~L-~f

--

" '.

'~",=-~1

,.

-' -.;.

Offprint.from "Archil'efor Ra/ional Mechanics and Analysis",

Volume 61, Number 4, /977, P. 317 - 401
(!) by Springer. Verlag /977
Printed in Germany

"

Convexity Conditions and Existence Theorems

in Nonlinear Elasticity
JOHN M. BALL
CO/ll/JlI/IIi('tI/ed

hy S.S. ANTMAN & C. M. DAFERMOS

Table of Contents
O. Introduction.
. . . .,
.'
..'
1. Boundary.value problems of non-linear hyperclasticity .
~. Technical preliminaries. . . . : . . , . . . .
3. Quasiconvcxity and thc Le~en,lre-F!adamard wndition
4. Sufficient ,'onditions for quasiconvesity .
5. '''''flIpic convcs and polywn'es functi,>ns . . .'
.
..
6. Sequential weak continuity of marrin~s on Orlicl-Sohole\ spaces
7. Esistence theorems.
..,..,....
S. Applications h> specific models of cl.lSt;" materials.
. , .'.
9. An example of non-uniljueness: bucklin~ of a fI,d . . . .
10. C()nclud;n~ remarb.
. .
Aekn()wled~ement . . . . .. . . .
References. . , '."
...,..

.
. . . . . . .
. . .
. . . . .,
. .
. . . . . .
. . .
. .

' . 337
. . 342
. . 347
349
. . 356
. . 363
. .'367
. . 373
. . 388
. . . . . 3q~
. . . . . 39M
. . . . . .,
399
. 399

O. Introduction

The: purpose of this arlicle is 10 presenl e:xisle:ncetheorems for various

equilibrium boundary-valuc problems of nonlinear elasticity in one. two and
threc dimcnsions under rcalislic hypolheses on the male:rial responsc. Although
some of Ihe results may be cxlcnded I,) covcr Cauchy elaslicily. we shall reslricl
our discussion to hyperelastic (Green elastic) malcrials. that is, to clastic mate~ .
rials possessing a slored-energy fundion. We ignore Ihermal en'ccts. For such
materials a Iypical boundary-value: problem lakes Ihe form of linding a vector
field uo: Q gin making the integral
/(u. Q)=

Sf(x.

(0.1)

u(x). Vu(x)) dx

!I

stationary in a suitable class of functions t. Here Q is a non-empty. bounded, open

subset

" = 1. 2, 3. The

of :JIf'n,

integrand

f will usually have the form

f(x. U. Vu):' 'I('(:c. Vu)+rp(x,

, For.traclion boundary-valuepn>hlcmsthere

(0.2)

11).

will also he a surface

int~~ral

term.

5.
...""

"

:..'

.~

'."\
Exi~tencc

J, M. RAI.I,

where '11"is the stored.energy. function and q, is a body force potential. In this
introduction we assumc that (0.2) holds and that 11 3. "
We attack this problem hy using the dircct methodof thecalculusof variations
to establish the existence of minimizers for (u..Q) in the class considered. Such
a programme has been successfully carried out hy ANTMAN[1-8J in an important
series of papers on the existence of equilibrium solutions in various problems
arising from theories of nonlinear elastic rods and axisymmetric shells (with or
without Cosserat structure). In these papers ANTMAN emphasizes the crucial
importance of choosing hypotheses on the material 'response that both ensure
the successof the analysis and are reasonable physically. In his work. and in mine.
the problem of c~istcnt'C is inextricably linked with that of finding satisfactory
constitutivc inequalities for nonlinear elasticity l TRUESI>ELL[I]).
'
As an illustration. consider the effect of imposing the constitutive requirement
that 'If' be convex with respect to Vu. This mat~ematically simple hypothesis.
whcn augmented with suitable smoothness nnd growth assumptions. ensures the
existencc of minimi7.crs for (0.1). (0.2). Existence theorems under this assumption
have been given by severnl nuthors (e.g., BEJIi[1.2]. OI>EN[I]). Unfortunately
these results arc only of IT1nthematical interest since convexity of 'If' with respect
to Vu is unncceptable physically t. Firstly. as was shown by COLEMAN& NOLL
[I] (sce also TRliESI>EU. & NOLI. [I. p, 163]) such convexity connicts with the
requirem'ellt that '11' be objective kt: (1.12)), Secondly. consider. for example.
a mixed displacement. dead load traction boundary-value problem for such a
material. Any equilibrium solution for such a problem must necessarily be an
nbsolute minimizcr for /(u, Q);' in pariicular. if a strict absolute minimizer exists
then it is the only equilibrium solution <I. This fact. which i~ an elementary
consequcnce of the theory of convex functions kt: MOREi\U [I], EKELAND &
TEMAM [IJ). arid for the truth of which '11' need not' be strictly convex. rules
out the nonuniqucness essential for the description of buckling <I<I.,Some less

restrictive 0Ondition on 'If' is therefore required.

A suitable condition. termed I/lIasi('o/ll'('xil.l'. wa~ introduced by MORREY[IJ
in a fundamental paper in 1952. 'If' is said to.be ,/lIasi<'mll'('x if

339

Theorems in Non-Linear Elasticity

homogeneous strain. we require that this homogeneous strain be' an absolute

minimizer for the total energy. Note that if in the above we admitted for consideration inhomogeneous bodies, or if we considered mixed displacement traction
boundary-value problems, then the condition would be unacceptable. as we
should expect certain buckled states to have lower total energy than the homo.geneous strain. As stated, however. the condition has a certain plausibility.
MORREY showed that if I(', 11..) is quasiconvex for every u. and if certain
continuity and growth hypotheses are satisfied. then for various boundary-value
problems there exist minimizers for feu. Q). Conversely. if 11is a minimizer for
I(u,.Q) among C1(.Q) functions satisfying given Dirichlet boundary conditions.
and if XoeQ, Fo= Vu (xo). then (0.3) holds. This fact may be used (sce Theorem 3.2)
to motivate quasiconvexity by showing that it is a necessary condition for the
existence of sufficiently regular minimizcrs for a class of displacement boundaryvalue problems. The degree of regularity required is: Qowever.' fairly severe.
Furthermore. if if" is quasiconvex and twice continuously diITerentiable, then
"If/' satisfies the Legendre-Hadamard or elliptidty condition t:
(12 if'"

" ~).')/

vII.. (Ill.P

"

'

(0.4)

JI.Jlp?;'O for all )..Jlel~ .

,

(It is not known whether the converse holds.) Because we have chosen to impose
quasiconvexity as a constitutive restriction, we must therefore regard the LegendreHadamard condition also as a constitutive restriction tt.
The statement above that quasiconvexity is sufficient for existence must now
be qualified. In fact. MORREY'Sremarkable existence theorem fails to apply
directly to nonlinear elasticity. For compressible materials his growth conditions
are too stringent; in particular. they prohibit any singular behaviour of

'11<

such

as the natural condition

ir(x. F) ro as detF-.O.

(0.5)

'

'

"J 'If '(x".1=;,

+ V '(.\')) dx

~ ".
J 'If'{xo.

1=;,)

dx = m(D)

if'(x(I'

Fo)

(0.3)

holds for each fixed x"eQ. for each constant 3 x 3 matrix Fo. for each bounded
open subset /)<;;;,:1f.I.
and for all 'eO'(D). Here m denotes Lebesguemeasure
and P(D) is the sct of all infinitely differentiable functions with compact support
contained in D. Wc regard quasiconvexity as a constitutive restriction on '11'"*.It
may be interpretcd as follows: For any homogeneous body made from the material
found at any point of Q. and for any displacement boundary-value problem
with zen~.body force for such a body that admits as a possible displacement a
t

For hyperela~tic md~ und ~hells ANTMANmnkes certain convexity hypotheses on the stored.

energy

...

rllnction~,

Bec"lI~e or the cool'dinute~

ject ions m~lde helnw.

,

employed

the~e hypothe~es

are not ~ubject

to the ob-

'

HII,I.[I] wus thc rir~t10ohserve thut "trkt convexitywith respect to PIIimplie~uniqueness.

The nonllni<lllenc~s estubli~hed hy ANTMhNari~e~ frum the pre~ence of lower order terms.
In the cU~e whcn if is not everywhere derined thc cnndition mu~t b,e sli!!htly modiried. See

~eel inn 3,

Moreover. his work gives no indication of how to treat the unilateral constraint
deWu > 0 tI. Incompr~ssible materials require the constraint det Vu = I. which
also poses problems.
. -..----..--.--.-t Throughout thi~ article wc employ the ~umn1ation convention fnr rcpcatcd sumce~,
tt This contra~ts with the views expresscd hy TRtlliSllliU. & Nol.!. [I. p, 275]. who ~ugge~ted that
the Legendre-Hlldum:lrd condition ~hould he regardcd nol as a con~titulivc rcstriction. but as a
stability condition, They conjcctured that vinlution or the Legendre-Had:lmard condition at u point
would lead to wave motion tending to move an elustic bndy from an unstuble to a ~table equilibrium
conriguration and that this proce~s muy help explain buckling, While the violution or the LegendreHadamard condition at a poinl may well result in ccrtain kinds of instubilitics ;I: ERICKSEN[3]). it
is by no mean~ necessary for bucklin!!, Indeed. in Section 9 we show thut buckling can occur when
the Legendre-Hadamurd condition hold~ cverywhere. Moreover. other kind~ of instabilities. sueh as
neck in!!. muy well be compatihle with the Legendre.H:.damurd condition (<:( ANntAN [6]), Anotber
suggestion of TRUESDELL& NOLI. [I, p, 129] concerning internal buckling of u rod would. if true.
directly contradict qua~iconvexity. but the hehaviour de~cribed by them a~ implau~ible ~eems typicul
of buck ling.
ANTMAN'Smuterial hYPolhe~e~ for rud~ und shell~ arc tho~e uppropriute under the a~sumption
that (004)holds in the three-dimen~ional theory,
The analogous problem ror rods has heen stlldied by ANTM"N [2-5. 7. 8].

.)

.:t..

"'1

Exi~lcnce Thcorem~ in Noo.Linear

J.M. BAl.!.

340 r

To overcome these dimcultie~ we investigate in Section 6 sequential weak

continuity properties of functions, defined on Orlicz-Sobolev spaces, having
the form
(0.6)

(): u~1>{I7u(')).

where cP is a continuous real-valued function defined on the set of all 3 x 3

matrices. This map is sequentially continuous from WI, "'(D) with the weak.
topology to 11(.Q)with the weak topology if and only if 0 has the form
(0.7)

O(u)= A + BHVu)~+ q(adj Vu)~+D deWu,

where A, Bf. q, D are constants and adj Vu is the transpose of the matrix of
cofactors of VII. When the domain of 0 is a larger Orlicz-Sobolev space, the
problem is .more delicate. In this case we give various theorems guaranteeing
sequential continuity or closure of 0 relative to various weak topologies.
We combine these results with standard techniques of the calculus of
variations to establish the existence of minimizers for l(u,.Q) in various classes
of functions when'lr has the form
'11'(x, F)

= K(X, F. adj

F. det F)

(0.8)

with K(X,',',') convex for each x. We call such functions 'II' pnlyco/ll''x. Note
that F, adj F and det F govern the deformations of line. surface and volume
elements respectively. If "If- is polyconvex, then'lr
is quasiconvex; in fact
polyconvexity is equivalent to a sufficient condition for quasiconvexity
given by MORREY.However our existence theorems are valid under weaker
growth conditions than MORREY'S.'Moreover, we can handle the pointwise
constraints on det Vu mentioned above by using our sequential weak continuity
results. Since there are few known examples of quasiconvex functions that are
not polyconvex. the restriction to polyconvex functions is not serious. It appears
that neither the quasiconvexity ~or the polyconvexity condition has been considered previously in the context of elasticity.
'
A wide variety of realistic models of nonlinear elastic materials satisfy the
hypotheses of our existence theorems. In particular, these include the MooneyRivlin material and certain stored-energy functions similar to. arid for incom-'
pressible materials identical to, those of OGDEN[2., 3]. That these stored-energy
functions are polyconvex follows from sufficient conditions for tlie polyconvexity
of isotropic functions given in Section 5. where some related results, are also
discussed.

.\

"

Our existence theorems apply to displacement. mixed displacement traction.

pure traction. mixed displacement pressure. and pure pressure boundary-value
problems. Similar methods work for more general classes of mixed boundary
conditions. For the most part we consider only polynomial growth hypotheses.
which result in a theory based on Sobolev spaces. For stored-energy functions
of slower growth an Orlicz-Sobolev space setting is required; for brevity we
treat only the displacement boundary-value problem for such functions. An
example is given of a stored-energy function requiring this more elaborate theory.

341

Elasliciry

Many of the results may be extended to give new existence theorems for nonlinear elliptic systems in higher dimensions: some results in this direction are
given in BALL[2].
In Section 9 I apply existence theorems to establish nonuniqueness for the
mixed displacement, zero traction boundary-value problem for a MooncyRivlin rod under compression. My main result is that non uniqueness occurs
for sufficiently long compressed rods of arbitrary uniform cross-section. Despite
the intuitively obvious nature of this result. such nonuniqueness has not previously been established for any mixed boundary-value problem of nonlinear

elasticity.

