 ) C h 2 k + 2 + C τ 2 + ε ( Π e w n 2 + P e u n 2 ) + C ( P e u n - 1 2 + Π e p n 2 ) ,

where the last step depends on Lemma 4.3, and 𝐶 depends on 𝜀, u t L ( ( 0 , T ) ; H k + 1 ( Ω ) ) , s n H k + 1 ( Ω ) , p n H k + 1 ( Ω ) , w n H k + 1 ( Ω ) , u n H k + 2 ( Ω ) , p n L ( Ω ) and u t t L ( ( 0 , T ) ; L 2 ( Ω ) ) . Summing up the above inequality from 1 to 𝑛, we complete the proof of this lemma.

B.3 Proof of Lemma 4.5

Considering equations (4.2a), (4.2e), (4.3a), (4.3e) and taking ρ = δ t P e u n , ϕ = Π e s n , respectively, we have

( δ t e u n , δ t P e u n ) K = - ( e s n , δ t ( P e u n ) ) K + ( e s ^ n ν , δ t P e u n ) K + ( δ t u n - u t n , δ t P e u n ) K ,
( δ t e w n , Π e s n ) K = - ( δ t e u n , Π e s n ) K + ( δ t e u ^ n , ν Π e s n ) K .
It is easy to see that

( δ t P e u n , δ t P e u n ) = - ( δ t ( u n - P u n ) , δ t P e u n ) + ( δ t u n - u t n , δ t P e u n ) - ( δ t ( u n - P u n ) , Π e s n ) + K ( δ t ( u ^ n - P u ^ n ) , Π e s n ν ) K - ( δ t ( w n - Π w n + Π e w n ) , Π e s n ) .

From Lemma 4.3, we deduce that

δ t P e u n 2 C h 2 k + 2 + C τ 2 + ε δ t Π e w n 2 + C ( P e u n - 1 2 + Π e p n 2 ) ,

where 𝐶 depends on 𝜀,

u L ( ( 0 , T ) ; H k + 4 ( Ω ) ) , u L ( ( 0 , T ) ; W 3 , ( Ω ) ) , u t L ( ( 0 , T ) ; H k + 2 ( Ω ) ) , u t t L ( ( 0 , T ) ; L 2 ( Ω ) ) and w t L ( ( 0 , T ) ; H k + 1 ( Ω ) ) .

B.4 Proof of Lemma 4.6

From equations (4.1c)–(4.1d) and (2.2c)–(2.2d), we have the error equations

( e p n , η ) K = - ( e q n , η ) K + ( e q ^ n , ν η ) K + ( ( u n r n - u n - 1 r n ) , η ) K ,
( e q n , φ ) K = ( u n - 1 r n - u h n - 1 r h n , φ ) K + γ ( e w n , φ ) K - γ ( e w ^ n ν , φ ) K .
Choosing η = γ Π e w n , φ = - P e q n , respectively, with the definition of the projection Π, we have

( P e q n , P e q n ) = γ ( p n - p h n , Π e w n ) + γ ( q n - P q n , Π e w n ) - γ K ( q ^ n - P q ^ n , Π e w n ν ) K - γ ( ( u n r n - u n - 1 r n ) , Π e w n ) - ( q n - P q n , P e q n ) + ( u n - 1 r n - u h n - 1 r h n , P e q n ) .

Depending on the equality of u n - 1 r n - u h n - 1 r h n = u n - 1 r n - u h n - 1 r n + u h n - 1 r n - u h n - 1 r h n , we get

P e q n 2 C h 2 k + 2 + C h 2 k + 2 u h n - 1 2 + C τ 2 + ε Π e w n 2 + C ( P e u n - 1 2 + u h n - 1 2 P e r n 2 + Π e p n 2 ) ,

where 𝐶 depends on 𝜀, u L ( ( 0 , T ) ; H k + 4 ( Ω ) ) , u t L ( ( 0 , T ) ; H 1 ( Ω ) ) and u .

B.5 Proof of Lemma 4.7

Choosing the test functions

ρ = 2 δ t P e q 1 , θ = - 2 δ t Π e p 1 , η = 2 Π e s 1 , φ = - 2 δ t P e u 1 , ϕ = 2 γ δ t Π e w 1 , ξ = δ t P e r 1

