• Aucun résultat trouvé

Convergence analysis for the boundary value problem

For the second term in (B.4.21) we notice that [[(E−Gh)×n]] = −[[Gh×n]] owing to the regularity of E. The by proceeding as above, we arrive at

({{κ∇×Fh}},[[(E−Gh×n]])Σ∪Γ≤ch12k[[Gh×n]]kL2(Σ∪Γ)kκ∇×FhkL2(Th)

≤ch12k[[(E−Gh)×n]]kL2(Σ∪Γ)kκ∇×FhkL2(Th), where we used the inverse inequalities

hk∇×κ∇×FhkL2(Th) ≤ckκ∇×FhkL2(Th), hskκ∇×FhkHs(Th) ≤ckκ∇×FhkL2(Th). The desired result is obtained by gathering the above estimates.

The following result will be useful in order to prove a convergence result inL2(Ω). Proposition B.4.4 (Adjoint continuity). For any s ∈ 0,12

, there is c > 0, uniform in h such that the following holds for every (E, p),(F, q) ∈ Zs×H01(Ω), Fh ∈ Xh ∩H0,curl(Ω), qh∈Mh and(Gh, dh)∈Xh×Mh:

c ah((E−Gh, p−dh),(F−Fh, q−qh))

kE−Gh, p−dhkh ≤ kF−Fh, q−qhkh+hα−1kF−FhkL2(Ω)

+hskκ∇×(F−Fh)kHs(Th)

+hk∇×κ∇×(F−Fh)kL2(Th)

(B.4.22)

+h−αkq−qhkL2(Ω)+h(12−α)kq−qhkL2(Σ). Proof. The proof proceeds similarly as in the proof of Proposition B.4.3. The only dierence here is that we have ({{∇×(E−Gh)}},[[(F−Fh)×n]])Σ∪Γ = 0.owing to the assumption on Fh. This identity makes the analysis of the consistency term (B.4.21) tractable.

Remark B.4.3. Note that the coercivity and the continuity of ah have been established for anyα∈[0,1].

Theorem B.5.1. Let g ∈ L2(Ω) and denote (E, p) the solution of (B.2.10). Let τ <

min(τε, τµ) where τε and τµ are dened in Theorem B.2.1. Denote (Eh, ph) the solution of (B.4.13). Then, for anyα ∈`(1−τ)

`−τ ,1

, there exists c >0, uniform inh, such that (B.5.1) kE−Eh, p−phkh 6chrkgkL2(Ω),

with r= min 1−α, α−1 +τ 1−αr .

Proof. We rst recall that, owing to Theorem B.2.1, we haveE∈Hτ(Ω)∩Hτ0,curl(Ω), together with the estimates

kEkHτ(Ω)+k∇×EkHτ(Ω)+k∇×(κ∇×E)kL2(Ω)+k∇pkL2(Ω) ≤ckgkL2(Ω). We establish (B.5.1) by using the triangular inequality

kE−Eh, p−phkh 6kE− KδE,0kh+kKδE− ChKδE, p− Phpkh +kChKδE−Eh,Php−phkh,

for some δ >0to be dened later, and by bounding from above the three terms separately.

Using the denition ofk · kh together with the approximation properties ofKδ, cf. (B.3.16)-(B.3.17)-(B.3.18), we infer:

kE− KδE,0kh ≤cδτk∇×EkHτ(Ω)+chαδτ−1kEkHτ(Ω)+hα−12kKδEkL2(Σ).

Note that the estimate (B.3.17) is critical to obtain a bound that depends only onk∇×EkHτ(Ω)

instead of kEkH1+τ(Ω). To estimate the last term in the above inequality, we apply (B.7.6) withΘ = 2(1−τ)1−2τ ,

hα−12kKδEkL2(Σ)≤chα−12kKδEk1−ΘHτ(Ω)kKδEkΘH1(Ω)

≤chα−12δΘ(τ−1)kEkHτ(Ω)≤chα−12δτ−12kEkHτ(Ω). Finally, we arrive at

(B.5.2) kE− KδE,0kh≤c

δτ+hαδτ−1+hα−12δτ−12

kgkL2(Ω).