'

In Sections 3 and 4 the conditions of quasiconvexity. polyconvexity and

ellipticity are examined in detail. In particular. necessary and sufficient conditions
are .given for polyconvexity based upon the work of ~USEMANN,EWALD&
SHEPHARD.It would be interesting to have a statical interpretation of polyconvexity. None of these three constitutive restrictions has at present a microscopic or thermodynamic motivation. in contrast, for example. to the situation
pertaining to the Navier-Stokes equations. for which the Clausius-Duhem
inequality gives conditions closely related to those sufficient for existence.
The reader interested in the constitutive theory of elasticity. but not in the
details of existence theorems, can omit Sections 2. 6 and 7 without much loss.
With the exception of the existence theorems themselves. the parts of this article
which bear most directly on the relevance of the quasiconvexity and polyconvexity conditions to realistic models of elastic materials are Theorems 3.1.
3.2,3.4.4.5.5.1,5.2 and the discussion in Section 8.
.
Existence theorems for linear elasticity (cf FICHERA[I]) have a different
character from those presented here on account of the geometrical approximation
made. In particular we have no need of an analogue of KORN'Sinequality. The
existence theorems of STOI'PEI.LI[1] :lnd VANBUREN[I) for nonlinear elasticity
(c:j:TRuEsDELL
& NOLL[I]. WANG& TRUESDELL
[I]) are based on those of the
linear theory. In these theorems the inverse function theorem is used to establish
the existence and uniqueness of small solutions to boundary-value problems
with small body forces and boundary data. The material response is assumed
to be such that existence. uniqueness and regularity theorems hold fo~ the
equilibrium equations linearized about the zero data solution. In the case of the
pure traction boundary-value problem the degeneracy of the linearized problem
forces the authors to assume that the surface and body forces possess no ax.is
of equilibrium. This hypothesis is unnecessary under the assumptions of our
existence theorem for the pure traction problem. Note. however, that STOPPELLI
and VANBURENmake assumptions about the material response only for strains
close to those of the zero data solution.
This article by no means exhausts the problem of the existence of equilibrium
solutions even for hypereleasticity. The most notable shortcoming of this work
is that in the two important cases of incompressible materials and compressible
materials satisfying (0.5) I have so far been unable to show that the minimizers
whose existence has been established are smooth enough even to satisfy a weak
form of the equilibrium equations. I hope to discuss this question in later work.

,~

"

~'

343

Exist~nce Theorems in Non-Linear Elasticity

342

J.M. BI\I.I.

I. Boundary-Value Problems of Nonlinear Hyperelasticity

This section begins with a brief presentation of the basic equations of non linear
elasticity. We then go on to consider the various boundary-value problems for
which we later prove existence theorems. The calculations here are purely formal.
in the sense that we assume that the various quantities appearing have sufficient
smoothness to justify any operations required (such as integration by parts).
The theory of nonlinear elasticity is discussed at length in the books by GREEN
& ZERNA[IJ. TRuEsDELL& NOLL [IJ and WANG& TRuEsDELL[lJ. and the
reader is referred to these texts when clarification is necessary.
We consider a material body f,4whose particles are labelled by their positions
x =(x.)=(XI' X2' X3) with respect to a rectangular Cartesian co-ordinate system
in a refmmce configuration Q which is a bounded open subset of {JfJ.Q need not
be homeomorphic to an open ball. In a given motion the position of the particle x
at time t is denoted u(x. t)t. The d(~/ormationgradient F is defined by
('11'

F=P'u: F.1=--=u'.
Px' ..
We suppose that u: Q - .cJf.1
is orientation-preserving

(1.1)

and locally invertible, so that

J=detF>O.

respectively, where Nand n denote the unit outward normals to the boundaries
8Q and I'U(Q.t) respectively.
The pointwise form of the balance laws of linear and angular momentum are
given by
(1.9)
DivTR+PR b=PR ii.

Consideration

,[

(Div TR)j~ aYR/ex)

('iX-"'

The symmetric. positive-definite right and lefi stretch te/lso/'s U. V and the
right and left Cauchy-Green tensors C, B are defined by

unit mass.

Throughout this article we assume that the material is hYPC!relastic;i.e., there

exists a real-valued stored-energy function ir(x. F) such that
iJ'ir

'Ir is o~iective if and only if

'

C=U2=FT F.

B= V2=FFT.

(1.3)

(1.11)

ir(x. QF) = 'Ir(x. F)

(1.12)

'IfI'(x, F)= '/r(x, U),

F=RU=VR.

V=RURT,

(1.4)

where R is the orthogonal rotatio/l tensor. The eigenvalues VI' vi. V3of U and V
are positive and are termed the principal stretches of the deformation, The principal

il/r(/rilm/sof n nnd C are given by

111l=nc=I'~d+dd+d

d.

each particle x the material of the body is elastic,so that a

constitutive equation of the form

(1.6)

holds. where TR denotes the .first Piola-Ki/'ch/1(?fr stress tensor. TR is related to

the C((lICIt.\'
stress tel/so/' T by
.
TJI,=JT(F-I)'I',

and it follows from (1.11) that (1.10) ,is satisfied identically. '11"is isotropic if
and only if if/' is objective and

'Inx, QFQ7')= inx. F)

'Ir(x, F)=l[>(X,I'!,IJ2,I'J)'

(1.5)

1I11l=IIIc=dd d.

TR(x. t)= tR(F(:c, n, x)

(1.13)

(1.14)

for all orthogonal matrices Q. In this case

IIl=Ic=d+l'~+d.

at

To--'-r'
R, - ciF;.

for all proper orthogonal matrices Q. If '11"is objective then

The following relations hold:

Wc suppose that

.. ,

PR(X) is the density in the reference configuration, and b is the body force per

of the stronger requirement that u be globally one-to-one is

beyond the scope of this article.

(1.10)

T=TT.
where

(1.2)

,I
I

The surface traction tR measured per unit area in the rcference configuration.
and the actual stress vector t measured per unit area of the deformed configuration, are given by
t= Tn
(1.8)
tR= TRN;

(1.7)

, We ch""s~ the n"tation common in rartial dilT~rential equations rather than that'of continuum
m~chanics where our x. u arc customarily denoted X. x respectiv~ly.

where <Pis symmetric in the I'i'

.
Deformations of illco/lll'ressihle materials are restricted
constraint
J=detF=
I.

( 1.15)

by.' the pointwisc

(1.16)

For incompressible materials the above theory has to be modilied by replacing T

by the extra stress

Tf;

= T + p I.

(1.17)

where I' is an inddcrrninate hydrostatic pressure. The stored-energy function

iF' for an incompressible material need be defined only for F satisfying (1.16).

~
''I...

Existence Theorems in Non-Linear

J.M. BALL

344
I

(1.18)

Condition (1.25) says that the total force on the body due to external loads is
zero (:j:TRUESDELL
& NOLL[I, p. 127J). To describe the effect of this condition
we consider two situations corresponding to different types of existence theorems
proved in Section 7.

(1.19)

the set of functions Uosatisfying(1.25) will be nonempty. so that under certain

We shall be concerned only with equilibrium configurations of ~. rf if/' is

objective we see from (1.9) that u. is an equilibrium configuration if and only if
'A:fu~.p+qj+PRhl=O.
where

.)

A~P (x , F)

der C2 if'-(x

. F'

=- afl

(;~

F)

)
/l

c)2if/(x,

F)

(1fl.

q/=-W

1. b(lI) is not a constant vector. In this case. under suitable hypotheses on b

conditions t it is likely that a function Uosuch as to render Jo stationary subject
to (1.21) exists. If so, then Uois a solution to the traction boundary-value problem.

A. The mixed displacement-traction boundary-value problem

jin' a compressiblematerial

{.

2. b(u)=bo. bo constant. In this case a is independent of uo' so that (1.25)

is a condition on the data of the problem. It proves convenient to consider JO(II)

'

as a functional defined on functions u satisfying the constraint

In this problem we seek u satisfying (1.18) in Q and satisfying the boundary

conditions
( 1.20)
u(X)=U(x)
for XE(1QI'

for

XE(1Q2'

..

Su dx=e"

'

tR(X')=tR(X)

345

Elasticity

( 1.21)

( 1.27)

where e is an arbitrary constant vector. The constraint (1.27) removes the

indeterminacy resulting from a possible rigid-body translation of u(Q). Jo is

where (1Q=(1QIVtJQ2' (1Qln(1Q2=rJ>,and U: (1QI-+rJI3,tR: tJQ2-+:~3 are given

functions. The boundary condition (1.21) is a condition of dr!.adloading, i.e."
the loads acting on U(aQ2)have fixed direction and fixed magnitude per unit
area of tQ 2' If ()Q2 = then we have a pure displacement hOlllldary-t.alueproblem,
while if N'll = rJ>we have a traction houndary-value problem.
Suppose that the body force b is conservative. so that

stationary

at u = "0 subject to (1.27) if and only if (1.21) holds and

Div TR + PR bo

(1.28)

= a.

rJ>

b = - grad 'l' ,

//(x)"R(x)dS.
<'!1,

(1.24)

where
der

(1.25)

a=;;;(Q) (jPRb(uo)dx+
n

JfRdS).
"n

'

playing thc role of the Lagrange

( 1.30)

is a solution to the dynamic traction boundary-valuc prohlem (1.9). (1.21),

Note that any equilibrium
solution
// must also satisry the zero moment

S11/\ PR ho dx+
n

.
j 11 /\ tR dS=O.

(1.31)

ell

unlike (1.25), depends explicitly on the unknown

and so cannot be imposed a priori.

This condition,

only if 1/0 is a solution to the mixed boundary-value problem.

If ('ft\ = rjJ then in general solutions to the boundary value problem will not
exist. since a necessary condition for a function "0 to render Jostationary subject

a=O

(1.29)

converse state,ment is a direct application of the

dcf
12
u(x, t)=//o(X)+T
a

condition

Let tQI * cjJ. Then a standard formal calculation shows that 10(Ilo) is
stationary with respect to Il satisfying (1.20) if and only if the Euler-Lagrange
equations (1.18) and the natural boundary conditions (1.21) hold, i.e., if and

to (1.21) is that

Sv dx = O.The

(1.23)

Consider the functional

.I,,(II)'';;fj'f,(x,//(x).F(x))dx-

multiplier corresponding to the constraint (1.27). A rigorous proof may also

be constructed by using a result of SCHWARTZ[I, p. 59].
If a =0 then //0 is an equilihrium solution. If a * then

Define ./; (x, u. F) by

1[1(11).

which is zero if

multiplier rule for isoperimetricproblems with

F)= 'II/'(x, F)+PR(X)

der d

c5Jo(llo)(V)=- d ' Jo(uo+cv)

= - j a. vdx+
(tR-tR)dS.
I.
n
N1
I ,=c

(1.22)

where 'l' = 'l'(u) is a real-valued potential, and where

derc'l'
(grad
=::;-T'
(,/I
'/;(X.II.

To prove this note first that if (1.21) and (1.28) hold. then

B. TIle mixed displacement pressure houndary-value

function

//,

problem

for a c(}1I1pressihle /1/ataial

In this problem we seck

conditions

(1.26)
t

II(X)=U{X)

11 satisfying

for

(1.18) in Q and satisfying the boundary

( 1.32)

XE(~QI'

(Tn)(x)= -p,n
for XE(1Q,
(1'=2,
if '[, takesthe formof a powerfulpotentialwell.for e,ample,

M).

(1.33)

. :"I.
J.M. BAU.

346

For such a deformation

where rlQ =

Ill ,I

U(IQ"
,si

f)Qk ('\

t!Q,= 4>-(k td),jj:

OQI'" 213 is a given function, and

exists a CI(Q) function p: Q

r=2, "', M.
Consider the functional

-~

A(!/~1

In one-dimensionalllniaxial
(\.34)

in (1.l6).

J.(II)=
Suppose

fl

By the divergence

11dx+

F=
( \.36)

N,dS.

SJf.

) N, dS
(Pr.jjk" ,pu .J
if.q

(1.37)

The fourth integral in (1.37) is zero by Kelvin's theorem applied to each cQ"
and the last integral is zero since r pAN = 0 on cQ" Thus
=-

Ill:

t:- 0

S(DivTR+PRb).vdx+
0

S .(t+pn).vdS.

.'I:2),1I2(XI' X2)' ;.X3)

11', 0 0
0 /1 0 .
0

(1.43)

the relevant boundary-

CO(O)

;.>0 constant.

locally integrable function / defines a distribution 1f through the equation

1f(4))= S /4> dx, <PE!?1(Q).
A sequence T,.- Tin 9'(Q) if and only if T,.(4))-+ T(<p)

(1.38)

for all <PEf2'(Q).(Here and throughout this work we consider only convergence

Thus Jt (u,,) is stationary if and only if Uo is a solution to the mixed displacement

pressure boundary-value problem. The calculation above is a slightly simplified
version of that of SEWELL [I]; see also BEATTY [I].
If (IQ, = 0 then we proceed in a fashion similar to A.
.The functionals that we study in later sections include functionals of the
form J" and J\. For the purposes of the existence theorems it is not necessary
to assume that 'If' is objective. If. however. this assumption is not made, the
resulting minimizers, if smooth. will not necessarily satisfy (1.10).
For incompressible materials the admissible functions are restricted by the
additional constraint (1.16).
Finally in this section we discuss brieOy the analogous problems of one and
t\~o-dimensional hyperelasticity. These problems arise from special deformations
of three-dimensional elastic bodies. In two-dimensional plane strain we consider
deformations having the form
U=(II\(XI'

(1.42)

For ease of referen,ce we list here various well known spaces used in this
article. As general references we cite FRIEDMAN [I], KRASNOSEL'SKII& RUTlCK!I
[I], SCHWARTZ[I] and for functional analytic aspects DUNFORD& SCHWARTZ[I].
Throughout this work Q denotes a nonempty, bounded. open subset of fJII'"
with Lebesgue measure d:c, where, except in Section 3, III takes the value 1,2 or 3.
C""(Q) denotes the space of infinitely-difTerentiable real-valued functions defined
on Q. ~(Q) consists of those elements of C'''.(Q) with compact support contained
in Q. We give 9,(Q) the strict inductive limit topology of SCHWARTZ. The dual
space of !?1(Q) is denoted 1'(Q), and' its elements are called distrihlltions. Any

- NIS ! (r.pq,P.qN,)(I:/jk 1/. Pukpj) dS,

~Jl(II+/:V)

AX3)'

2. Technical Preliminaries
StR.vdS
ca

'

i'fI

(1.41)

I.