in (4.2), with the property of Π, we have the error equation

(B.2) 2 ( b ( u h 0 ) Π e p 1 , δ t Π e p 1 ) + 2 γ δ t Π e w 1 2 + δ t P e r 1 2 = - 2 ( δ t ( u 1 - P u 1 ) , δ t P e q 1 ) + 2 ( δ t u 1 - u t 1 , δ t P e q 1 ) + 2 ( s 1 - Π s 1 , δ t Π e p 1 ) - 2 ( ( b ( u 1 ) - b ( u h 0 ) ) p 1 , δ t Π e p 1 ) - 2 ( b ( u h 0 ) ( p 1 - Π p h 1 ) , δ t Π e p 1 ) - 2 ( δ t ( p 1 - Π p 1 ) , Π e s 1 ) - 2 ( δ t ( q 1 - P q 1 ) , Π e s 1 ) + 2 K ( δ t ( q ^ 1 - P q ^ 1 ) , Π e s 1 ν ) K + 2 ( ( u 1 r 1 - u 0 r 1 ) τ , Π e s 1 ) + 2 ( δ t ( q 1 - P q 1 ) , δ t P e u 1 ) - 2 ( u 0 δ t r 1 - u h 0 δ t r h 1 , δ t P e u 1 ) - 2 γ ( δ t ( w 1 - Π w 1 ) , δ t Π e w 1 ) - 2 γ ( δ t ( u 1 - P u 1 ) , δ t Π e w 1 ) + 2 γ K ( δ t ( u ^ 1 - P u ^ 1 ) , δ t Π e w 1 ν ) K - ( δ t ( r 1 - P r 1 ) , δ t P e r 1 ) + ( δ t r 1 - r t 1 , δ t P e r 1 ) + ( 2 u 1 u t 1 - 2 u 0 δ t u 1 + 2 u 0 δ t u 1 - 2 u h 0 δ t u h 1 , δ t P e r 1 ) := ϱ 1 + ϱ 2 + ϱ 3 + ϱ 4 ,

where

ϱ 1 = - 2 ( u 0 δ t r 1 - u h 0 δ t r h 1 , δ t P e u 1 ) + ( 2 u 0 δ t u 1 - 2 u h 0 δ t u h 1 , δ t P e r 1 ) ,
ϱ 2 = - 2 ( ( b ( u 1 ) - b ( u h 0 ) ) p 1 , δ t Π e p 1 ) - 2 ( b ( u h 0 ) ( p 1 - Π p 1 ) , δ t Π e p 1 ) - 2 ( δ t ( u 1 - P u 1 ) , δ t P e q 1 ) + 2 ( δ t u 1 - u t 1 , δ t P e q 1 ) + 2 ( s 1 - Π s 1 , δ t Π e p 1 ) ,
ϱ 3 = - 2 ( δ t ( p 1 - Π p 1 ) , Π e s 1 ) + 2 ( δ t ( q 1 - P q 1 ) , δ t P e u 1 ) - 2 γ ( δ t ( w 1 - Π w 1 ) , δ t Π e w 1 ) - ( δ t ( r 1 - P r 1 ) , δ t P e r 1 ) + ( δ t r 1 - r t 1 , δ t P e r 1 ) + ( 2 u 1 u t 1 - 2 u 0 δ t u 1 , δ t P e r 1 ) + 2 ( ( u 1 r 1 - u 0 r 1 ) τ , Π e s 1 ) ,
ϱ 4 = - 2 ( δ t ( q 1 - P q 1 ) , Π e s 1 ) + 2 K ( δ t ( q ^ 1 - P q ^ 1 ) , Π e s 1 ν ) K - 2 γ ( δ t ( u 1 - P u 1 ) , δ t Π e w 1 ) + 2 γ K ( δ t ( u ^ 1 - P u ^ 1 ) , δ t Π e w 1 ν ) K .
A simple mathematical inequality yields

(B.3) b ( u h 0 ) 1 2 Π e p 1 2 - b ( u h 0 ) 1 2 Π e p 0 2 τ + 2 γ δ t Π e w 1 2 + δ t P e r 1 2 = 2 ( b ( u h 0 ) Π e p 1 , δ t Π e p 1 ) - 1 τ b ( u h 0 ) 1 2 Π e p 1 - b ( u h 0 ) 1 2 Π e p 0 2 + 2 γ δ t Π e w 1 2 + δ t P e r 1 2 2 ( b ( u h 0 ) Π e p 1 , δ t Π e p 1 ) + 2 γ δ t Π e w 1 2 + δ t P e r 1 2 = ϱ 1 + ϱ 2 + ϱ 3 + ϱ 4 ,

where the last equality depends on (B.2). Next we estimate ϱ 1 , ϱ 2 , ϱ 3 and ϱ 4 gradually.

Since

ϱ 1 = - 2 ( ( δ t r 1 - δ t r h 1 ) u h 0 + ( u 0 - u h 0 ) δ t r 1 , δ t P e u 1 ) + 2 ( ( δ t u 1 - δ t u h 1 ) u h 0 + ( u 0 - u h 0 ) δ t u 1 , δ t P e r 1 ) = - 2 ( ( δ t r 1 - δ t P r 1 ) u h 0 + ( u 0 - u h 0 ) δ t r 1 , δ t P e u 1 ) + 2 ( ( δ t u 1 - δ t P u 1 ) u h 0 + ( u 0 - u h 0 ) δ t u 1 , δ t P e r 1 ) ,

together with (3.11) and Lemma 4.5, we get

| ϱ 1 | C h 2 k + 2 + C P e u 0 2 + ε ( δ t P e u 1 2 + δ t P e r 1 2 ) C h 2 k + 2 + C ( P e u 0 2 + Π e p 1 2 ) + ε ( δ t Π e w 1 2 + δ t P e r 1 2 ) .

