J. Math. Pures Appl.

81 (2002) 311341
Implicit partial differential equations and the constraints
of nonlinear elasticity
Bernard Dacorogna

, Chiara Tanteri
Dpartement de Mathmatiques, EPFL, 1015 Lausanne, Switzerland
Manuscript received 2 July 2001
Abstract
We study a Dirichlet problem associated to some nonlinear partial differential equations under additional constraints that are
relevant in nonlinear elasticity. We also give several examples related to the complex eikonal equation, optimal design, potential
wells or nematic elastomers. 2002 ditions scientiques et mdicales Elsevier SAS. All rights reserved.
Keywords: Rank one convex hull; Implicit partial differential equations
1. Introduction
We consider here the following Dirichlet problem (as well as some higher-order versions of the problem)
F
i
_
x, u(x), Du(x)
_
=0, i =1, 2, . . . , I, a.e. x ,
u =, on ,
(1)
where R
n
is a bounded open set, u: R
n
and therefore Du R
nn
, F
i
: R
nn
R, i = 1, . . . , I, are quasiconvex
functions and , the boundary datum, is given.
This problem has been intensively studied and we refer to Dacorogna and Marcellini [6] for a discussion of these implicit
equations. We will be interested here in considering the case where we require that the solutions satisfy some constraints that are
natural in nonlinear elasticity. The rst one is the noninterpenetration of matter that is expressed mathematically by det Du >0
and the second one is the incompressibility which reads as det Du =1. These two questions were raised in [6] and are discussed
by Dacorogna, Marcellini and Tanteri [7] for the rst one and by Mller and Sverak [14] for the second one, using a different
approach based on the method of convex integration of Gromov.
We will discuss here some theoretical results related to the rst (Section 2) and second (Sections 3, 4) cases and deal with
several relevant examples (Sections 5 to 10). We will also make some general considerations concerning polyconvex hulls
(Section 11) and we will conclude in Appendix A with some well known properties of singular values of matrices.
We now describe six examples that we will investigate here, but we rst recall that we respectively denote by coE, PcoE,
RcoE, the convex, polyconvex, rank one convex hull of a given set E R
mn
.
The rst example has already been considered when n =2 by Dacorogna and Marcellini [6].
Example 1.1 (Complex eikonal equation). The problem has been introduced by Magnanini and Talenti [12] motivated by
problems of geometrical optics with diffraction. Given R
n
a bounded open set, f : R R R, f = f (x, u, v),
a continuous function, we wish to nd a complex function w W
1,
(; C),
w(x) =u(x) +iv(x)
*
Correspondence and reprints.
E-mail addresses: [Link]@[Link] (B. Dacorogna), [Link]@[Link] (C. Tanteri).
0021-7824/02/$ see front matter 2002 ditions scientiques et mdicales Elsevier SAS. All rights reserved.
PII: S0021- 7824( 01) 01235- 1
312 B. Dacorogna, C. Tanteri / J. Math. Pures Appl. 81 (2002) 311341
such that
n

i=1
w
2
x
i
+f
2
=0, a.e. in ,
w =, on ,
(2)
where w
x
i
=w/x
i
. The problem is then equivalent to
|Dv|
2
=|Du|
2
+f
2
, a.e. in ,
Dv; Du =0, a.e. in ,
w =, on .
We will solve, in fact, a more restrictive problem, namely:
|Dv|
2
=r
2
+f
2
, |Du|
2
=r
2
, a.e. in ,
Dv; Du =0, a.e. in ,
w =, on ,
for an appropriate r >0. In algebraic terms (at this point we can consider f to be constant) we have, letting s =
_
r
2
+f
2
,
E =
_
=
_
a
b
_
R
2n
: |a| =r, |b| =s and a; b =0
_
.
Letting
A=
_
1/r 0
0 1/s
_
R
22
we will prove that
PcoE =RcoE =
_
R
2n
:
1
(A),
2
(A) 1
_
,
where
1
(A),
2
(A) are the singular values of the matrix A R
2n
(cf. Appendix A for more details).
The second example is important for optimal design and is related to the method of confocal ellipses of Murat and Tartar [17]
and of the results of Dacorogna and Marcellini [4,6]. However the existence part will be obtained without the use of the confocal
ellipses method, contrary to the one in [4,6].
Example 1.2 (Optimal design). Let R
2
be a bounded open set and consider the DirichletNeumann problem:
w(x) {0, 1}, a.e. x ,
det D
2
w(x) 0, a.e. x ,
w(x) =(x), Dw(x) =D(x), x .
The associated algebraic problem is (denoting the set of 2 2 symmetric matrices by R
22
s
) when we let
E =
_
R
22
s
: trace {0, 1}, det 0
_
,
to nd that
RcoE =coE =
_
R
22
s
: 0 trace 1, det 0
_
.
We will next consider two more academic examples but that exhibit some interesting features. The rst one shows how we
can handle some problems depending on singular values under a constraint on the positivity of the determinant.
Example 1.3. Let R
2
be an open set and consider

1
(Du) +
2
(Du) =1, a.e. in ,
det Du >0, a.e. in ,
u(x) =(x), x .
B. Dacorogna, C. Tanteri / J. Math. Pures Appl. 81 (2002) 311341 313
The associated algebraic problem is: given
E =
_
R
22
:
1
() +
2
() =1, det 0
_
,
to prove that
PcoE =RcoE =
_
R
22
:
1
() +
2
() 1, det 0
_
.
The fourth problem that we want to discuss is the following second-order problem.
Example 1.4. Let R
2
be a bounded open set and consider the DirichletNeumann problem:

2
u
x
i
x
j

=1, a.e. x , i, j =1, 2,

u(x) =(x), Du(x) =D(x), x .
The algebraic problem is then
E =
_
=(
ij
) R
22
s
: |
ij
| =1, i, j =1, 2
_
and we will nd that
PcoE =RcoE =
_
=(
ij
) R
22
s
: |
ij
| 1, i, j =1, 2, |
11

22
| det
_
.
The last two examples concern the incompressibility constraint. The rst one is the problem of two potential wells in two
dimensions that was resolved by Mller and Sverak in [13] and by Dacorogna and Marcellini in [5,6] for the case det A=det B
and by Mller and Sverak in [14] for the case det A = det B. In Section 9 we will show that our results also apply to this
problem.
Example 1.5 (Potential wells). Let R
n
be open and
E =SO(2)A SO(2)B
with det A =det B >0. Let
int RcoE,
where int RcoE stands for the interior (relative to the manifold det = det A = det B) of the rank one convex hull of E (cf.
Section 9 for the characterization of RcoE). Then there exists u W
1,
(; R
2
) such that
Du(x) E, a.e. in ,
u(x) =x, on .
The last example is related to some recent work of DeSimone and Dolzmann [9] on nematic elastomers; we refer to this
article for the description of the physical model.
Example 1.6 (Nematic elastomers). Let r < 1 (this is called the oblate case while r >1 is called the prolate case and it can be
handled similarly), let 0
1
(A)
n
(A) denote the singular values of a matrix A R
nn
and
E =
_
A:

(A) =r
1/(2n)
, 1 n 1,
n
(A) =r
(1n)/(2n)
, det A=1
_
.
We will prove that
RcoE =
_
A:
n

i=

i
(A) r
(1)/2n
, 2 n, det A=1
_
=
_
A:

(A)
_
r
1/2n
, r
(1n)/2n
_
, 1 n, det A=1
_
(this representation formula, under the second form, has been established in [9] when n = 2, 3; actually we will consider
below a slightly more general case). Our analytical result is then: given int RcoE and R
n
an open set, there exists
u +W
1,
0
(; R
n
) ((x) =x) such that
314 B. Dacorogna, C. Tanteri / J. Math. Pures Appl. 81 (2002) 311341

1
_
Du(x)
_
= =
n1
_
Du(x)
_
=r
1/2n
, a.e. x ,

n
_
Du(x)
_
=r
(1n)/2n
, a.e. x ,
det Du(x) =1, a.e. x .
2. Inequality constraints
The results of this section are inspired by those of Dacorogna, Marcellini and Tanteri [7]. We recall rst some notations and
denitions introduced in [6].
Notations. (1) Let N, n, m1 be integers. For u: R
n
R
m
we write
D
N
u =
_

N
u
i
x
j
1
x
j
N
_
1im
1j
1
,...,j
N
n
R
mn
N
s
.
(The index s stands here for all the natural symmetries implied by the interchange of the order of differentiation.) When N =1
we have
R
mn
s
=R
mn
while if m=1 and N =2 we obtain
R
n
2
s
=R
nn
s
,
i.e., the usual set of symmetric matrices.
(2) For u: R
m
we let
D
[N]
u =
_
u, Du, . . . , D
N
u
_
stand for the matrix of all partial derivatives of u up to the order N. Note that
D
[N]
u R
mM
s
=R
m
R
mn
R
mn
2
s
R
mn
(N1)
s
,
where
M =1 +n + +n
(N1)
=
n
N
1
n 1
.
Hence
D
[N]
u =
_
D
[N1]
u, D
N
u
_
R
mM
s
R
mn
N
s
.
We now dene the main property, called the relaxation property (cf. [6]), in order to get existence of solution.
Denition 2.1 (Relaxation property). Let E, K R
n
R
mM
s
R
mn
N
s
. We say that K has the relaxation property with
respect to E if for every bounded open set R
n
, for every u