We leave to the reader the routine task of formulating

two cases.
value problems in these
\

'.

- d JI(U+I:V) =-S(Div\TR+PRb).vdx+
d I.
".0
0
S tpl:ijkl;1'q'I/.pllj.qVkN,dS
/'0
.
j
k
I,qp

1)=

strpin we let

( \.35)

theorem

S !Pf.jjkgPq'I/,pllj,qllk

MI

112.2 -111.2112,

U=(UI(X,),/IX2'

now that CQI =t=

4>.Then for v satisfying v=O on DQI we obtain

(1.40)

where 11and A are positive constants. The corresponding deformation gradient is

pq' F i F j k
derr
J \
p q 11,
2 (X, U, F) = J I ( X, U, F) + P + b p, ,f.jjk f.

and where J is defined

.2 0
0 ;.)

112

taking the value p, on DQ, for each

JI(U)= S.f2(x,u(x),F(x)dx,
0
where

111.2 0

F = !/ci.
and the incompressibilitycondition is

p,(r=2,...,M)
are constant
pressures. We assume that for r~2 DQ, is either
a closed surface or is bounded by a closed curve lying in DQ\. Suppose also that
there

347

Existence Theorems in Non.Linear Eh.sticity

(1.39)

(7T

I
I

of sequences, rather fhan nets.) If TE1'(Q), then we may define (f'(-; by

uT
n(1)
-- .
-(4))=-Tc;
.
Dx"
I'X'
)
.
.
"
aT
'
Then TH -;;-- maps .1):(Q ) mto itseIf and IS contmuous.

ex'"

C(Q). Cl (Q) denote respectively

the spaces of continuous and continuously

difTerentiable real-valued functions defined on Q with the usual supremum
norms. f}P+ denotes the non-negative real numbers. gj the extended real line
with the usual topology. If E is a subset of tW then Co E denotes the convex
hull of E.

The spaces l!(Q), 1; p; w, of (equivalence elasses 00 integrable real.

valued functions are defined in the standard way. The Sobolev space
W 1.P(Q). 1; P; '-X',consists of those functions !/ belonging to l!'(Q) with weak

"

".
348

.~

..

Existence

J. M. BAll

derivatives ~u2 (I ~et~/11) belonging to IJ(Q). W1,/'(Q) is a Banach space under

the norm ex
m
." r -.., .,
I

111111wo. I' IfIJ

= lIullJ.Plf/J+2-1
L

;;-'.
(X

11

li

LPW,

in Non-Line,,,

349

Elasticity

noted 11IIAand is equivalent to 11Ill",),Holder's inequality is valid iri the form

Suvdx~lIulI(A)IIl'IIA
n

'

hI

Theorems

(2.1)

(2.6)

for all ueLA(Q), veLA(Q).

The closure of ~(Q) in Wf.P(Q) is denoted W01,P(Q).H-;~~I(Q)denotes the space

The Orlicz-Sobolev space WI L...(Q)(WI E...(Q)is defined as the set offunctions

of function: u which together with their weak derivatives :~: (1 ~ et~ m) are
locally integrable. When dealing with the spaces in this paragraph we assume
that f' < 00 unless otherwise stated..
.
Weak and weak * convergence of sequences are denoted iby ---"- and ..,
respectively. In the case of the Banach space Wt. ""(Q)we de.finethe weak * topology
to be that induced by the natural imbedding of WI. ""iQ) in the product space
(L'(Q)j1 +m,where each factor has the weak * topology. Thus a sequence un..u in

ueLA(Q) (EA(Q)) such that the weak derivatives :x~eLA(Q) (EA(Q)(1 ~et~m).
WI LA(Q) is a Banach space under the norm

lIullw'L...ln,=Ilull(A)+

LA(Q) = EA(Q)=IJ(Q),

If A is a real-valued, continuous, even, convex function of tem satisfying

A(t)
A(t)
A(tO for t>O, -->0
as t->O, -->.x;
as t--+oo then we call A an N.
t
t
function. If A is an N-function, its conjugate function A is defined by A(t)=
sup{ts-A(s):
se&f}. A is also an N-function and satisfies A=A. Furthermore
Young's inequality,
ts~A(t)+A(s),
(2.2)
holds for all s, te(~. If A. Bare N-functions then we write A-<B if and only if
there exist positive numbers to and k such that

if

A(t)~B(kt)

(2.3)

B if and only if A -<B and B-<A. and A ~B if and only

(2.7)

11

WJ LA(Q).

and A(t)-IW,

where

WI L...(Q)= WI EA(Q)= Wt.P(Q),

1 I
-+-;= I.

P P
Throughout this article we shall be dealing with vector and matrix functions 'P.
When we write 'PeX, where X is anyone of the spaces introduced above, we mean

that each c~mponent of 'P belongs to X, and we define 11'J'llx to be

L 11~llx

where the sum is over all components of 'Piof''J' .

i
Finally we define some regularity conditions for Q.
(i) Q has the segment property if there exists a locally finite open covering
{8j} of cQ and corresponding vectors {yJ such that x + t }liEQ for all
xeQ (')8i and for all te(O, I).
(ii) Q satisfiesthe co/!eco/!ditio/!if there existsa fixed cone knS;;~m such that
each point x EDQis the vertex of a cone kfi(X) that lies in Q and is congruent
to kf)'

lim~=O
t-<"

'

(2.4)

BP.t)

for every i.>0. If A is an N-function the Orlicz class .PA(Q)consistsof all (equivalence classes of) real-valued

1/

Wt. "(Q) if and only if Un.L.U in ~X>(Q)and ~ u: .L. ~'~ in ~"'(Q) (1 ~ et~ m).
uX
(,X
:.

for all t ~ to. We write A

.-1 :u.
X (A'
.
.
and similarly for WI EA(Q). The Closure of ~(Q) in W1 LA(Q).is written
In the special case when A(t) -It( (f' > I) we have the equalities

measurable

SA(u(x)dx

functions

(iii) Q satisfies a stro/!g Lipschitz conditioll if each xe('Q has a neighbourhood

.:lit"such that in some co-ordinate system. with origin at x. Q (').fIt"is represented in ')7/"by (m< F(n. (' = (I'
(m-I) with F a Lipschitz continuous function.

u on Q such that

< 00.

(2.5)

3. Quasiconvexityand the Legendrc-HadamardCondition

We consider integrals of the form'

The Orlicz space LA(Q) is the linear hull of 'pA(Q).LA(Q) is a Banach space with
respect to the Luxemburg norm

/(u, Q)=

IluIlIA,=inf{k>O; S A(u/k)dx~t}.
n

If A-<Bthen LJ!(Q)s;;L,~(Q).while if A~B then LB(Q)~LA(Q). The space EA(Q)

is definedas the closureof the bounded functionsin the LA(Q)norm. We havethat
EA(Q)S;; .Y-'...(Q)
S;;LA(Q).The dual of EA(Q)can be identified by means of the scalar
product n ul,dx with LA(Q). The norm on LA(Q) dual to 11II(A)on EA(Q) is de-

'

where

(
x,U(X)EQX.~n,

the real-valued

Sfix,

f)

iJl/(X)

u(x), l7u(x)dx.
.

( l7u(X)~=--axa=lI~.(X)
)

function fis defined and continuous

(I~i~/!,I~~~II1),

and where

on a given relatively open sub-

set S of Q x .?Inx Mn x m. Here Mn xm denotes the space of real I! x m matrices, with

the induced norm of 9Pm~We assume that for each xEQ there exist ue.tJi'".
FeMnxm

with (x, u, F)eS.

.')

~.

j'

Dcfinition 3.1 ('f. MORREY[I].) Let U be an open subset of M".m. Let g: U

be continuous. Then g is said to be quasiconvex t at Foe U if and' only if
S

fJ1I
(3.1)

g(Fo + V ,(y)dy ~ g(Fo)m(D)

for every bounded open subset Dc;;~ and for every 'e~(D) which satisfies
Fo+ V'(y)e U for all yeD. g is quasiconvex on U ifit is quasiconvex at each Foe U.

'\

Note that ifg is quasiconvex at Foe U, and if'e WJ' "'(D) satisfies F;,+V'(y)eK
for almost all yeD and for some compact subseVf of U, then (3.1) holds for ,.
In fact by the definition of WJ' "'(D) there exists a"sequence of functions 're~(D)
that converges to in Wt. ~(D). Since K is compact there exists an,integer N and a
compact set Kt with Kc Kt c U such that Fo+ V'r(y)eKt for all yeD and all
I' ~ N. If I' ~ N then (3.1) holds for 'r' By the compactness of Kt, the continuity of

g and the bounded convergencetheorem we obtain (3.1)for ,.

Let

Existence

J. M. BALL

350

A = {we ~~~ t{Q): (x, w(x), VW{X)ES for almost

all xeQ,

and

I(w, Q)

exists and is finite}.

Dividing by em and letting c

convergence theorem- 0

Theorems

in Non-Linear

351

Elasticity

O we obtain (3.1) for Fo= V u(xo) by the bounded

Corollary 3.1.1. (C! MORREY[I, p.43] and MEYERS[I, p.128].) Let Uc;;M",m

beopen,letg: U fJ1Ibe cominuous,let Foe U andlet (3.1)holdfor a given bounded

open subset Do r;;;. gr and for every' e!2(Do) which satisfies F0 + J7'(y)E U for all
yeDo. Then(3.1) holds jar all such D, " i.e., g is quasiconvex at Fo.

Proof. Apply the theorem to the integrand g(F) with Q = Do and UI(X)= (Fo):X'. 0
. Theorem 3.1 is es~entjally. the same as a result stated
following earlier work of BUSEMANN & SHEPHARD [I "p. 31].

by SIL VERMAN

[I],

We next show that forintegrands

that are independent of x and 11the existence
of a sufficiently regular minimizer to certain Dirichlet problems implies that the
quasiconvexity condition holds.
. .,'

Theorem 3.2. Let Ur;;;.M"xmbe open and let g: U->.rJPbe C()ntiIllIOIIS.

Suppose
that either

(i) /1 = I, Q is arbitrary,

1 ;;?;IX;;?;
m), say. Let "0: Q

or (ii) Q is a hypel'cllbe,

= {XE;jf"': 0 <x'

< I,

ijf" be! dejlne!d by

Theorem 3.1. Let u EA be such that

u~(x) = F~x. + :;i,

I(u, Q);;?; I(w, Q)

.f(n'all wEA with W-U~{Q) and Ilw-ul/clm sujJlciently small. Let xoEQ and suppose that u and Vu have representatives, again denoted by u and Vu, that are continuous at Xo with (xo, u(xo), Vu(xo)eS. Let U = {F: (xo, u(xo), F)eS}. Then
f(xo, u(xo),') is quasiconvex at Vu(xo)eU.
Proof. Let u satisfy the hypotheses of the theorem, let D be a bounded open subset
of Jr and let' E~(D) satisfy (xo, u(xo), Vu(xo) + V'(y)eS for all yeD. For
c>O define u,: Q"""i~n by
X-X

(~ c )

u,(x)=u(x)+c'
= u(x)

if

x-x
c 2.ED

otherwise.

For c small enough the set xo+DD,on which u and u, differ, is contained in Q,
and thus u,-uEiJ(Q). Also for XEXo+DDwe have
.
f(x.II,(x). VU,(x))=f(X,U(X)+c,

(X~xo).

J7u(x)+J7' (~Xo)),

which by our continuity assumptions and our assumptions about' is bounded

uniformly above on the set XI)+ I:D for c small enough. Thus u,eA and 1(11,Q);;?;
1(11"Q). Making the change of
. variables y= x -xo
I: we obtain
S

f(xo + D}',u(xo +I:y)+e'(y),

~ Sf(xo+/:y,u(xo+cy),
D

Vu(xo + cy) + V,(y)emd}'

Vu(xo+ey))r.mdy.

, This is MOKKEY'soriginal terminology. In his book [2] he calls such functions (for U = M"'"')
strongly <juasiconvex. retaining the term quasiconvex for functions satisfying the weakened form of
the Legendre-Hadamard condition that we shall term rank 1 convexity. The reader is warned that
<juasiconvexity has other meanings in the literature on convex' analysis.

where FE U and ZE~" are constants. Let

)(u)=

fI

g(l7u{x)) dx

and let AI ={UECt(Q): J7u(x)eU.for all xEQ. J{II) I!xists and is finite. and u=uo
0/1cQ). Supposethere exists VEAI such that
J(V);;?;J(II)

jar all uEAI.

Then g is quasiconrex at FE U.
Proof. Let W=V-lIo' Then weCt(Q) and w=o on c'Q. In Case (i) there exists
xoeQ with J7w(x())=O.Therefore J7v(xo)=F and the result follows from Theorem
3.1.In Case (ii)we have that J7w(xr)--+ 0 as r -->rx for any sequence {x,) r;;;. Q with
x
0 as I' -, rJj.ByTheorem 3.1,g is quasiconvexat F + Vw(x,)eU. Taking the
li~it I' -> 00 in the quasiconvexitycondition givesthe result. 0 .
Remarks. For n = I or m= I quasiconvexity is equivalent to convexity (see
MORREY[I, 2J). The analogue of Theorem 3.2 for n> I. Q arbitrary, is false. As
an example, let m=n=2. Q={(xI.xl):xf+x~<q.
Define g:M1xl
9P by
g(F)=I'{r), where r=IFI=;,tr(FF1)! and where 1': ,jf+ --+:~ is zero for r~1 and
positive for O;;?;I'< I. We show that for any Uoe Cl (Q) there exists an absolute
minimizer for )(11)= S g{J7u(x))dx among Cl (Q) functions 11satisfying u = uo on
.

fI

f!Q. Let 'ECI(Q) satisfy '=0 on DQ and let W'(X)/~I:>O for all xEQ; e.g., we
may take '(x)=(ht(x),hl(x)+h2(x)),
where hi and h2 are C1(Q) functions satisfying hi =h2 =0 on (iQ, J7hj{x)=O if and only if x=aj U= I. 2) with al q"a2' For
large enough k > 0 and for all XEQ we have
W{lIo+k')(x)l~h-llI7uollcm

I,

"I.