Let us now turn our attention tokKδE− ChKδE, p− Phpkh. Owing to the denition ofCh and the regularity of KδE, we have ChKδE ∈H0,curl(Ω), so that we only have four terms to bound (the jumps of ChKδE across the mesh interfaces and the tangent trace on Γ are zero, cf. Remark B.4.1). Using the properties of Kδ and Ch together with (B.7.5) we deduce that:

12∇×(KδE− ChKδE)kL2(Ω)≤ch`−1kKδEkH`(Ω)≤ch`−1δτ−`kEkHτ(Ω), hαk∇·(ε(KδE− ChKδE))kL2(Ω)≤chα+`−1kKδEkH`(Ω) ≤chα+`−1δτ−`kEkHτ(Ω),

h1−α12∇(p− Php)kL2(Ω)≤ch1−αkpkH1 0(Ω),

hα−12k[[ε(KδE− ChKδE)·n]]kL2(Σ)≤chα−12kKδE− ChKδEkL2(Σ)

≤chα−12kKδE− ChKδEk1−

1

L2(Ω)kKδE− ChKδEk

1

Hα(Ω)

≤chα−12h`(1−1)h(`−α)1 kKδEkH`(Ω)

≤chα+`−1δτ−`kEkHτ(Ω).

These estimates lead to

(B.5.3) kKδE− ChKδE, p− Phpkh ≤c

h`−1δτ−`+h1−α

kgkL2(Ω).

The last term, kChKδE−Eh,Php−phkh, is a little more subtle to handle. We start from the coercivity ofah (B.4.19) and use both the Galerkin orthogonality (B.4.15) and the continuity ofah (B.4.20) withs= 1−α to get the following estimate:

kChKδE−Eh,Php−phkh

≤cah((ChKδE−Eh,Php−ph),(ChKδE−Eh,Php−ph)) kChKδE−Eh,Php−phkh

≤cah((ChKδE−E,Php−p),(ChKδE−Eh,Php−ph)) kChKδE−Eh,Php−phkh

≤c kChKδE−E,Php−pkh+hα−1kE− ChKδEkL2(Ω)

+h1−αkκ∇×(E− ChKδE)kH1−α(Ω)+h−αkp− PhpkL2(Ω)

+hk∇×κ∇×(E− ChKδE)kL2(Th)+h12−αkp− PhpkL2(Σ)

.

We now handle each term in the right hand side separately. Using the triangle inequality kChKδE−E,Php−pkh ≤ kChKδE− KδE,Php−pkh +kKδE−E,0kh and the estimates (B.5.2)-(B.5.3), we obtain

kChKδE−E,Php−pkh ≤c

δτ+hαδτ−1+hα−12δτ−12 +h`−1δτ−`+h1−α

kgkL2(Ω). Similarly, we obtain

hα−1kE− ChKδEkL2(Ω)≤c

hα−1δτ+hα+`−1δτ−`

kgkL2(Ω), h1−αkκ∇×(E− ChKδEkH1−α(Ω)≤c

h1−αδτ+α−1+h`−1δτ−`

kgkL2(Ω).

Note that the previous computation is valid since 1 −α ≤ τ owing to the assumption α ∈`(1−τ)

`−τ ,1

. For the last term involving E we use the commuting property ¯δ∇×KδE = Kδ∇×E, see (B.3.19) as follows:

hk∇×κ∇×(E− ChKδE)kL2(Th) ≤hk∇×κ∇×EkL2(Th)+hk∇×κ∇×KδEkL2(Th)

+hk∇×κ∇×(KδE− ChKδE)kL2(Th)

≤c

hkgkL2(Ω)+hk∇×KδEkH1(Ω)+h`−1kKδEkH`(Ω)

≤c

hkgkL2(Ω)+hkKδ∇×EkH1(Ω)+h`−1δτ−`kEkHτ(Ω)

≤c

h+hδτ−1+h`−1δτ−`

kgkL2(Ω).

For the remaining terms involvingp, we use (B.7.5) together with the approximation properties ofCh:

h−αkp− PhpkL2(Ω) ≤ch1−αkpkH1

0(Ω)≤ch1−αkgkL2(Ω), h12−αkp− PhpkL2(Σ) ≤h12−αkp− Phpk1−

1

L2(Ω)kp− Phpk

1

Hα(Ω)

≤h12−αh1−1 h(1−α)1 kpkH1

0(Ω)≤ch1−αkgkL2(Ω).

Gathering all the above estimates together with (B.5.2) and (B.5.3), we nally obtain (B.5.4) kE−Eh, p−phkh ≤c δτ +h1−α+h+hδτ−1+h`−1δτ−`+hα−1δτ

+h1−αδτ+α−1+hαδτ−1+hα−12δτ−12

kgkL2(Ω).