, a polynomial of degree N with D

N
u

(x) =, satisfying
_
x, D
[N1]
u

(x), D
N
u

(x)
_
int K,
there exists a sequence u

C
N
piec
(

; R
m
) such that
u

+W
N,
0
_
; R
m
_
,
u

in W
N,
,
_
x, D
[N1]
u

(x), D
N
u

(x)
_
E int K, a.e. in ,
_

dist
__
x, D
[N1]
u

(x), D
N
u

(x)
_
; E
_
dx 0 as .
The following theorem is the main abstract existence theorem.
B. Dacorogna, C. Tanteri / J. Math. Pures Appl. 81 (2002) 311341 315
Theorem 2.2. Let R
n
be open. Let F
i
: R
mM
s
R
mn
N
s
R, F
i
=F
i
(x, s, ), i =1, 2, . . . , I, be continuous with
respect to all variables and quasiconvex with respect to the variable . Let E, K R
n
R
mM
s
R
mn
N
s
be such that
E = {(x, s, ): F
i
(x, s, ) =0, 1 i I
1
, F
i
(x, s, ) 0, I
1
+1 i I} ,
K
_
(x, s, ): F
i
(x, s, ) 0, 1 i I
_
.
Assume that the set in the right-hand side of the inclusion is bounded uniformly for x and whenever s vary on a bounded
set of R
mM
s
and that K has the relaxation property with respect to E. Let C
N
piec
( ; R
m
) such that
_
x, D
[N1]
(x), D
N
(x)
_
E int K, a.e. in ;
then there exists (a dense set of ) u +W
N,
0
(; R
m
) such that
F
i
_
x, D
[N1]
u(x), D
N
u(x)
_
=0, i =1, 2, . . . , I
1
, a.e. x ,
F
i
_
x, D
[N1]
u(x), D
N
u(x)
_
0, i =I
1
+1, . . . , I, a.e. x .
Remark 2.3. (1) This result has been proved by Dacorogna and Marcellini [6] (Theorem 6.3) when I =I
1
. With substantially
the same proof it was found by Dacorogna, Marcellini and Tanteri [7].
(2) An interesting case of constraints is when I =I
1
+1 and
F
I
1
+1
() =det
(i.e., det 0). It is actually this constraint that will be used in two of the examples below.
If the set E is given by only one equation the theorem takes a simpler form.
Theorem 2.4. Let R
n
be open. Let F : R
n
R
nn
R be continuous and quasiconvex. Assume that {
R
nn
: F(x, s, ) 0, det > 0} is bounded in R
nn
uniformly with respect to x and s in a bounded set of R
n
. If
C
1
piec
( ; R
n
) is such that
F
_
x, (x), D(x)
_
0, a.e. x ,
det D(x) >0, a.e. x ,
then there exists (a dense set of ) u +W
1,
0
(; R
n
) such that
F
_
x, u(x), Du(x)
_
=0, a.e. x ,
det Du(x) >0, a.e. x .
Remark 2.5. The fact that we can treat strict inequalities follows from the observation that, by hypothesis, we can nd > 0
such that det D > since C
1
piec
. The remaining part of the proof follows from the next theorem.
We have a generalization of the above theorem.
Theorem 2.6. Let R
n
be open. Let F, : R
mM
s
R
mn
N
s
R be continuous and respectively quasiconvex and
quasiafne. Assume that
_
R
mn
N
s
: F(x, s, ) 0, (x, s, ) 0
_
is bounded in R
mn
N
s
uniformly with respect to x and s in a bounded set of R
mM
s
. If C
N
piec
( ; R
m
) is such that
F
_
x, D
[N1]
(x), D
N
(x)
_
0, a.e. x ,

_
x, D
[N1]
(x), D
N
(x)
_
<0, a.e. x ,
then there exists (a dense set of ) u +W
N,
0
(; R
m
) such that
F
_
x, D
[N1]
u(x), D
N
u(x)
_
=0, a.e. x ,

_
x, D
[N1]
u(x), D
N
u(x)
_
0, a.e. x .
316 B. Dacorogna, C. Tanteri / J. Math. Pures Appl. 81 (2002) 311341
Proof. We will do the proof when is afne and when there is no dependence on lower-order terms, i.e.,
E =
_
R
mn
N
s
: F() =0, () 0
_
.
The general case follows as in [6].
Step 1: We rst prove that
RcoE =
_
R
mn
N
s
: F() 0, () 0
_
.
Indeed call X the right-hand side. It is clear that E X and that X is rank one convex; we therefore have RcoE X. We now
show the reverse inclusion. Let X be xed and assume that F() < 0, otherwise E and the result is trivial. Since is
rank one afne, we have for every , a matrix of rank one, that for every t R
( +t ) =() +t
_
D();
_
.
We therefore choose a matrix of rank one so that
_
D();
_
=0
(in the preceding theorem () =det + and D() =adj
n1
) and this leads to the desired identity
( +t ) =(), t R.
By compactness of E we deduce that we can nd t
1
<0 <t
2
so that
F( +t ) <0, t (t
1
, t
2
), F( +t
i
) =0, i =1, 2.
We can therefore rewrite
=
t
2
t
2
t
1
( +t
1
) +
t
1
t
2
t
1
( +t
2
)
which leads to RcoE. Note for further reference that we easily obtain
int RcoE
_
R
mn
N
s
: F() 0, () <0
_
.
Step 2: We wish now to show that RcoE has the relaxation property with respect to E, i.e., that for every bounded open set
R
n
, for every u

, an afne function with D

N
u

(x) =, satisfying
D
N
u

(x) int RcoE,

there exists a sequence u

C
N
piec
( ; R
m
) such that
u

+W
N,
0
_
; R
m
_
,
u

in W
N,
,
D
N
u

(x) int RcoE, a.e. in ,

_

F
_
D
N
u

(x)
_
dx 0 as .
If F() =0, we choose u

=u

. So from now on we can assume that F() <0 and () <0. Using then the compactness
assumption we can nd as in Step 1, a matrix of rank one, t
1
<0 <t
2
such that (we let
t
= +t )
( +t ) = () <0,
F( +t ) = F( +t ) <0, t (t
1
, t
2
),
F( +t
1
) = F( +t
2
) =0.
The approximation lemma (cf. Lemma 6.8 of Dacorogna and Marcellini [6]) with A=
t
1
+
and B =
t
2

for small enough

and
=
t
2

t
2
t
1
2
A+
(t
1
+)
t
2
t
1
2
B
B. Dacorogna, C. Tanteri / J. Math. Pures Appl. 81 (2002) 311341 317
leads immediately to the result. Note that in this lemma since rank[AB] =1
(A) =(B) =() <0
and the constructed function satises
dist
_
D
N
u

(x), co{A, B}
_
a.e. in
we deduce that by choosing sufciently small we have (Du

) <0 and F(Du

) <0 which implies that

D
N
u

(x) int RcoE

as wished. We may then apply Theorem 2.2 to get the result.
The more difcult question is to know when the relaxation property holds if the set E is given by more than one equation.
One such case is the following theorem that uses the notion of approximation property (cf. Theorem 6.14 in [6]).
Denition 2.7 (Approximation property). Let E K(E) R
n
R
mM
s
R
mn
N
s
. The sets E and K(E) are said to have the
approximation property if there exists a family of closed sets E

and K(E

), >0, such that

(1) E

K(E

) int K(E) for every >0;

(2) for every >0 there exists
0
=
0
() >0 such that dist(; E) for every E

and [0,
0
];
(3) if int K(E) then K(E

) for every >0 sufciently small.

Theorem 2.8. Let E R
n
R
mM
s
R
mn
N
s
be closed and bounded uniformly with respect to x R
n
and whenever s vary on
a bounded set of R
mM
s
and RcoE has the approximation property with K(E

) =RcoE

, then it has the relaxation property

with respect to E.
As a corollary we obtain (cf. Corollary 6.18 in [6] or [7]).
Corollary 2.9. Let R
n
be open. Let F
i
: R
mn
N
s
R, i =1, 2, . . . , I , be quasiconvex and let
E =
_
R
mn
N
s
: F
i
() =0, i =1, 2, . . . , I
1
, F
i
() 0, i =I
1
+1, . . . , I
_
.
Assume that RcoE is compact and RcoE =coE. Let C
N
piec
( ; R
m
) verify
D
N
(x) E int RcoE, a.e. x ,
or W
N,
(; R
m
) satisfy
D
N
(x) compactly contained in int RcoE, a.e. x .
Then there exists (a dense set of ) u +W
N,
0
(; R
m
) such that
F
i
_
D
N
u(x)
_
=0, a.e. x , i =1, . . . , I
1
,
F
i
_
D
N
u(x)
_
0, a.e. x , i =I
1
+1, . . . , I.
3. The incompressible case
Comparable results to those of the present section are obtained by Mller and Sverak [14] using the ideas of Gromov on
convex integration; here we show how the method of Dacorogna and Marcellini in [6] can also be applied.
In the present section and in Section 9 and 10 we will consider subsets E or K of the manifold det =1, so when we will
write int K we will mean the interior relative to the manifold.
We now adapt the denitions of the relaxation and the approximation properties to the present context. Here we give the rst
one under a slightly more restrictive form in order to avoid some technicalities. We rst let for >0 and R
n
an open set,
W

be the set of functions u C

1
piec
( ; R
n
) such that there exists an open set

so that meas(

) < and u is
piecewise afne in

. We could consider a more general set but the proof is then more involved.
318 B. Dacorogna, C. Tanteri / J. Math. Pures Appl. 81 (2002) 311341
Denition 3.1 (Relaxation property). Let E, K R
n
R
n
R
nn
. We say that K has the relaxation property with respect to
E if for every bounded open set R
n
, for every u

, an afne function with Du

(x) =, satisfying
_
x, u

(x), Du

(x)
_
int K,
there exists a sequence u

W
1/
such that
u

+W
1,
0
_
; R
n
_
,
u

in W
1,
,
_
x, u

(x), Du

(x)
_
E int K, a.e. in ,
_

dist
__
x, u

(x), Du

(x)
_
; E
_
dx 0 as .
The following theorem is the main abstract existence theorem.
Theorem 3.2. Let R
n
be open. Let F
i
: R
n
R
nn
R, F
i
=F
i
(x, s, ), i =1, 2, . . . , I, be continuous with respect
to all variables and quasiconvex with respect to the variable . Let E R
n
R
n
R
nn
be such that
E =
_
(x, s, ) R
n
R
n
R
nn
: F
i
(x, s, ) =0, i =1, 2, . . . , I
1
, F
i
(x, s, ) 0, i =I
1
+1, . . . , I, det =1
_
.
Assume that RcoE has the relaxation property with respect to E and that it is bounded uniformly for x and whenever s
vary on a bounded set of R
nn
. Let be an afne function such that
_
x, (x), D(x)
_
E int RcoE, in ;
then there exists (a dense set of ) u +W
1,
0
(; R
n
) such that
F
i
_
x, u(x), Du(x)
_
=0, i =1, 2, . . . , I
1
, a.e. x ,
F
i
_
x, u(x), Du(x)
_
0, i =I
1
+1, . . . , I, a.e. x ,
det Du(x) =1, a.e. x .
Proof. We will make the proof when there is no dependence on lower order terms, otherwise use the standard procedure in [6].
Step 1: We rst observe that R
n
can be assumed bounded, without loss of generality. We then let V be the set of
functions u so that there exists u

