10 1137@S003614290037174X

SIAM J. NUMER. ANAL.
c 2001 Society for Industrial and Applied Mathematics

Vol. 39, No. 3, pp. 902–931
A PRIORI ERROR ESTIMATES FOR FINITE ELEMENT

METHODS BASED ON DISCONTINUOUS APPROXIMATION
Downloaded 12/30/12 to 129.173.72.87. Redistribution subject to SIAM license or copyright; see http://www.siam.org/journals/ojsa.php
SPACES FOR ELLIPTIC PROBLEMS∗

BÉATRICE RIVIÈRE† , MARY F. WHEELER† , AND VIVETTE GIRAULT‡
Abstract. We analyze three discontinuous Galerkin approximations for solving elliptic problems
in two or three dimensions. In each one, the basic bilinear form is nonsymmetric: the first one has a
penalty term on edges, the second has one constraint per edge, and the third is totally unconstrained.
For each of them we prove hp error estimates in the H 1 norm, optimal with respect to h, the mesh
size, and nearly optimal with respect to p, the degree of polynomial approximation. We establish
these results for general elements in two and three dimensions. For the unconstrained method, we
establish a new approximation result valid on simplicial elements. L2 estimates are also derived for
the three methods.
Key words. discontinuous Galerkin methods, elliptic equations, error estimates, penalties
AMS subject classifications. 35J25, 65N15, 65N30
PII. S003614290037174X
1. Introduction. Over the past five years, discontinuous Galerkin methods have
become very widely used for solving a large range of computational fluid problems.
They are preferred over more standard continuous methods because of their flexibility
in approximating globally rough solutions, their local mass conservation, their pos-
sible definition on unstructured meshes, their potential for error control and mesh
adaptation, and their straightforward applications to anisotropic materials. Another
interesting aspect of these methods is that they can employ polynomials of degree k
(Pk ) on quadrilaterals, whose dimension is substantially lower than that of Qk , the
standard space of functions on quadrilaterals. Similarly, they lead to smaller systems
of equations than the locally conservative mixed finite element methods and do not
require the solution of saddle-point problems.
In this paper, we discuss three numerical algorithms for elliptic problems which
employ discontinuous approximation spaces. The three methods are called the non-
symmetric interior penalty Galerkin (NIPG) method, the nonsymmetric constrained
Galerkin (NCG) method, and the discontinuous Galerkin (DG) method. The three
algorithms are closely related in that the underlying bilinear form for all three is the
same and is nonsymmetric.
In the NIPG method, we modify the bilinear form of the interior penalty Galerkin
method treated by Douglas and Dupont [9], Wheeler [18], Arnold [2], and Wheeler
and Darlow [19]. Here, we antisymmetrize the bilinear form by changing the sign of
one term and as a consequence we require only a positive penalty, whereas the proofs
in [18] assume that the penalty is bounded below by a problem-dependent constant.
We choose the penalty term in such a way that we can establish both H 1 and L2
optimal convergence rates.
∗ Received by the editors May 1, 2000; accepted for publication (in revised form) March 8, 2001;
published electronically July 31, 2001.

http://www.siam.org/journals/sinum/39-3/37174.html
† The Center for Subsurface Modeling, Texas Institute for Computational and Applied Mathemat-
ics, The University of Texas, Austin, TX 78712 (riviere@ticam.utexas.edu, mfw@ticam.utexas.edu).

‡ Université Pierre et Marie Curie, Paris VI, Laboratoire Analyse Numérique, 4, place Jussieu,
F-75230 Paris Cedex 05, France (girault@ann.jussieu.fr).

902
ERROR ESTIMATES 903
In the NCG method, we impose that the jumps on each edge (or face) of the
subdivision have integral average zero. One constraint per edge (or face) is sufficient
for again proving optimal rates of convergence in the H 1 and L2 norms.
The DG method is totally unconstrained and for that reason its error analysis
is more delicate. We recover full accuracy in the H 1 norm by introducing a special
interpolant, local to each element, whose average flux is orthogonal to constants on
each edge (or face). This interpolant is constructed first for constant tensors and then
extended to W 1,∞ tensors by an approximation result.
The results presented here extend significantly the work in [15]. To our knowledge
these results are new.
The DG method with this bilinear form was first introduced by Baumann [6], and
Oden, Babus̆ka, and Baumann in [13]. In [6], Baumann showed that the method is
elementwise conservative and he proved a stability result in one dimension for poly-
nomials of at least degree three. However, his proof is based on an inf-sup condition,
which seems problematic in higher dimensions. Our approach uses special approxi-
mation results and trace theorems but no isomorphisms.
This paper consists of five sections after this introduction. In section 2, we list the
notation, state the problem, and describe the formulation of the three methods. In
sections 3, 4, and 5, the proofs of the error estimates of the three methods described
in section 2 are respectively given. A numerical example is presented in section 6. In
the last section, we present some conclusions.
2. Statement of the problem. Finite element methods.
2.1. Notation and approximation properties. Let Ω be a polygonal domain

in Rn . Everything here applies also to n = 1, but to simplify we consider only n = 2
or 3. Let Eh = {E1 , E2 , . . . , ENh } be a nondegenerate quasi-uniform subdivision of
Ω, where Ej is a triangle or a quadrilateral if n = 2, or a tetrahedron if n = 3.
The nondegeneracy requirement (also called regularity) is that there exists ρ > 0
such that if hj is the diameter of Ej , then Ej contains a ball of radius ρhj in its
interior. However, for quadrilaterals, the requirement is a little bit stronger: the
quadrilateral is convex and each of its subtriangles contains a ball of radius ρhj .
Let h = max {hj , j = 1, . . . , Nh }. The quasi-uniformity requirement is that there is
τ > 0 such that hhj ≤ τ for all j ∈ 1, . . . , Nh . This quasi-uniformity assumption is
used for deriving error estimates in terms of the degree of polynomials (i.e., for the
p-version). For deriving error estimates in terms of h (i.e., for the h-version), we need
only
a regular subdivision. We denote the edges (resp., faces for n = 3) of Eh by
e1 , e2 , . . . , ePh , ePh+1 , . . . , eMh , where ek ⊂ Ω, 1 ≤ k ≤ Ph , and ek ⊂ ∂Ω, Ph + 1 ≤
k ≤ Mh . With each edge (or face) ek , we associate a unit normal vector ν k . For
k > Ph , ν k is taken to be the unit outward vector normal to ∂Ω. For s ≥ 0 and p ≥ 1,
let
W s,p (Eh ) = {v ∈ Lp (Ω) : v|Ej ∈ W s,p (Ej ), j = 1, . . . , Nh },
and we denote it by H s (Eh ) when p = 2.

We now define the average and the jump for φ ∈ H s (Eh ), s > 12 . Let 1 ≤ k ≤ Ph ;
for ek = ∂Ei ∩ ∂Ej with ν k exterior to Ei , set
1 1
{φ} = (φ|Ei )|ek + (φ|Ej )|ek , [φ] = (φ|Ei )|ek − (φ|Ej )|ek .
2 2
904 B. RIVIÈRE, M. F. WHEELER, AND V. GIRAULT
We consider a matrix-valued function K = (kij )1≤i,j≤n and a nonnegative scalar

function α. We assume
K ∈ W 1,4 (Eh ), α ∈ L∞ (Ω).

We also assume that K is symmetric positive definite in Ω̄ uniformly with respect to
x. This means that if γmin (x) ≤ γmax (x) are the smallest and the largest eigenvalues
at any point x of K, then there exist γ0 > 0 and γ1 > 0 such that
(2.1) ∀ x ∈ Ω̄, γmin (x) ≥ γ0 , γmax (x) ≤ γ1 .
The L2 inner product is denoted by (., .). The usual Sobolev norm of H s on E ⊂ Rn
is denoted by · s,E . We define the following “broken” norms for positive s:
  12
Nh
|||Φ|||s =  Φ 2s,Ej  .
j=1
The reader can refer to Adams [1] and to Lions and Magenes [12] for the properties
of Sobolev spaces.
Let r be a positive integer. The finite element subspace is taken to be
Nh

Dr (Eh ) = Pr (Ej ),
j=1
where by Pr (Ej ) we mean that the polynomials are defined on Ej and not on a
reference set Ê. This distinction is unnecessary when Ej is a triangle, a tetrahedron,
or a parallelogram because the transformation from Ê to Ej is affine. But it is
important when Ej is a quadrilateral: in this case, it is shown by [3] that Pr (Ê) does
not have optimal approximation properties, whereas it is shown by [10] that Pr (Ej )
has optimal approximation properties.
We use the following hp approximation properties, proven in [4, 5]. Let Ej ∈ Eh
and φ ∈ H s (Ej ). Then there exists a constant C depending on s, τ, ρ but independent
of φ, r, and h and a sequence zrh ∈ Pr (Ej ), r = 1, 2, . . . , such that for any 0 ≤ q ≤ s
hµ−q
j
(2.2) φ − zrh q,Ej ≤C φ s,Ej , s ≥ 0,
rs−q
µ− 12
hj 1
(2.3) φ− zrh 0,γi ≤C φ s,Ej , s> ,
1
rs− 2 2
where µ = min(r + 1, s) and γi is an edge or a face on ∂Ej . Using the same technique
as in [4], it can be shown that we have the additional approximation result
µ− 32
hj 3
(2.4) φ− zrh 1,γi ≤C φ s,Ej , s> .
3
rs− 2 2
Finally, we shall often use the following bounds with respect to h that extend to three-
dimensional formulae (2.4), (2.5) of [2]:
1
(2.5) ∀ φ ∈ H 1 (Ej ), φ 2
0,γi ≤C |φ|2 + hj |φ|21,Ej ,
hj 0,Ej
2
∂φ 1
(2.6) ∀ φ ∈ H 2 (Ej ), ≤C |φ|2 + hj |φ|22,Ej .
∂ν 0,γi hj 1,Ej
ERROR ESTIMATES 905
We shall also use the following inverse inequalities.