We want to useδ = hβ for some β ∈ (0,1), i.e., δh−1 → +∞ as h→ 0. Once the negligible terms are removed in (B.5.4), we derive the following estimate:

kE−Eh, p−phkh ≤c hα−1δτ +h1−α+h`−1δτ−`

kgkL2(Ω). Usingδ =h1−α` implies thathα−1δτ =h`−1δτ−` and we arrive at

kE−Eh, p−phkh ≤c(hα−1+τ(1−α`) +h1−α)kgkL2(Ω), which leads to (B.5.1) withr:= min 1−α, α−1 +τ 1−α`

. Note that the assumed lower bound on α ensures that we have a convergence result as h→0.

Remark B.5.1. Note that the best choice forα is such that 1−α =α−1 +τ 1−α` . This choice gives the following convergence rate τ2(1−1`)< r=τ2`−τ`−1 < τ2.

We now derive a convergence estimate assuming that the solution of (B.2.10) is smooth.

Theorem B.5.2. Let g ∈ L2(Ω) and denote (E, p) the solution of (B.2.10). Assume that E ∈Hk+1(Ω) and p∈ Hk+α(Ω) for some 0 < k≤`−1 and α ∈[0,1]. Denote (Eh, ph) the solution of (B.4.13). Then there exists c >0, uniform in h, such that

(B.5.5) kE−Eh, p−phkh ≤c hk

kgkL2(Ω)+kEkHk+1(Ω)+kpkHk+α(Ω)

.

Proof. The proof is similar to that of Theorem B.5.1. We start from the triangular inequality kE−Eh, p−phkh ≤ kE− ChE, p− Phpkh+kChE−Eh,Php−phkh.

We bound the two terms in the right hand side separately. For the rst one, we use the approximation properties of Ch to get:

kE− ChE, p− Phpkh ≤c

hkkEkHk+1(Ω)+hk+αkEkHk+1(Ω)+h12kE− ChEkL2(Σ∪Γ)

+h1−αhk+α−1kpkHk+α(Ω)+hα−12kE− ChEkL2(Σ)

. Using (B.7.5) for anyσ∈ 12,1

, we have kE− ChEkL2(Σ)≤ckE− ChEk1−

1

L2(Ω)kE− ChEk

1

Hσ(Ω)≤c hk+12kEkHk+1(Ω). As a result, we obtain

(B.5.6) kE− ChE, p− Phpkh ≤chk

kEkHk+1(Ω)+kpkHk+α(Ω)

.

Now we turn our attention tokChE−Eh,Php−phkh. We use the coercivity ofah, the Galerkin orthogonality and the continuity of ah (for anyσ∈ 0,12

) to get

kChE−Eh,Php−phkh≤c kE− ChE, p− Phpkh+hα−1kE− ChEkL2(Ω)

+hσkκ∇×(E− ChE)kHσ(Th)

+hk∇×κ∇×(E− ChE)kL2(Th)

+h−αkp− PhpkL2(Ω)+h12−αkp− PhpkL2(Σ)

. Using the approximation properties ofCh together with (B.5.6), we infer

kE− ChE, p− Phpkh ≤chk

kEkHk+1(Ω)+kpkHk+α(Ω)

, hα−1kE− ChEkL2(Ω) ≤chk+αkEkHk+1(Ω),

hσkκ∇×(E− ChE)kHσ(Th) ≤chkkEkHk+1(Ω), h−αkp− PhpkL2(Ω) ≤chkkpkHk+α(Ω). For the last term involving p, we use (B.7.5) for some σ∈ 12,1

: h12−αkp− PhpkL2(Σ)≤ch12−αkp− Phpk1−

1

L2(Ω)kp− Phpk

1

Hσ(Ω)

≤ch12−αhk+α−12kpkHk+α(Ω) =chkkpkHk+α(Ω).

For the last term involving E, we distinguish two cases depending whether k < 1 or k ≥1.

If k < 1, we use an inverse inequality together with the approximation properties of Ch to deduce that

hk∇×κ∇×(E− ChE)kL2(Th)≤hk∇×κ∇×EkL2(Ω)+chkChEkH2(Th)

≤hkgkL2(Ω)+hkkEkHk+1(Ω). If k≥1, we use the approximation properties ofCh to get

hk∇×κ∇×(E− ChE)kL2(Th)≤chkE− ChEkH2(Th)≤chkkEkHk+1(Ω). In both cases, we have:

hk∇×κ∇×(E− ChE)kL2(Th)≤chk

kEkHk+1(Ω)+kgkL2(Ω)

. Gathering all the above estimates and using (B.5.6) gives the desired result (B.5.5).