W
1/
, u

= on and Du

E int RcoE a.e. such that u

u in L

() as
.
Note that V and V is a complete metric space when endowed with the C
0
norm. Note that by weak lower semicontinuity
we have
V
_
u +W
1,
0
(; R
n
): F
i
(Du(x)) 0, i =1, 2, . . . , I, a.e. x , det Du(x) =1, a.e. x
_
.
Step 2: Let for u V
L(u) =
I
1

i=1
_

F
i
_
Du(x)
_
dx.
Observe that by quasiconvexity of F
i
we have for every u V
liminf
u
s

u,u
s
V
L(u
s
) L(u). (3)
We next immediately see that for every u V (recall that in V we have det Du =1 and F
i
(Du) 0, i =I
1
+1, . . . , I )
L(u) =0 Du(x) E, a.e. in . (4)
We then let
V
k
=
_
u V: L(u) >
1
k
_
.
B. Dacorogna, C. Tanteri / J. Math. Pures Appl. 81 (2002) 311341 319
We have that V
k
is open (cf. (3)). Furthermore it is dense in V . This will be proved in Step 3. If this property has been established
we deduce from Baire category theorem that
_
V
k
is dense in V . Thus the result by (4).
Step 3: So it remains to prove that for any u V and any >0 sufciently small we can nd u

V
k
so that
u

.
We will prove this property under the further assumption that, for some >0, small, u W

and
Du(x) E int RcoE, a.e. in .
The general case will follow by denition of V . By working on each piece where u is afne and by setting u

=u on

we can assume that u is afne in . The result now follows at once from the relaxation property.
As usual if the set E is dened by only one equation the relaxation property is easier to establish and we therefore have as a
rst consequence of the theorem.
Theorem 3.3. Let R
n
be open. Let F : R
n
R
nn
R be continuous and quasiconvex and coercive, with respect to
the last variable , in any direction, uniformly with respect to x and s in a bounded set of R
n
. If is afne and is such that
F
_
x, (x), D(x)
_
0, a.e. x ,
det D(x) =1, a.e. x ,
then there exists (a dense set of ) u +W
1,
0
(; R
n
) such that
F
_
x, u(x), Du(x)
_
=0, a.e. x ,
det Du(x) =1, a.e. x .
Proof. We rst let

0
=
_
x : F
_
x, (x), D(x)
_
=0
_
,

1
=
0
=
_
x : F
_
x, (x), D(x)
_
<0
_
and observe that, by continuity,
0
is closed and hence
1
is open; we therefore need only to work on this last set, since in
0
we can choose u =.
We may now apply the abstract theorem with
E =
_
(x, s, ) R
n
R
n
R
nn
: F(x, s, ) =0, det =1
_
,
K = RcoE =
_
(x, s, ) R
n
R
n
R
nn
: F(x, s, ) 0, det =1
_
.
The proposition below ensures that all the hypotheses of the abstract theorem are satised and therefore the theorem is
proved.
Proposition 3.4. Let F : R
n
R
n
R
nn
R be continuous and rank one convex and coercive, with respect to the last
variable , in any direction, uniformly with respect to x and s in a bounded set of R
n
. Let
E =
_
(x, s, ) R
n
R
n
R
nn
: F(x, s, ) =0, det =1
_
.
Then
RcoE =
_
(x, s, ) R
n
R
n
R
nn
: F(x, s, ) 0, det =1
_
.
Furthermore RcoE has the relaxation property with respect to E.
Proof. We will do the proof when there is no dependence on lower-order terms, i.e.,
E =
_
R
nn
: F() =0, det =1
_
.
Step 1: We now prove that
RcoE =
_
R
nn
: F() 0, det =1
_
.
320 B. Dacorogna, C. Tanteri / J. Math. Pures Appl. 81 (2002) 311341
Indeed call X the right-hand side. It is clear that E X and that X is rank one convex; we therefore have RcoE X. We now
show the reverse inclusion. Let X be xed and assume that F() <0, otherwise E and the result is trivial. We can then
nd R
nn
a matrix of rank one so that
det( +t ) =det =1
for every t R. This is easy and follows from the following observation ( being a matrix of rank one)
det( +t ) =det +t adj
n1
; ;
so choose so that
adj
n1
; =0. (5)
By compactness of E we deduce that we can nd t
1
<0 <t
2
so that
F( +t ) <0, t (t
1
, t
2
),
F( +t
i
) =0, i =1, 2.
We can therefore rewrite
=
t
2
t
2
t
1
( +t
1
) +
t
1
t
2
t
1
( +t
2
)
which leads to RcoE.
Step 2: We wish now to show that for every bounded open set R
n
, for every u

, an afne function with Du

(x) =,
satisfying
Du

(x) int RcoE,

there exists a sequence u

W
1/
, such that
u

+W
1,
0
(; R
n
),
u

in W
1,
,
Du

(x) int RcoE, a.e. in ,

_

F
_
Du

(x)
_
dx 0 as .
If F() =0, we choose u

=u

. So from now on we can assume that F() < 0 (and det =1). Using then the coercivity
assumption we can nd as in Step 1, a matrix of rank one, t
1
<0 <t
2
such that (we let
t
= +t )
det( +t ) = det =1,
F( +t ) = F( +t ) <0, t (t
1
, t
2
),
F( +t
1
) = F( +t
2
) =0.
So let be large enough. We can then nd =() >0 so that
F(
t
) , t [t
1
+1/, t
2
1/].
Call A =
t
1
+1/
, B =
t
2
1/
, (x) =u

(x) and observe that

det A=det B =1, rank{AB} =1 and D = =
t
2
1/
t
2
t
1
2/
A+
(t
1
+1/)
t
2
t
1
2/
B.
By continuity of F we can nd

( =()) >0 so that

dist
_
, co{A, B}
_

F() /2.
Therefore apply approximation Lemma 4.1 with <min{1/,

} and call u

the function that is found in the lemma, to get the

result.
The next question we discuss is to know when the relaxation property holds if the set E is dened by more than one equation.
The question is more involved and we need, as in Section 2 or as in [6], the so called approximation property.
B. Dacorogna, C. Tanteri / J. Math. Pures Appl. 81 (2002) 311341 321
Denition 3.5 (Approximation property). Let E K(E) R
n
R
n
R
nn
. The sets E and K(E) are said to have the
approximation property if there exists a family of closed sets E

and K(E

), >0, such that

(1) E

K(E

) int K(E) for every >0;

(2) for every >0 there exists
0
=
0
() >0 such that dist(; E) for every E

and [0,
0
];
(3) if int K(E) then K(E

) for every >0 sufciently small.

Theorem 3.6. Let E R
n
R
n
R
nn
be closed and bounded uniformly with respect to x R
n
and whenever s vary on
a bounded set of R
n
and RcoE has the approximation property with K(E

) =RcoE

, then it has the relaxation property with

respect to E.
Proof. We will make the proof when there is no dependence on lower-order terms. We therefore are given R
n
, a bounded
open set and u, an afne function with Du(x) =, with int RcoE and we wish to show that there exists a sequence u

such that
u

u +W
1,
0
_
; R
n
_
,
u

u in W
1,
,
Du

(x) E int RcoE, a.e. in ,

_

dist
_
Du

(x); E
_
dx 0 as 0.
(6)
By the approximation property we have, for some , that RcoE

and hence that R

J
coE

for a certain J. We then

proceed by induction on J.
Step 1: We start with J =1. We can therefore write
Du = =t A+(1 t )B, rank{AB} =1, with A, B E

.
We then use the approximation Lemma 4.1 to nd (setting

=
A

B
) u

u near ,
u

,
Du

(x) =
_
A in
A
,
B in
B
,
det Du

(x) =det A=det B, in ,

dist
_
Du

(x); Rco E

_
, in ,
where we have used the fact that
co{A, B} RcoE

.
The claim (6) follows by choosing and smaller if necessary.
Step 2: We now let for J >1
R
J
coE

.
Therefore there exist A, B R
nn
such that
=t A+(1 t )B, rank{AB} =1, A, B R
J1
coE

.
We then apply the approximation Lemma 4.1 and nd that there exist a vector-valued function v W
/2
and
A
,
B
disjoint
open sets such that
meas
_
(
A

B
)
_
/2 meas ,
v u near ,
v u

/2,
Dv(x) =
_
A in
A
,
B in
B
,
dist
_
Dv(x); RcoE

_
, in .
322 B. Dacorogna, C. Tanteri / J. Math. Pures Appl. 81 (2002) 311341
We now use the hypothesis of induction on
A
,
B
and A, B. We then can nd

A
,

B
, v
A
W
/4
in
A
, v
B
W
/4
in

B
satisfying
meas
_

A
_
, meas
_

B
_
/4 meas ,
v
A
v near
A
, v
B
v near
B
,
v
A
v

/2 in
A
, v
B
v

/2 in
B
,
dist
_
Dv
A
(x); E

_
, a.e. in

A
,
dist
_
Dv
B
(x); E

_
, a.e. in

B
,
dist
_
Dv
A
(x); RcoE

_
, a.e. in
A
,
dist
_
Dv
B
(x); Rco E

_
, a.e. in
B
.
Letting

=

B
and
u

(x) =
_
v(x) in (
A

B
),
v
A
(x) in
A
,
v
B
(x) in
B
we have indeed obtained (6) by choosing and smaller if necessary, and thus the result.
4. The approximation lemma
The following result is due to Mller and Sverak [14] and is an extension of a classical lemma (cf. for example, Lemma 6.8
in [6]) to handle constraint on the determinants. For the convenience of the reader we will give the proof of Mller and Sverak
with however a slight variation.
Lemma 4.1. Let R
n
be an open set with nite measure. Let t [0, 1] and A, B R
nn
with rank{A B} = 1 and
det A=det B >0. Let be such that
D(x) =t A+(1 t )B, x .
Then, for every >0, there exist u C