J.M. BALL

352

xeQ.

u corresponding

to "0 may be chosen

so that det J7u(x)~c>O

Condition (iv) was derived by GRAVES[IJ and has recently been studied by
ERICKSEN [2]. Motivated by (iv) we say that g is rank 1 coltl'ex at He U if the
inequality (3.2) holds 'whenever the right-hand side is defined. In this case it is
easy to see that the Legendre-Hadamard condition holds at H. The prototype
for the following theorem wasdiscoveredby HADAMARD[I, 2J, the first rigorous
proof being that of GRAVES[I]. For other' proofs and relevant literature see
DUHEM[IJ, MCSHANE[IJ, CATTANEO
[IJ, VANHOVE[1, 2J, TRuESDEll NOlL
[1, p.253J and MORREY [2, p. 10]. The proof here is based on MORREY [I, p.45]
and on Theorem 3.1.

for all

Definition
3,2. Let U be an open subset of MO x m. A function g: U -+ (}l is rank 1
cOIu'ex on U if it is convex on all closed line segments in U with end points dif-

&

fering by a matrix of rank I, i.e.,

g(F+(1-l)a@b}~lg(F)+(1-l)g(F+a@b)

"

Theorem 3.4. Let 11,Xo satisfy the hypotheses of Theorem 3.1. Let

It'

foraIlFeV.le[O,IJ,
Qe.'?-?"',be.rJr, with F+tla@beU
(Q@b)~d,;falb (1 ;;;;;n. I ;;oc;;m).

for all tIe [0, I]. Here

.

Vo= {F: (xo' I/(xo). F)ES:.

Thellf(xo, ,i(xo), .} is rank I

Theorem 3.3. Let V be CIIlopen subset of MOxm and let g: V -+ .'?P.The following
conditions (i)-(iv) are equil'alent:
.
(i) g ;.~rank 1 cOIwex on V:
(ii) for each fixed FeM"xm, be9l'" the function

Qi-+g(F+a@b)

closedlinesegmentsin the set (a: F +a @be U}:

is convex on all

holdsfor

all Ae[O, 1) andfor

all HeM"

(H-.l-l

xm, ce(}l",

C@d)

g(F +a@b)~g(F)+Ar(F) alb.

-k-I C@PI
h -1 ,c,
17'=
C\01Pt
_p-I c@/'II
/ p-te@/'II

(3.3)

I I'

(I <fJ;;m),

IYpl~p

where yp=x'Pp. Let'F;-, F,+, Fp-, F/ (l<fJ~m) be the faces YI=-k,

Yt=h,
Yp = -, p, Yp= 0, respectively. Let Ttt (1;;fJ~ m) b~ the pyramid with ba~eF/
and with vertex at the origin. Let' be defined on D to be continuous on D. zero

allle[O, 1]. .

c12g(F) I1 1 h 0
'Tf ;r-~F.j Cla. '. 11=

;;11,

on oD, and linear on each Ttp and Tt;. with '(0)

Ifge Cl (U), then AHF) =():~),

Ifge C2(V), then (i)-(v) /I'eequivalent to
(vi) (Legendre-HadamClrdcondition)

-k;:aYI

de9l'" satisfying H + jlc@deU

(v) for each Fe V the/'e exists A(F)eMmX" such that

k= l-:-l
A h. Choose vectorsPp(1 <fJ5.m) such that (Pt' ..., Pm)is an orthonormal

(3.2)

L~

Un.

set in ~m. Let D denote the rectangular parallelepiped

~'dll
tIe
1 ,1].
If ge t( V), then (i)-(iv) are equivalent to

whenever.F +la@beUfor

at Vu(xo)e

Let H=J7u(xo)' Let Ae[O,I),

A
let e, d be such that H+e@d and H- I-l
c@d belong to Vo and assume
without loss..of generality that d=l=O,A=1=0.Let p>O, p,=d/ldl, h=I/ld1 and

(iv) the inequality

('Olll''''>':

Proof. Define g: Vo-+(}l by g(F)=I(xo,u(xo),F).

(iii) for each fixed FeM"xm, aetJ.f" the function bi-+g(F+a@b) is convex on all
closedlinesegmentsin the set {b: F + a@beU}: .

g(H)~).g(H+c@d)+(I-l)g

353

ExistenceTheorems in Non-Linear Elasticity

so that J(uo+k')=O and uo+k' is an absolute minimizer. But g is not quasiconvex,asmay be seenby putting "0 = F x in th~ aboveargument,where Fe M2 x 2
is constant with IFI < I. With a little more work one can show that the absolute
minimizer

"

.~

= - c.Then
..

on 1t1

on 1t1
on 1til
on 1tlt.

Provided fI is large enough Theorem 3.t and the remark following Definition 3.1
imply that

Jg(H + J7'(x)} clx~g(H)

."""
,11!m F U
,j or a II a em, b em,
e .

m(D)= g(H) 2m-1 pm-I(h

+k).

Hence

Proof. The equivalence of (i), (ii) and (iii) is clear (cf. SllVERMAN [1, Thm.4]).
The equivalence of (i) and (iv) is proved by making the change of variables c=
a, d = b, H = F + (1 - A)a@b. Let ge C( V). That (v) implies (ii) follows by a
well known condition for convexity. To show that (ii) implies (v) one' can use the
arguments of MORRr:V [I. p.47J to establish (3.3) for a@b belonging to some
neighbourhood of zero in M" xm, and then deduce(v) from the convexity of the
function ai-+g(F+a@b).
The remaining assertions of the theorem are obvious.
(). -I)

2ft

2/11 h+k

g(H+

/ -1
1

2k

C@/'l)+ h+k g(H-k

i:

11: 2

(g(1{_p-1

C@/'/J)+g(lI+p-1

-I

C@I'I)
L'@/'p)} ] ~g(1/).

Letting p -+ 00 we obtain (3.2). 0

The proof above shows that, roughly speaking, quasiconvexity implies the
Legendre-Hadamard

condition. Whether or not the converse holds is an im-

.i

.,

1'\

354

portant

'~

1. M. BALL

Existence

open question (cJ.the COAlments

of MORREY
[2, p. 122]).The conditions

in Non-Linear

Elasticily

355

AI

are known to be equivalent only in certain special cases, for example in the quadratic case f(F) =uf! F; FJ with af! constant and ni, n arbitrary (MORREY[1,2],
VANHOVE[I]), and for certain parametric integrands when n=m+1 (MORREY
[1,2]). In particular nothing interesting is known about the case m = n >1, which
occurs in non linear elasticity.
To discuss this problem, consider a continuous integrand fWu), defined on
all of M" xm,and independent of x and u. Suppose that f is rank 1 convex on

"3

Mn x m.It is well known that if D is a bounded open set in fJr"then any function
{e.@(D) can be approximated
in Wd.OO(D)by piecewise affine functions (EKElAND
& TF.MAM[I, p.286]). Thus a natural method of attack is to follow the lead of
the proof of Theorem 3.4 and to seek domains D with a partition into a finite
number of disjoint open sets Dk and a set of measure zero, such that the quasi-

Theorems

convexity condition

J.r(Fo + V'(x)dx~f(Fo) m(D)

(3.4)

holds for any FoEM"xm and for any 'EWd'''''(D) that is affine on each Dk (cl
.
SllVERMAN[I, Thm. 2]).
For ease of illustration we consider the case m=2, n arbitrary; similar comments apply for m ~ 3. First let D be the interior of a triangle in al2 with vertices
ai, a2' aJ' and let e be an interior point of D. Let DI' D2' DJ be the interiors of
the triangles a2 eaJ' aJ eal. a. ea2 respectively. Let nl be the unit outward normal
to (!D on the side a2 aJ. let I1= la2 aJI, and let n2. nJ,12', IJ be defined analo.
gously. Let 'EWJ''''(D) be affine on each Dk with '(e)=c. Then.
.
It
.17'= 2m(Dk) cQ9nk on Dk'
and (3.4) becomes

I
I
. I
I

A2

A3

IAd(Fo+2
~kD) CQ9nk)~f(Fo),
k-I
m kC
where Ak=m(Dk)/m(D). But this inequality follows from rank I convexity of f
because

AkIknk -0.
k-I 2m(Dk)
.
A similar argument shows that (3.4) holds for piecewise affine functions if D is
the interior of a convex polygon and the Dk'S are triangles formed by joining a
single interior point of D to adjacent vertices of the polygon.
A different situation arises if we introduce more interior nodes into the parti.
tion of D. For example let D be an equilateral triangle AI A2 AJ of side I partitioned into 16 congruent equilateral subtriangles of side (see Fig. I). Let nl' n2'
nJ be the unit outward normals shown. and let el' e2' eJ be the position vectors
of the three interior nodes B.. B2 and BJ' Let Cl' C2'cJ be given and Ic;t' E Wd' "'(D)

.~
Cl' The values of V, in each subv3
.

be affine on each subtriangle with nel) =

"1
.Fig.I

triangle are shown in Fig. 1. The corresponding quasiconvexity inequality is

easy to write down. but does not seem to fol1ow from rank I convexity of;: This
suggests that the Legendre-Hadamard condition does not imply quasiconvexity.
Unfortunately the search for a counter-example is hampered by the fact that, as
we have mentioned, any such f cannot be quadratic in Vu. It is conceivable that
if m=n then the Legendre-Hadamard condition implies quasiconvexity for objective functions.
Finally we mention in passing an implication of thc configuration in Figure I
for finite element methods. Let D be as in Figure I and suppose that one wishes
to solve a boundary-value problem for u subject to the pointwise constraint
det Ji'1l(X)
= I for all xED (the two-dimensional analogue of the incompressibility
constraint (1.16)) and. for example, the boundary conditions u(x) =x for all
xeiiD. Then the only function {e W;:'""(D)that is affine on each subtriangle with
det V(x +{(x)) = I for all xeD is '=0. Indeed.any such map XI--x +C(x)deforms
each subtriangle into another triangle with equal area. It follows that, for example,
B. can be displaced only along the line through 8( parallel to AI AJ, and also
along the line through 81 parallel to Al A2. Hence 81 is fixed by the map, and

",.

o"

Exi~tence

J. M. BALl

356

Theorem~ in Non.Linear

357

Ela~tidty

etc.. so that rPo, rPl' 4>2' cPJ are constants, 01=OI(Fl, Fi). O2= 01(FIJ. Fl) and
OJ=OJ(Fl. FI1).Applying the conditions A:~ = -A~7, ('tc., we can reduce rPto
- '
02 t/J

similarly for Bl' BJ' A similar argument applies when the number of congruent
subtriangles in the partition of D is.increased. Thus either nonlinear interpolation
functions or nonconforming elements must be used. A related difficulty for incompressible fluids is discussed by TEMAM[I].

the form rP(F)=t/J(F)+q(adjF)~+DdetF, where oFloF} =0 for all i,j,IX,p.

The result follows.
Ji

4. Sufficient Conditions for Quasiconvexity

For a general continuous rP we use a mollifier argument. Let pefiJ(MJx3)

p(F)dF=l. For c>O let p,(F)=C-9p(Fjc).
satisfy P60, p(F),,:,"O if 1F161,

M"3

The quasiconvexity condition is not a pointwise condition on the function f,

and is therefore difficult to' verify in particular cases. In this section we shall be
concerned with more accessible conditions that are sufficient for quasiconvexity.
These conditions apply to functions f for which the Legendre-Hadamard con.
dition is not known to be equivalent to quasiconvexity.
Throughout the rest of the article we assume, unless the contrary is stated, that
m=n=I,20r3.'

"

"

Then

But since rP is continuous,

,~

..

'

rP

if n= I,
if n= 2,
if n= 3,

is C2. Then by Theorem 3.3 (vi), (4.1) is equivalent to

A.lf
i} (F)

dcr (,1 rP(F)

= Fi 1 FT' I t
(, . (o'/1
~

r
.P
101Iows t h at A i}

".

i
l S rP(Fo+tV'(y)) dy=
etl D
and the result follows.

.
I.e.,

= ('rP2=crPJ
(,F]

of]

"

= ao! =0

OF]

the dimension of the convex hull Co M of M is s (i.e., the linear subspace spanned
by M is .efl").Wedo not aSSlCmethat M is convex. Let :IF:M ->i.Jf.For variable r 6 I
we denoted by"A = {AI' ..., A,] a variable set of non-negative real numbers Ai
,
Ai= I.
with

A .P
i} =

i-I

F;I +rPJ(F)FIIFl

I'

Definition 4.1. :IFhas a convex lower bound if and only if there exists a real-valued

convexfunction C(z) definedon Co M such that :IF(z)f:;C(z) in M. (Without loss

of generality C may be assumed to be affine.)

where t denotes the matrix of off-diagonal elements of F, and where the functions
rP/' 01, X are C2. Since A:f =0. etc.. we obtain the equations

c'F]

(~P.(I dy=O.

convexfunctionsdefinedon non-convcxsets. Let sf:; I and let M;!.Jf"be such that

+OI(F) Fi +Ol(F) FI +OJ(F) F] + X<Ft,

('rPo

- DSAf!(Fo+tV,)

We next recall some results of BUSEMANN.