By b ( u 1 ) - b ( u h 0 ) = b ( u 1 ) - b ( u 0 ) + b ( u 0 ) - b ( u h 0 ) , the interpolation properties of the projections 𝑃, Π and the error estimate of P e u 0 in Theorem 3.1, we obtain

| ϱ 2 | C ( h k + 1 + τ ) δ t P e q 1 + C ( h k + 1 + τ ) δ t Π e p 1 .

For the estimate of ϱ 3 , it is obvious to see that

| ϱ 3 | C h 2 k + 2 + C τ 2 + ε ( Π e s 1 2 + δ t P e u 1 2 + δ t Π e w 1 2 + δ t P e r 1 2 ) + C Π e s 1 .

Using Lemma 4.3 and Lemma 4.5, we deduce that

| ϱ 3 | C h 2 k + 2 + C τ 2 + C ( P e u 0 2 + Π e p 1 2 ) + ε ( δ t Π e w 1 2 + δ t P e r 1 2 ) + C Π e s 1 .

In one dimension, ϱ 4 = 0 . In multi-dimension,

| ϱ 4 | C h 2 k + 2 + ε ( Π e s 1 2 + δ t Π e w 1 2 ) .

According to Lemma 4.3, we obtain

| ϱ 4 | C h 2 k + 2 + C τ 2 + C ( P e u 0 2 + Π e p 1 2 ) + ε δ t Π e w 1 2 .

Inserting the estimates of ϱ 1 , , ϱ 4 into (B.3), with Theorem 3.1, we have

b ( u h 0 ) 1 2 Π e p 1 2 - b ( u h 0 ) 1 2 Π e p 0 2 τ + 2 γ δ t Π e w 1 2 + δ t P e r 1 K 2 C h 2 k + 2 + C τ 2 + C Π e p 1 2 + ε ( δ t Π e w 1 2 + δ t P e r 1 2 ) + C Π e s 1 + C ( h k + 1 + τ ) δ t P e q 1 + C ( h k + 1 + τ ) δ t Π e p 1 .

Recalling Theorem 3.1, Lemma 4.3, Lemma 4.6 and (3.11), there holds

Π e p 1 2 + τ δ t Π e w 1 2 + τ δ t P e r 1 2 C h 2 k + 2 + C τ 2 + C τ Π e p 1 2 + ε τ ( δ t Π e w 1 2 + δ t P e r 1 2 ) + C τ Π e s 1 + C ( h k + 1 + τ ) ( P e q 1 + Π e p 1 ) + P e q 0 2 + Π e p 0 2 C h 2 k + 2 + C τ 2 + C τ Π e p 1 2 + ε τ ( δ t Π e w 1 2 + δ t P e r 1 2 ) + ε ( P e q 1 2 + Π e p 1 2 + Π e s 1 2 ) C h 2 k + 2 + C τ 2 + C τ Π e p 1 2 + ε τ ( δ t Π e w 1 2 + δ t P e r 1 2 ) + ε ( P e r 1 2 + Π e w 1 2 + Π e p 1 2 ) .

From Lemma 4.2, it follows that

Π e p 1 2 + τ δ t Π e w 1 2 + τ δ t P e r 1 2 + Π e w 1 2 + P e r 1 2 C h 2 k + 2 + C τ 2 + ε ( P e r 1 2 + Π e w 1 2 ) + ε τ δ t P e r 1 2 + C τ P e r 1 2 + P e r 0 2 + ε τ δ t Π e w 1 2 + C τ Π e w 1 2 + Π e w 0 2 .

Noticing the initial data in Theorem 3.1, with sufficiently small 𝜏, we have

Π e p 1 2 + τ δ t Π e w 1 2 + τ δ t P e r 1 2 + Π e w 1 2 + P e r 1 2 C h 2 k + 2 + C τ 2 ,

and together with Lemma 4.3, Lemma 4.4, Lemma 4.5 and Lemma 4.6, we get the following estimates:

Π e s 1 2 + P e u 1 2 + P e q 1 2 + τ δ t P e u 1 2 C h 2 k + 2 + C τ 2 ,

where 𝐶 depends on 𝜀, u L ( ( 0 , T ) ; H k + 4 ( Ω ) ) , u L ( ( 0 , T ) ; W 3 , ( Ω ) ) , u t L ( ( 0 , T ) ; H k + 4 ( Ω ) ) and u t t L ( ( 0 , T ) ; L 2 ( Ω ) ) .