( ; R
n
) and disjoint open sets
A
,
B
, so that
|meas
A
t meas |, |meas
B
(1 t ) meas | ,
u near ,
u

,
Du(x) =
_
A in
A
,
B in
B
,
det Du(x) =det A=det B, in ,
dist
_
Du(x), co{A, B}
_
, in .
Remark 4.2. By co{A, B} =[A, B] we mean the closed segment joining A to B.
Proof. We divide the proof into two steps.
Step 1: We rst assume that
D =t A+(1 t )B =I
and hence det A =det B =1. We also assume (these assumptions will be removed in Step 2) that the matrix has the form
AB = e
1
where e
1
=(1, 0, . . . , 0), =(0,
2
, . . . ,
n
) R
n
, i.e.,
AB =
_
_
_
0 0 . . . 0

2
0 . . . 0
. . . . . . . . . . . .

n
0 . . . 0
_
_
_ R
nn
.
B. Dacorogna, C. Tanteri / J. Math. Pures Appl. 81 (2002) 311341 323
We can express as union of cubes with faces parallel to the coordinate axes and a set of small measure. Then, by posing
u on the set of small measure, and by homotheties and translations, we can reduce ourselves to work with equal to the
unit cube.
Let

be a set compactly contained in and let C

0
() and L>0 be such that
meas(

) /2,
0 (x) 1, x ,
(x) =1, x

D(x)

L/, x

D
2
(x)

L/
2
, x

.
(7)
Let us dene a C

function v : [0, 1] R in the following way: given >0, divide the interval (0, 1) into three nite unions
I
A
, I
B
, J of disjoint open subintervals such that
I
A
I
B
J =[0, 1],
meas I
A
=t , meas I
B
=1 t ,
v

(x
1
) =
_
(1 t ) if x
1
I
A
,
t if x
1
I
B
,
v

(x
1
)
_
t , (1 t )
_
, x
1
(0, 1),

v(x
1
)

(x
1
)

, x
1
(0, 1).
We then let

A
={x

: x
1
I
A
},
B
={x

: x
1
I
B
}
and observe that by choosing small enough we have

meas
A
t meas

meas
B
(1 t ) meas

.
We next dene V : R
n
R
n
by
V(x) =v

(x
1
)(x)(0,
2
, . . . ,
n
) +v(x
1
)
_

i=2

x
i
,
2

x
1
,
3

x
1
, . . . ,
n

x
1
_
.
Note that V C

and has the following properties (where has been chosen sufciently small)
divV 0 in ,
V 0 near ,
|DV v

e
1
|, |V |
2
in ,
DV =
_
(1 t ) e
1
in
A
,
t e
1
in
B
.
We can now dene u as the ow associated to the vector eld V (this is the usual procedure to construct a volume preserving
map), i.e.,
d
ds
u(s, x) =V
_
u(s, x)
_
, s [0, 1], u(0, x) =x. (8)
The map u(x) =u(1, x) has all the claimed properties, as will now be shown.
(1) Indeed since V 0 near , we have by uniqueness of the solution of the differential system that
u(s, x) x, s [0, 1]
and hence the boundary condition for u is satised (recall that by hypothesis we are considering the case (x) =x).
(2) Since we have |V|
2
we deduce that

u(s, x) x

s
_
0
V
_
u(, x)
_
d

2
s, s [0, 1], x . (9)
324 B. Dacorogna, C. Tanteri / J. Math. Pures Appl. 81 (2002) 311341
(3) If x
A

B
then by uniqueness of solutions we nd
u(s, x) =x +sv

(x
1
), s [0, 1],
and, hence,
Du(x) =D
x
u(1, x) =I +v

(x
1
) e
1
=
_
A in
A
,
B in
B
.
(4) Since divV 0 in we have automatically (cf., for example [2], p. 28)
det D
x
u(s, x) 1, s [0, 1], x .
(5) Finally, it remains to show that
dist
_
Du(x), co{A, B}
_
a.e. in .
We rst set
L(x) =v

(x
1
)(x) e
1
.
Returning to (8) we get
d
ds
D
x
u(s, x) =DV
_
u(s, x)
_
D
x
u(s, x), s [0, 1], D
x
u(0, x) =I
and we compare the solution of this system with the one of
d
ds
F(s, x) =L(x)F(s, x), s [0, 1], F(0, x) =I.
Using the properties of V and (9) we get that

F(s, x) D
x
u(s, x)

, s [0, 1], x .
The conclusion then follows from the observation that
F(s, x) =e
sL(x)
=I +sv

(x
1
)(x) e
1
, s [0, 1],
and the facts that [0, 1], v

[t , (1 t )].
Step 2: We now consider the general case. Since AB is a matrix of rank one we can nd a, b R
n
(replacing a by |b|a
we can assume that |b| =1) such that
C
1
AC
1
B =a b,
where C = D = t A + (1 t )B. We can then nd R = (r
ij
) SO(n) R
nn
(i.e., a rotation) so that b = e
1
R and hence
e
1
=bR
t
. We then set

=R and

A=RC
1
AR
t
and

B =RC
1
BR
t
.
We observe that by construction, setting =Ra, we have
t

A+(1 t )

B =I,

A

B = e
1
.
Note that this implies in particular that
1
= 0 (since det

A = det

B = 1) and hence = (0,

2
, . . . ,
n
). We may therefore
apply Step 1 to

and to (y) =RC
1
(R
t
y) and nd

A
,

B
and u C

(

; R
n
) with the claimed properties. By setting
u(x) =CR
t
u(Rx), x ,

A
=R
t

A
,
B
=R
t

B
we get the result by recalling that
Du(x) =CR
t
Du(Rx)R.
B. Dacorogna, C. Tanteri / J. Math. Pures Appl. 81 (2002) 311341 325
5. The complex eikonal equation
We now discuss Example 1.1. We will rst derive a theorem on the rank one convex hull and then go back to the differential
equation. First recall the notation of Appendix A. For a matrix R
mn
,
=
_
_
_

1
1

1
n
.
.
.
.
.
.

m
1

m
n
_
_
_=
_
_
_

1
.
.
.

m
_
_
_=(
1
, . . . ,
n
)
we let 0
1
()
mn
() be its singular values.
We then have the following
Theorem 5.1. (1) Let mn, r
1
, . . . , r
m
>0 and
E =
_
R
mn
:
_

i
;
j
_
=r
i
r
j

ij
_
,
where
ij
is the Kronecker symbol. Let
A=diag
_
1
r
1
, . . . ,
1
r
m
_
R
mm
then
E =
_
R
mn
:
i
(A) =1, i =1, . . . , m
_
, RcoE =coE =
_
R
mn
:
m
(A) 1
_
.
(2) Let mn, r
1
, . . . , r
n
>0 and
E =
_
R
mn
:
i
;
j
=r
i
r
j

ij
_
,
where
ij
is the Kronecker symbol. Let
A=diag
_
1
r
1
, . . . ,
1
r
n
_
R
nn
then
E =
_
R
mn
:
i
(A) =1, i =1, . . . , n
_
, RcoE =coE =
_
R
mn
:
n
(A) 1
_
.
Remark 5.2. (1) The rst case with m=2 will apply to the complex eikonal equation.
(2) If we consider the second case with m=3, n =2, r
1
=r
2
and
=Du(x, y) =
_
_
u
1
x
u
1
y
u
2
x
u
2
y
u
3
x
u
3
y
_
_
then Du E means, in geometrical terms, that the surface has been parametrized globally by isothermal coordinates.
(3) The case m = n = 2 has been established in [6]. Recently Bousselsal and Le Dret [1] (still when m = n = 2), in the
context of nonlinear elasticity, found that (cf. their Theorem 3.11 with =0) if r
1
=r
2
=1 then
F =
_
R
22
:

2
+

1
;
2
_

1, i =1, 2
_
RcoE.
This is of course compatible with the result of [6] and of the above theorem. Note however that F =RcoE since, for example,
=
_
_
1

3
0

3
0
_
_
RcoE, / F.
Proof. Obviously the two cases are transposed one from each other and we therefore will only deal with the second one. We
start by letting
X =
_
R
mn
:
n
(A) 1
_
.
Step 1: We rst prove that RcoE coE X. The rst inclusion RcoE coE always holds and the second one follows
from the following two observations.
326 B. Dacorogna, C. Tanteri / J. Math. Pures Appl. 81 (2002) 311341
First we note that the set X is convex since the function
n
(A) is convex (cf., for example, Lemma 7.10 in [6]). Next
observe that the inclusion E X holds. Indeed if E (note that A=(
1
/r
1
, . . . ,
n
/r
n
)) then
E
_
(A)
i
; (A)
j
_
=
ij
A O(m, n)

(A) =1, 1 n.
Step 2: We now discuss the reverse inclusions X RcoE coE. Let X. Replacing by A we can assume, without
loss of generality, that A=I
nn
. Applying Theorem A.4, we can nd R O(n) such that
R =

=
_

1
, . . . ,

n
_
, with
_

i
;

j
_
=

ij
,
i
() =

.
Since the sets E and X are invariant under the (right) action of O(n) we will assume, without loss of generality, that
=(
1
, . . . ,
n
), with
i
;
j
=0, i =j,
i
() =|
i
| 1, 1 i n.
It is therefore sufcient to show that such belongs to RcoE.
Assume rst that |
i
| >0, i =1, . . . , n and write
=(
1
, . . . ,
n
) =
1 +|
1
|
2