EWAl.D& SIIEPIIARD
[I] conccrning
IS a Iternatmg;

rP(F)=rPo(F)
Fi FI F] +rPl(F)FI Fj\+rP2(F) F]

'

'

and Fi, so that

F so that (adj F)~= t, (adj F>=Oif j =t=i or (J =t=a, we obtain similarly

(4.2)

A1t= - Ajf and Af! =0 if IX=fJ or i=j. Since AI: = Ag = An =0 it followsthat

cP(F) isaffinein eachof Fi. Fl

Proof. If rPand -rP are quasiconvex then by Theorems 3.3 and 3.4 both rPand
-rP are rank I convex on M"X", and so by the theorem rP has the form (4.2).
Conversely any rPof the form (4.2) is such that, in the notation of the preceding
proof, Af! is alternating. Then for any Foe M" X",for any bounded open set D~ fJ1I.",
and for any 'e&(D) we have

and for all FeMJx3,

=A .PI}(F)

Corollary 4.1.1. Let rP:M"X"->!IIbe continuous. Thell both rP(/lId -rP are quasi.
convex on M" X"ifand ollly if rPhas the form (4.2).

Proof. We just treat the case n = 3; the cases n = 1,2 are easier. Suppose first that

h
were

as

(4.1)

the adjugate matrix of F (i.e., the transpose of the matrix of cofactors).

for all a,be.!IlJ

g, (t)= A (I:)+ 8':(1:) t

xJ

.. ..

that q(F.) -> Cf, D(c)-> D as c -> O.The result for continuous rPfollows.

where a, b, ai' I1f,1',A, Bf. Cf, D are arbitraryconstants, and where adj F denotes

A:l(F)aia}b.bp=O

on bounded subsets of MJ
the functions

converge uniformly #on compact subsets of !IIto a function g(t), which is easily
shown to have the form g(t)=A+B~t. Thus A(I;)->A and Bf(I:)->Bf as c->O.
By choosing

are

for all FeM"X", a, befll.", Ae[O,l]. Then rPhas the farm

rP

rP,-->rP uniformly

..

rP(F+a0b)

rP(F)= al + (J~F; + I' det F

rP(F)= A + Bf F;+ q(adj F)~+ D det F

satisfies (4.1), is C", and thus

Theorem 4.1. Let rP:M"X"->.91be contil3uousand such that both rP and

rP(F)=a+bF

p, '"rPclearly

I: -> O. For fixed i, IXlet F; = t, Fj =0 if j =t=

i or (J=t=
a. Then

is a special case of ERICKSEN[1], EDELEN[1, 2] and RUND[1, 2].

rP(F+(1-..1.) a 0b)=ArP(F)+(1-A)

rP, ~

rP,(F)=A(c)+ Bf(c) F; + Cf(c)(adj F)~+D(i:) det F.

We first study those functions rP(F)'thatbelong to the null-spaceof the EulerLagrange operator; i.e., those functions for which the corresponding EulerLagrange equations are identically satisfied. For smooth rP the following result

rank 1 convex on M" X",so that

:IFis said to be convex on M if it is the restriction to M of a real-valued convex

function (in the usual sense) defined on Co M; equivalently, .'F may be extended
to a convex function on Co M.

t,

"
. 359

Existence Theorems in Non-Linear Elasticity

J.M, BALL

358

Theorem

4.2 (BUSEMANN;EWALD & SHEPHARD[I]).

Proof. We give the proof for n=3, that for n=2 being similar. For ke:1l define
v,.<;;Et by
v,.= {F. adjF): FeM3d,
detF=k}.

if al1d ol1lyy'it has a convex lower ball/Idami the inequality

. (i) :F is COilvex 0/1 M

It suffices to show that Co v,.= El for all k. Suppose not. Then there is a closed
half-space
7t= {(F, A)eEI: F.IGf+A~H~~J.l},

:F(z&)~ LA/:F(ZI)
I-I
.
holds for all Zl' ..., z. tlnd Z&=

to Co M is given by
gF(Z)= inf
0-<,11-1

L AIZj lying

. I-I
)'I:F(Zj),

in M. A suitable convex, extensiol1

(G,HHO, with v,.<;;7t(ROCKAFELLAR

[I, p.99]). If RI'R1eM3d
are proper
orthogonal then
-I ..
I .
'r
T
T T '1<:
f. Gj -rA. HI =tr[(R1 FR2)(R2 GR, )+(R2 ARI )(R. flR2)].

zjeM, I ~r<oo.
,

Since adj(RIFR2)=Ri'(adj F) Ri', det (RI FR2)=det F, we may without loss of

generality suppose that H is diagonal. Suppose that H =t0 and assume without

(ii) Let Co M be open. 711e1l

either of thefollowingconditio/Isis necessaryand
sufficient fa/' :F to be convex on M:,
(a) :F has a convex lower bound alld the inequality

'

lossofgeneralitythat H~""0. Let

F=diag(kN-1

.+1
"(z&)~ L)'I:F(z,)
.+ I

L Zj Iyillg

I-I

.
L

i/I M.

"

(b) fill' etlch point zoeM there exist numbers tll(ZO)(i= 1, ..., s)such that
"(z)

~ "(zo)+

I-I

. 'For g: U --+:11we may define a function G: T(V)--+.q,f by G(T(F))=g(F), i.e..

for al/ zeM.

defined by (4.3)is convexon T(U). .

If n = I and V is convex.then polyconvexityof g is the same as convexityin the
usual sense. If n = 2 or n = 3. polyconvexity is characterized by the following

E=Etx:1l,
where

'EI=M3x3XM3x3

if n= 1,
if n=2,
if n=3.

Theorem 4.4. Let U be such thllt Co T( U) is opell.

71fe,!

~ is polY('IJIll'exit and

ollly it olle 01' the .fill/owing three e(/l/ipalell' ('IlIu/itiolJS holds:

(i) Ifn=2:

Thus E may be identified with .q,p,.'ln);

where s(1)= I, s(2)=5 and s(3)= 19.
Define the map T: Mn' n---E by'
'
T(F)=F
T(F)=(F, det F)
T(FJ=(F. adj F,det F)

(4.3)

Definition4.2.A functiong: U---3l is polycollt:exif and only if G: T( V) ---R

We now define finite-dimensional Euc1idean spaces E and Et by

EI=M1x2

if n = I,
if n=2.
if 11=3.

g= G .
g(F)=G(F.detF)
g(F)=G(F.adjF,detF)

ClI(ZO)(Z/_,%~),

El is empty

"

. Then adj F=diag(N sgn H:.kN-t sgn Hi, kN-t)and det F=k. Hence (F,adj F)e
Vk, but for N > large enough (F. adj F)17t. If H = 0 then we may assume that
G:""O, let F=(kNsgnG:,N-tsgnG:.N-t)
and proceed similarly. Hence
Vk$1t and this contradiction proves the result. 0

I-I

holds for all Zt. ..., Z,,+ I tlnd %If =

sgn Hi. Nt sgn Hi. NI).

there exists a cOIll'ex Jimctiol1 C(F, (i) Oil Co T( U) with

for aI/ FeU.

g(F)~ C(F,.sict F)

if 11= 1,
if 11=2,
if 11=3.

ami the illequality

f>

f>

g ,L 1 A;FIi)) ~L,~ I AIg(F(i))

le

Let V <;;M"' n. B)',1hothcoremo,n th'e inv~rial1c~,ofdomain-the ~et T(U)<;;

E.,is
epen, if and qnty.Jf U is'opm I1q~.
T( V) is not in general convexeven if U
is con vex (except in the case 11= 1). The following result shows, in particular,

that

f>

in certain important cases when V is open, so is Co T(U).

Theorem

4.3. Let K <;;.'1fbe IlOl1e/llpty tlml COIll'('X.Cllldlet U = {F e Mn xn: det FeK}.

711('11
Co T(V)=Et x K.

"
holds .fill' all Aj~O with L: Aj= 1. lmd .fil/' lIll Fli)e V sati4j'illg
1=1

'

i~ AjdetF1i)=det (~I AjFIi)).

-'

(4.4)

360

It

..'"

361

Existence Theorems in Non-Linear Elaslicily

J. M. BALL

If n

Theorem 4.5. Let U be sI/ch thut Co T( U) is opel1. If g is polycOlwex, thell g is

= 3:

'/I/ISicol1l'ex

there exists a convex fUlldioll C(F, A, <5)on Co T(U) with

g(F);?;;C(F,adj F, det F)
20

20

(I

A, F(I)

IMI

I-I.

A, g(F(/)

The converse to Theorem 4.5 is false if 11= 3. In fact when U= MJ. J and

20

.I Aladj FI/)=adj
.-1
'.
20
LAI

and the quasiconvexity of g follows from (4.7). 0

AI= I, and for all F(lIeU satisfying

20

Il1d

J ,~. dy= DJ[(adj(Fo + Vm~-(adj Fo)~]d)'= DJ[det(Fo + V')-det Fo]dy=O,

20

holds for all AI;?;;

0 with

I
.~ IMI

(I-II

A, F(I)

g(F)=ai! 1-:11-1.
'-,.11

(4.5)

det FIll

=det (I
I-I

A, FIll

I'

(ii) If 11=2:
for each Fe U there exist numbers a7(F), a(F) such that
(4.6)

Ill FeU.

(4.11)

forallnonzero

).,.jle.o;J,

(4.12)

so that g is quasiconvex. TERPsTRAalso showed that for 11=2 any quasiconvex g

of the form (4.10) satisfies (4.11)for suitable 81. and hence is polyconvex. I know of

no counterexample to the converse of Theorem4.5 if n=2. Conditions of the

rr 11=3:

'.

for each FeU

type (4.11) were studied by CLEBSCH[I] and HADAMARD

[1,3].

there exist lIumbers ,,;(F), b;{F), cIF) such that

g(F);?;;g(F)+a;(F)(I-~- r;.i)+biW)((adj F)~-(adj F)~)+c(F)(det F -det F)

for all Fe U.

. (4.7)

(iii) If n=2:
for each FE U there.exist numbers A7W), a(F) such that
g(F +n);?;;g(F)+AHF) 7t~+a(F) det n

(4.8)

ji". all F +nE U.

rr 11=3:
for each FEU there exist .ItlImbersA/(F). 8i (F), cIF) such that
g(F +n);?;;g(F)+ A7{F) n~+ 8i(F)(adj n)~+ cIF) det n
.tbr all F + nE U.

(4.9)

Proof. That polyconvexity of g is equivalent to (i) or (ii) follows immediately

from Theorem 4.2 (ii) and Thcorem 4.3. That (ii) and (iii) are equivalent follows

= F +n

for all F.

As was pointed out by MORREY[I, p. 26]. an example of TERPsTRA[I] shows

that if 11= 3 there exist constants ail such that (4.11) is violated for any 8i, even
ai!A';JJI.Jlp>O

by setting f

F)~~O

though

g(F);?;;g(F)+a7(FJiF.I~F:)+a(F)(det F -:-det F)

for

g(F)-87(adj

).)

(4.10)

with ail constant, it is easily seen that g is polycon~ex ir arid' only if there are

constants Bi with

20

I-I

U.

Proof. We give the proof for n = 3. Since G is the restriction of a convex function
to the set T(U). it is continuous and hence so is g. Let D be a bounded open subset
ot atJ, let Foe U and let' E!}j(D) satisfy Fo+ V, (y)e U for all ye D. By Corollary
4.1.1 we have

for all Fe U,

and the illequality

g

Oil

and rewriting

the right-hand

the same value in both conditions.) 0

sides of (4.7) and (4.9). (c(F) has

The conditions of quasiconvexity and polyconvexitymay be contrasted as

foHows.Consideringfor illustration the case in which11=3. let Ajand F(I)be given
satisfying (4.5).There is then no reason to suppose that a domain D, and a partition
of D into 20 open subsets Dj of volurpc ;.01and a set of measure zero, can be found
such that there is a continuous piecewise affine function on D whose gradient takes
the value F(/)on Dj (cf. the discussion at the end of Section 3). Only for such D, Dj
can the corresponding polyconvexity inequulity be deduced from the quasiconvexity condition. I remark that Theorem 4.4 (i) offers a way of proving that a
given function is not polyconvex.
By sacrificing the pointwise nature or the condition imposed on g. we may'
obtain a sufficient condition for quasiconvexity generalizing polyconvexity.

For brevity we discuss this new condition just when1f=3.

We consider continuous runctions G: Co T( U) -+:?fwith U ; M.lx J arbitrary.
Roughly speaking we require that the intcgral of G. rather than Clitselr. be convex.
To be precise. let )';?;;I.II;?;;I, v~ I and Ict ue W"(Q} satisfy Vu(:c)eU for almost
all xeQ. Suppose further that adj J7uel!'(Q)". det VueC'(Q). Thus if I(u) denotes
.

(Vu. adj VII.dct Vu), thcn I(u)ef14 d~ 1.J(Q)"x [!I(Q)" x C(Q). Let

R. = (I(u H): 'E.<Y(Q}}

Of course, if G is C'. thcn thc coefficients on the right-hand sides of (4.6) and
(4.7) are given by the derivatives of G with respect to its arguments. Condition
(iii) is the form given by MORREY[2. p. 123], who proved the following theorem:

and let (6~denote the closed affine subspace of ffdspanned by R.. Let
.)f~= (O'e(6~:O'(:c)eCo T( U) almost everywhere in Q}.

..

"I

<j;

Existence Theorems in Non-Linear Elaslicity

J. M. BALL

362

Define
J..(er)=

. To complete this section I remark that many of the results may be extended
without undue difficulty to arbitrary m and n; the polyconvexity condition is then
a requirement of convexity with respect to the basis elements of the null space of
the Euler-Lagrange operator. (See BALL[2].)

JG{er(x)) dx.
n

Suppose that J.. exists and is finite or +00 for all C1E~. Le~ G be such that J. is

a convex functionon the convex set~.