B.6 Proof of Lemma 4.8

Similarly to the case of n = 1 , taking

ρ = 2 δ t P e q n , θ = - 2 δ t Π e p n , η = 2 Π e s n , φ = - 2 δ t P e u n , ϕ = 2 γ δ t Π e w n , ξ = δ t P e r n

in (4.3), there holds

(B.4) 2 ( b ( u h n - 1 ) Π e p n , δ t Π e p n ) + 2 γ δ t Π e w n 2 + δ t P e r n 2 = - 2 ( δ t ( u n - P u n ) , δ t P e q n ) + 2 ( δ t u n - u t n , δ t P e q n ) + 2 ( s n - Π s n , δ t Π e p n ) - 2 ( ( b ( u n ) - b ( u h n - 1 ) ) p n , δ t Π e p n ) - 2 ( b ( u h n - 1 ) ( p n - Π p n ) , δ t Π e p n ) - 2 ( δ t ( p n - Π p n ) , Π e s n ) - 2 ( δ t ( q n - P q n ) , Π e s n ) + 2 K ( δ t ( q ^ n - P q ^ n ) , Π e s n ν ) K + 2 ( ( u n r n - u n - 1 r n ) - ( u n - 1 r n - 1 - u n - 2 r n - 1 ) τ , Π e s n ) + 2 ( δ t ( q n - P q n ) , δ t P e u n ) - 2 ( u n - 1 r n - u n - 2 r n - 1 τ - u h n - 1 r h n - u h n - 2 r h n - 1 τ , δ t P e u n ) - 2 γ ( δ t ( w n - Π w n ) , δ t Π e w n ) - 2 γ ( δ t ( u n - P u n ) , δ t Π e w n ) + 2 γ K ( δ t ( u ^ n - P u ^ n ) , δ t Π e w n ν ) K - ( δ t ( r n - P r n ) , δ t P e r n ) + ( δ t r n - r t n , δ t P e r n ) + ( 2 u n u t n - 2 u n - 1 δ t u n + 2 u n - 1 δ t u n - 2 u h n - 1 δ t u h n , δ t P e r n ) := ϱ 5 + ϱ 6 + ϱ 7 + ϱ 8 ,

where

ϱ 5 = - 2 ( u n - 1 r n - u n - 2 r n - 1 τ - u h n - 1 r h n - u h n - 2 r h n - 1 τ , δ t P e u n ) + ( 2 u n - 1 δ t u n - 2 u h n - 1 δ t u h n , δ t P e r n ) ,
ϱ 6 = - 2 ( ( b ( u n ) - b ( u h n - 1 ) ) p n , δ t Π e p n ) - 2 ( b ( u h n - 1 ) ( p n - Π p n ) , δ t Π e p n ) - 2 ( δ t ( u n - P u n ) , δ t P e q n ) + 2 ( δ t u n - u t n , δ t P e q n ) + 2 ( s n - Π s n , δ t Π e p n ) ,
ϱ 7 = - 2 ( δ t ( p n - Π p n ) , Π e s n ) + 2 ( δ t ( q n - P q n ) , δ t P e u n ) - 2 γ ( δ t ( w n - Π w n ) , δ t Π e w n ) - ( δ t ( r n - P r n ) , δ t P e r n ) + ( δ t r n - r t n , δ t P e r n ) + ( 2 u n u t n - 2 u n - 1 δ t u n , δ t P e r n ) + 2 ( ( u n r n - u n - 1 r n ) - ( u n - 1 r n - 1 - u n - 2 r n - 1 ) τ , Π e s n ) ,
ϱ 8 = - 2 ( δ t ( q n - P q n ) , Π e s n ) + 2 K ( δ t ( q ^ n - P q ^ n ) , Π e s n ν ) K - 2 γ ( δ t ( u n - P u n ) , δ t Π e w n ) + 2 γ K ( δ t ( u ^ n - P u ^ n ) , δ t Π e w n ν ) K .
Using a method similar to (B.3), we have

(B.5) ( b ( u h n - 1 ) Π e p n , Π e p n ) - ( b ( u h n - 2 ) Π e p n - 1 , Π e p n - 1 ) τ + 2 γ δ t Π e w n 2 + δ t P e r n 2 = b ( u h n - 1 ) 1 2 Π e p n 2 - b ( u h n - 1 ) 1 2 Π e p n - 1 2 τ + ( b ( u h n - 1 ) - b ( u h n - 2 ) τ Π e p n - 1 , Π e p n - 1 ) + 2 γ δ t Π e w n 2 + δ t P e r n 2 2 ( b ( u h n - 1 ) Π e p n , δ t Π e p n ) + ( b ( u h n - 1 ) - b ( u h n - 2 ) τ Π e p n - 1 , Π e p n - 1 ) + 2 γ δ t Π e w n 2 + δ t P e r n 2 ϱ 5 + ϱ 6 + ϱ 7 + ϱ 8 + C δ t u h n - 1 L Π e p n - 1 2 ,

where the last inequality depends on (B.4) and the smoothness assumption of 𝑏 in (1.2).