Lemma 2.1. Let E be an element in Rn (n = 2, 3) of diameter hE , let ek be
an edge or a face of E, and let ν k be a unit vector normal to ek . Then, if χ is a
polynomial of degree r on E, there exists a constant C independent of E and r such

that
−1
(2.7) χ 0,ek ≤ CrhE 2 χ 0,E ,
− 12
(2.8) ∇χ · ν k 0,ek ≤ CrhE ∇χ 0,E .
Proof. Let us briefly recall the proof for the reader’s convenience. In all the
proofs, C will be a generic constant with different values on different places, that is,
independent of h and r. We consider the case where n = 2; the proof extends easily
−1
to three dimensions. The factor hE 2 results from a straightforward scaling argument,
and we have only to exhibit the factor r in a reference setting. First, let us consider
the case where E is the reference square [−1, 1]×[−1, 1], e is the segment [−1, 1]×{1},
and p is a polynomial of Qr , i.e., a polynomial of degree at most r in each variable.
Then, expanding p in terms of Legendre polynomials

p(x1 , x2 ) = αm,n Lm (x1 )Ln (x2 ),
m,n
we obtain on one hand

2
αm,n
2
p 0,E = .
m,n
(m + 12 )(n + 12 )
On the other hand,

p(x1 , 1) = αm,n Lm (x1 ).
m,n
Therefore,
1
2 2
p 0,e = αm,n 1
m n
m+ 2
and by the Cauchy–Schwarz inequality

1 αm,n
2 1
2
p 0,e ≤ 1 1 n+ ≤ Cr2 p 2
0,E .
m
m+ 2 n
n+ 2 n
2
Hence,
(2.9) p 0,e ≤ Cr p 0,E .
This inequality carries over to a polynomial of Pr in a quadrilateral because the image

of Pr on the reference square belongs to Qr .
Finally, in the case of triangles, we choose for reference triangle
√ E the equilateral
triangle with basis e = [−1, 1] × {0} and third vertex at (0, 3). Then we truncate
this triangle by removing the smaller triangle above the line x2 = 1. The remaining
area is a trapezoid, say, T̃ , that is mapped onto the reference square [−1, 1] × [−1, 1]
by an invertible bilinear mapping. The image of p by this transformation, say, p̃,
belongs to Qr , and therefore (2.9) is valid for p̃ in the square. Then passing back to
the trapezoid T̃ , we obtain
p 0,e ≤ Cr p 0,T̃ ≤ Cr p 0,E .
2.2. Problem and nonsymmetric bilinear form. Let the boundary of the
domain ∂Ω be the union of two disjoint sets ΓD and ΓN . We denote ν the unit normal
1
vector to each edge of ∂Ω exterior to Ω. For f given in L2 (Ω), p0 given in H 2 (ΓD ),
and g given in L2 (ΓN ), we consider the following elliptic problem:

−∇ · (K∇p) + αp = f in Ω,
(2.10) p = p0 on ΓD ,
K∇p · ν = g on ΓN .
With the above assumptions on K and α, problem (2.10) has a unique solution in
H 1 (Ω) when |ΓD | > 0 or when α = 0. On the other hand, when ∂Ω = ΓN and
α = 0, problem (2.10)
has a solution in H 1 (Ω) which is unique up to an additive
constant, provided Ω f = − ∂Ω g (see [11]) . For K in W 1,4 (Eh ) and ψ, φ ∈ H 2 (Eh ),
we consider the nonsymmetric bilinear form
Nh

aN S (ψ, φ) = (K∇ψ∇φ + αψφ)
j=1 Ej
Ph
Ph

− {K∇ψ · ν k } [φ] + {K∇φ · ν k } [ψ]
k=1 ek k=1 ek

− (K∇ψ · ν k )φ + (K∇φ · ν k )ψ.
ek ∈ΓD ek ek ∈ΓD ek
We define the linear form

L(φ) = (f, φ) + (K∇φ · ν k )p0 + φg.
ek ∈ΓD ek ek ∈ΓN ek
2.3. Finite element schemes. First we introduce the following interior and
boundary penalty term:
Ph
rσk
rσk
J0σ,β (φ, ψ) = [φ][ψ] + φψ,
|ek |β ek |ek |β ek
k=1 ek ∈ΓD
where σ is a discrete positive function that takes the constant value σk on the edge
or face ek and is bounded below by σ0 > 0, above by σm , |ek | denotes the measure of
ek , and β ≥ 12 is a real number. The Galerkin approximation P NIPG in Dr (Eh ) solves
the following discrete problem:
aN S (P NIPG , v) + J0σ,β (P NIPG , v) = L(v)

(2.11) rσk
+ p0 v ∀ v ∈ Dr (Eh ).
|ek |β ek
ek ∈ΓD
The next lemma establishes the consistency of this scheme.

Lemma 2.2. Under the above assumptions on the data (including K), and if the
solution p of problem (2.10) satisfies p ∈ H 2 (Eh ), then p satisfies
σ,β
rσk
(2.12) aN S (p, v) + J0 (p, v) = L(v) + p0 v ∀ v ∈ H 2 (Eh ).
|ek |β ek
ek ∈ΓD
ERROR ESTIMATES 907
Conversely, if p ∈ H 1 (Ω) ∩ H 2 (Eh ) satisfies (2.12), then p is the solution of problem

(2.10).
Proof. Since (2.10) is a linear problem, we can proceed as follows. First we observe
that (2.12) has at most one solution p in H 2 (Eh ) (or H 2 (Eh )/R if |ΓD | = 0 and α ≡ 0).
Indeed, if p solves (2.12) with zero data, then p is constant in each Ej and [p] = 0 on
each interior edge ek . Hence p is constant in Ω and if |ΓD | > 0 or α = 0, then this
constant is zero. Otherwise, p is unique in H 2 (Eh )/R. Thus, it suffices to prove that
if the unique solution p of (2.10) belongs to H 2 (Eh ), then it also solves (2.12). For
this, let v be an element in H 2 (Eh ). We multiply the first equation of (2.10) by v,
integrate on Ej , and sum all over j.
Nh
Ph

(K∇p∇v + αpv) − {K∇p · ν k }[v] − (K∇p · ν)v = (f, v).
j=1 Ej k=1 ek ∂Ω
Using the Neumann boundary conditions, we get

Nh
Ph

(K∇p∇v + αpv) − {K∇p · ν k } [v] − (K∇p · ν k )v
j=1 Ej k=1 ek ek ∈ΓD ek

= (f, v) + gv.
ek ∈ΓN ek

We add ek ∈ΓD ek (K∇v ·ν k )p and ek ∈ΓD |erσk |kβ ek pv to both sides, use the Dirich-
let boundary condition, and note that [p] = 0. We clearly have (2.12).
We note that on each element, the mass conservation for the NIPG method can
be written as

rσk
αP NIPG − {K∇P NIPG · ν ∂Ej } + [P NIPG ][1]
Ej ∂Ej \ΓN |ek |β ek
ek ∈∂Ej \∂Ω

rσk NIPG rσk
+ P = f + g + p0 .
|ek |β ek Ej ∂Ej ∩ΓN |ek |β ek
ek ∈∂Ej ∩ΓD ek ∈∂Ej ∩ΓD
The constrained discrete space is defined as follows:

 
 Nh

Dr∗ (Eh ) = v ∈ Pr (Ej ) : [v] = 0 ∀ k = 1, . . . , Ph .
 ek 
j=1
The discrete approximation P NCG in Dr∗ (Eh ) satisfies
(2.13) aN S (P NCG , v) = L(v) ∀ v ∈ Dr∗ (Eh ).
The consistency of this scheme is a consequence of Lemma 2.2.

The discontinuous Galerkin approximation P DG in Dr (Eh ) satisfies the formula-
tion
(2.14) aN S (P DG , v) = L(v) ∀ v ∈ Dr (Eh ).
The fact that this scheme is consistent with problem (2.10) has been shown by
Baumann [6].
Lemma 2.3. If r ≥ 1, the discrete solution of each of the three methods exists
and is unique, with one exception: if α is not bounded away from zero, then in the
DG scheme, we assume that r ≥ 2 and the elements are not quadrilaterals.
Proof. Indeed, since it is a square system of linear equations in finite dimension,

it suffices to show uniqueness of the solution.
In the NIPG case, choose f = 0, p0 = 0, g = 0, and v = P NIPG in (2.11). Then
NIPG
P is piecewise constant on Eh . The “interior” penalty terms imply that P NIPG is
one constant over Ω and the boundary penalty term allows us to conclude that P NIPG
is the zero constant if |ΓD | > 0. If |ΓD | = 0 and α = 0, then we can also conclude
that P NIPG is zero. Otherwise, P NIPG is unique up to an additive constant.
In the DG case, choose f = 0, p0 = 0, g = 0, and v = P DG in (2.14). Then, P DG
is piecewise constant on Eh . If α is bounded below by a positive constant, then P DG
is the zero constant on Ω. If α is not bounded below by a positive constant, then in
order to prove that P DG is globally constant in Ω,
we need to construct a test function
v on a given element Ej such that the quantity ek K∇v · ν k is given on one edge (or
face) of Ej and vanishes on the other edges (or faces). If K is constant in each Ej ,
one can construct such a test function on a triangle, parallelogram, or tetrahedron,
by following a similar argument as in Theorem 5.1. If K is not constant in each Ej ,
we assume that h is small enough and the existence of this test function follows from
Corollary 5.2. Then, the proof ends as in the NIPG case.
Finally, in the NCG case, choose f = 0, p0 = 0, g = 0, and v = P NCG in (2.13).
Then, P NCG is piecewise constant on Eh . The constraint on the discrete space implies
that P NCG is globally constant over Ω and we finish the proof as above.
Note that if |ΓD | = 0 and α ≡ 0, then for the three methods, the discrete solution
is unique up to an additive constant.
Remark. The statement of Lemma 2.3 is possibly true for the DG scheme on arbi-
trary quadrilaterals when α is not bounded away from zero; however, the construction
of an adequate test function is more problematic. On the other hand, the statement
is not true for the DG scheme with r = 1 and α not bounded away from zero.
3. A priori error estimates for NIPG method. In this section, we derive a
priori energy error estimates for the problem with mixed boundary conditions. These
estimates are optimal both in h and r if the continuous polynomials are contained
in the discrete “broken” space (which is the case for triangles in two dimensions and
tetrahedra in three dimensions). In the case of the Neumann problem (ΓN = ∂Ω) on
triangles or tetrahedra, one can obtain fully optimal hr convergence results for any
β ≥ (n − 1)−1 , n = 2, 3. In Theorem 3.2, L2 error estimates are proved in the case of
pure Neumann boundary conditions and they are optimal in h and suboptimal in r.
Theorem 3.1. Under the assumptions of Lemma 2.2 and if p ∈ H s (Eh ) and
β = (n − 1)−1 , we have the following: If α ≡ 0, then
1 1 hµ−1
|||K 2 ∇(P NIPG − p)|||0 ≤ C ,K |||p|||s .
σ0 rs−1−δ
If α ≥ α0 > 0, then
1 hµ−1
|||P NIPG − p|||1 ≤ C , K, α ∞ |||p|||s ,
σ0 rs−1−δ
h2µ−2
(3.1) J0σ,β (P NIPG − p, P NIPG − p) ≤ C |||p|||2s ,
r2s−2−2δ
ERROR ESTIMATES 909
where µ = min(r + 1, s), r ≥ 1, s ≥ 2, and δ = 0 in the case of triangles or tetrahedra

when |ΓD | = 0. In the general case, δ = 1/2. In the case of the pure Neumann
problem on triangles or tetrahedra, these results are valid for any β ≥ (n − 1)−1 .
Proof. From Lemma 2.2, we have

rσk
aN S (p, v) + J0σ,β (p, v) = L(v) + p0 v ∀ v ∈ Dr (Eh ).
|ek |β ek
ek ∈ΓD
Let p̃ be an interpolant of p having optimal hp-approximation errors (2.2)–(2.4) and

denote χ = P NIPG − p̃. We have, by consistency and definition,
aN S (χ, χ) + J0σ,β (χ, χ) = aN S (p − p̃, χ) + J0σ,β (p − p̃, χ)
Nh