Definition4.3. If g: U -+!Jtthen g is said to satisfy condition (~'/l'

y)

at u if and

g(F)=G(F. adj F. detF)

The purpose of this section is to give a method for producing a wide variety
of nontrivial isotropic polyconvex functions. These functions will prove valuable
in Section 8 when we apply our existence theorems to certain models which have
been proposed for rubbers. We begin by discussing isotropic cOllvex functions of
n x n matrices for arbitrary n ~ I. We recall that the singular vC/luesof an n x n
matrix F are by definition the eigenvalues of the' positive' s~trljdefinite symmetric
matrix VFP. Woen F is the deformation gradient these eigenvalues are the
principal stretches of the deformation. When examining the results below the
reader should bear in mind equation (1.15).

(4.13)

for all FeU.

. Theorem 4.6. Let U he slid, that Co T( U) is opell. Let )I.JI.Vbe arbitrary.
.
(i) Let K: U -+ ,qt he 1'lycO/wex and boullded helow. alld let u be as above. 71len
K satisJies Py./I.. at 11..

..

with FoEU ami zeJ.f.l both cOllstant. Let gsatis.f)'

Py./I. ,. at 11. 7111.'11

g is quasi'm/l!ex

at FoE U.

Notation. Vectors in &toare de~oted by X=(XI' ...,x.) and the inner product
x, YE~o is written (x, y). !Jt':.denotes the positive orthant

Proof. (i) Sinceg is polyconvexthere exists a convex function G: Co T( U)-+ 9t

satisfying (4.13).Since Co T(U) is open G is continuous, and hence G{er(')) is
measurable foreach ere~. ByTheorem 4.2(i)we may supposethat G is bounded
below on Co T(U). Thus J. exists and is finite or + 00 for all C1e~. and J.. is
,I

. of two vectors

{x: xl~O for I ~i~n}

of !!l. 9,. denotes the permutation group on n symbols (an element P of &"acts
on an 11vector by permuting its entries).
We shall prove the following theorem:

clearly convex. .
.
(ii) Suppose first that G is Cion Co T( U), Let 'e~(Q) satisfy Fo+ r"x)e U
for all xeQ. Let C1= I(II), a= I(u +'). Then er,aE~. A standard argument shows

. nSG{C1+t(a-q))

that if

e(t)=

dx,

.1
I

. then e e CI([O,I]) with the obvious derivative. By the convexity of J.. we have
that e(O)~ eo) - e'(o). H.ence. .
.
aG
.
aG.
Jg(Fo) dx ~ Jg(Fo + r,) dx
n
n
.

fiG.

+
:1

- nS[ u~
:"'-/(I(u))

,~. + "'_-': r:\/ (I(II)) (adj r,)~

.

_. (I(Il))det .r, dx.

and so Kisquasiconvex
at Foe U.

..

The result followsfor general continuous G by a mollifierargument. 0

Condition

Py./IoV

does not in general imply polyc!1vexity.Indeed let g(F)

be given by (4.10) with CI:f' satisfying(4.12)but not (4.11)for any B:. so that g is
not polyconvex, Let )1=2 with JIand varbitrary and let U = M3" 3.Define G: E -+!Jt
by G(F, A. (5)=g(F) for all (F. A, (5)eE, Let u be as above. For any 'E WoulD) we
.
have (VANHOVE[I]. MORREY[2])
J a:fC',. '~p dx ~O.

It follows that J.. is convex on ~, and hence g satisfies Py./loY at U,

There are some grounds, connected with the results of Section 6, for believing
that condition Py,IIoY
at affine u and the quasiconvexity condition are equivalent.
but I have been unable to prove or disprove this.
.

5. Isotropic Convex and Polyconvex Functions

(';

only ifthere'exists G=G.: Co T(U)-+!Jt with the above. properties ard such that

(ii) Let 1I'(X)=F.;,x'+='

363

Theorem 5.1. Let n~ 1. Let cl>(t'l , ...,1'0) be a symmctric real-valued fU/lctioll

defined on !Jt':..For FE MO"0defille
W(F)=cl>(Vj' ..., vo).

(5.1)

where VI' :.,. voare the singular values of F. Thell

. (i) W is cOnt'ex on K=(VeMO"o: V is positive-semidefinite and symmetric}
ifand ollly ifcl>is COli
vex.
(ii) W L~convex 011MO"Oif and only if cl>is convex alUlnondecreasingin each
variable vI'
Remarks: Part (i), which is' probably known to matrix theorists, was stated
by HILL [3] for n = 3. HILL'Sproof relieson a property of the trace of the product
of two symmetric matrices. a recent proof of which has been given by THEOBALD
[I]. HILLassumed that W is differentiable everywhere, an assumption which rules
out several simple and useful functions and which can be surprisingly tedious to
verify. Proofs using differentiability obscure the geometri~ nature of the result.
The harder implication in (ii) is due to THOMPSON
& FREEDE[3]. Our method
of proof is broadly similar in approach. but rather different in detail. The result
extends work of VONNEUMANN[I] on gauge-functions and matrix norms (see
especially his Remark 5). For further information and references on singular value
inequalities see AMIR-MoEZ [I], AMIR-MoEZ& HORN [I], THOMPSON[I].
THOMPSON & FREEDE [1-3].

.,

...

-I

J.M. BALL

364

Lemma 5.1. (VONNEUMANN[1]; ~ee also MIRSKY[1,2]). Let A,BeM.x.

sillgular

values

IX,!;; 1X2!;; ..: !;; ex.!;; 0 a1ld p,!;; fJ.2"!?,.... "!?,.

P.!;; O. Th,en

(y.x)~YI.xl + ...+ .Vk,Xk

~CI XI +'''+Ck'~k
~/l.

(5,2)

Lemma 5.2. (VONNEUMANN[1]). Under the hypotheses

Q,R

be the coordinates of y in some order. Then

have

'

Itr (A B)I ~ (IX.PJ.

'max

of Lemma 5.1

(5.3)

tr(AQBR)=(IX,P),

so that ye1C. Hence' Lt =L.

M is a convex set containing the points Pc. Thus M, f,;M. Let 1Ccontain MI,
let yeM and let YI!i;;Y2"!?,.'"
~y.!i;;O be the coordinates of y in some order. Then

where the m(lximum i.~Wkell over a'Upairs Q. R of orthogonal matrices.

Proof. There exist orthogonal matrices QI' Q2. RI, R2'such that A =QI diag(lX)RI,
B = Q2 diag(PJ R2. Choose Q = Ri QI. R = RI Qf. Then
tr (AQBR)=(IX. P).

But for any orthogonal Q. R the matrices AQ and BR have singular values IXand
respectively. Hence tr (AQBR)~(IX. P) by Lemma 5.1. D

(y,x)~

2: .v,Xi

I-'
.-1

I-I

Proof. In Lemma 5.2put A =F, B=diag r. Then

(y,x)~I-'2:

II

I .

Thusye1C and MI =M.

Proof of Theorem 5.1.

'

k
I

in Lemma5,3 it follows that for

Lemma

5.4.

Let Cl !i;;C2 ~ ... "!?,.

c.!i;; O. Define

L={)'e.gp::(r.Py)~(r,c)
M= )'eL:

LI =Co
MI =Co
71len L=LI

t y,= t c,} .

I-I

C,. O.

.. ..

I-I

..

C,(Xi-X.)+,~. 2: C,= 2: C,Xi~J1..

i-I
I-I

u, !;;U2!i;;..."!?,.u.!i;;0.
Ihe set s

. for all rl!;;r2"!?,.'..."!?,.r.!;;0

andall Pe91:,},

'

i-I

(0. P(CI' C2'"''

.~~

be convexand symmetric.Let U, VeK have singular values(eigenvalues)

~ k ~ n, 2: v, is a convex function of F.

(i) That the convexity of W implies the convexity of tPis immediate. Thus let tP

place

I-I

'

.-1

and each tr (FQBR)is a convexfunction

of F. 0
.
,1,0,...,0)
k'h

Y,

Choosing rj=x,-x.(1 ~i<n), r.=O in the definitionof L we obtain

'

. (5.4)

0-1

= 2: Yi(Xi-,XiI)+'X. 2: Ci'

Lemma5.3. Let rl!i;;r2!i;;"'"!?,.r.!;;0. 71le1l'(r.v) is a cOlwex fUllction of F, where

t'l !i;;v2!i;;... "!?,.
v.!;;0 are the singulart,aluesof F.

(r, v)=max
Q.R tr(FQBR).

= I-'2: Yi,Xi+ i-I2: c,- -I-I

2:
.-1
.
.

Remark. By letting r=(I

365

Existence Theorems in Non-Linear Elasticity

Let Ae [0, I] and let A

Let C=AU +(1-A)

v,!i;;v2!;;...!;;V.~0.

~ AU +(1 - A.)V have singular values al !;;a2!;;'" !i;;a.~O.

v. Note that for some orthogonal

0); Pe&:,. 1 ~l~n},

Q. R we have,

2: ai= 1,)-1 (AQfjuj+(I-A)Rfjl')=

.

i-I

'

2: ci.

i.1

'

(5.5)

By (5.5) and Lemma 5.3 we have a EM. Hence by Lemma 5.4 a e MI. Therefore

(Pc: Pe,~}.
alld M=MI.'

'

Proof. L is a convex set containing 0 and the points P(CI,C2'''''C,,0

0).
Thus Lt!:;;L. To show that Lf,; L, we prove that any closed half-space in ~.
containing LI also contains the closed convex set L. Let such a half-space be
1C=Iye!ai.: (Y.X)~/l}, where xe.C}t., /le.C}tare fixed. Let the coordinates ,of x in
some order be
XI"!?,.X2!i;;...~.Xk!;;0>,Xk+I!i;;...!i;;X.,
'

where

p'e91:" A, !;; 0,

~.VII"!?"O

1.1

W(A)=4'>(a)=<P.

A,= 1. Thus

(ti-I A.;1~c),

(5.6)

W(A)~ 2: A,tP(P'c)= 2: A{cP(c)= tP(c)~ A4'>(U)+(1-A) tP(v),

I-I
I-I
where we have used the symmetry of tP. Hence.
W(A)~A W(U)+(t -A) W(V)

where the symbols to the right of 0 are omitted if k = n. Let ye L and let
.1'1 !i;;.v2!i;;...

.
I
.

as required.

(5.7)

,t,

d.r
g(V)

'

Theorem 5.2. If n = 2 let

W(F)=t/l(vl'

'

,',

367

Existence Theorems in Non-Linear Elasticity

(H)Let W be convexon M"~. Then clearrytP is convex.Also for fixednonnegative VI' oo., Vk-I' Vk+I' oo., V.

J. M. BALL

366

.t

~,

V2' VI V2)'

, (5.8)

'

= W(d lag(vl,,,,,Vk-I,V,Vk+1'

,,",v.

is convex in v. But g(v)= tP(vl' .oo,Vk-Jt lvi,Vk+1'

v.) and any even convex
function of te,~ is nondecreasing for te:;O. Hence tP is nondecreasing in each
variable.
'

'

Let <f>
be convex,symmetricand nondecreasingin eachvariable.Let).e[O,1]
and let F,GeM"', A = 1F +(1 -1) G have singular values
U.e:;U2e:;oo.e:;U.e:;O,VIe:;V2~...e:;V.e:;O,

ale:;a2e:;...e:;a.e:;O
'

The geometrical basis for Theorem 5.1 is easily seen by considering the special
cases n = 2 and n = 3 (see Fig. 2).
2

'

(a)
(b)

t/J(XI' X2' D)= t/J(X2' xl' D) for all XI' x2 eai' + ,'DeK,
t/J(XI' X2' D) is nondecreasing in XI' X2'
Ifn=3let
W(F)=t/J(vl'

and satisfies
(a) t/J(Px,Py,D)=t/J(x,y,~) for al/ P, Pe:Jl'3and all x,ye9f~,oeK,
(b) t/J(xl'x2,x3'YI'Y2'Y3,D) is 1I00ldecreasin$
in each xl,}oj.'
TIlen W is polycol1vex on U.
We give the proof for n = 3. Define

Proof.

(5.9)

V2' vJ' V2 vJ' vJ VI' VI V2' VI V2 V3)'

where VI' V2' V3are the singular t'alues of FEU, and where t/J:9f~ x K

respectively. Let c=lu +( 1-1) 11.By Lemma 5.3 a EL. Thus aeL1. and since tP
is nondecreasing in each variable we have
W(A)=tP(a)~c1>(c)~l W(F)+(I -1) W(G).

where VI' V2are the singular values of FeU, and where t/J:ai'~ x K -+ ai' is convex
and satisfies

G(F, A, b)=

t/J(VI'

-+9f

is conveX

G: E. x K -+ Eft by

v2. V3' al' a2, aJ' D),

(5.10)

V3

where VI'al are the singular values of F, A. G is well-defined by (a). Clearly

W(F)

= G (F, adj

(5.1l)

F, det F)

for all FeU. It remains to show that G is convex. Let F,H,A,BeM3x3,D'f.1eK,

1e[O,I]. Let F,H, A, B,U +(1-1) H,lA +(1-2)B have singular values VIe:;
V2 e:;V3' hi e:;h2 e:;h3, a1 e:; a2 e:;aJ' bl e:; b2 e:;b3, UI e:; U2 e:;U3' dl e:;d2 e:;d3,

respective-

ly. Let w = 111+ (I -1) h, C= 1 a + (1- ).)b. Using Lemma 5.3, Lemma 5.4, (a) and
Cb)we see that
G(lF +(1-1)

H,lA +(1-1)B,

VI

f.1)=t/J(u,d.1b+(l-1)f.1)
'~t/J(w,d,H+(1-1)JI)

~t/J(w, c, H+(1-1)JI)

V2

Fig. 2. (i) 11" 2

H+(l-l)

~ 1 t/J(I1,a. 15)+(1 -1) t/J(h. b, JI)

Fig, 2. (ii) n=3

=lG(F.