In view of

u n - 1 r n - u n - 2 r n - 1 τ - u h n - 1 r h n - u h n - 2 r h n - 1 τ = u n - 1 δ t r n + r n - 1 δ t u n - 1 - u h n - 1 δ t r h n - r h n - 1 δ t u h n - 1 = ( u n - 1 - u h n - 1 ) δ t r n + u h n - 1 ( δ t r n - δ t r h n ) + ( δ t u n - 1 - δ t u h n - 1 ) r n - 1 + δ t u h n - 1 ( r n - 1 - r h n - 1 )

and

u n - 1 δ t u n - u h n - 1 δ t u h n = ( u n - 1 - u h n - 1 ) δ t u n + u h n - 1 ( δ t u n - δ t u h n ) ,

together with Lemma 4.5 and the bound condition in (4.4), we get

ϱ 5 = - 2 ( ( u n - 1 - u h n - 1 ) δ t r n + u h n - 1 ( δ t r n - δ t r h n ) , δ t P e u n ) - 2 ( ( δ t u n - 1 - δ t u h n - 1 ) r n - 1 + δ t u h n - 1 ( r n - 1 - r h n - 1 ) , δ t P e u n ) + 2 ( ( u n - 1 - u h n - 1 ) δ t u n , δ t P e r n ) + 2 ( u h n - 1 ( δ t u n - δ t P u n ) , δ t P e r n ) C h 2 k + 2 + C P e u n - 1 2 + ε ( δ t u h n - 1 2 P e r n 2 + δ t P e r n 2 ) + C ( δ t P e u n - 1 2 + δ t P e u n 2 ) + C δ t u h n - 1 2 h 2 k + 2 C h 2 k + 2 + ε ( δ t u h n - 1 2 P e r n 2 + δ t P e r n 2 ) + C ( P e u n - 1 2 + P e u n - 2 2 ) + C ( Π e p n - 1 2 + Π e p n 2 ) + ε ( δ t Π e w n - 1 2 + δ t Π e w n 2 ) + C δ t u h n - 1 2 h 2 k + 2 ,

where 𝐶 depends on 𝜀, u L ( ( 0 , T ) ; H k + 4 ( Ω ) ) , u L ( ( 0 , T ) ; W 3 , ( Ω ) ) , u t L ( ( 0 , T ) ; H k + 2 ( Ω ) ) , u t , u t t L ( ( 0 , T ) ; L 2 ( Ω ) ) .

Noting

( b ( u h n - 1 ) ( p n - Π p n ) , δ t Π e p n ) = - ( δ t ( b ( u h n - 1 ) ( p n - Π p n ) ) , Π e p n - 1 ) + δ t ( b ( u h n - 1 ) ( p n - Π p n ) , Π e p n ) ,

we obtain

ϱ 6 = 2 ( δ t ( b ( u h n - 1 ) ( p n - Π p n ) ) , Π e p n - 1 ) - 2 δ t ( b ( u h n - 1 ) ( p n - Π p n ) , Π e p n ) + 2 ( δ t ( ( b ( u n ) - b ( u h n - 1 ) ) p n ) , Π e p n - 1 ) - 2 δ t ( ( b ( u n ) - b ( u h n - 1 ) ) p n , Π e p n ) - 2 ( δ t ( s n - Π s n ) , Π e p n - 1 ) + 2 δ t ( s n - Π s n , Π e p n ) + 2 ( δ t ( δ t ( u n - P u n ) ) , P e q n - 1 ) - 2 δ t ( δ t ( u n - P u n ) , P e q n ) - 2 ( δ t ( δ t u n - u t n ) , P e q n - 1 ) + 2 δ t ( δ t u n - u t n , P e q n ) .

Through a simple mathematical calculation, we have

δ t ( b ( u h n - 1 ) ( p n - Π p n ) ) = b ( u h n - 1 ) - b ( u h n - 2 ) τ ( p n - Π p n ) + b ( u h n - 2 ) p n - Π p n - ( p n - 1 - Π p n - 1 ) τ ,

and

(B.6) δ t ( ( b ( u n ) - b ( u h n - 1 ) ) p n ) = b ( u n ) - b ( u n - 1 ) - ( b ( u h n - 1 ) - b ( u h n - 2 ) ) τ p n + ( b ( u n - 1 ) - b ( u h n - 2 ) ) p n - p n - 1 τ = b ( u n - 1 ) - b ( u n - 2 ) - ( b ( u h n - 1 ) - b ( u h n - 2 ) ) τ p n + ( b ( u n - 1 ) - b ( u h n - 2 ) ) p n - p n - 1 τ + b ( u n ) - b ( u n - 1 ) - ( b ( u n - 1 ) - b ( u n - 2 ) ) τ p n .