= (K∇(p − p̃)∇χ + α(p − p̃)χ)
j=1 Ej
Ph
Ph

− {K∇(p − p̃) · ν k }[χ] + {K∇χ · ν k }[p − p̃]
k=1 ek k=1 ek

− (K∇(p − p̃) · ν k )χ + (K∇χ · ν k )(p − p̃)
Ph
rσk
rσk
(3.2) + [p − p̃][χ] + (p − p̃)χ.
|ek |β ek |ek |β ek
k=1 ek ∈ΓD
The first two terms in (3.2) can be bounded in the following way:

Nh

K∇(p − p̃) · ∇χ ≤ C|||∇(p − p̃)|||0 |||K 2 ∇χ|||0 ,
1

j=1 Ej
1 1
≤ |||K 2 ∇χ|||20 + C|||∇(p − p̃)|||20 .
6
Nh
1
α(p − p̃)χ ≤ α 2
p − p̃
1
α 2 χ 0,
∞ 0
j=1 Ej
1 1 2 1
≤ α2 χ 0 + α ∞ p − p̃ 20 .
2 2
The third term is bounded by
1 1
|ek |β 2
rσk 2
{K∇(p − p̃) · ν k }[χ] ≤ {K∇(p − p̃) · ν k } [χ] 0,ek ,
rσk
0,ek
|ek |β
ek
P
h 1 Ph
|ek |β

{K∇(p − p̃) · ν k }[χ] ≤ J0σ,β (χ, χ) + C {K∇(p − p̃) · ν k } 2
0,ek .
ek 8 rσk
k=1 k=1
As ek is an interior edge, we assume that ek = ∂E 1 ∩ ∂E 2 , where E 1 and E 2 are

elements of Eh ; we set Ek12 = E 1 ∪ E 2 and h̄k = max(diam(E 1 ), diam(E 2 )). Using the
approximation result (2.4), we induce
Ph
Ph
|ek |β 2 C h̄2µ−3
{K∇(p − p̃) · ν k } 0,ek ≤ |ek |β k2s−2 |||p|||2s,E 12 .
rσk σ0 r k
k=1 k=1
Thus, the final bound for the third term in (3.2) is

P
1 C
Ph
h̄2µ−3
h

(3.3) {K∇(p − p̃) · ν k }[χ] ≤ J0σ,β (χ, χ) + |ek |β k2s−2 |||p|||2s,E 12 .
ek 8 σ0 r k
k=1 k=1
In the case of triangles or tetrahedra, we can choose a continuous p̃ and for the
pure Neumann problem, there are no other terms; thus we can conclude by choosing
β ≥ (n − 1)−1 . In the general case, to bound the fourth term in (3.2), we deduce
similarly by using the approximation result (2.3) and the inverse inequality (2.8)

{K∇χ · ν k }[p − p̃] ≤ {K∇χ · ν k } 0,e [p − p̃] 0,e
k k
ek
µ− 12
r 1 h̄k
≤C 1 |||K ∇χ|||0,Ek12
2
1 |||p|||s,Ek12 ,
h̄k2 rs− 2
P
h 1 h2µ−2
1
(3.4) {K∇χ · ν k }[p − p̃] ≤ |||K 2 ∇χ|||20 + C 2s−3 |||p|||2s .
ek 6 r
k=1
The interior penalty terms in (3.2) are bounded by using (2.3)

P
h
rσk Ph
rσk 1

[p − p̃][χ] ≤ C [p − p̃] 2
0,ek + J0σ,β (χ, χ)
|ek |β ek |ek |β 8
k=1 k=1
Ph
h̄2µ−1 1
(3.5) ≤C k
|ek |−β |||p|||2s,E 12 + J0σ,β (χ, χ).
r2s−2 k 8
k=1

The boundary terms | ek ∈ΓD ek (K∇(p − p̃) · ν k )χ|, | ek ∈ΓD ek (K∇χ · ν k )(p −

p̃)|, and | ek ∈ΓD |eσkk|β ek (p − p̃)χ| have similar bounds as the interior terms, using,
respectively, (3.3), (3.4), and (3.5). By combining the bounds together, we obtain in
the general case
1 1 h2µ−2 h2µ
|||K 2 ∇χ|||20 + α 2 χ 2
0 + J0σ,β (χ, χ) ≤ C 2s−2
|||p|||2s + C α ∞ 2s |||p|||2s ,
r r
Ph
2µ−3
h̄ h̄2µ−1
+C |ek |β k2s−3 + |ek |−β k2s−2 |||p|||2s,E 12
r r k
k=1
h2µ−3 2µ−1
−β hk
(3.6) +C |ek |β k
+ |ek | |||p|||2s,Ek .
r2s−3 r2s−2
ek ∈ΓD
Thus, if β = (n − 1)−1 , then we obtain optimal convergence rates with respect

to h.
Remark. When |ΓD | > 0 and p0 is a polynomial of degree r, for instance, p0 = 0,
in the case of triangles or tetrahedra, the continuous interpolant p̃ can be constructed
so that p̃ = p0 on ΓD . Then, the statement of Theorem 3.1 is valid for δ = 0 and any
β ≥ (n − 1)−1 .
In the following theorem, we assume that ΓN = ∂Ω and that the subdivision of
Ω consists of triangles or tetrahedra.
ERROR ESTIMATES 911
Theorem 3.2. Assume that Ω is convex, ΓN = ∂Ω, and K is sufficiently smooth

so that for any f ∈ L2 (Ω), the solution φ of the dual problem
−∇ · (K∇φ) + αφ = f in Ω,
K∇φ · ν = 0 on ∂Ω
belongs to H 2 (Ω), with continuous dependence on f . Then, in the case of triangles or

tetrahedra, for any β ≥ (n − 1)−1 ,
1 3 β
P NIPG − p 0,Ω ≤C 1 hmin(µ,µ− 2 + 2 (n−1)) |||p|||s ,
rs− 2
where µ = min(r +1, s) for r ≥ 1, s ≥ 2, and C is independent of h, r, p. In particular,
optimal L2 rates of convergence with respect to h are obtained if β ≥ 3 for n = 2 and
if β ≥ 32 for n = 3.
Proof. Consider the dual problem
−∇ · (K∇φ) + αφ = P NIPG − p in Ω,
K∇φ · ν = 0 on ∂Ω.
By assumption, φ ∈ H 2 (Ω) and there is a constant C that depends on Ω such that
(3.7) φ 2,Ω ≤ C P NIPG − p 0,Ω .
Note that if α ≡ 0, then since both P NIPG and p are defined up to an additive
constant, this constant can be chosen so that P NIPG − p has zero mean value. This
compatibility condition guarantees that the dual problem has a unique solution.
Denote χ = P NIPG − p. Then
2
χ 0,Ω = (−∇ · (K∇φ) + αφ, χ).
Integrating by parts on each element yields

Nh
Nh

2
χ 0,Ω = (K∇φ∇χ + αφχ) − (K∇φ · ν)χ
j=1 Ej j=1 ∂Ej
Nh
Ph

= (K∇φ∇χ + αφχ) − {K∇φ · ν k }[χ]
j=1 Ej k=1 ek
because the regularity of φ and K implies that the jumps [K∇φ · ν k ]|ek = 0. By
subtracting the orthogonality equation for any φ∗ ∈ Dr (Eh )
aN S (χ, φ∗ ) + J0σ,β (χ, φ∗ ) = 0,
and using the symmetry of K, the regularity of φ and K, and choosing for φ∗ a
continuous interpolant of φ satisfying (2.2)–(2.4), we obtain
Nh

2
χ 0,Ω = (K∇(φ − φ∗ )∇χ + α(φ − φ∗ )χ)
j=1 Ej
Ph
Ph

+ {K∇(φ − φ∗ ) · ν k }[χ] − 2 {K∇φ · ν k }[χ].
k=1 ek k=1 ek
The first two terms are bounded in the following fashion by using (2.2) and (3.7):

Nh Nh


K∇(φ − φ∗
)∇χ ≤C 1
φ − φ∗ 1,Ej K 2 ∇χ 0,Ej

j=1 Ej j=1
h 1
φ 2,Ω |||K 2 ∇χ|||0
≤C
r
h 1
≤ C χ 0,Ω |||K 2 ∇χ|||0 ;
r
Nh
2
α(φ − φ∗
)χ ≤ C α ∞ h φ 2,Ω χ 0,Ω
r2
j=1 Ej
h2
≤C α ∞ 2 χ 0,Ω |||χ|||1 .
r
By the Cauchy–Schwarz inequality, by (3.1), for any β ≥ (n − 1)−1 , and by properties
(2.4) and (3.7), we have
P
h Ph
|ek |β
1 1
2−2
∗
{K∇(φ − φ ) · ν k }[χ] ≤ {K∇(φ − φ∗ ) · ν k } 0,ek [χ] 0,ek
ek rσk
k=1 k=1
1
Ph
2
1 |ek |β
≤ J0σ,β (χ, χ) 2 {K∇(φ − φ∗ ) · ν k } 2
0,ek
rσk
k=1
β
µ− 12 2 (n−1)
h h
≤C |||p|||s χ 0,Ω .
rs
Similarly, by Cauchy–Schwarz, (3.1), and (2.6),
P
h Ph
|ek |β
1
2
rσk
1
2
{K∇φ · ν k }[χ] ≤ {K∇φ · ν k } 0,ek [χ] 0,ek

ek rσk |ek |β
k=1 k=1
1
Ph
2
1 |ek |β
≤ J0σ,β (χ, χ) 2 {K∇φ · ν k } 2
0,ek
rσk
k=1
β
µ− 32 2 (n−1)
h h
≤C 1 |||p|||s χ 0,Ω .
rs− 2
Combining the previous bounds and Theorem 3.1 applied in the pure Neumann case,
we have for any β ≥ (n − 1)−1 that
3 β
hµ− 2 + 2 (n−1) hµ hµ
χ 0,Ω ≤C + |||p|||s + C α ∞ |||p|||s ,
1
rs− 2 rs rs
which concludes the proof.