+
+
1 |
1
|
2

=
1 +|
1
|
2
_

1
|
1
|
,
2
, . . . ,
n
_
+
1 |
1
|
2
_

1
|
1
|
,
2
, . . . ,
n
_
.
Note that rank{
+

} =1 and that if

=(

1
,

2
, . . . ,

n
) then
_

i
;

j
_
=0, i =j,

=1,

1, i =2, . . . , n.
Iterating the procedure with the second component and then with the other ones we conclude that can be written as a rank one
convex combination of elements of E, i.e., RcoE.
The case where some of the |
i
| =0 is handled similarly. For example if
1
=0 we write
=(
1
, . . . ,
n
) =
1
2

+
+
1
2

=
1
2
(e
1
,
2
, . . . ,
n
) +
1
2
(e
1
,
2
, . . . ,
n
),
where e
1
is any vector of R
m
such that
|e
1
| =1,
i
; e
1
=0, i =2, . . . , n.
Iterate again the procedure to deduce that RcoE, as claimed. This concludes the proof.
We can nally apply the results in Section 2 to obtain the following existence theorem for the complex eikonal equation (the
case n =2 is already in [6]).
Corollary 5.3. Let R
n
be a bounded open set, f : R R R, f = f (x, u, v), a continuous function and
W
1,
(; C). Then there exists w W
1,
(; C) satisfying
n

i=1
w
2
x
i
+f
2
=0, a.e. in ,
w =, on ,
(10)
where w
x
i
=w/x
i
. Or, in other words, there exists (u, v) W
1,
(; R
2
) such that
|Dv|
2
=|Du|
2
+f
2
, a.e. in ,
Dv; Du =0, a.e. in ,
(u, v) =(
1
,
2
), on .
Proof. In fact we solve a more restrictive problem of the type of the above theorem, i.e.,
|Du|
2
=r
2
, a.e. in ,
|Dv|
2
=r
2
+f
2
, a.e. in ,
Dv; Du =0, a.e. in ,
(u, v) =(
1
,
2
), on ,
B. Dacorogna, C. Tanteri / J. Math. Pures Appl. 81 (2002) 311341 327
where r >0 is chosen so large that

2
(AD) 1
for >0, i.e.,
D(x) compactly contained in int RcoE, a.e. x .
We may then apply Corollary 2.9 with F
1
=
1
+
2
2, F
2
=
2
1 (in case f is constant, otherwise proceed as in [6]).
6. A problem of optimal design
We will denote, in this section, the set of 2 2 symmetric matrices by R
22
s
. Our algebraic result is as follows.
Theorem 6.1. Let
E =
_
R
22
s
: trace {0, 1}, det 0
_
,
then
RcoE =coE =
_
R
22
s
: 0 trace 1, det 0
_
, int RcoE =
_
R
22
s
: 0 <trace <1, det >0
_
.
Remark 6.2. Note that it is slightly surprising that the rank one convex hull is in fact convex since the function det is not
convex.
Proof. We call
X =
_
R
22
s
: 0 trace 1, det 0
_
, Y =
_
R
22
s
: 0 <trace <1, det >0
_
.
Step 1: We rst prove that
RcoE coE X.
The rst inclusion always holds and the second one follows from the fact that E X and that X is convex. Indeed let , X,
0 t 1 we wish to show that t + (1 t ) X. It is clear that the rst inequality in the denition of X holds since
trace is linear. We now show the second one. Observe rst that since det =
11

22

2
12
, det =
11

22

2
12
0 and
trace , trace 0 then
11
,
22
,
11
,
22
0 and we therefore have (assume below that
11
,
11
> 0 otherwise the inequality
below is trivial)

;
11

22
+
11

22
2
12

12

11

2
12

11
+
11

2
12

11
2
12

12
=
(
11

12

11

12
)
2

11

11
0.
We therefore deduce that
det
_
t +(1 t )
_
=t
2
det +t (1 t )

; +(1 t )
2
det 0.
Step 2: We now prove that
X RcoE.
Since X is compact, as usual, it is enough to prove that X RcoE. However it is easy to see that
X =E
_
R
22
s
: 0 trace 1, det =0
_
and therefore the proof will be completed once we will show that the second set in the right-hand side is in RcoE. Assume that
is such that 0 <t =trace <1 and det =0. We can then write
=
_
x

x(t x)

x(t x) t x
_
=t
1
+(1 t )
2
=t
_

(1 )

(1 ) 1
_
+(1 t )
_
0 0
0 0
_
,
where x =t . The result follows from the facts that
1
,
2
E and det(
1

2
) =0.
Step 3: The fact that Y =int RcoE is easy.
328 B. Dacorogna, C. Tanteri / J. Math. Pures Appl. 81 (2002) 311341
Combining the above theorem with Corollary 2.9 we get
Theorem 6.3. Let R
2
be an open set and C
2
piec
( ) satisfy
0 (x) 1, a.e. x ,
det D
2
(x) >0, a.e. x ,
or W
2,
() such that
(x) 1 , a.e. x ,
det D
2
(x) , a.e. x ,
for some >0; then there exists w +W
2,
0
()
w(x) {0, 1}, a.e. x ,
det D
2
w(x) 0, a.e. x .
Remark 6.4. The above theorem has been proved (except the case with W
2,
boundary data) in Theorem 3.12 in [6] using the
method of confocal ellipses of Murat and Tartar. However the proof we have here (cf. also [7]) relies on the abstract existence
result of Section 2 and on the algebraic theorem above. Of course the use of the abstract theorem is more exible, because we
could imagine, for example, to replace 1 by a function a(x, u, Du), which is out of reach by the explicit method.
7. A rst academic example
Inspired by the two preceding examples we look to the problem (recalling that we denote the singular values of a matrix
R
22
by 0
1
()
2
()).
Theorem 7.1. Let 0 , <1 and
E =
_
R
22
:
1
() +
2
() =1 , det
_
then
RcoE =
_
R
22
:
1
() +
2
() 1 , det
_
,
int RcoE =
_
R
22
:
1
() +
2
() <1 , det >
_
.
Remark 7.2. Note that if (1 )
2
4 < 0 then E = . It can also be proved that if f () =
1
() +
2
() and g() =
||
2
2det then
coE =
_
R
22
: f() 1 , g() (1 )
2
4
_
,
which in the case =0 takes the simpler form
coE =
_
R
22
:
1
() +
2
() 1
_
.
Proof. We let
X =
_
R
22
:
1
() +
2
() 1 , det
_
.
The fact that RcoE X is elementary since E X and the functions
1
() +
2
() and det are polyconvex. So
we only need to show the converse inclusion. The compactness of X implies that the result will be proved if we can show that
X RcoE. Since
X =E
_
R
22
:
1
() +
2
() 1 , det =
_
,
we only need to show that any R
22
with
1
() +
2
() < 1 and det = belongs to RcoE. Choose R
22
be
any matrix of rank one such that

;
11

22
+
11

22

12

21

21

12
=0.
B. Dacorogna, C. Tanteri / J. Math. Pures Appl. 81 (2002) 311341 329
Dene then for t R

t
= +t
and observe that by construction det
t
= det =. Using again the compactness argument we can nd t
1
< 0 < t
2
such that

t
1
,
t
2
E, i.e.,

1
(
t
i
) +
2
(
t
i
) =1 , i =1, 2;
the result then follows at once (the representation formula for int RcoE is easily deduced).
As a corollary we obtain
Corollary 7.3. Let R
2
be an open set and C
1
piec
( ) be such that

1
(D) +
2
(D) <1, det D >0, a.e. in .
Then there exists u +W
1,
0
() satisfying

1
(Du) +
2
(Du) =1, det Du >0, a.e. in .
Proof. This theorem follows either directly from Theorem 2.4 or via the approximation property as done below. First nd

0
>0 so that
det D
0
>0.
Let
0
and
E

=
_
R
22
:
1
() +
2
() =1 , det
_
(E =E

0
) hence according to the above theorem we have
RcoE

=
_
R
22
:
1
() +
2
() 1 , det
_
.
We may then combine Theorem 2.2 and Theorem 2.8 to get the result.
8. A second academic example
We will now consider Example 1.4 and we will compute its rank one convex hull. Recall that the set of 2 2 symmetric
matrices is denoted by R
22
s
.
Theorem 8.1. Let a
ij
>0, i, j =1, 2, with a
12
=a
21
. Let
E =
_
=(
ij
) R
22
s
: |
ij
| =a
ij
, i, j =1, 2
_
,
then
coE =
_
=(
ij
) R
22
s
: |
ij
| a
ij
, i, j =1, 2
_
.
(1) If a
11
a
22
a
2
12
<0 then
RcoE =
_
=(
ij
) R
22
s
: |
12
| =a
12
, |
11
| a
11
, |
22
| a
22
_
.
(2) If a
11
a
22
a
2
12
=0 then
RcoE =
_
=(
ij
) R
22
s
: |
ij
| a
ij
, i, j =1, 2, |a
22

11
a
11

22
| det =
11

22
+
2
12
_
.
(3) If a
11
a
22
a
2
12
>0 then
RcoE
_
=(
ij
) R
22
s
: |
ij
| a
ij
, i, j =1, 2, |a
22

11
a
11

22
| a
11
a
22
a
2
12
det
_
.
Remark 8.2. (1) If we consider the 2 2 matrix 0 then it is clear that in Case 1: 0 / RcoE, while in Case 2: 0 RcoE but
0 / int RcoE. It can be shown, however, that in Case 3: 0 int RcoE.
330 B. Dacorogna, C. Tanteri / J. Math. Pures Appl. 81 (2002) 311341
(2) To apply the above theorem to partial differential equations one needs that int RcoE =; this does not happen in Case 1,
contrary to the two other cases. However we need also the approximation property (cf. Denition 2.7) of the rank one convex
hull that we are not able, at the moment, to prove.
(3) In Case 3 we were not able to nd a complete characterization of RcoE; the set given in the right-hand side of the
inclusion is too large.
Proof. The representation formula for the convex hull is trivial.
Case 1: We denote by
X =
_
=(
ij
) R
22
s
: |
12
| =a
12
, |
11
| a
11
, |
22
| a
22
_
.
(1) It is clear that X RcoE. Indeed write any X (assume without loss of generality that
12
=a
12
) as
=
_