Due to the smoothness assumption of 𝑏 and (B.6), we get

(B.7) δ t ( ( b ( u n ) - b ( u h n - 1 ) ) p n ) = b ( μ 1 n - 2 ) ( u n - 1 - u n - 2 ) - b ( μ 2 n - 2 ) ( u h n - 1 - u h n - 2 ) τ p n + b ( μ 3 n - 1 ) ( u n - u n - 1 ) - b ( μ 1 n - 2 ) ( u n - 1 - u n - 2 ) τ p n + ( b ( u n - 1 ) - b ( u h n - 2 ) ) p n - p n - 1 τ = ( b ( μ 1 n - 2 ) - b ( μ 2 n - 2 ) ) ( u n - 1 - u n - 2 ) - b ( μ 2 n - 2 ) ( u n - 1 - u n - 2 - ( u h n - 1 - u h n - 2 ) ) τ p n + ( b ( μ 3 n - 1 ) - b ( μ 1 n - 2 ) ) ( u n - u n - 1 ) - b ( μ 1 n - 2 ) ( u n - u n - 1 - ( u n - 1 - u n - 2 ) ) τ p n + ( b ( u n - 1 ) - b ( u h n - 2 ) ) p n - p n - 1 τ ,

where

μ 1 n - 2 = u n - 2 + λ 1 n - 2 ( u n - 1 - u n - 2 ) ,
μ 2 n - 2 = u h n - 2 + λ 2 n - 2 ( u h n - 1 - u h n - 2 ) ,
μ 3 n - 1 = u n - 1 + λ 3 n - 1 ( u n - u n - 1 ) , 0 < λ 1 n - 2 , λ 2 n - 2 , λ 3 n - 1 < 1 .
By
μ 1 n - 2 - μ 2 n - 2 = u n - 2 - u h n - 2 + ( λ 1 n - 2 - λ 2 n - 2 ) ( u n - 1 - u n - 2 ) + λ 2 n - 2 ( u n - 1 - u n - 2 - ( u h n - 1 - u h n - 2 ) ) ,
μ 3 n - 2 - μ 1 n - 2 = u n - 1 - u n - 2 + ( λ 3 n - 2 - λ 1 n - 2 ) ( u n - u n - 1 ) + λ 1 n - 2 ( u n - u n - 1 - ( u n - 1 - u n - 2 ) ) ,
Taylor expansion for b ( μ 1 n - 2 ) - b ( μ 2 n - 2 ) and b ( μ 3 n - 1 ) - b ( μ 1 n - 2 ) in (B.7), the boundedness of b and b ′′ in (1.2) and Lemma 4.5, we conclude that

| ϱ 6 | C h 2 k + 2 + C τ 2 + C h k + 1 δ t u h n - 1 L Π e p n - 1 + C h k + 1 Π e p n - 1 + C τ Π e p n - 1 + C P e u n - 2 Π e p n - 1 + C δ t P e u n - 1 Π e p n - 1 + ε P e q n - 1 2 + L L C h 2 k + 2 + C τ 2 + C h k + 1 δ t u h n - 1 L Π e p n - 1 + C ( P e u n - 1 2 + P e u n - 2 2 + Π e p n - 1 2 ) + ε δ t Π e w n - 1 2 + ε P e q n - 1 2 + L L ,

where

L L = | - δ t ( b ( u h n - 1 ) ( p n - Π p n ) , Π e p n ) - δ t ( ( b ( u n ) - b ( u h n - 1 ) ) p n , Π e p n ) + δ t ( s n - Π s n , Π e p n ) - δ t ( δ t ( u n - P u n ) , P e q n ) + δ t ( δ t u n - u t n , P e q n ) |

and 𝐶 depends on 𝜀,

u L ( ( 0 , T ) ; H k + 4 ( Ω ) ) , u L ( ( 0 , T ) ; W 3 , ( Ω ) ) , u t L ( ( 0 , T ) ; H k + 4 ( Ω ) ) , u t L ( ( 0 , T ) ; W 3 , ( Ω ) ) and u t t L ( ( 0 , T ) ; L 2 ( Ω ) ) .

For ϱ 7 , it is easy to see that

| ϱ 7 | C h 2 k + 2 + C τ 2 + ε ( Π e s n 2 + δ t P e u n 2 + δ t P e r n 2 + δ t Π e w n 2 ) .

Combining the above inequality with Lemma 4.3 and Lemma 4.5, we deduce that

| ϱ 7 | C h 2 k + 2 + C ( P e u n - 1 2 + Π e p n 2 ) + ε ( δ t Π e w n 2 + δ t P e r n 2 ) ,

where 𝐶 depends on 𝜀, u L ( ( 0 , T ) ; H k + 4 ( Ω ) ) , u L ( ( 0 , T ) ; W 3 , ( Ω ) ) , u t L ( ( 0 , T ) ; H k + 4 ( Ω ) ) and u t t L ( ( 0 , T ) ; L 2 ( Ω ) ) .

Similarly to the estimates of ϱ 4 , we have ϱ 8 = 0 in one dimension. In multi-dimension,

| ϱ 8 | C h 2 k + 2 + ε ( Π e s n 2 + δ t Π e w n 2 ) .

According to Lemma 4.3, we obtain

| ϱ 8 | C h 2 k + 2 + C ( P e u n - 1 2 + Π e p n 2 ) + ε δ t Π e w n 2 ,

where 𝐶 depends on 𝜀, u L ( ( 0 , T ) ; H k + 4 ( Ω ) ) , u L ( ( 0 , T ) ; W 3 , ( Ω ) ) and u t L ( ( 0 , T ) ; H k + 4 ( Ω ) ) .