Remark. On parallelograms or quadrilaterals, we cannot use a continuous inter-
polant, and therefore the values of β in Theorems 3.1 and 3.2 are incompatible. Of
course, we can have a continuous interpolant on the function space Qk , but the scheme
is more costly.
ERROR ESTIMATES 913
On the other hand, for the pure Dirichlet problem, even on triangles or tetrahedra,
we must choose β = (n − 1)−1 in order to establish Theorem 3.1, and these are
incompatible with the values of β in Theorem 3.2. However, if the Dirichlet datum
p0 on ∂Ω is a polynomial of degree r, in the case of triangles or tetrahedra, then the

statement of Theorem 3.2 is valid.
Finally, it is well known that the duality argument of Theorem 3.2 is not well
adapted to mixed boundary conditions (except in very particular cases) because the
dual problem may have solutions that do not belong to H 2 (Ω).
4. A priori error estimates for the NCG method. In this section, we derive
an error estimate for the H 1 norm and the L2 norm that are both h-optimal for the
constrained Galerkin method applied to the pure Neumann problem.
Theorem 4.1. Under the assumptions of Lemma 2.2 and, if p ∈ H s (Eh ), we
have
hµ−1
|||P NCG − p|||1 ≤ C(K, α ∞ ) s− 5 |||p|||s ,
r 2
where µ = min(r + 1, s), C is independent of h, r, p, and r ≥ 1, s ≥ 2.
Proof. We have the following orthogonality equation:
aN S (P NCG − p, v) = 0 ∀ v ∈ Dr∗ (Eh ).
We can show [14] that there is an interpolant p̃ ∈ Dr∗ (Eh ) that satisfies the approxi-
mation properties (2.2)–(2.4). Let χ = P NCG − p̃; then
aN S (χ, χ) = aN S (p − p̃, χ)
Nh

= (K∇(p − p̃)∇χ + αχ(p − p̃))
j=1 Ej
Ph
Ph

(4.1) − {K∇(p − p̃) · ν k }[χ] + {K∇χ · ν k }[p − p̃].
k=1 ek k=1 ek
The first two terms are easily bounded by Cauchy–Schwarz and by using (2.2):

Nh

K∇(p − p̃)∇χ ≤ |||K 2 ∇χ|||0 |||K 2 ∇(p − p̃)|||0
1 1

j=1 Ej
  12
Nh
h2µ−2
j
≤ C|||K 2 ∇χ|||0  p 2s,Ej  .
1
(4.2) 2s−2
j=1
r
In a similar manner, we have

Nh

αχ(p − p̃) ≤ α 2 χ
1 1
α 2 (p − p̃)
0 0
j=1 Ej
  12
1
Nh
h2µ
j
(4.3) ≤C α 2
∞
1
α2 χ 0
 p 2
s,Ej
 .
j=1
r2s
Now we estimate the third term in (4.1), which we denote by A. Let ck be any
constant:
Ph
Ph

A≡ {K∇(p − p̃) · ν k }[χ] = {K∇(p − p̃) · ν k }[χ − ck ].
k=1 ek k=1 ek
We have by Cauchy–Schwarz and (2.4)

1
3 Ph
2
hµ− 2 2
(4.4) |A| ≤ C 3 |||p|||s [χ − ck ] 0,ek .
rs− 2 k=1
The other factor is bounded in the following way; here again, we assume that ek is
the interface between two elements E 1 and E 2 of Eh :
[χ] 0,ek = [χ − ck ] 0,ek ≤ (χ − ck )E 1 0,ek + (χ − ck )E 2 0,ek .
Since P NCG belongs to Dr∗ (Eh ), we have

(χ)E 1 dσ = (χ)E 2 dσ.
ek ek
Therefore, it suffices to estimate (χ − ck )E 1 0,ek with the choice

1
ck = (χ)E 1 dσ.
|ek | ek
1

Passing to the reference element, we note that ck = |ê| ê
χ̂dσ̂ and that the mapping

ˆ ˆ
f → f − 1 ˆ 1
f dσ̂ is continuous on H (ê) and vanishes on constant functions. Thus,
|ê| ê
χ̂ − ck 0,ê
ˆ ê χ̂
≤ Ĉ ∇ 0,ê ,
where ∇ˆ ê denotes the tangential gradient on ê. However, since ∇ˆ ê χ̂ belongs to a

finite-dimensional space, on which all norms are equivalent, and since the subdivision
of Ω is regular, we get the following by applying the inverse estimate (2.7) on the
reference element:
χ̂ − ck ˆ ê χ̂
≤ Ĉr ∇
0,ê 0,Ê .
Therefore,
1 1 1
(χ − ck )E 1 0,ek ≤ Ĉ|ek | 2 h1 |E 1 |− 2 r ∇χ 0,E 1 ≤ Ĉh12 r ∇χ 0,E 1 .
Thus, summing on k, we have

Ph
Ph
Nh

2 2
(4.5) [χ] 0,ek = [χ − ck ] 0,ek ≤C hj r2 ∇χ 2
0,Ej .
k=1 k=1 j=1
Combining (4.4) and (4.5), we obtain a bound for A:
hµ−1 1
(4.6) |A| ≤ C 5 |||p|||s |||K 2 ∇χ|||0 .
rs− 2
ERROR ESTIMATES 915
The last term in (4.1) is bounded as in Theorem 3.1:

P
h hµ−1

1
(4.7) {K∇χ · ν k }[p − p̃] ≤ C|||K 2 ∇χ|||0 s− 3 |||p|||s .
ek r 2
k=1
Combining (4.2), (4.3), (4.6), and (4.7), we obtain

hµ−1 1 1 hµ 1
aN S (χ, χ) ≤ C |||p|||s |||K 2 ∇χ|||0 + C α 2
∞ α 2 χ 0 |||p|||s ,
r s− 25 rs
which concludes the proof.
Remark. If we add the condition on P NCG ,

NCG
(4.8) P = p0 ∀ ek ∈ ΓD ,
ek ek
then we can construct p̃ such that

(P NCG − p̃) = 0 ∀ ek ∈ ΓD ,
ek
and therefore the error terms arising from the Dirichlet boundary condition have the
same estimates as the interior jump terms. Hence, if we impose (4.8), the statement
of Theorem 4.1 is also valid for the problem with mixed boundary conditions.
The next theorem shows that the NCG scheme gives optimal results with respect
to h in the L2 norm. For simplicity, we do not specify the dependence with respect
to r.
Theorem 4.2. Under the assumptions of Theorem 3.2, we have
P NCG − p 0,Ω ≤ Chµ |||p|||s
for r ≥ 1, s ≥ 2, and C independent of h and p.
Proof. As in the proof of Theorem 3.2, we consider the following dual problem:

−∇ · (K∇ψ) + αψ = P NCG − p in Ω,
K∇ψ · ν = 0 on ∂Ω.
Denote χ = P NCG − p. Thus, we have
2
χ 0,Ω = (−∇ · (K∇ψ) + αψ, χ)
Nh
Ph

(4.9) = (K∇ψ∇χ + αψχ) − {K∇ψ · ν k }[χ].
j=1 Ej k=1 ek
Let ψ ∗ be in Dr∗ (Eh ) ∩ C 0 (Ω̄). The orthogonality condition implies that

Nh
Ph
∗ ∗
(4.10) 0= (K∇χ∇ψ + αχψ ) + {K∇ψ ∗ · ν k }[χ].
j=1 Ej k=1 ek
Now, we subtract (4.10) from (4.9) and use the regularity of K∇ψ · ν k on ek :
Nh

χ 20,Ω = (K∇(ψ − ψ ∗ )∇χ + α(ψ − ψ ∗ )χ)
j=1 Ej
Ph
Ph

(4.11) −2 (K∇ψ · ν k )[χ] + {K∇(ψ − ψ ∗ ) · ν k }[χ].
k=1 ek k=1 ek
As in the proof of Theorem 4.1, the first two terms are easily bounded by using
Cauchy–Schwarz, Hölder inequalities, and the approximation property (2.2):

Nh

K∇(ψ − ψ )∇χ ≤ Ch χ 0,Ω |||K 2 ∇χ|||0 ,
∗ 1

j=1 Ej

Nh

α(ψ − ψ )χ ≤ C α ∞ h2 χ 0,Ω |||χ|||1 ,
∗

j=1 Ej
1
and we apply Theorem 4.1 to bound the factor |||K 2 ∇χ|||0 in the first term and |||χ|||1,Ω
in the second term. To bound the third term in (4.11), let p̃ be an element of Dr∗ (Eh )∩
C 0 (Ω̄) and 1c be any constant vector; we can write
Ph
Ph

2 (K∇ψ · ν k )[χ] = 2 (K∇ψ · ν k )[P NCG − p̃]
k=1 ek k=1 ek
Ph

=2 ((K∇ψ − 1c) · ν k )[P NCG − p̃].
k=1 ek
As it was proved in Theorem 4.1, (4.5), and from the approximation result (2.2), we
have
Ph
Nh

[P NCG − p̃] 2
0,ek ≤C hj ∇(P NCG − p̃) 2
0,Ej
k=1 j=1
Nh
Nh

NCG 2 2
≤C hj ∇(P − p) 0,Ej +C hj ∇(p − p̃) 0,Ej
j=1 j=1
≤ Ch2µ−1 |||p|||2s .
On the other hand, by choosing

1
1c = 1 K∇ψ,
|E | E1
we have by applying (2.6)

1
(K∇ψ − 1c) · ν k 0,ek ≤ (K∇ψ − 1c)|E 1 0,ek ≤ Ch12 |K∇ψ|1,E 1 .
By assumption, K ∈ W 1,4 (Eh ) and ψ ∈ H 2 (Eh ). Therefore, K∇ψ belongs to H 1 (Eh )
with
|||K∇ψ|||1 ≤ C χ 0,Ω ;
then
P
h

2 (K∇ψ · ν k )[χ] ≤ Chµ χ 0,Ω |||p|||s .
ek
k=1
Let A denote the last term in (4.11):

1 1
Ph
2 Ph
2
|A| ≤ C {∇(ψ − ψ ∗ ) · ν k } 2
0,ek [χ] 2
0,ek .
k=1 k=1
ERROR ESTIMATES 917
As previously, we have by (2.4) and regularity of ψ

Ph

{∇(ψ − ψ ∗ ) · ν k } 2
0,ek ≤ Ch|||ψ|||22 ≤ Ch χ 2
0,Ω ,
k=1
and according to (4.5),

Ph

2
[χ] 0,ek ≤ Ch|||∇χ|||20 ≤ Ch2µ−1 |||p|||2s .
k=1
Thus, we obtain
|A| ≤ hµ χ 0,Ω |||p|||s .
The theorem is obtained by combining all the above results.