11
a
12
a
12

22
_
=
a
11
+
11
2a
11
_
a
11
a
12
a
12

22
_
+
a
11

11
2a
11
_
a
11
a
12
a
12

22
_
and similarly
_
a
11
a
12
a
12

22
_
=
a
22
+
22
2a
22
_
a
11
a
12
a
12
a
22
_
+
a
22

22
2a
22
_
a
11
a
12
a
12
a
22
_
to deduce that RcoE.
(2) We now show the reverse inclusion. Observe rst that trivially E X. Therefore to get the claimed result it is enough
to show that X is a rank one convex set. So let , X with det( ) = 0 and 0 < t < 1. Note that since , X then
(
12

12
)
2
is either 0 or 4a
2
12
. The second case cannot happen since we would have
0 = det( ) =(
11

11
)(
22

22
) (
12

12
)
2

_
|
11
| +|
11
|
__
|
22
| +|
22
|
_
4a
2
12
4a
11
a
22
4a
2
12
<0,
(11)
which is absurd. So the only case that can happen is
12
=
12
(with |
12
| =a
12
). The claimed result t +(1 t ) X is then
immediate.
Case 2: As before we call
X =
_
=(
ij
) R
22
s
: |
ij
| a
ij
, i, j =1, 2, |a
22

11
a
11

22
| det =
11

22
+
2
12
_
.
(1) We easily see that E X and that X is a rank one convex (in fact even polyconvex) set since all the functions involved
with the inequalities are polyconvex and thus rank one convex. We therefore have RcoE X.
(2) We now discuss the inclusion X RcoE. We start by observing that if we can show (cf. below) that X RcoE then
the result will follow. Indeed if int X, since X is compact, we can nd for every R
22
s
with rank = 1, t
1
< 0 < t
2
,
such that
+t
1
, +t
2
X
and hence since X RcoE we get that RcoE.
We now wish to show that if X then RcoE. Note rst that the last inequality in the denition of X is equivalent,
bearing in mind that a
11
a
22
a
2
12
=0, to
0 a
2
12

2
12
(a
11

11
)(a
22
+
22
), 0 a
2
12

2
12
(a
11
+
11
)(a
22

22
). (12)
Observe that if either |
11
| =a
11
or |
22
| =a
22
then by (12) necessarily |
12
| =a
12
. However if |
12
| =a
12
then, by the same
argument as in Case 1, we deduce that RcoE.
So we now assume that |
ij
| < a
ij
and (since X) one of the inequalities in (12) is an equality and without loss of
generality say the rst one while the second one is a strict inequality. If we call
V
1
=
_
=(
ij
) R
22
s
: a
2
12

2
12
=(a
11

11
)(a
22
+
22
)
_
,
V
2
=
_
=(
ij
) R
22
s
: a
2
12

2
12
=(a
11
+
11
)(a
22

22
)
_
,
Y
1
= X V
1
then relint Y
1
(the relative interior of Y
1
). We can then choose
=
_

2
1

1

2

2
2
_
B. Dacorogna, C. Tanteri / J. Math. Pures Appl. 81 (2002) 311341 331
with
1
,
2
=0 so that
+t V
1
, t R;
this is always possible by choosing

1
=1,
2
=
a
12
+
12
a
11

11
or more generally any nonzero solution of

2
2
(a
11

11
) +2
1

12

2
1
(a
22
+
22
) =0.
Then since relint Y
1
and Y
1
is compact we can nd t
1
<0 <t
2
, such that
+t
1
, +t
2
Y
1
.
But

Y
1
means that either |

ij
| =a
ij
for a certain i, j and this case has already been dealt with or

V
2
. Hence the only
case that requires still to be analyzed is when |
ij
| < a
ij
and V
1
V
2
, i.e. when in (12) the two inequalities are actually
equalities. Note that any V
1
V
2
is of the form
=
11
_
1

a
22
/a
11

a
22
/a
11
a
22
/a
11
_
and thus if we denote by
=
_
1

a
22
/a
11

a
22
/a
11
a
22
/a
11
_
,
which is a matrix of rank one, we nd that
+t V
1
V
2
, t R.
The usual argument then applies, namely if |
ij
| <a
ij
and V
1
V
2
we can nd t
1
<0 <t
2
, such that one of the inequalities
|
ij
| <a
ij
becomes an equality; in which case we conclude that RcoE by the previous steps.
Case 3: The claimed inclusion follows for the same reasons as in Case 2.
9. The case of potential wells
We now discuss how to apply Theorem 3.2 to the case of two potential wells under incompressibility constraint (this result
has recently been obtained by Mller and Sverak [14] and we show here how our method gives the same result). We start, as
usual, with some algebraic considerations.
Proposition 9.1. Let 0 < <1,
=
_
0
0 1/
_
and
E =SO(2)I SO(2).
Let
F() =
_
(
11

22
)
2
+(
12
+
21
)
2
+
_
_
1

11

22
_
2
+
_

12
+
1

21
_
2
,
G() =(
11

22
)
2
+(
12
+
21
)
2
+
_
1

11

22
_
2
+
_

12
+
1

21
_
2
.
Then F and G are convex and invariant under the (left) action of SO(2). Moreover
332 B. Dacorogna, C. Tanteri / J. Math. Pures Appl. 81 (2002) 311341
E =
_
R
22
: F() =
1

, G() =
_
1

_
2
, det =1
_
,
RcoE =
_
R
22
: F()
1

, det =1
_
,
int RcoE =
_
R
22
: F() <
1

, det =1
_
.
Remark 9.2. Note that
1
= diag(1 , 1/(1 )) int RcoE and
2
= diag(/(1 ), (1 )/) int RcoE for > 0
sufciently small.
Proof. The result follows from the representation obtained by Sverak and Corollary 8.3 of [6].
(1) The fact that F and G are convex and invariant under the (left) action of SO(2) is easy.
(2) Let us show now that if
X =
_
R
22
: F() =
1

, G() =
_
1

_
2
, det =1
_
then E =X. The inclusion E X is easy since
F(I) =F() =
1

, G(I) =G() =
_
1

_
2
, det I =det =1
and F, G and det are invariant under the (left) action of SO(2). We now discuss the reverse inclusion. Let X then either
_
(
11

22
)
2
+(
12
+
21
)
2
_
=0
which implies that SO(2) or
__
1

11

22
_
2
+
_

12
+
1

21
_
2
_
=0
which implies that SO(2). In either cases we nd that E.
(3) Call
Y =
_
R
22
: F()
1

, det =1
_
.
We now show that RcoE =Y. To prove this we use the representation formula established by Sverak (cf. [6]), i.e.,
RcoE =
_
=
_
y
1
y
2
y
2
y
1
__
1 0
0 1
_
+
_
z
1
z
2
z
2
z
1
__
0
0 1/
_
,
_
y
2
1
+y
2
2
+
_
z
2
1
+z
2
2
1, det =1
_
.
Expressing y
1
, y
2
, z
1
, z
2
in terms of
ij
we nd
_

11
=y
1
+z
1

12
=
_
y
2
+
1

z
2
_

21
=y
2
+z
2

22
=y
1
+
1

z
1

_
_
1

_
y
1
=
1

11

22
_
1

_
y
2
=
12
+
1

21
_
1

_
z
1
=(
11

22
)
_
1

_
z
2
=(
12
+
21
)
.
Since
F() =
_
1

_
_
_
y
2
1
+y
2
2
+
_
z
2
1
+z
2
2
_
we get immediately the result.
B. Dacorogna, C. Tanteri / J. Math. Pures Appl. 81 (2002) 311341 333
(4) We now discuss the representation formula for int RcoE. Call Z the right-hand side in the formula. The inclusion
Z int RcoE is clear and so we show the reverse one. Let int RcoE; then up to a rotation we can always assume that

12
=0 and hence since det =1 we deduce that if

t
=
_

11
0

21
+t 1/
11
_
,
then =
0
and
det
t
1, t R.
Since int RcoE we nd that
t
RcoE for all t small enough. Observe nally that the function t F(
t
) is strictly convex
(note however that the function F() is not strictly convex) and therefore if t =0 is small enough we have
F() <
1
2
F(
t
) +
1
2
F(
t
)
1

which is the claimed result Z.

Theorem 9.3. Let R
n
be open and
E =SO(2)A SO(2)B
with det A=det B >0. Let
int RcoE
then there exists u W
1,
(; R
2
) such that
Du(x) E, a.e. in ,
u(x) =x, on .
Remark 9.4. If det A=det B this result was already obtained by Mller and Sverak [13] and Dacorogna and Marcellini [5].
Proof. Step 1: We start with some algebraic considerations. Observe that there is no loss of generality if we assume that
A=I and B =
_
0
0 1/
_
.
Indeed rst diagonalize BA
1
, i.e., nd R
a
, R
b
SO(2) so that
R
a
BA
1
R
b
==
_
0
0 1/
_
we therefore deduce that
R
b
EA
1
R
b
=SO(2)I SO(2).
Step 2: We dene for (0, 1]
I