Remark B.5.2. Note that the error estimate (B.5.5) is optimal since it implies thatk∇×(E− Eh)kL2(Ω) ≤ c hk, which is the best that can be expected from the piece-wise polynomial approximation of degreek. Note also that there is no lower bound on α to get convergence when the solution of (B.2.10) is smooth, i.e., anyα in the range[0,1] is acceptable.

B.5.2 Convergence in the L2-norm.

Before proving that the discrete solution converges to the exact solution in the L2-norm, we prove a global version of Lemma B.7.4 that will be useful in the proof of Theorem B.5.3.

Lemma B.5.1. Lets∈ 0,12

. Then there existsc >0, uniform inh, such that the following holds, for any ψ ∈Hcurl(Ω)∩Hs(Ω) and any Fh ∈Xh:

(B.5.7) |(ψ,[[Fh×n]])Σ∪Γ|

≤c h12k[[Fh×n]]kL2(Σ∪Γ) hskψkHs(Ω)+hk∇×ψkL2(Ω)

. Proof. Let us considerψ ∈Hcurl(Ω)∩Hs(Ω)and Fh∈Xh. Notice that the left hand side is well dened owing to Lemma B.7.4. We start from

ψ,[[Fh×n]]

Σ∪Γ

≤ |(ψ− Kδψ,[[Fh×n]])Σ∪Γ|

| {z }

:=I1

+|(Kδψ,[[Fh×n]])Σ∪Γ|

| {z }

:=I2

,

for someδ to be dened later. We handle the two termsI1,I2 separately. For the rst one, we apply Lemma B.7.4 withv= [[Fh×n]],φ=ψ− Kδψ and σ=s, and we sum over all the facesF ∈Σ∪Γ. This leads to

I1 ≤c h12k[[Fh×n]]kL2(Σ∪Γ) hskψ− KδψkHs(Th)

+hk∇×(ψ− Kδψ)kL2(Th)+kψ− KδψkL2(Ω)

≤c h12k[[Fh×n]]kL2(Σ∪Γ) hskψ− KδψkHs(Th)

+hk∇×ψkL2(Th)+hk∇×KδψkL2(Th)+kψ− KδψkL2(Ω)

. Using the approximation properties ofKδ (B.3.16) and (B.3.18), we arrive at

I1 ≤c h12k[[Fh×n]]kL2(Σ∪Γ) hskψkHs(Ω)

+hk∇×ψkL2(Ω)skψkHs(Ω)+hkKδψkH1(Ω)

≤c h12k[[Fh×n]]kL2(Σ∪Γ) (hss+hδs−1)kψkHs(Ω)+hk∇×ψkL2(Ω)

. We handleI2 by using the Cauchy-Schwarz inequality on every∂Ωi,i= 1,· · · , N.

I2 ≤c h12k[[Fh×n]]kL2(Σ∪Γ) N

X

i=1

h12kKδψkL2(∂Ωi). We use (B.7.6) on everyΩi withΘ := 2(1−s)1−2s , this leads to

I2 ≤c h12k[[Fh×n]]kL2(Σ∪Γ) N

X

i=1

h12kKδψk1−ΘHs(Ωi)kKδψkΘH1(Ωi)

≤c h12k[[Fh×n]]kL2(Σ∪Γ)h12kKδψk1−ΘHs(Ω)kKδψkΘH1(Ω),

where the constant c depends on N, which we recall is a xed number. Using again the approximation properties ofKδ we infer that

I2 ≤c h12k[[Fh×n]]kL2(Σ∪Γ)h12δ(s−1)ΘkψkHs(Ω)

≤c h12k[[Fh×n]]kL2(Σ∪Γ)h12δs−12kψkHs(Ω).

Then (B.5.7) is obtained by gathering the above estimates and settingδ=h.

Remark B.5.3. The proof of Lemma B.5.7 can done by using the decomposition ψ = ψ− Chψ+Chψ instead of ψ =ψ− Kδψ+Kδψ.

Theorem B.5.3. Let g ∈ L2(Ω) and denote (E, p) the solution of (B.2.10). Let τ <

min(τε, τµ) where τε and τµ are dened in Theorem B.2.1. Denote (Eh, ph) the solution of (B.4.13). Then, for any α ∈`(1−τ)

`−τ ,1

, there exists c >0, uniform inh, such that (B.5.8) kE−EhkL2(Ω)≤c h2rkgkL2(Ω),

withr := min 1−α, α−1 +τ 1−α`

. If in additionEandpare smooth, sayE∈Hk+1(Ω) andp∈Hk+α(Ω)for some 0< k < `−1, then the following holds:

(B.5.9) kE−EhkL2(Ω) ≤c hk+r

kgkL2(Ω)+kEkHk+1(Ω)+kpkHk+α(Ω)

.