Remark. In the case of the pure Dirichlet problem, i.e., p = p0 on ∂Ω, we can
obtain the same order of convergence in the L2 norm if we impose (4.8) on P NCG .
Indeed, the corresponding dual problem is

−∇ · (K∇ψ) + αψ = P NCG − p in Ω,
ψ = 0 on ∂Ω.
Then, we interpolate ψ by ψ ∗ in Dr∗ (Eh ) ∩ H01 (Ω) and (4.11) is replaced by

Nh

2
χ 0,Ω = (K∇(ψ − ψ ∗ )∇χ + α(ψ − ψ ∗ )χ)
j=1 Ej
Ph
Ph

−2 (K∇ψ · ν k )[χ] + {K∇(ψ − ψ ∗ ) · ν k }[χ]
k=1 ek k=1 ek

−2 (K∇ψ · ν k )χ + {K∇(ψ − ψ ∗ ) · ν k }χ.

Since p̃ can be constructed so that ek χ = 0 for all ek in ΓD , the above boundary
terms have the same estimates as the interior jump terms.
5. A priori error estimate for the DG method. In this section, we derive
an a priori optimal error estimate in two or three dimensions for the problem with
Neumann, or Dirichlet, or mixed boundary conditions, on triangles, parallelograms,
or tetrahedra. We make several additional assumptions:
• K ∈ [W 1,∞ (Ej )]n×n ∀ j = 1, . . . , Nh , uniformly in Ej , i.e., ∃ γ2 > 0,
  12
n
max  ∇kim 2∞,Ej  ≤ γ2 .
Ej
i,m=1
• We denote K̄ = (k̄ij ), where

1
(5.1) k̄ij = kij ∀ E ∈ Eh .
|E| E
It is easy to prove that K̄ is also symmetric positive definite and that the
largest eigenvalue of K̄ is ≤ γ1 and its smallest eigenvalue is ≥ γ0 , where γ1
and γ0 are the constants of (2.1).
a3
e1
e2
E
a2
e3
a1
Fig. 5.1. Triangular element.
We first prove a local approximation result that holds for n = 2 or 3.

Theorem 5.1. Let E be an element of the subdivision Eh , that is, a triangle or a
parallelogram in two dimensions or a tetrahedron in three dimensions. Let hE be the
diameter of E, let p ∈ H s (E), for s ≥ 2, let r ≥ 2, and let K̄ be the constant tensor
defined by (5.1). There exists an interpolant of p, p̃I ∈ Pr (E) satisfying

(5.2) K̄∇(p̃I − p) · ν k = 0 ∀ ek ∈ ∂E,
ek
hµ−i
(5.3) ∇i (p̃I − p) 0,E ≤C E
3 p s,E for i = 0, 1, 2,
rs− 2 −δ
µ− 32
I hE
(5.4) |K̄∇(p̃ − p) · ν k | ≤ C 3 p s,E ,
ek rs− 2
where δ = 0 for i = 0, 1, δ = 12 for i = 2, µ = min(r + 1, s), and C is independent of

hE and r.
Proof. The case of triangles. Let E be a triangle with vertices a1 , a2 , a3 , opposite
sides e1 , e2 , e3 , and unit exterior normal vectors ν 1 , ν 2 , ν 3 as in Figure 5.1. Let
λ1 , λ2 , λ3 be the barycentric coordinates of a1 , a2 , and a3 in E. First, we will show
that
given f in H s (E) with s ≥ 2, there is a polynomial q1 in P2 (E) such that
e1
K̄∇(q1 − f ) · ν 1 = 0, e2 K̄∇q1 · ν 2 = 0, and e3 K̄∇q1 · ν 3 = 0. For this, consider
the polynomial
q1 = 4q1 (a12 )λ1 (1 − λ1 ),
where a12 is the midpoint of e3 = [a1 , a2 ]. It is easy to check that each component of
∇q1 = 4q1 (a12 )∇λ1 (1 − 2λ1 ) has zero mean-value on e2 and e3 , and (∇λ1 )λ1 vanishes
on e1 . Therefore, q1 (a12 ) is determined by the conditions

4q1 (a12 ) K̄∇λ1 · ν 1 = K̄∇f · ν 1 .
e1 e1
However,
ν 1 |e1 |
(5.5) ∇λ1 = − .
2 |E|
Therefore,

1 |E| 1
q1 (a21 ) = − 2
K̄∇f · ν 1 .
2 |e1 | (K̄ν 1 , ν 1 ) e1
ERROR ESTIMATES 919
a3
a4
a2
a1 E
Fig. 5.2. Tetrahedron element.
Hence,

C γ1 1 ∂f
|q1 (a12 )| ≤ K̄∇f · ν 1 ≤ C hE2 .
γ0 e1 γ0 ∂n 0,e1
Therefore, for i = 0, 1, 2,
1 1 1
−i ∂f
(5.6) ∇i q1 0,E ≤ C|E| 2 h−i
E |q1 (a12 )| ≤ C|E| hE
2 2
.
∂n 0,e1
Similarly, we construct polynomials q2 and q3 in P2 (E) such that

K̄∇qδ · ν δ = K̄∇f · ν δ for δ = 2, 3,
eδ eδ

K̄∇q2 · ν 1 = K̄∇q2 · ν 3 = 0,
e1 e3

K̄∇q3 · ν 1 = K̄∇q3 · ν 2 = 0,
e1 e2
and such that the inequalities of (5.6) hold for q2 and q3 , with respect to e2 and e3 .
Let q = q1 + q2 + q3 , constructed with f = p − pI , where pI is an approximation of p
satisfying (2.2)–(2.4), and set p̃I = q + pI . Then p̃I satisfies (5.2), and applying (2.4)
to (5.6) we derive, for i = 0, 1, 2,
|p̃I − p|i,E ≤ |q|i,E + |pI − p|i,E
hµ−i
≤C E
3 p s,E + |pI − p|i,E ,
rs− 2
which has the same order of approximation as |pI − p|i,E . Similarly,

|K̄∇(p̃I − p) · ν k | ≤ |K̄∇q · ν k | + |K̄∇(pI − p) · ν k |,
ek ek ek
and applying (2.4) to the second term, (2.6) to the first term, and the above estimate
for |q|i,E , we obtain (5.4).
The case of tetrahedra. The situation is much the same for tetrahedra. Let E be
a tetrahedron with vertices a1 , a2 , a3 , a4 , opposite faces e1 , e2 , e3 , e4 , and unit exterior
normal vectors ν 1 , ν 2 , ν 3 , ν 4 as in Figure 5.2. Let λ1 , λ2 , λ3 , λ4 be the barycentric
s
coordinates of a1 , a2 , a3 , and a4 in E. Again, we will show that given f in H (E)
with s ≥ 2, there is a polynomial q1 in P2 (E) such that e1 K∇(q1 − f ) · ν 1 = 0,

eδ
K∇q1 · ν δ = 0 for δ = 2, 3, 4. For this, consider the polynomial
q1 = q1 (a1 )λ1 (3λ1 − 2).

υ2 a3
a4 e2
υ1
e1
υ3
E
e3
e4 a2
υ4
a1
Fig. 5.3. Parallelogram element.
It is easy to check that each component of ∇q1 ,
∇q1 = q1 (a1 )∇λ1 (6λ1 − 2),
has zero mean-value on e2 , e3 , e4 and (∇λ1 )λ1 vanishes on e1 . Therefore, q1 (a1 ) is

determined by the condition

−2q1 (a1 ) K̄∇λ1 · ν 1 = K̄∇f · ν 1 ,
e1 e1
and since (5.5) is valid in three dimensions, this holds if

|E| 1
q1 (a1 ) = K̄∇f · ν 1 .
|e1 |2 (K̄ν 1 , ν 1 ) e1
Hence
γ1 ∂f
|q1 (a1 )| ≤ C ,
γ0 ∂n 0,e1
and for i = 0, 1, 2, we obtain the analogue of (5.6):
1 1 ∂f
∇i q1 0,E ≤ C|E| 2 h−i −i
E |q1 (a1 )| ≤ C|E| hE
2 .
∂n 0,e1
Then, the proof finishes as in the triangular case.

The case of parallelograms. Let E be a parallelogram with vertices a1 , a2 , a3 , a4 ,
opposite sides e1 , e2 , e3 , e4 , and unit exterior normal vectors ν 1 , ν 2 , ν3 , ν 4 as in Fig-
ure 5.3. We want to construct a polynomial q ∈ P2 (E) such that e3 K̄∇q · ν 3 =

e3
K̄∇f · ν 3 and eδ K̄∇q · ν δ = 0 for δ = 1, 2, 4. For this, we replace the barycen-
tric coordinates with the functions φi ∈ P1 (E), 1 ≤ i ≤ 4, defined by φi |ei = 0 for
1 ≤ i ≤ 4 and φ1 |e3 = 1, φ2 |e4 = 1, φ3 |e1 = 1, φ4 |e2 = 1. Then, we set
4
1
q= 2qi φi φi − , qi ∈ R,
i=1
2
and we want to adjust the numbers qi so that q satisfies the above conditions. Con-
sidering that E is a parallelogram, it is easy to check that
|ei |
∇φi = − ν i.
|E|
ERROR ESTIMATES 921
Set c3 = |e3 |(K̄ν 3 , ν 3 ), c4 = |e4 |(K̄ν 4 , ν 4 ), c34 = |e3 |(K̄ν 3 , ν 4 ), c43 = |e4 |(K̄ν 3 , ν 4 ).
Then the numbers qi must satisfy the following system of equations:

|e3 |
(3c3 q1 + c43 q2 + c3 q3 − c43 q4 ) = K̄∇f · ν 3 ,
|E| e3
c3 q1 − c43 q2 + 3c3 q3 + c43 q4 = 0,
−c34 q1 + c4 q2 + c34 q3 + 3c4 q4 = 0,
c3 q1 + 3c4 q2 − c34 q3 + c4 q4 = 0.
In order to solve this system of equations and derive an upper bound for qi , set
α34 = (K̄ν 3 , ν 3 )(K̄ν 4 , ν 4 ) − (K̄ν 3 , ν 4 )2 ,
and observe that
1 1
α34 = K̄ 2 ν 3 2
K̄ 2 ν 4 2
sin2 θ,
1 1
where θ is the angle between K̄ 2 ν 3 and K̄ 2 ν 4 . Owing to the regularity of Eh , this
angle is bounded below for all E ∈ Eh by a fixed angle θ0 > 0:
1 1
∀ E ∈ Eh , α34 ≥ K̄ 2 ν 3 2
K̄ 2 ν 4 2
sin2 θ0 ≥ γ02 sin2 θ0 .
Then, the solution of the above system is
b |E| (K̄ν 3 , ν 4 )
q4 = , q2 = −q4 ,
2 |e3 ||e4 | α34
b |E| 1 (K̄ν 3 , ν 4 )2 1
q3 = − 2
+ ,
2 |e3 | (K̄ν 3 , ν 3 ) α34 4
b |E| 1 (K̄ν 3 , ν 4 )2 3
q1 = 2
+ ,
2 |e3 | (K̄ν 3 , ν 3 ) α34 4

where b = e3
K̄∇f · ν 3 . We can easily check that
C γ12 1 1 ∂f
|q2 | = |q4 | ≤ 2 2 hE2 ,
2 γ0 sin θ0 ∂n 0,e3
C γ1 12 γ12 1 ∂f
|q3 | ≤ h + ,
2 γ0 E 2 2
γ0 sin θ0 4 ∂n 0,e3
C γ1 12 γ12 3 ∂f
|q1 | ≤ h + .
2 γ0 E γ02 sin2 θ0 4 ∂n 0,e3
From these bounds, we immediately deduce that ∇i q satisfies (5.6) and the proof
finishes as in the triangular case.
The next corollary extends the result of Theorem 5.1 to nonconstant tensors.
Corollary 5.2. We retain the notation of Theorem 5.1. Let K be a tensor-
valued function in [W 1,∞ (E)]n×n . There exists h0 > 0 independent of E such that
for all hE ≤ h0 , there exists an interpolant of p, p̃I ∈ Dr (Eh ) satisfying

K∇(p̃I − p) · ν k = 0 ∀ ek ∈ ∂E,
ek
hµ−i
∇i (p̃I − p) 0,E ≤C E
3 p s,E , i = 0, 1, 2,
rs− 2 −δ
µ− 32
I hE
|K∇(p̃ − p) · ν k | ≤ C 3 p s,E ,
ek rs− 2
where δ = 0 if i = 0, 1, δ = 12 if i = 2, µ = min(r + 1, s), and C is independent of h

and r.
Proof. Let K̄ be defined by (5.1). Let
K − K̄ ∞,E = sup K(x) − K̄ ,

x∈E
where · denotes the matrix norm subordinated to the Euclidean norm. It is easy
to check that
sup K(x) − K̄ ≤ sup K(x) − K̄ S,

x∈E x∈E
n
where K(x) 2S = i,j=1 kij 2
(x). Since k̄ij is also the average of k̂ij = kij ◦ F on the
reference square Ê, the mapping k̂ij → k̂ij − k̄ij vanishes if k̂ij is a constant function.
Thus,
kij − k̄ij ∞,E

ˆ k̂ij
≤C ∇ ≤ ChE ∇kij ∞,E ∀ E ∈ Eh .
∞,Ê
Therefore,
(5.7) K − K̄ ∞,E ≤ C∞ hE ∇K ∞,E .
In the course of the proof of Theorem 5.1, for any f ∈ H s (E), we have constructed a
polynomial q ∈ P2 (E) such that

∀ e ∈ ∂E, K̄∇q · ν = K̄∇f · ν.
e e
We propose to achieve the same result, with K instead of K̄, by means of the following
iterative procedure. Starting with q 0 = 0, construct q m ∈ P2 (E), as in Theorem 5.1,
the solution of

(5.8) ∀ e ∈ ∂E, K̄∇q · ν = K∇f · ν − (K − K̄)∇q m−1 · ν.
m
e e e
m
Theorem 5.1 guarantees the existence of q ; moreover, combining (5.6), (5.7), and
(5.8) we obtain in the two-dimensional (2D) case
1 −1
(5.9) ∇q m 0,E ≤ C2 |E| 2 hE 2 ( ∇f · ν 0,∂E + C∞ hE ∇K ∞,E ∇q m−1 · ν 0,∂E )
and in the three-dimensional (3D) case

1
∇q m 0,E ≤ C3 |E| 2 h−1
E ( ∇f · ν 0,∂E + C∞ hE ∇K ∞,E ∇q m−1 · ν 0,∂E )
with constants independent of m, h, and E. Let us prove by induction on m that for

small enough hE ,

1 −1
(5.10) m
∇q 0,E ≤ 2C 2 |E| 2 h 2 ∇f · ν
E 0,∂E in two dimensions,
1 −1
2C3 |E| hE ∇f · ν 0,∂E in three dimensions.
2
This is trivially true for m = 0. Assume that (5.10) holds for m − 1. Applying (2.8),
we have in the 2D case
1 −1 1
∇q m 0,E ≤ C2 |E| 2 hE 2 ( ∇f · ν 0,∂E + CC∞ hE2 ∇K ∞,E ∇q m−1 0,E )
ERROR ESTIMATES 923
with a similar inequality in the 3D case. Therefore, in both cases, there exists a
constant C0 that depends on the above constants but is independent of h, m, and E,
such that if
1
(5.11) h0 =
2C0 ∇K ∞,E
and hE ≤ h0 , then the perturbation is bounded by ∇q m−1 0,E /2. But either
∇q m−1 0,E ≤ ∇q m 0,E , in which case (5.10) is established for ∇q m 0,E , or
∇q m 0,E ≤ ∇q m−1 0,E
and the induction hypothesis implies (5.10). Therefore, in both cases, (5.10) is proven
by induction. Furthermore, substituting (5.10) and (2.8) into the analogue of (5.9)
for ∇i q m , we easily derive that, for i = 0, 1, 2,
1
1 −i
i m
∇ q 0,E ≤ C|E| 2 h2
E ∇f · ν 0,∂E in two dimensions,
1 −i
C|E| hE ∇f · ν 0,∂E in three dimensions.
2
It follows from these bounds, uniform with respect to m, that a subsequence of q m

(still denoted by q m ) converges to a polynomial of P2 (E), and passing to the limit
with respect to m in (5.8), q satisfies

∀ e ∈ ∂E, K∇q · ν = K∇f · ν.
e e
The estimates for q follow from the previous bounds and the argument of Theorem
5.1.
Corollary 5.3. Let h ≤ h0 , where h0 is defined by (5.11), let Eh consist of
triangles, parallelograms, or tetrahedra (in three dimensions), and let p ∈ H s (Eh ), for
s ≥ 2, and let r ≥ 2. There exists an interpolant of p, p̃I ∈ Dr (Eh ) satisfying

(5.12) {K∇(p̃I − p) · ν k } = 0 ∀ k = 1, . . . , Ph ,
ek

(5.13) K∇(p̃I − p) · ν k = 0 ∀ ek ∈ ΓD ,
ek
hµ−i
(5.14) |||∇i (p̃I − p)|||0 ≤ C 3 |||p|||s , i = 0, 1, 2,
rs− 2 −δ
Ph
Mh
3
hµ− 2
(5.15) |{K∇(p̃I − p) · ν k }| + |K∇(p̃I − p) · ν k | ≤ C(K) 3 ,
k=1 ek k=Ph +1 ek rs− 2
where δ = 0 if i = 0, 1, δ = 12 if i = 2, and µ = min(r + 1, s), and C is independent

of h and r.
Proof. The proof is a straightforward application of Corollary 5.2.
Theorem 5.4. Assume Ω ⊂ Rn for n = 2, 3. If α ≡ 0, then there is a constant
C independent of h, r, p such that for s ≥ 2 and r ≥ 2
1 hµ−1
|||K 2 ∇(p − P DG )|||0 ≤ C(K) 5 |||p|||s .
rs− 2
If α ≥ α0 > 0, then the following inequality holds:
hµ−1
|||p − P DG |||1 ≤ C(K, α ∞ ) s− 5 |||p|||s ,

r 2
where µ = min(r + 1, s).
Proof. The difference between P DG and p satisfies
aN S (P DG − p, v) = 0 ∀ v ∈ Dr (Eh ).
We take v = P DG − p̃I , where p̃I is the interpolant of p constructed in Corollary 5.3

and denote χ = P DG − p̃I . We have
Nh

aN S (χ, χ) = (K∇(p − p̃I ) · ∇χ + α(p − p̃I )χ)
j=1 Ej
Ph
Ph

− {K∇(p − p̃I ) · ν k }[χ] + {K∇χ · ν k }[p − p̃I ]
k=1 ek k=1 ek

− (K∇(p − p̃I ) · ν k )χ + (K∇χ · ν k )(p − p̃I ).
In view of (5.12) and (5.13), we can write

Nh

aN S (χ, χ) = (K∇(p − p̃I ) · ∇χ + α(p − p̃I )χ)
j=1 Ej
Ph
Ph

− {K∇(p − p̃I ) · ν k }([χ] − ck ) + {K∇χ · ν k }[p − p̃I ]
k=1 ek k=1 ek

I
(5.16) − (K∇(p − p̃ ) · ν k )(χ − ck ) + (K∇χ · ν k )(p − p̃I ),
where ck is chosen as follows: as previously, we assume that ek = ∂E 1 ∩ ∂E 2 , where

E 1 and E 2
are elements of Eh adjacents to ek ; then we take ck = c1 − c2 , where
1
ci = |E i | E i χ, i = 1, 2. Then
[χ] − ck 0,ek ≤ χ|E 1 − c1 0,ek + χ|E 2 − c2 0,ek
and applying (2.5) and (2.2),

− 12 1 1
χ|E 1 − c1 0,ek ≤ C(h1 χ − c1 0,E 1 + h12 |χ|1,E 1 ) ≤ Ch12 |χ|1,E 1 .
Hence,
1 1
[χ] − ck 0,ek ≤ C(h12 |χ|1,E 1 + h22 |χ|1,E 2 ).
Therefore, (5.15) gives

Ph
hµ−1 1
{K∇(p − p̃I ) · ν k }([χ] − ck ) ≤ C(K) 3 |||K 2 ∇χ|||0 |||p|||s .
k=1 ek rs− 2
ERROR ESTIMATES 925
The first two terms and the fourth term are estimated as in Theorem 4.1 (the same
result as in (3.4) except for the dependence with respect to r that becomes r2s−5 ).
Finally, the boundary terms are estimated as the interior terms, and the theorem is
obtained by combining all the results together.