=
_
1 0
0 1/(1 )
_
,

=
_
/(1 ) 0
0 (1 )/
_
.
Observe that I

int RcoE and if

E

=SO(2)I

SO(2)

and accordingly F

, G

, we then have
E

int RcoE and K(E

) =RcoE

int RcoE.
Therefore E and RcoE have the approximation property with K(E

) = RcoE

. Hence combining Proposition 9.1 with

Theorem 3.6 and Theorem 3.2 we get the result.
334 B. Dacorogna, C. Tanteri / J. Math. Pures Appl. 81 (2002) 311341
10. The case of nematic elastomers
The problem considered here has been introduced by DeSimone and Dolzmann [9]. We begin with the computation of the
rank one convex hull; this follows from [6,8] (in the case n =2, 3 cf. [9]).
Theorem 10.1. Let 0
1
(A)
n
(A) denote the singular values of a matrix A R
nn
and
E =
_
A:
i
(A) =a
i
, i =1, . . . , n, det A=
n

i=1
a
i
_
,
where 0 <a
1
a
n
. The following then holds:
PcoE =RcoE =
_
A:
n

i=

i
(A)
n

i=
a
i
, =2, . . . , n, det A=
n

i=1
a
i
_
.
Moreover if 0 <a
1
< <a
n
and is sufciently small so that
0 <a

1
=(1 )
1n
a
1
a

2
=(1 )a
2
a

n
=(1 )a
n
,
then E and RcoE have the approximation property with K(E

) =RcoE

, where
E

=
_
A:
i
(A) =a

i
, i =1, . . . , n, det A=
n

i=1
a
i
_
.
Remark 10.2. (i) Note that when n = 3 and a
1
= a
2
= r
1/6
, a
3
= r
1/3
, r < 1, we recover the result of DeSimone and
Dolzmann, namely:
PcoE =RcoE =
_
A R
33
:
i
(A)
_
r
1/6
, r
1/3
_
, det A=1
_
.
(ii) The hypothesis that all the a
i
are different is too strong and can be weakened; it is enough to assume that the a
i
are not
all equal. It is clear also that if all the a
i
are equal, then int RcoE =, since then E =RcoE.
Proof. Let us denote by
X =
_
A R
nn
:
n

i=

i
(A)
n

i=
a
i
, =2, . . . , n, det A=
n

i=1
a
i
_
.
Step 1: The fact that RcoE PcoE X is easy, since E X and the functions
A
n

i=

i
(A)
n

i=
a
i
, =2, . . . , n,
are polyconvex and Adet A

n
i=1
a
i
is quasiafne.
Step 2: As usual by compactness of X it is enough to prove that X RcoE. We show the result by induction.
(1) n =1. This is trivial.
(2) n 2. Any A X can, without loss of generality, be assumed of the form
A=
_
_
_
x
1
.
.
.
x
n
_
_
_
with 0 x
1
x
2
x
n
,

n
i=
x
i

n
i=
a
i
, = 2, . . . , n, and

n
i=1
x
i
=

n
i=1
a
i
. Since A X we deduce that

n
i=
x
i
=

n
i=
a
i
, for a certain {2, . . . , n}. We can then apply the hypothesis of induction to
{x
1
, . . . , x
1
} and {a
1
, . . . , a
1
}
and to
{x

, . . . , x
n
} and {a

, . . . , a
n
}.
Indeed for the second one this follows from the hypotheses
B. Dacorogna, C. Tanteri / J. Math. Pures Appl. 81 (2002) 311341 335
n

i=
x
i

n

i=
a
i
, = +1, . . . , n,
n

i=
x
i
=
n

i=
a
i
while (note that if =2 then necessarily x
1
=a
1
and this part is trivial , so we will assume that 3) for the rst one we have
1

i=
x
i
=
n

i=
x
i
_
n

i=
x
i
_
1
=
n

i=
x
i
_
n

i=
a
i
_
1

i=
a
i
_
n

i=
a
i
_
1
=
1

i=
a
i
, =2, . . . , 1,
and
1

i=1
x
i
=
1

i=1
a
i
.
We can therefore deduce, by hypothesis of induction, that A RcoE.
Step 3: We now observe that the approximation property follows from the fact that (if 0 <a
1
< <a
n
)
int RcoE =
_
A:
n

i=

i
(A) <
n

i=
a
i
, =2, . . . , n, det A=
n

i=1
a
i
_
.
The main result is then, adopting the notations of the above theorem (assume here that

n
i=1
a
i
=1).
Theorem 10.3. Let R
n
be open, 0 <a
1
< <a
n
and
E =
_
A R
nn
:
i
(A) =a
i
, i =1, . . . , n, det A=
n

i=1
a
i
=1
_
.
Let be an afne function (D =) such that
n

i=

i
() <
n

i=
a
i
, =2, . . . , n,
det =1;
then there exists (a dense set of ) u +W
1,
0
(; R
n
) such that

i
(Du(x)) =a
i
, i =1, 2, . . . , n, a.e. x ,
det Du(x) =1, a.e. x .
Remark 10.4. Of course a similar result holds if the a
i
are not constants but depend on (x, u).
Proof. Observe rst that
E =
_
A R
nn
:
n

i=

i
(A) =
n

i=
a
i
, =2, . . . , n, det A=1
_
and that A

n
i=

i
(A), = 2, . . . , n, are quasiconvex. The result follows then by combining Proposition 10.1 with
Theorem 3.2 and Theorem 3.6.
11. Gauges, Choquet functions and Minkowski theorem for polyconvex sets
We conclude this article with some general considerations about polyconvex sets. In classical convex analysis the gauge of
a convex set, the Choquet function that characterizes extreme points or the Minkowski theorem (often known as KreinMilman
theorem which is its innite-dimensional version) are important tools. We generalize these notions to polyconvex sets.
We rst recall some notations and denitions and we refer to [3,6] for more details.
Denition 11.1. (1) For a matrix A R
mn
we let
T (A) =(A, adj
2
A, . . . , adj
mn
A),
336 B. Dacorogna, C. Tanteri / J. Math. Pures Appl. 81 (2002) 311341
where adj
s
A stands for the matrix of all s s subdeterminants of the matrix A, 1 s mn =min{m, n}, and
(m, n) =
mn

s=1
_
m
s
__
n
s
_
,
where
_
m
s
_
=
m!
s!(ms)!
.
(2) The different envelopes of a given function are dened as
Cf =sup{g f : g convex},
Pf =sup{g f: g polyconvex},
Rf =sup{g f : g rank one convex},
they are respectively the convex, polyconvex and rank one convex envelope of f .
(3) We say that a set K R
mn
is polyconvex if for every t
i
0 with

(m,n)
i=1
t
i
=1 and every A
i
K with
(m,n)

i=1
t
i
T (A
i
) =T
_
(m,n)

i=1
t
i
A
i
_
,
then
(m,n)

i=1
t
i
A
i
K.
We start with a theorem dening the gauge of a polyconvex set.
Theorem 11.2. Let K R
mn
be a nonempty polyconvex set and let

K
(x) =
_
0, if x K,
+, if x / K
be its indicator function. Let H : R
(m,n)
R=R{+} be dened as
H() = sup
xK
__
T (x);
__
.
The following then hold:
(1) H is lower semicontinuous, convex and positively homogeneous of degree one.
(2) If K is closed and if H

: R
(m,n)
R=R{+} is the conjugate function of H (i.e., H

) =sup{

; H()})
then

K
(x) =H

_
T (x)
_
, K =
_
x R
mn
: H

_
T (x)
_
0
_
.
(3) If 0 K then H() H(0) =0. And if K is compact then H takes only nite values.
(4) If 0 int K and if K is compact then
H() =0 =0;
and in this case
K =
_
x R
mn
: H
0
_
T (x)
_
1
_
,
where H
0
is the polar of H (called the gauge of K), i.e.,
H
0
(

) = sup
=0
_

;
H()
_
.
B. Dacorogna, C. Tanteri / J. Math. Pures Appl. 81 (2002) 311341 337
Remark 11.3. (1) When m=n =2 we have H: R
22
RR=R{+} is given by
H(, ) = sup
xK
_
x; + det x
_
and K =
_
x R
22
: H

(x, det x) 0
_
.
(2) Note that H
0
is positively homogeneous of degree one but of course this is not the case for the function x H
0
(T (x)).
Example 11.4. Let for R
22
, 0
1
()
2
() denote its singular values and
K =
_
R
22
:
2
() a
2
,
1
()
2
() a
1
a
2
_
,
which is a polyconvex set (cf. [6]).Then
H
0
(

) =max
_

2
(

)
a
2
,
|

|
a
1
a
2
_
is a gauge for K.
Proof. (1) Since K is nonempty then H > . H being the supremum of afne functions, it is convex and lower
semicontinuous. The fact that H is positively homogeneous of degree one is easy.
(2) Observe rst that according to a result in [3] (cf. pp. 199202) we have
H() =
p
K
(),
K
(x) =
pp
K
(x) =H

_
T (x)
_
hence the result.
(3) This is obvious.
(4) We now show that if 0 int K and if K is compact then
H() =0 =0.
The implication () follows from (1) and we therefore discuss only the converse one. Let x R
mn
be an arbitrary point.
Since 0 int K we deduce that for every sufciently small then x/|x| K and therefore
0 =H()
_
T
_
x
|x|
_
;
_
(13)
since x R
mn
is arbitrary the above inequality implies that =0, as claimed. We prove this last fact only when m=n =2,
the general case being proved similarly. The inequality (13) reads then (writing =(x

, ))
0 =H()

|x|
x; x

+
2

det x
|x|
2
, x R
22
.
We therefore get, using the arbitrariness of
x; x

=0, x R
22
,
det x 0, x R
22
,
hence (x

, ) =(0, 0).
The last identity
K =
_
x R
mn
: H
0
_
T (x)
_
1
_
is easy.
The next step is to dene a function that characterizes the extreme points. In the convex case this is known as the Choquet
function (see, for example, work of Pianigiani [15]); but rst let us dene the following.
Denition 11.5. Let K R
mn
be polyconvex; we say that X K is an extreme point in the polyconvex sense of K if
T (X) =

I
i=1
t
i
T (A
i
)
t
i
>0 with

I
i=1
t
i
=1, A
i
K
_

_
A
i
=X, i =1, . . . , I.
The set of extreme points in the polyconvex sense of K is denoted by K
p
ext
.
338 B. Dacorogna, C. Tanteri / J. Math. Pures Appl. 81 (2002) 311341
Theorem 11.6. Let K R
mn
be a nonempty compact polyconvex set and K
p
ext
be its extreme points in the polyconvex sense.
Then there exists : R
mn
R=R{+} a polyconvex function so that
K
p
ext
=
_
x K: (x) =0
_
, (x) 0 x K.
Proof. We rst dene
f(x) =
_
|x|
2
, if x K,
+, otherwise
and (x) =
_
Pf (x) f(x), if x K,
+, otherwise.
In the convex case it is the function that is the Choquet function. Observe that : R
mn
R=R{+} is polyconvex and
that
(x) 0, if x K, (x) =0, if x K
p
ext
.
Indeed the inequality is clear since in K the function f is nite and, by denition, Pf is always not larger than f . We now
show that
(x) =0 x K
p
ext
.
Note that if x K then
(x) =|x|
2
+ inf
x
i
K
_