Inserting into (B.5) these estimates ϱ 5 ϱ 8 , we have

( b ( u h n - 1 ) Π e p n , Π e p n ) - ( b ( u h n - 2 ) Π e p n - 1 , Π e p n - 1 ) τ + 2 γ δ t Π e w n 2 + δ t P e r n 2 C h 2 k + 2 + C τ 2 + C ( P e u n - 1 2 + P e u n - 2 2 + Π e p n 2 + Π e p n - 1 2 ) + C δ t u h n - 1 L Π e p n - 1 2 + ε ( δ t Π e w n 2 + δ t Π e w n - 1 2 + δ t P e r n 2 ) + ε δ t u h n - 1 2 P e r n 2 + C δ t u h n - 1 2 h 2 k + 2 + ε P e q n - 1 2 + L L ,

which complete the proof of this lemma.

References

[1] M. Baccouch, Optimal energy-conserving local discontinuous Galerkin method for the one-dimensional sine-Gordon equation, Int. J. Comput. Math. 94 (2017), no. 2, 316–344. 10.1080/00207160.2015.1105364Search in Google Scholar

[2] J. W. Barrett and J. F. Blowey, Finite element approximation of a model for phase separation of a multi-component alloy with a concentration-dependent mobility matrix, IMA J. Numer. Anal. 18 (1998), no. 2, 287–328. 10.1093/imanum/18.2.287Search in Google Scholar

[3] J. W. Barrett and J. F. Blowey, Finite element approximation of the Cahn–Hilliard equation with concentration dependent mobility, Math. Comp. 68 (1999), no. 226, 487–517. 10.1090/S0025-5718-99-01015-7Search in Google Scholar

[4] J. W. Barrett, J. F. Blowey and H. Garcke, Finite element approximation of the Cahn–Hilliard equation with degenerate mobility, SIAM J. Numer. Anal. 37 (1999), no. 1, 286–318. 10.1137/S0036142997331669Search in Google Scholar

[5] J. W. Cahn and J. E. Hilliard, Free energy of a nonuniform system I. Interfacial free energy, J. Chem. Phys. 28 (1958), 258–267. 10.1002/9781118788295.ch4Search in Google Scholar

[6] W. Chen, C. Wang, X. Wang and S. M. Wise, Positivity-preserving, energy stable numerical schemes for the Cahn–Hilliard equation with logarithmic potential, J. Comput. Phys. X 3 (2019), Article ID 100031. 10.1016/j.jcpx.2019.100031Search in Google Scholar

[7] S. M. Choo and Y. J. Lee, A discontinuous Galerkin method for the Cahn–Hilliard equation, J. Appl. Math. Comput. 18 (2005), no. 1–2, 113–126. 10.1007/BF02936559Search in Google Scholar

[8] B. Cockburn, G. Kanschat, I. Perugia and D. Schötzau, Superconvergence of the local discontinuous Galerkin method for elliptic problems on Cartesian grids, SIAM J. Numer. Anal. 39 (2001), no. 1, 264–285. 10.1137/S0036142900371544Search in Google Scholar

[9] B. Cockburn and C.-W. Shu, The local discontinuous Galerkin method for time-dependent convection-diffusion systems, SIAM J. Numer. Anal. 35 (1998), no. 6, 2440–2463. 10.1137/S0036142997316712Search in Google Scholar

[10] B. Dong and C.-W. Shu, Analysis of a local discontinuous Galerkin method for linear time-dependent fourth-order problems, SIAM J. Numer. Anal. 47 (2009), no. 5, 3240–3268. 10.1137/080737472Search in Google Scholar

[11] C. M. Elliott and D. A. French, A nonconforming finite-element method for the two-dimensional Cahn–Hilliard equation, SIAM J. Numer. Anal. 26 (1989), no. 4, 884–903. 10.1137/0726049Search in Google Scholar

[12] X. Feng, Fully discrete finite element approximations of the Navier–Stokes–Cahn–Hilliard diffuse interface model for two-phase fluid flows, SIAM J. Numer. Anal. 44 (2006), no. 3, 1049–1072. 10.1137/050638333Search in Google Scholar

[13] X. Feng and O. A. Karakashian, Fully discrete dynamic mesh discontinuous Galerkin methods for the Cahn–Hilliard equation of phase transition, Math. Comp. 76 (2007), no. 259, 1093–1117. 10.1090/S0025-5718-07-01985-0Search in Google Scholar

[14] X. Feng, Y. Li and Y. Xing, Analysis of mixed interior penalty discontinuous Galerkin methods for the Cahn–Hilliard equation and the Hele–Shaw flow, SIAM J. Numer. Anal. 54 (2016), no. 2, 825–847. 10.1137/15M1009962Search in Google Scholar

[15] X. Feng and S. Wise, Analysis of a Darcy–Cahn–Hilliard diffuse interface model for the Hele–Shaw flow and its fully discrete finite element approximation, SIAM J. Numer. Anal. 50 (2012), no. 3, 1320–1343. 10.1137/110827119Search in Google Scholar