So far, we do not know how to derive a sharper error estimate for the DG method
in the L2 norm. As we have seen in Theorem 5.4, if α is bounded away from zero, the
error in the H 1 norm gives automatically a bound for the error in the L2 norm, but
we lose a power of h with respect to the interpolation error.
The next proposition extends this result in two dimensions for the pure Neumann
problem to the case where α is not bounded away from zero. The same argument can
be applied to the pure Dirichlet problem. In the case of mixed boundary conditions,
the same result also holds, provided the corresponding dual problem satisfies the
regularity assumption of Proposition 5.5.
Proposition 5.5. Let the dimension n = 2. In addition to the assumptions
4
made at the beginning of this section, suppose K is such that for any f in L 3 (Ω), the
solution φ ∈ H 1 (Ω) of the dual problem defined as
(5.17) −∇ · K∇φ + αφ = f in Ω,
(5.18) K∇φ · ν = 0 on ∂Ω
4
belongs to W 2, 3 (Eh ), with continuous dependence on f . Then,
hµ−1 1
P DG − p 0,Ω ≤C 5 (C1 + C2 h 2 r)
rs− 2
for r ≥ 2, s ≥ 2; C1 and C2 are independent of h, r, p.
Proof. Let p̃ be an interpolant of p in Dr (Eh ) ∩ C 0 (Ω̄) that satisfies the approx-
imation properties (2.2)–(2.4) and denote χ = P DG − p̃. Consider the dual problem
4
(5.17)–(5.18) with f = χ. By assumption, φ belongs to H 1 (Ω) ∩ W 2, 3 (Eh ) and there
exists a constant C that depends on Ω such that
|||φ|||2, 43 ≤ C χ 4
3 ,Ω
,
where ||| · |||k,m denotes the Sobolev broken norm on W k,m (Eh ). We observe that the
4
arguments in the proof of Lemma 2.2 are valid for φ in H 1 (Ω) ∩ W 2, 3 (Eh ):

aN S (φ, v) = χv ∀ v ∈ H 2 (Eh ).
Ω
Therefore, we obtain
2
χ 0,Ω = aN S (φ, χ).
We note that by symmetry of K, the following property holds for the nonsymmetric
bilinear form aN S :
Nh
Nh

aN S (φ, χ) + aN S (χ, φ) = 2 K∇χ∇φ + 2 αχφ.
j=1 Ej j=1 Ej
Thus, we obtain
Nh

2
(5.19) χ 0,Ω =2 (K∇χ∇φ + αχφ) − aN S (χ, φ).
j=1 Ej
We now choose φ∗ in D2 (Eh ) an interpolant of φ constructed in Corollary 5.3, and we

add the orthogonality equation aN S (χ, φ∗ ) − aN S (p − p̃, φ∗ ) = 0:
Nh

2
χ 0,Ω =2 (K∇φ∇χ + αχφ) − aN S (χ, φ − φ∗ ) − aN S (p − p̃, φ∗ )
j=1 Ej
Nh
Nh
Nh

∗
=2 (K∇φ∇χ + αχφ) − K∇χ∇(φ − φ ) − αχ(φ − φ∗ )
j=1 Ej j=1 Ej j=1 Ej
Nh
Nh

− K∇(p − p̃)∇φ∗ − α(p − p̃)φ∗
j=1 Ej j=1 Ej
Ph
Ph

+ {K∇χ · ν k }[φ − φ∗ ] − {K∇(φ − φ∗ ) · ν k }[χ]
k=1 ek k=1 ek
Ph

(5.20) + {K∇(p − p̃) · ν k }[φ∗ ].
k=1 ek
4 3
Using the fact that φ belongs to W 2, 3 (Eh ) 7→ H 2 (Eh ), we bound the volume terms
as follows:

Nh

2 (K∇φ∇χ + αφχ) ≤ C(|||K 12 ∇χ|||0 + |||α 12 χ|||0 ) χ 0 ,

j=1 Ej

Nh

(K∇χ∇(φ − φ ∗
) + αχ(φ − φ ∗
))
1 1 1
≤ Ch (|||K ∇χ|||0 + h|||α χ|||0 )|||φ||| 32

2 2 2

j=1 j
E
1 1 1
≤ Ch 2 (|||K 2 ∇χ|||0 + h|||α 2 χ|||0 )|||φ|||2, 43
1 1 1
≤ Ch 2 (|||K 2 ∇χ|||0 + h|||α 2 χ|||0 ) χ 0,Ω ,

Nh
hµ−1 h
(K∇(p − p̃)∇φ∗ + α(p − p̃)φ∗ ) ≤ C s−1 |||p|||s |||∇φ∗ |||0 + |||φ∗ |||0
r r
j=1 Ej
hµ−1 h
≤C 1+ |||p|||s χ 0,Ω .
rs−1 r
The first edge term in (5.20) is bounded by the inverse estimate (2.8), the trace
theorem (2.5) combined with Corollary 5.3, with the same notation as in Theorem 3.1:
P
h Ph

−1 1
{K∇χ · ν k }[φ − φ∗ ] ≤ C h̄k 2 r K 2 ∇χ 0,Ek12 h̄k |||φ||| 32 ,Ek12
ek
k=1 k=1
1 1
(5.21) ≤ Crh 2 |||K 2 ∇χ|||0 χ 0,Ω .
Similarly as in Theorem 5.4, we define a constant on each edge (or face) ek by

1 1
ck = χ− χ.
|E 1 | E1 |E 2 | E2
ERROR ESTIMATES 927
1 4
To bound the next term in (5.20), we use the imbedding W 4 , 3 (ê) 7→ L2 (ê) and the
trace theorem fˆ 14 , 34 ≤ Ĉ fˆ 1, 34 . Thus, by Corollary 5.3, we derive
W (ê) W (Ê)
P
h Ph
∗
{K∇(φ − φ ) · ν k }[χ] ≤ {K∇(φ − φ∗ ) · ν k } 0,ek [χ] − ck 0,ek
ek
k=1 k=1
Ph
1 1
≤C |||φ|||2, 43 ,E 12 h̄k2 |||K 2 ∇χ|||0,Ek12
k
k=1
1
Ph
2
1 1
≤ Ch |||K ∇χ|||0
2 2 |||φ|||22, 4 ,E 12
3 k
k=1
1 1
≤ Ch 2 |||K 2 ∇χ|||0 |||φ|||2, 43
1 1
≤ Ch 2 |||K 2 ∇χ|||0 χ 0,Ω .
(5.22)
Using the approximation properties, we can bound the last term in (5.20)
P
h Ph
∗
{K∇(p − p̃) · ν k }[φ ] ≤ {K∇(p − p̃) · ν k } 0,ek [φ∗ − φ] 0,ek
ek
k=1 k=1
3
Ph
µ−
h̄ 2 k
≤C 3 |||p|||s,Ek12 h̄k |||φ||| 32 ,E 12
k=1
rs− 2 k
µ− 12
h
(5.23) ≤C 3 |||p|||s χ 0,Ω .
rs− 2
The theorem is obtained by combining all the bounds above.
Remark. In three dimensions, the duality argument used in proving Proposi-
tion 5.5 is more delicate. The regularity of the dual problem (5.17)–(5.18) should be
3
as follows: for any f ∈ L2 (Ω), the solution φ of (5.17)–(5.18) belongs to W 2, 2 (Eh )
with continuous dependence on f . But, as it is shown in [8], for K sufficiently smooth,
φ belongs to W 2,p (Ω) for any p < p0 and p0 ≥ 43 , and this is substantially smaller
4
than 32 . To simplify the discussion, let us assume that φ belongs to W 2, 3 (Eh ). Then
4 5
on one hand W 2, 3 (E) 7→ H 4 (E) and on the other hand, the trace of ∇φ on a face
1 4 3
e belongs to W 4 , 3 (e) 7→ L 5 (e). Thus, by using Hölder’s inequality, we can modify
the proof of Proposition 5.5 and derive in three dimensions that, under the very mild
4
regularity assumption φ ∈ W 2, 3 (Eh ), we have
hµ−1 1
P DG − p 0,Ω ≤C 5 (C1 + C2 h 2 r).
rs− 2
6. Numerical example. The reader will find in [6, 13] several examples of
numerical computations with the DG method that confirm our theoretical results in
the case of smooth solutions. In contrast, the experiment in this section investigates
the robustness of the DG method when solving problem (2.10) with discontinuous
coefficients and a nonsmooth solution. The domain Ω = (−1, 1)2 is divided into
four subdomains Ωi as shown in Figure 6.1. The tensor K is a constant tensor Ki
over each subdomain and we assume that K1 = K3 and K2 = K4 . We consider
111111
000000
000000
111111
000000
111111
Ω2Ω
111111
000000 1
111111
000000
00000
11111
00000
11111
00000
11111
Ω Ω
00000
11111
3 4
00000
11111
Fig. 6.1. Initial domain for the problem with discontinuous coefficients.
the case where K1 = 5Id and K2 = Id. We seek an analytical solution of the form
rα (ai sin(αθ) + bi cos(αθ)), where (r, θ) are the polar coordinates of a given point in
Ω and ai , bi are constants that depend on the subdomain Ωi .
We can easily show that with the choice α = 0.53544094560, the analytical so-
lution pex satisfies the usual interface conditions: pex and K∇pex · ν are continuous
across the interfaces and −∇ · K∇p = 0 on Ω. In addition, the solution has a singu-
larity at the origin, so that it belongs to H 1 (Ω) but not to H 2 (Ω). We first analyze
the convergence of the DG solution on a sequence of uniformly refined meshes. The
coarse mesh consists of the four subdomains Ωi . The relative error in the L2 norm,
DG
defined as p−Pp 0 0 , is plotted against the number of degrees of freedom in Figure 6.2.
First, we choose finite elements of degree two, and then of degree three. We see, for
these elements and in this example, that the convergence rate does not depend on the
degree of the approximation, but that it depends on the regularity of the solution. In
that case, the convergence rate in the L2 norm is O(h2α ) = O(h1.07 ), where h is the
0
10
r=2
r=3
−1
10
Relative L2 error
−2
10
0.53
1
−3
10
1 2 3 4 5
10 10 10 10 10
Number of degrees of freedom
Fig. 6.2. Relative error in the L2 norm versus the number of degrees of freedom for uniform
meshes in the case r = 2, 3.
ERROR ESTIMATES 929
0
10
r=2
r=3
Relative H error
1
0
−1
10
0.26
1
−2
10
1 2 3 4 5
10 10 10 10 10
Fig. 6.3. Relative error in the H 1 norm versus the number of degrees of freedom for uniform
meshes in the case r = 2, 3.
mesh size, or O(N α ), where N is the number of degrees of freedom. We also study
the numerical error in the energy norm and we see in Figure 6.3 that the convergence
α
rate of the error in the energy norm is O(hα ) = O(N 2 ), which is a little bit better
than expected.
We now study the influence of adaptivity on the numerical DG solution. We refer
to [16] for more details on the mesh adaptation strategy that we use here. Here,
the error indicators are the norms of solutions of a discrete problem, for which the
approximation order is r + q. Therefore, we denote by q the degree of enrichment.
We begin with a uniform mesh of size h = 1 and we adaptively refine the mesh.
We fix r = 2 and q = 5 and we see that adaptive mesh refinements lead to optimal
convergence rates (Figure 6.4).
7. Conclusion and perspectives. The energy estimates we have derived in