Proof. We are going to use a duality argument à la Nitsche-Aubin. In the following we denotea1h the extension to

(Zτ(Ω) +Xh)×H01(Ω)2

of the bilinear form dened on[Xh×Mh]2 in (B.4.12) withθ= 1. Then the following symmetry property holds:

a1h((F, q),(G, r)) =a1h((G,−r),(F,−q)). for all ((F, q),(G, r)) ∈

(Zτ(Ω) +Xh)×H01(Ω)2

. Let (w, q) ∈ H0,curl(Ω)×H01(Ω) be the solution of the following (adjoint) problem:

∇×κ∇×w−ε∇q=ε(E−Eh), ∇·(εw) = 0.

Recall that Theorem B.2.1 implies thatw∈Zτ(Ω)∩Hτ(Ω)for anys < τ and that kwkHτ(Ω)+kκ∇×wkHτ(Ω)+k∇×κ∇×wkL2(Ω)≤ckE−EhkL2(Ω). (B.5.10)

The denition of the pair(w, q)implies that(ε∇q,∇ϕ)=−(ε(E−Eh,∇ϕ)for allϕ∈H01(Ω), and the following identities hold:

12(E−Eh)k2L2(Ω) =a1h((w,−q),(E−Eh, ph−p)) +cαh2(1−α)(ε∇q,∇(ph−p))

=a1h((E−Eh, p−ph),(w, q)) +cαh2(1−α)(ε(E−Eh),∇(p−ph))

=ah((E−Eh, p−ph),(w, q)) +cαh2(1−α)(ε(E−Eh),∇(p−ph)) + (1−θ) ({{κ∇×w}},[[−Eh×n]])Σ∪Γ

We now use the Galerkin orthogonality and introduceChKδw,Phq, withδ=h1−α`: (B.5.11) kε12(E−Eh)k2L2(Ω) =ah((E−Eh, p−ph),(w− ChKδw, q− Phq))

+cαh2(1−α)(ε(E−Eh),∇(p−ph))−(1−θ) (κ∇×w,[[Eh×n]])Σ∪Γ. Note that we replaced{{κ∇×w}}byκ∇×wsince the tangent component ofκ∇×wis contin-uous across the interfaces owing to∇×(κ∇×w)∈L2(Ω).

We now handle the three terms in the right hand side separately. For the rst one, we use Proposition B.4.4 withs= 1−α, F=wand Fh=ChKδw(note that Fh ∈Xh∩H0,curl(Ω)

sinceKδw∈ C0(Ω)); we then infer that

ah (E−Eh, p−ph),(w− ChKδw, q− Phq) ≤

ckE−Eh, p−phkh kw− ChKδw, q− Phqkh

+hα−1kw− ChKδwkL2(Ω)+h−αkq− PhqkL2(Ω)+h12−αkq− PhqkL2(Σ)

+hk∇×κ∇×(w− ChKδw)kL2(Th)+h1−αk∇×(w− ChKδw)kH1−α(Ω)

. The right hand side has already been estimated in the proof of Theorem B.5.1. We then have (B.5.12)

ah (E−Eh, p−ph),(w− ChKδw,Phq−q)

≤ckE−Eh, p−phkhhrkE−EhkL2(Ω).

The second term in (B.5.11) is estimated by using the Cauchy-Schwarz inequality, the deni-tion of the norm k · kh and inequalityr≤1−α,

h2(1−α)(ε(E−Eh),∇(p−ph))

≤c h1−αk∇(p−ph)kL2(Ω)h1−αkE−EhkL2(Ω)

≤ckE−Eh, p−phkhhrkE−EhkL2(Ω). (B.5.13)

The last term in (B.5.11) is estimated by using Lemma B.5.1 with ψ:=κ∇×wand s:=τ:

(1−θ) κ∇×w,[[Eh×n]]

Σ∪Γ

≤ckE−Ehkh hτkκ∇×wkHτ(Ω)+hk∇×(κ∇×w)kL2(Ω)

≤ckE−EhkhhrkE−EhkL2(Ω), (B.5.14)

where we have used (B.5.10) and r ≤ τ2 < τ. Upon inserting (B.5.12)-(B.5.13)-(B.5.14) in (B.5.11) we obtain

12(E−Eh)k2L2(Ω) ≤chrkE−EhkL2(Ω)kE−Eh, p−phkh. Owing to the uniform positivity of ε, this leads to:

kE−EhkL2(Ω)≤chrkE−Eh, p−phkh.