[16] X. Feng and H. Wu, A posteriori error estimates for finite element approximations of the Cahn–Hilliard equation and the Hele–Shaw flow, J. Comput. Math. 26 (2008), no. 6, 767–796. Search in Google Scholar

[17] R. Guo, Y. Xia and Y. Xu, Semi-implicit spectral deferred correction methods for highly nonlinear partial differential equations, J. Comput. Phys. 338 (2017), 269–284. 10.1016/j.jcp.2017.02.059Search in Google Scholar

[18] R. Guo and Y. Xu, Efficient solvers of discontinuous Galerkin discretization for the Cahn–Hilliard equations, J. Sci. Comput. 58 (2014), no. 2, 380–408. 10.1007/s10915-013-9738-4Search in Google Scholar

[19] R. Guo and Y. Xu, Semi-implicit spectral deferred correction method based on the invariant energy quadratization approach for phase field problems, Commun. Comput. Phys. 26 (2019), no. 1, 87–113. 10.4208/cicp.OA-2018-0034Search in Google Scholar

[20] R. Guo, Y. Xu and Z. Xu, Local discontinuous Galerkin methods for the functionalized Cahn–Hilliard equation, J. Sci. Comput. 63 (2015), no. 3, 913–937. 10.1007/s10915-014-9920-3Search in Google Scholar

[21] L. Ji and Y. Xu, Optimal error estimates of the local discontinuous Galerkin method for Willmore flow of graphs on Cartesian meshes, Int. J. Numer. Anal. Model. 8 (2011), no. 2, 252–283. Search in Google Scholar

[22] D. Kessler, R. H. Nochetto and A. Schmidt, A posteriori error control for the Allen–Cahn problem: Circumventing Gronwall’s inequality, M2AN Math. Model. Numer. Anal. 38 (2004), no. 1, 129–142. 10.1051/m2an:2004006Search in Google Scholar

[23] W. H. Reed and T. R. Hill, Triangular mesh method for the neutron transport equation, Technical report LA-UR-73-479, Los Alamos Scientific Laboratory, Los Alamos, 1973. Search in Google Scholar

[24] H. Song and C.-W. Shu, Unconditional energy stability analysis of a second order implicit-explicit local discontinuous Galerkin method for the Cahn–Hilliard equation, J. Sci. Comput. 73 (2017), no. 2–3, 1178–1203. 10.1007/s10915-017-0497-5Search in Google Scholar

[25] R. Temam, Navier–Stokes Equations, Theory and Numerical Analysis, AMS Chelsea, Providence, 2001. 10.1090/chel/343Search in Google Scholar

[26] V. Thomée, Galerkin Finite Element Methods for Parabolic Problems, Springer, Berlin, 2006. Search in Google Scholar

[27] S. M. Wise, Unconditionally stable finite difference, nonlinear multigrid simulation of the Cahn–Hilliard–Hele–Shaw system of equations, J. Sci. Comput. 44 (2010), no. 1, 38–68. 10.1007/s10915-010-9363-4Search in Google Scholar

[28] Y. Xia, Y. Xu and C.-W. Shu, Local discontinuous Galerkin methods for the Cahn–Hilliard type equations, J. Comput. Phys. 227 (2007), no. 1, 472–491. 10.21236/ADA464873Search in Google Scholar

[29] Y. Xu and C.-W. Shu, Local discontinuous Galerkin methods for high-order time-dependent partial differential equations, Commun. Comput. Phys. 7 (2010), no. 1, 1–46. Search in Google Scholar

[30] Y. Xu and C.-W. Shu, Optimal error estimates of the semidiscrete local discontinuous Galerkin methods for high order wave equations, SIAM J. Numer. Anal. 50 (2012), no. 1, 79–104. 10.1137/11082258XSearch in Google Scholar

[31] J. Yan and C.-W. Shu, A local discontinuous Galerkin method for KdV type equations, SIAM J. Numer. Anal. 40 (2002), no. 2, 769–791. 10.1137/S0036142901390378Search in Google Scholar

[32] X. Yang and D. Han, Linearly first- and second-order, unconditionally energy stable schemes for the phase field crystal model, J. Comput. Phys. 330 (2017), 1116–1134. 10.1016/j.jcp.2016.10.020Search in Google Scholar

[33] X. Yang, J. Zhao and Q. Wang, Numerical approximations for the molecular beam epitaxial growth model based on the invariant energy quadratization method, J. Comput. Phys. 333 (2017), 104–127. 10.1016/j.jcp.2016.12.025Search in Google Scholar

Received: 2020-04-29
Revised: 2020-11-19
Accepted: 2021-03-09
Published Online: 2021-04-20
Published in Print: 2021-07-01

© 2021 Walter de Gruyter GmbH, Berlin/Boston

Downloaded on 16.12.2024 from https://www.degruyter.com/document/doi/10.1515/cmam-2020-0066/html
Scroll to top button