(m,n)+1

i=1
t
i
|x
i
|
2
: T (x) =
+1

i=1
t
i
T (x
i
), t
i
0 with
+1

i=1
t
i
=1
_
.
Therefore if x K
p
ext
, we deduce, by denition, that in the inmum the only admissible x
i
are x
i
= x; and hence we have
(x) =0. We now show the converse implication, i.e., (x) =0 x K
p
ext
. From the above representation formula we obtain,
since (x) =0 and x K, that
|x|
2
= sup
x
i
K
_
(m,n)+1

i=1
t
i
|x
i
|
2
: T (x) =
+1

i=1
t
i
T (x
i
)
_
.
Combining the above with the convexity of the function x |x|
2
we get that
|x|
2

(m,n)+1

i=1
t
i
|x
i
|
2

(m,n)+1

i=1
t
i
x
i

2
=|x|
2
;
the strict convexity of x |x|
2
implies then that x
i
=x. Thus x K
p
ext
.
We now have the following version of Minkowski theorem.
Theorem 11.7. Let E R
mn
be a non empty compact set. Let E
p
ext
be the extreme points in the polyconvex sense of PcoE,
then
PcoE =PcoE
p
ext
.
Proof. We adapt here an idea of Zhang [18].
Step 1: We rst prove that if K is a compact and polyconvex set then it has at least one extreme point in the polyconvex
sense, i.e., K
p
ext
= . Let coK be the convex hull of K, which is a compact and convex set. It is a well established fact in
convex analysis that coK has then at least one extreme point (in the convex sense). Since, by denition, any extreme point (in
the convex sense) is an extreme point in the polyconvex sense, we deduce the result.
Step 2: We next let
K =PcoE; L=PcoE
p
ext
.
The only nontrivial inclusion is K L. We then dene
g(X) =
_
dist(X; L), if X K,
+, otherwise,
f (X) =Pg(X) 0. (14)
We could also have taken, if we want a function that is nite everywhere,
B. Dacorogna, C. Tanteri / J. Math. Pures Appl. 81 (2002) 311341 339
g(X) =
_
dist(X; L)
_
(mn)+1
, f (X) =Pg(X) 0.
Observe that
L=
_
X K: f(X) =0
_
(15)
(this follows from the polyconvexity of L). Let
a =max
_
f (X): X K
_
0. (16)
We will show that a =0 which by (15) implies K L as claimed. Let
E
a
=
_
X K: f (X) =a
_
=,

K =PcoE
a
K. (17)
Since f is polyconvex and nonnegative we get

K
_
X K: 0 f (X) a
_
. (18)
It follows from Step 1 that
=

K
p
ext
E
a
. (19)
Assume for a moment that we can show
=

K
p
ext
E
p
ext
(20)
then combining (19), (20) and the fact that
E
p
ext
L=
_
X K: f (X) =0
_
we would then deduce that
=

K
p
ext
E
a
L.
This implies at once that a =0 and hence K L as claimed.
So it only remains to show (20). We thus let

K
p
ext
be such that
T () =
I

i=1
t
i
T (
i
), t
i
>0 with
I

i=1
t
i
=1,
i
K,
and we wish to show that

i
=, i =1, . . . , I, (21)
which implies that E
p
ext
and hence (20). Ordering differently the
i
K, if necessary, we have

i
E
a
_
f (
i
) =a
_
i =1, . . . , I
1
,

i
KE
a
_
f (
i
) <a
_
i =I
1
+1, . . . , I.
We rst show that I
1
=I . If this were not the case we would have from (19) and the polyconvexity of f that
a =f ()
I

i=1
t
i
f (
i
) =a
I
1

i=1
t
i
+
I
1
+1

i=1
t
i
f (
i
) <a
which is absurd, thus I
1
= I . However since

K
p
ext
(where

K = PcoE
a
) we therefore deduce that (21) holds and hence
E
p
ext
which is the claimed inclusion (20).
Acknowledgements
This research has been partially nanced by Fonds National Suisse (21-50472.97). We would like to thank A. DeSimone,
G. Dolzmann and P. Marcellini for interesting discussions.
340 B. Dacorogna, C. Tanteri / J. Math. Pures Appl. 81 (2002) 311341
Appendix A. Singular values
We recall here the denition and some properties of singular values of matrices (cf. work of Horn and Johnson [11, p. 152]
and [6, p. 171]). First we write any A R
mn
as
A=
_
_
_
a
1
1
a
1
n
.
.
.
.
.
.
a
m
1
a
m
n
_
_
_=
_
_
a
1
.
.
.
a
m
_
_
=(a
1
, . . . , a
n
).
We start with the following
Denition A.1. We denote by O(m, n) the set of orthogonal matrices R R
mn
, i.e.,
R
t
R =I
nn
where I
nn
denotes the identity matrix in R
nn
. When m=n, we write O(n) =O(n, n).
Remark A.2. If m=n, then in general
R
t
R =I
nn
RR
t
=I
mm
(i.e., R O(m, n) R
t
O(n, m)); while if m=n then these two properties are equivalent (i.e., R O(n) R
t
O(n)).
We now give the denition of the singular values.
Denition A.3. (1) Let mn and A R
mn
. The singular values of A, denoted by 0
1
(A)
m
(A), are dened to
be the square root of the eigenvalues of the symmetric and positive semidenite matrix AA
t
R
mm
.
(2) Let mn and A R
mn
. The singular values of A, denoted by 0
1
(A)
n
(A), are dened to be the square
root of the eigenvalues of the symmetric and positive semidenite matrix A
t
A R
nn
.
The following theorem is the standard decomposition theorem (cf. Horn and Johnson [10, Lemma 7.3.1]).
Theorem A.4. (1) Let mn, A R
mn
and 0
1
(A)
m
(A) be its singular values then there exists R O(m) such
that
RA=

A=
_
_
a
1
.
.
.
a
m
_
_
, with
_
a
i
;a
j
_
=

a
i

a
j

ij
,
i
(A) =

a
i

.
Furthermore there exists Q O(n, m) R
nm
(i.e., Q
t
Q=I
mm
) such that
RAQ=
_
_
_

1
(A) 0
.
.
.
.
.
.
.
.
.
0
m
(A)
_
_
_.
(2) Let mn, A R
mn
and 0
1
(A)
n
(A) be its singular values then there exists R O(n) such that
AR =

A=(a
1
, . . . ,a
n
), with a
i
;a
j
=|a
i
||a
j
|
ij
,
i
(A) =|a
i
|.
Furthermore there exists Q R
nm
with QQ
t
=I
nn
(i.e., Q
t
O(m, n)) such that
QAR =
_
_
_

1
(A) 0
.
.
.
.
.
.
.
.
.
0
n
(A)
_
_
_.
References
[1] M. Bousselsal, H. Le Dret, Remarks on the quasiconvex envelope of some functions depending on quadratic forms, Preprint, 1999.
[2] E.A. Coddington, N. Levinson, Theory of Ordinary Differential Equations, MacGraw-Hill, New York, 1955.
B. Dacorogna, C. Tanteri / J. Math. Pures Appl. 81 (2002) 311341 341
[3] B. Dacorogna, Direct Methods in the Calculus of Variations, Springer, Berlin, 1989.
[4] B. Dacorogna, P. Marcellini, Existence of minimizers for non quasiconvex integrals, Arch. Ration. Mech. Anal. 131 (1995) 359399.
[5] B. Dacorogna, P. Marcellini, Sur le problme de CauchyDirichlet pour les systmes dquations non linaires du premier ordre, C. R.
Acad. Sci. Paris Ser. I Math. 323 (1996) 599602.
[6] B. Dacorogna, P. Marcellini, Implicit Partial Differential Equations, Birkhuser, Boston, 1999.
[7] B. Dacorogna, P. Marcellini, C. Tanteri, Equations de type implicite avec contraintes, C. R. Acad. Sci. Paris Sr. I Math. 330 (2000)
271274.
[8] B. Dacorogna, C. Tanteri, On the different convex hulls of sets involving singular values, Proc. Roy. Soc. Edinburgh A 128 (1998) 1261
1280.
[9] A. DeSimone, G. Dolzmann, Material instabilities in nematic elastomers, Phys. D 136 (12) (2000) 175191.
[10] R.A. Horn, C.A. Johnson, Matrix Analysis, Cambridge University Press, 1985.
[11] R.A. Horn, C.A. Johnson, Topics in Matrix Analysis, Cambridge University Press, 1991.
[12] R. Magnanini, G. Talenti, On complex valued solutions to a 2D eikonal equation. Part one, qualitative properties, in: G.Q. Chen,
E. DiBenedetto (Eds.), Contemporary Mathematics Series, Vol. 238, American Mathematical Society, 1999, pp. 203229.
[13] S. Mller, V. Sverak, in: J. Jost (Ed.), Attainment Results for the Two-well Problem by Convex Integration, International Press, 1996,
pp. 239251.
[14] S. Mller, V. Sverak, Convex integration with constraints and applications to phase transitions and partial differential equations, J. Eur.
Math. Soc. 1 (1999) 393422.
[15] G. Pianigiani, Differential inclusions. The Baire category method, in: A. Cellina (Ed.), Methods of Nonconvex Analysis, Lecture Notes in
Math., Springer, Berlin, 1990, pp. 104136.
[16] R.T. Rockafellar, Convex Analysis, Princeton University Press, 1970.
[17] L. Tartar, Estimations nes des coefcients homogniss, in: P. Kre (Ed.), Ennio De Giorgi Colloquium, Research Notes in Math.,
Vol. 125, Pitman, 1985, pp. 168187.
[18] K. Zhang, On the structure of quasiconvex hulls, Ann. Inst. H. Poincar, Anal. Non Linaire 15 (1998) 663686.