A. (5)+(1-1)
,

Taking the case /1= 2 (Fij!. 2 (i)) first, we see that the proof ofTheorem 5.1 (i) shows
that a lies on the line segment PQ (note also that the points splitting A C, BD in the
ratio 1: 1-1 lie on PQ) and clearly tPis less than tP(c)on this segment by convexity.
In the case' (H) a lies ,in the shaded area L. If n = 3 (Fig. 2 (ii) the set M is the hexa-

gonal area enclosed by the points PQRSTIJ and lying on the plane
VI +V2+VJ=CI

+C2+('J'

G(H, B. JI).
0

A special case of a function t/Jsatisfying the hypotheses of Theorem 5.2 for n =3 is

t/J (v. a. b)=

t/JI (v) + t/J2 Ca) + t/J3 (b),

(5.12)

where the t/Ji are convex, and where t/JI' t/J2 are symmetric and nondecreasing in
each variable. We use this example in Section 8.

"

while L is the convex hull of M and its projections onto the coordinate planes, so'
that it is part of the cube of side Cl shown.
We now give some sufficient condition for polyconvexity. Let n=2 or 3. To
keep things simple we consider stored-energy'functions W(F) defined on se'tsof
"

the form U= {FeM"': det FeK}, where K<;;f:f.+

is convex.

6. Sequential Weak Continuity of Mappings on Orlicz-Sobolev Spaces

Definition 6.1. Let X and Y be real Banach spaces. A map f: X -+ Y is sequentially
weakly continuous if and only if xr~x in X implies f(xr)~f(x)
in Y.
(In general a nonlinear sequentially weakly continuous map f: X -+ Y is not
continuous with respect to the weak topologies on X and Y (see BALL[I]).)

,.,

1.M. BAll.

36g

Existence

. Sequentially weak * continuotIs maps are defined analogously. In this section

we study the case when X is an Orlicz-Sobolev space and Y = IJ(Q). The reader
unfamiliar with the theory of Orlicz and Orlicz-Sobolev spaces can replace them
everywhere by the corresponding Lebesgue and Sobolev spaces, the results for
which are indicated in parentheses in soine of the theorems which follow. There
will be no great loss of generality in doing so, at least as far as most of the examples
in Section 8 are concerned. Some of the results proved here in the framework of
Orlicz-Sobolev spaces are proved for ordinary Sobolev spaces in BALL[2].
We first obtain necessary conditions.
Theorem 6.1. (MORREY [I ]). Let m and n be arbitrary. Let t/J: gIIm X gII"

SrJt(x,

ii

(6.2)

=2:

If u e WI. 2 (0) then det 17u e LI (0) and formula. (6.1) holds

S(det Vu) rPdx= S(Ul U2.t rP.2 -ut U2.2 rP.t)dx

a
n

SwJrP.Jdx=O
.

for allrPegj1(Q).

17uX

We
(6.5)

If ue Coo(D), then dlv W=0 and (6.5) holds. Since Coo(Q) is dense in WI. 2(0),'
(6.5) holds for any ue wt EA(Q).
.
To show that (6.3) holds in S7JJ'(O)
it is thus sufficient to prove that

.
.

=(adj

To prove (b) we first note that ul.JeEA(Q), wJeE,{(D) so that detl7ueV(Q).

next show that divw=O in a weak sense; i.e.,

The arbitrariness ofae implies that O=rP(I7u(').and hence rP(I7ur(')-L..rP(I7u(')

rPis affine.For details see BALL[2].

(6.4)

(ii) The proof of (a) is identical to that of (i). Let w be defined by wJ

S rP(I7u(x)ae(x)dx.

Remark: MORREY'Sproof of Theorem 6.1 may be easily adapted to show that

if D is a bounded open subset of ~ (m arbitrary) and if rP:gII-+iJl, then the map
()""1>((}(')) is sequentially weakly continuous from lJ'(Q) -+ LI(Q) if and only if

for all rPeS7JJ(Q).

But (6.4) holds trivially if ue COO(Q),and C"'(D) is dense in WI,2(O) in its norm
topology. Since both sides of (6.4) are continuous functions of ueW1.2(Q), (6.4)

holds fo~ue wt. 2(Q).

Proof. If the given map is sequentially weak * continuous then by the theorem
both rP and rP are quasiconvex,so that by Corollary 4.1.1 rP has the form (4.2).
Conversely. let rPhave the form (4.2) and let ur-L..u in WI.",,(O). The sequence
rP(rur(.) is bounded in L""(O),so that there exists a subsequence Uuof Ursuch that
rP(Vuu(.)-L..O in L""(Q). Let 11.:gfJ -+ gII be contintlous and define rPI(x, F)=
:trP(F) IX(X)so that rPI is quasiconvex. By the theorem

Proof. The hypotheses imply that u -+ J(u, Q) is a sequentially weak * continuous

map from WI:"'(Q) qt. 0

in S7JJ'(0).

Proof. (i) That det Vue Lt (Q) is obvious. Formula (6.1) holds in S7JJ'(Q)
if and only if

(63)
..

U3 )] .3

(ii) n = 3: (a) If u e Wl. 2(0) then adj 17u e Lt (0) and formu1a0(6.2)holds in S7JJ'
(Q).
(b) Let A,B be N-functions with A~t2, A~B. If ueWIEA(O) and
adj l7ueEB(O) (c.g., ueWI.P(O), p~2, and adj l7uel!'(!2)) then
det Vue LI(0) and formula (6.3) holds in S7JJ'(!l).
.

Corollary 6.1.2. Let n=1. 2 or 3 and let A be an N-function (cr. Section 2). Let
rP: M"'" -+.~ bi! continuous and such thut u H 4>(I7u( ) is a sequel1tially continuous
mup (!{Wt LA(D) with the weuk * topology into l!(Q) with the weak topology. Then
rP Iws tlteform (4.2).
.

(adj l7u)f=(UI+2ul+I,o+ 1),0+2_(u/+2 UI+I.0+2),O+I

Lemma 6.1.

X M" x m -+ gII

a
.
to gIIif andonly if rPhas the form (4.2).

inL"(Q).TheresuItsfollows. 0

n =3:

(In (6.2) there is no implied summation, and the indices are to be taken modulo 3.)

(6.1)

2 U3 3 -U2 3 U3 2 )] I
".'"
+' .[ Ul(u2 .3.1
U3 - U2
U3 - U2,I U3,3)] ,2 + [ Ul(U2.1.2.2.1

Corollary 6.1.1. Let n = 1, 2 or 3 and let rP:M" X"-+ gII be continuous. Then the
map uHJ(U. 0)=
rP (Vu (x)dx
is a sequentially weak * continuous map from

SrP(l7uu(x)IX(x)dx

n=2: 'det I7U=(UIU2.2).t _(Ul U2,1).2

!
I

is sequentially weuk * lower .~emicoI1tinuouson WJ..",(0) (i.e.. "r -L..u in WJ. at.(0)
implies l(u,Q)~lim.
r-'" I(u"O)) if and only if rJt(Xo.llo,') is. quasicOlwex 011M"xm

369

Elasticity

det l7u= [ ul(adj I7U){

I] J =[UI(U2
.
'

"

II(X). 1711(x)) dx

for each xoegllm,uoegll".

in Non-Linear

(i) n
[(ut Q)=

Theorems

.' We turn now to sufficient conditions. Our results are based on the following
elementary identities for C2 functions u:

he continuous. Then'

WI'''''(Q)

"

"

Sut.JwrPdx= - aSUi wJrP.Jdx

for all rPeS7JJ(Q)

(6.6)

whenever weEB(Q) and satisfies (6.5).

By the results of DONALDSON& TRUDINGER [1. Thm. 22] there exists a sequence

u(k)eC""(D)with

U(k)-+U

in WIEA(D).Let pe~(.qt3),p~O.
Sp(x)dx=l, and
.
,,,.'

. define pkeS7JJ({Jt3)
by Pk(X)= kp(kx). Extend w by zero outside D, so that we EB(.qt3).
DONALDSON
& TRUDlNGER[1. Lemma 2.1] show that the convolutions pk*w
are in EB(gII3)and pk* W"" win. EB(9f'3)as k 00. Fix rPeS7JJ(Q).
Then if k is large
. enough, (6.5) implies that
.
div(pk*w)(x)=

SPk.J(x"":y)II,J(y)d)'=O

.,>

(6.7)

..,

,.,

Existence

J.M. BALL

370

'" wl)

q, dx =

Jdiv{uik)(Pk '" w)

q,)dx - J Uik)(Pk'" wJ)q,.Jdx

n

Since A -<B we obtain (6.6) by letting k -> 00. 0

'"

Remark. Note that A>t2 if and only if A>A. This may.beproveddirectly

from the definition of A.

Proof. (i) If

n = 3: (Adj Vu): = (UI+2U:~~I).o+2_(Ui+2U~~~2).0+1'

(6.9)

Det Vu = [UI(Adj Vu){J.J'

(6.10)

the imbedding theorem of DONALDSON& TRu-

then

III eLA.(Q).

The result followsby Young's inequality.

0

(ii) This is proved similarly.

Remark. If Q is an arbitrary bounded open set then Leinma 6.2 holds with

L'(Q)replaced by Llloc (Q).

The main result of this section is the following:

Theorem6.2.
(i) n=2:

when these distributions are meaningful. Obviously if u satisfies the hypotheses

of Lemma 6.1 then these distributions may be identified with the LI(Q) functions
det Vu (n=2). adj Vu and det Vu (n=3) respectively.
Let A be an N-function. Following DONALDsoN& TRUDINGER[1], we let

Let

'"
A be an N-function

and A ~ A"'. If

IIr-L-U

then DetVur->DetVu

satisfying

in WI LA (Q) (e.g.,

JgA(t)dt<oo,
0

(6.12)

in IJ(Q) with p>l,q>1

then we define the N -f~nction A'" by

(6.13)

Note that for A(t)=IW, 1<p<n, we have

I
gA(t)=tp-'jj-1

")

A*(t)=~

Lemma 6.2. Let Q satisfy the cone condition.

(i) 11=2: Let A be an N-fimction satisfying either

.p'
'"

(6.14)

J(

n
.

JgA(t)dt<oo,

p>1),

with either A sati.vying

JgA (t)

in W1.P(Q),
.

dt

< 00, or

Illr 112
r,2 _1/1

U2.2 )

A.dx=
'I'

J( r

Ul -1/1

n'

-kdx+
) 1/2
r. 2 'I'

JU1 (1/2r.2.-1/2

!J'

-I.dx
2 ) 'I'

tends to zero as r -> 00. Hence I/~";.2 -> Ul":2 in 0'(Q).Similarly II~U;.I -> IIIU:I
in ~'(Q). The result follows. The proof of (ii) is similar. 0

or both (6.12)

CorolIary 6.2.1.

A -<A*. If ue W. LA(Q) (e.g., if ue W1.t(Q)), then UIU~2and UIU:I

belollg to LI(Q), so that Det 17u exists as an element of 0'(Q).
and

(ii) 11=3: (a) Let A he as in (i). If ueW.LA(Q)

(e.g., if ueW1.t(Q)), then
. 1/+ 2Ui+1.0+1. and Ll +2UI+ I.0+2 belon g to LI (Q) so that (Ad
' Vu)O
. J
i exists

in W1.P(Q),

Proof. (i) Fix q,E0(Q) and let Q' be'an open set with Q:::>Q':::>suppq,and such
that the imbedqing theorems of DONALDSON& TRlJDlNGER hold for Q'.. Then
since Ilurllw'LAm')is bounded and "r -+u in 0 (Q'), it follows (see KRASNOSEl'SKII
& RUTICKII[I, p. 132]) that ur -+ u in L'"(Q') or L,(Q'). Therefore the Holder
inequality and the boundedness of IlurllwILA(n')imply that

It1

JgA(s)ds.
0

ur~u

and ..! +..! <1), then DetVur->DetVuin

p
q

0'(Q).

(A"')-I(ltD=

if

or both (6.12)

with A satisfying (6.12) and with B~ A*. If ur-L-u in WI LA(Q) and

Adj VUr-L-Adj Vu in LB(Q) (e.g., ifur~u in W1.P(Q), Adj Vur~Adj
Vu

JI gA(t)dt=oo,

. .

JI gA(t)dt<oo,

in 0'(Q).

(b) Let A, B be N-functions

where A-I denotes the inverse function to A on [0,00). If A satisfies'

'"

either

(ii) n=3: (a) Let A be as in (i). If ur-L-u in WI LA(Q) (e.g., if ur~u

p>t) then (Aaj VUr):-> (Adj Vu): in 0'(Q).
'"

(6.11)

t.;;;;0,

JgA(t) d t < 00

'theorem implies that

(6.8)

Det VU=(UI u:2).2;-(UI U:I).2'

ere-

[I, Thm.3.2] implies that UlEL"'(Q), while if A satisfies (6.12) the same

DINGER

The functions det Vu (n = 2), adj Vu and det Vu (n =3) can be given a meaning
as distributions under weaker conditions than those of Lemma 6.1. We thus
define the distributions Det Vu (n=2), Adj Vu and Det Vu (n=3) by

..! +..! ~1), then III(Adj VII){ ELl (Q), so that Det Vu exists as an
p
q
.
ment of 0' (Q).

'n

. A-I(t)
gA(t)= tl+l/" ,

371

Elasticity

or with A satisfying (6.12) and with B-<A*. If ueWI LA(Q) and

Adj VIIELB(Q) (e.g., if UE W1.P(Q), Adj VUELq(Q) with p> 1, q> 1 and

= - J U!k)(Pk '" wJ) q, .Jdx.

n=2:

in Non.Linear

'"
(b) Let A,B he N-functiolls with either A satisfying f gA(t)dt<oo.

for all XESUppq,. Therefore if S9P3 is an open ball containing

suppq" then
.
.