Now we consider two cases. Assuming only minimal regularity, Theorem B.5.1 gives a bound on kE−Eh, p−phkh that leads to (B.5.8). If E and p are smooth, then we can apply Theorem B.5.2 and we obtain (B.5.9).

Remark B.5.4. Let τ ∈ (0,12) and denote (E, p) the solution of (B.2.10). Assume that E ∈ Hτ(Ω)and E∈/ Hτ+(Ω)for all τ+ > τ. Then the best choice ofα is α= `(2−τ)2`−τ , which gives the convergence rate 2r =τ`−`−1τ

2; this convergence rate approaches the optimal rate, τ, when the approximation degree`is large.

Remark B.5.5. It is interesting to notice that the degree of the polynomials used for Mh is not involved in the convergence rate when minimal regularity is assumed. This means that we can use dierent degrees of polynomials for Xh and Mh, and that it is sucient to take polynomials of degree1 for Mh.

B.5.3 Numerical illustrations

We present in this section some numerical illustrations of the performance of the method on a boundary value problem. We consider the L-shaped domain

Ω = (−1,1)2\([0,+1]×[−1,0]). We assume thatΩis composed of three subdomains:

1 = (0,1)2, Ω2 = (−1,0)×(0,1), Ω3= (−1,0)2.

We useκ≡1inΩ,ε|Ω2 = 1 and ε|Ω1|Ω3 =:εr. Denotingλ >0 a real value such that tan

λπ 4

tan

λπ 2

r,

we dene the scalar potentialSλ(r, θ) = rλφλ(θ), where (r, θ) are the polar coordinates, and φλ is dened by

φλ(θ) = sin(λθ) if 0≤θ < π

2, φλ(θ) = sin λπ2

cos λπ4 cos

λ

θ−3π 4

if π

2 ≤θ < π, φλ(θ) = sin

λ

3π 2 −θ

if π ≤θ≤ 3π 2 . Then we solve the problem

∇×∇×E= 0, ∇·(εE) = 0, E×n|∂Ω =∇Sλ×n.

The exact solution is given byE =∇Sλ. We present two series of simulations. In table B.1, we useλ= 0.535, which leads toεr= 0.499±10−3. In table B.2, we useλ= 0.24, which leads toεr'7.55 10−2. In both case, we have computed the relative error in theL2-norm, and the column COC stands for the computed order of convergence. We have used several values of α, to show the eect of λ and α on the convergence rates. It seems that the convergence

h α= 0.4 α= 0.6 α= 0.9

rel.tol. coc rel.tol. coc rel.tol. coc

0.2 2.332E-1 - 1.444E-1 - 1.249E-1

-0.1 2.473E-1 -0.08 1.168E-1 0.31 8.846E-2 0.50 0.05 2.631E-1 -0.09 9.452E-2 0.31 6.186E-2 0.52 0.025 2.797E-1 -0.09 7.700E-2 0.30 4.289E-2 0.53 0.0125 2.968E-1 -0.09 6.312E-2 0.29 2.962E-2 0.53

Table B.1: L2-errors and computed order of convergence forλ= 0.535. We expect a conver-gence rate that is at most0.535: it is almost optimal withα= 0.9

rate improves when α is close to 1, which seems to be in contradiction with Remark B.5.4.

Actually, if we write the example used here in the form of (B.2.10), we can use g which is divergence free. Then one can notice in the convergence proofs that in this case, we have only

r=α−1 +τ 1−α

`

, which increases withα.

h α= 0.4 α= 0.6 α= 0.9 rel.tol. coc rel.tol. coc rel.tol. coc

0.2 5.773E-1 - 4.739E-1 - 4.426E-1

-0.1 6.209E-1 -0.11 4.507E-1 0.07 3.801E-1 0.22 0.05 6.711E-1 -0.11 4.413E-1 0.03 3.259E-1 0.22 0.025 7.180E-1 -0.10 4.452E-1 -0.01 2.788E-1 0.23 0.0125 7.564E-1 -0.08 4.602E-1 -0.05 2.380E-1 0.23

Table B.2: L2-errors and computed order of convergence for λ= 0.24. We expect a conver-gence rate that is at most0.24: it is almost optimal with α= 0.9