this paper are local in the H 1 norm, i.e., they involve only the local regularity of the
solution element per element, and not its global regularity. Of course, for recovering
full accuracy in the L2 norm, we need the global H 2 regularity of the dual problem.
But this is always necessary for a duality argument.
These discontinuous Galerkin methods have a very wide range of applications.
The most interesting one from the computational point of view is the DG method
because it is completely unconstrained. As a future extension, we propose to investi-
gate the numerical analysis of the DG method for problems of miscible displacement
in porous media, which involves the analysis of time-dependent convection-diffusion
equations. Numerical simulations with the DG method give very promising results
[17], especially compared to other methods. Another extension concerns subdivisions
with nonmatching grids. The advantage of the DG method is that it does not seem
to require the introduction of mortar elements or penalties.
0
10
unif r=2
adap r=2 q=5
Relative H error
1
0
−1
10
0.26
1
−2
10
1 2 3 4 5
10 10 10 10 10
Fig. 6.4. Relative error in the H 1 norm versus the number of degrees of freedom for uniformly
and adaptively refined meshes in the case r = 2, q = 5.
Acknowledgments. The authors wish to thank Professors J. Tinsley Oden, Ivo

Babus̆ka, Bernardo Cockburn, and Christine Bernardi, as well as the referees, for their
valuable suggestions and constructive comments.
REFERENCES
[1] R. A. Adams, Sobolev Spaces, Academic Press, New York, 1975.

[2] D. N. Arnold, An interior penalty finite element method with discontinuous elements, SIAM
J. Numer. Anal., 19 (1982), pp. 742–760.
[3] D. N. Arnold, D. Boffi, and R. S. Falk, Approximation by quadrilateral finite element,
Math. Comp., submitted.
[4] I. Babus̆ka and M. Suri, The h-p version of the finite element method with quasi-uniform
meshes, RAIRO Modél. Math Anal. Numér., 21 (1987), pp. 199–238.
[5] I. Babus̆ka and M. Suri, The optimal convergence rates of the p version of the finite element
method, SIAM J. Numer. Anal., 24 (1987), pp. 750–776.
[6] C. E. Baumann, An h-p Adaptive Discontinuous Finite Element Method for Computational
Fluid Dynamics, Ph.D. thesis, The University of Texas at Austin, Austin, 1997.
[7] P. G. Ciarlet, Handbook of Numerical Analysis, Finite Element Methods, Vol. II, North-
Holland, Amsterdam, 1991.
[8] M. Dauge, Neumann and mixed problems on curvilinear polyhedra, Integral Equations Op-
erator Theory, 15 (1992), pp. 227–261.
[9] J. Douglas and T. Dupont, Interior penalty procedures for elliptic and parabolic Galerkin
methods, in Computing Methods in Applied Sciences, Lecture Notes in Phys. 58, Springer-
Verlag, Berlin, 1976, pp. 207–216.
[10] V. Girault, A local projection operator for quadrilateral finite elements, Math. Comp., 64
(1995), pp. 1421–1431.
[11] P. Grisvard, Elliptic Problems in Nonsmooth Domains, Monogr. Stud. Math. 24, Pitman,
Boston, 1985.
[12] J. L. Lions and E. Magenes, Problèmes aux Limites non Homogènes et Applications, I,
Dunod, Paris, 1968.
[13] J. T. Oden, I. Babuska, and C. E. Baumann, A discontinous hp finite element method for
diffusion problems, J. Comput. Phys., 146 (1998), pp. 491–519.
[14] P. Percell and M. F. Wheeler, A local residual finite element procedure for elliptic equa-
tions, SIAM J. Numer. Anal., 15 (1978), pp. 705–714.
ERROR ESTIMATES 931
[15] B. Rivière, M. F. Wheeler, and V. Girault, Improved energy estimates for interior
penalty, constrained and discontinuous Galerkin methods for elliptic problems. Part I,
Comput. Geosci., 3 (1999), pp. 337–360.
[16] B. Rivière and M. F. Wheeler, A posteriori error estimates and mesh adaptation strategy
for discontinuous Galerkin methods applied to diffusion problems, Comput. Math. Appl.,
to appear.
[17] B. Rivière Discontinuous Galerkin Methods for Solving the Miscible Displacement Problem
in Porous Media, Ph.D. Thesis, The University of Texas at Austin, Austin, 2000.
[18] M. F. Wheeler, An elliptic collocation-finite element method with interior penalties, SIAM
J. Numer. Anal., 15 (1978), pp. 152–161.
[19] M. F. Wheeler and B. L. Darlow, Interior penalty Galerkin procedures for miscible dis-
placement problems in porous media, in Computational Methods in Nonlinear Mechanics,
J. Tinsley Oden, ed., North-Holland, Amsterdam, New York, 1980, pp. 485–506.

10 1137@S003614290037174X

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

10 1137@S003614290037174X

Uploaded by

Copyright:

Available Formats

SIAM J. NUMER. ANAL.

c 2001 Society for Industrial and Applied Mathematics

A PRIORI ERROR ESTIMATES FOR FINITE ELEMENT

SPACES FOR ELLIPTIC PROBLEMS∗

AMS subject classiﬁcations. 35J25, 65N15, 65N30

published electronically July 31, 2001.

ics, The University of Texas, Austin, TX 78712 (riviere@ticam.utexas.edu, mfw@ticam.utexas.edu).

F-75230 Paris Cedex 05, France (girault@ann.jussieu.fr).

2. Statement of the problem. Finite element methods.

2.1. Notation and approximation properties. Let Ω be a polygonal domain

W s,p (Eh ) = {v ∈ Lp (Ω) : v|Ej ∈ W s,p (Ej ), j = 1, . . . , Nh },

and we denote it by H s (Eh ) when p = 2.

We consider a matrix-valued function K = (kij )1≤i,j≤n and a nonnegative scalar

K ∈ W 1,4 (Eh ), α ∈ L∞ (Ω).

We shall also use the following inverse inequalities.

polynomial of degree r on E, there exists a constant C independent of E and r such

we obtain on one hand

and by the Cauchy–Schwarz inequality

This inequality carries over to a polynomial of Pr in a quadrilateral because the image

and g given in L2 (ΓN ), we consider the following elliptic problem:

We deﬁne the linear form

aN S (P NIPG , v) + J0σ,β (P NIPG , v) = L(v)

The next lemma establishes the consistency of this scheme.

Conversely, if p ∈ H 1 (Ω) ∩ H 2 (Eh ) satisﬁes (2.12), then p is the solution of problem

Using the Neumann boundary conditions, we get

The constrained discrete space is deﬁned as follows:

The discrete approximation P NCG in Dr∗ (Eh ) satisﬁes

(2.13) aN S (P NCG , v) = L(v) ∀ v ∈ Dr∗ (Eh ).

The consistency of this scheme is a consequence of Lemma 2.2.

(2.14) aN S (P DG , v) = L(v) ∀ v ∈ Dr (Eh ).

Proof. Indeed, since it is a square system of linear equations in ﬁnite dimension,

where µ = min(r + 1, s), r ≥ 1, s ≥ 2, and δ = 0 in the case of triangles or tetrahedra

Proof. From Lemma 2.2, we have

Let p̃ be an interpolant of p having optimal hp-approximation errors (2.2)–(2.4) and

As ek is an interior edge, we assume that ek = ∂E 1 ∩ ∂E 2 , where E 1 and E 2 are

Thus, the ﬁnal bound for the third term in (3.2) is

The interior penalty terms in (3.2) are bounded by using (2.3)

Thus, if β = (n − 1)−1 , then we obtain optimal convergence rates with respect

Theorem 3.2. Assume that Ω is convex, ΓN = ∂Ω, and K is suﬃciently smooth

belongs to H 2 (Ω), with continuous dependence on f . Then, in the case of triangles or

By assumption, φ ∈ H 2 (Ω) and there is a constant C that depends on Ω such that

(3.7) φ 2,Ω ≤ C P NIPG − p 0,Ω .

Integrating by parts on each element yields

aN S (χ, φ∗ ) + J0σ,β (χ, φ∗ ) = 0,

 {K∇φ · ν k }[χ] ≤ {K∇φ · ν k } 0,ek [χ] 0,ek

which concludes the proof.

p0 on ∂Ω is a polynomial of degree r, in the case of triangles or tetrahedra, then the

aN S (P NCG − p, v) = 0 ∀ v ∈ Dr∗ (Eh ).

In a similar manner, we have

We have by Cauchy–Schwarz and (2.4)

[χ] 0,ek = [χ − ck ] 0,ek ≤ (χ − ck )E 1 0,ek + (χ − ck )E 2 0,ek .

Since P NCG belongs to Dr∗ (Eh ), we have

Therefore, it suﬃces to estimate (χ − ck )E 1 0,ek with the choice

where ∇ˆ ê denotes the tangential gradient on ê. However, since ∇ˆ ê χ̂ belongs to a

Thus, summing on k, we have

Combining (4.4) and (4.5), we obtain a bound for A:

The last term in (4.1) is bounded as in Theorem 3.1:

Combining (4.2), (4.3), (4.6), and (4.7), we obtain

then we can construct p̃ such that

Let ψ ∗ be in Dr∗ (Eh ) ∩ C 0 (Ω̄). The orthogonality condition implies that

we have by applying (2.6)

Let A denote the last term in (4.11):

{K∇φ · ν k }[χ] ≤ {K∇φ · ν k } 0,ek [χ] 0,ek

≤ Ch (|||K ∇χ|||0 + h|||α χ|||0 )|||φ||| 32