JUlk)./h
as.

Theorems

;, (i) 11=2:

'

Let A he as in 71leorem6.2(i) and let UE WI LA(Q). Then

Det VU=(1/2U~I).2-(Ll2U~2)..

"

. We may take for {1' a finite subcover by 'Open balls of the closure of supp

as all element of I'(Q).

:'.

(6.15)

in I'(Q).
.p.

.,

'r:
372

'5
ExistenceTheorems in Non-Linear Elaslicily

Let rE(O, I), B,={lxl<r}

(ii) '11=3:

Det J7u=[u2(Adj Vu~lJ=[u3(Adj

Vu}D.J

But

in !Z!'(Q).

- J <pu(n.dS- J <p..uldx.
~s.

n s.

J <puln.dS=~nJ <p(rx)u/(rx)lI.r"-ldS.

(6.17)

Provided r"-I R(r)-+O as r-+O and u, J7uELt(Q) we therefore obtain from (6.21)
that
J<pu~.dx=- J4>..uidx,
n

{l

where we have used the dominated convergence theorem. Thus under these
conditions J7n, as defined by (6.19).is the matrix of weak derivatives of u.
In particular, letting R(r)= I +r, we find that UEWI.P(Q)for any P<II. Now
let n = 2. Let I/IECOO
([0, I]) take the values I and 0 in neighbourhoods of r =0 and
.
del
. .'
r= I respectively. Then <p(x)= 1/1(Ixl)' belongs to !Z!(Q), ana

[I] to show that if wELs(Q), then Pt * w.i..w in Ls(Q). We omit the

details.

I
.
J (I +r) I/I(r) dr.
0
I

J(detJ7n)4> dx=2rr
n

n = 2: The map Ul-+det Vu: WI. P(Q) -+ lJ"2(Q) is sequentially

...
tinuous if p> 2.

weakly

(6.21)

~s.

Proof. (i) Let <PE!z!(Q)and let Q' be an open set with Q::>Q'::>supp<p and satisfying the segment property. Then by the results of GOSSEZ [I, Thm.1.3] there
exists a sequence U,ECoo(Q') with u,.i..u
in WI LA(Q'). Clearly (Det Vu,)(<p)=
[(U;U~.I).2-(U;U~.2),I](<P)' Letting r-+OCJ we obtain from the theorem that
(Det Vu)(<p)=[(U2U~I).2-(U2U~2).I](<P),and the result follows.
.
(ii) The proof of (a) is identical to that of (i). The proof of (b) is similar to
that of Lemma6.I(iib),
the principal change being the use of Lemma 1.6 of

0
Corollary6.2.2.#

J <pu~.dx=
n s.

in !Z!'(Q). (6.16)

(b) Let A,B beas in Theorem 6.2(ii b) and let ue WI LA(Q), Adj Vue Ls(Q).
Then

GOSSEZ

and let <PE!z!(Q).Then

(a) Let A be as in, Theorem 6.2(iia) and let ue WI L,4(Q). TIlen

(Adj VU)~=(UI+IUI+2 "o+2).+I-(UI+IUI+2"+I).+2

(i)

373

J.M. BALL

conJ (Ul U:I <P. 2

-.

UIU:

2 cjJ .I)dx

= -

rr

J (1 + r)21/1'(r) d r.

(ii) n = 3: (a) The map ul-+adj V'u: WI..p(Q) -+ lJ'12(Q) is sequentially weakly continuous if p > 2.
.
(b) The map ul-+det Vu: W1.P(Q) -+ lJ'/3'(Q)is sequentially weakly continuous if p> 3.

. Hence formula (6.1) does not hold in this case. so that detVu4=DetJ7u. Note
also that if p<2 there is no sequ.ence of C'XJ(Q)functions u, such that U,--'-Uin
W1.P(Q) and det Vu,--'-det Vu in LI(Q), since such a sequence would satisfy

Proof. We just prove (iib). Let p>3. It is clear from the Holder inequality that
if UE W1.P(Q), then det VuelJ'/3 (Q). Let u,---"-uin W1.P(Q).Then adj Vu, is bounded

nJ det VU,dx=4rr.

whereas

~nthe reflexive space l!'2~Q), and henc~ by the theo~em (part (iia~ adj Vu,---"-adjVu
In lJ"2(Q). But det VU, IS bounded ID the teflexlve space lJ" (Q) and tbus by
part (iib) of the theorem det Vu,---"-detVu in lJ't3(Q). 0
.
Warning. The distributions det J7u (adj J7u) and Det Vu'(i\dj Vu) need not be
the same even if the former is a continuous function, as the followingexample.
shows.
'.
I

Example6.1, n= 2 or 3.

JdetVudx=3rr.

When n=3, a similar calculation shows that nE W1.P(Q) for p<3, adj VueL4(Q)
for q <!. but that (6.17)does not hold.
In the ahove example Det Vu has an atom at x=o. I do not know~ whether

det Vu = Det V u if Det J7n is a funet ion.

Let r=lxl and let R(r) be a smooth real-valued function on.:(O,1].

Let Q= {Ixl< I} and define u: Q-+~" by
R(r).
u(x)=- r x,
Then for r>O we have
(

u.-.

and

det Vu

. No/!'
added ill

r>O, u(O) arbitrary.

-~ <5.+
(rR'-R

r3

7. Existence Theorems

/ p
<5.p X x

R"-I R'
r"-I

We have already stated the result of MORREY.Theorem 6.1. which shows

that for a continuous integrand quasiconvexity is necessary and sufficient for
sequential weak * lower semicontinuity in Wt. ,t.(Q). MORREY[I. 2] has also
given sufficient conditions for sequential weak lower semicontinuity in WI.!(Q).
s ~ I. For purposes or comparison. and ror ruture use, we giverj an extension or
his results due to MEYERS[I].

(6.18)
. (6.19)
(6.20)

proof: This corollary

follows from resulls of Y. G. RESHETNYAK [1. Thm.4].
[2. Thm. 2]. Theorem 4 in BALL [2] is also essenlially a consequence of RESHETNYAK'Swork. 10 which
I would have referred had I seen it in time.

"

Theorem 7." Let f: Q x :jf"x M" X"-+:If he C(}lltimlOtls.all/I let f(x. n. .) he quasic/IIwex .fl}r all xEQ. 11elf". Suppose there exist real CO/1stantsKi>O (i= t, 2).
s~ I. O<y;;; 1 and a jimctio/1 hEL1(Q) sueh that

..

,,,.

374

.~

"'!

J. M. BALL

375

Existence Theorems in Non.Lincar Elasticity

(i) fIx, u, F)~b(x),

We shall use the following lower semicontinuity theorem, which is a special case
of a result given by 'EKELAND& TEMAM[I. Thm. 2.1, p.226]. For related results
see CESARt[1,2,3].

(ii) l.f(x, u+ v, F +H)- fIx, u, ~I ~ Kt {I +Oul +Ivl +IFI+IHI)'-Y}(ivl+IHIV,

(iii) If(x+ y, u, F)- fIx, u, F)I~K2 {l +IFI'} I/Oy!),

for all

values of the various arguments, where I/:Bl + -+ Bl+ is a continuous increasing junction with 1/(0)=0. (Here and elsewhere IFI denotes any fixed norm

on FeM"X".) Then

Theorem 7.2. Let GI: Q x (Bl"XBld)-+i:Hbe of Caratheodory type (Definition 7.1

with v=n+u) and satisfy

GI (x, 11,a)~

I(u, 0)= nSf(x, u(x),

Vu(x)dx
.

for someN-fill1ctionCP.Supposethat Gi(x,u,') is convex on f!I('for all xeQ, ue!:H".

Let 1,->-1
in LI(Q) and let {u,} be a sequence of measurable functions with u,-+U
almost everywhere in Q. Then

is sequentially we(/~ly lower semicontimlOus on Wt.'(Q).

Remarks.

SGI (x. u(x), a(x)dx~

/. Let xoeQ. Then conditions (ii) and (Hi)imply in particular that'

and

(7J)

If(x, u, F)- fIx, 0, 0)1~ ~I [I +Oul+ IFJ)'-Y](lul+)FIF

If(x, 0,0)- f(xo' 0, 0)1~ K21/(lx-xol).

(7.2)

(i)' If(x,u,F)I~Ko(I+(iYI+IFI)'},

'.

where KJ>O.

(a) for almost all xeD. G1(x. .) is continuous on .~". and

(b) for all le:!!". G1(', a) is measurable on D.

is said to be of

KI = + co, K2=

co (no constraint).

(Ht)### G is continuous with respect to /I, a on ,c/"

(H2)
for all ue..:JI'",a EE, G('. /I.a) is measurable on W -I (a).

(HJ)

with Ilr>s; we shall see that many such integrands belong to a physically interesting class included in our existence theorems.
be open. A map G1:Dx.~v-+ri

type and convex with respect to

leE. and where KI' K2: Q -+Ii are measurable. Examples of relevant choices of
Cf' Kf U= 1, 2) are
1. Cf is arbitrary,

The conditions (i)-(iv) are extremely restrictive with regard to applications to

non linear elasticity. Firstly, (ii) precludes any singular behaviour of f (for example (0.5)). Secondly (i)' and (iv) together rule out integrands typified by the
example
f(F)= IFIS+ Idet FI'
(7.3)

Definition7.!. Let D~k

ClIratlu!ot!ory type if

(7.6) .

2. (n =3)-C2(X. F, A. 5)=5,K2=0 (corresponding to the continuity condition

det Vu(xO and an additional unilateral constraint, absent if KI= + co, on the
.
measurl= of strain Cl(x, Vu, adj Vu, det V/I) #).
We now make continuity, growth and polyconvexity hypotheses on the
integrand. Because of the nature of the growth conditions, and because we wish
to consider situations in which the distribut,ions adj Vu and Adj VII. det Vu and
Det Vu may be different. we make these hypotheses on the associated function
G(x, u, a) (cf. (4.3).
Let t/' = (x, u, a)e Q x 91'"x E: a e W(x)}. Let G: .er -+ 91'be such that ##

In order to prove existence theorems by use of Theorem 7.1; it is necessary

to make, in addition to conditions (i)-(iii), a coercivity assumption on f. Typically
we might assume that s> I and'
.
.
(iv) f(x,u,F)~KJIFls+b(x),

(7.5)

We now describe the main ingredients of our existence'theory. starting first

where Cl; c2: !:Hx E -+ Bl are of Caratheodory

2. If s~n then the growth conditions with respect to u may b~ weakened by

use of the Sobolev imbedding theorems; the reader is referred'to MEYERS[I]
for details. An extension to an Orlicz-Sobolev space setting could also prob.

SG1(x. u,(x), a,(x)dx.

W(x) = {leE: Cl (x, a) < Kt (X), K2(X)< -c2 (X, a)}.

for all (x, u, F), where Ko>O .is a constant. Conditions (i)', (Ii) and (Hi) are
MEYERS'continuity and growth conditions for the function If b, while (i)

ably be made.

lim

'-'X. (I

with those relevant for compressible materials.

For each xEQ let W(x) be a nonempty convex open subset of E=.tJiS("),such
that for each lEE the set W-I(a)d~ (i-eQ: aeW(x)} is measurable. We impose.
. as a local constraint on our variational problem that, in a sense to be made precise
later. T{Vu(x)eW(x) almost everywhere in Q, where we are using the notation
of Section 4. In applications to nonlinear elasticity W(x) will often have the form

Combining (7.1) and (7.2) we see that'

impliesa furtherhypothesisof his theorem.

(7.4)

r!>(la!)

(H4)
(Hs)
-------.-

for almost all xeQ, G(x./I.a)-+ +co as a-+iJW(x). the convergence

being uniform with respect to /I in any bounded subset of .'1P.",
(Polyconvexity) for each xeQ. /le&f". G(x, u,') is convex on W(x),
(Coercivity)

Such constraints are by no means unrealistic. A unilalenll constraint at large strains might be
relevant. for example. for a mixture of elastic materials with onc or more constituents possessing
limited extensibility (,I: NIEDERER[I]). See also the comments in Section 10.
For simplicity wc suppose that G is defined for all UE.o/I".
...
Weaker conditions are possible (Il EKELANO& TEMAM[I]).

..

'c
J. M. BAtL

376

.', Existence Theorems in Non-Linear

.
forall (x,u,F)e.9';

n = I: there exist~ an N-function A, and a function be Lt (Q) such that

, G(x,u,F)~b(x)+A(F)'

to establish the existence of a minimizer in

J(U)d;!
.

G(x,u,F,H, c5)
for all (x,u,F.H,c5)e.9',

'

n=3: .w={ueWILA(Q):AdjJ7ueLB(Q),
DetJ7ueLc(Q), (ru(x),
Det J7u(x)e W(x) almost everywhere in Q}.
The equivalence classes in .cI under tne equivalence relation

(AdjJ7u)(x),
,

are termed the Dirichlet classes'in si.

Adj J7uj-L-H

and

...

J7Uj~J7uo in V(Q),

in LB(Q),

Adj J7l1j~H

Det 17u)-L-c5 in. LC<Q),

<

Det J7uj~c5

in V(Q),

in V (Q).

'

J(lIo)~ iim J(II).

)-'X>

Since GI(x,u,a)=+oo
if a~W(x) it follows that (J7uo(x), AdjJ7uo(x),
Det l7uo(.x)eW(x) almost everywhere. Thus uodi' and the result follows. 0
We now give a modified version of Theorem 7.3 for the case in which n =3
and G is independent of 15.The proor is similar and is omitted.

'

(6.12) and A-A*,

if n=I,

J7u(x))dx
.

DetJ7u(x)dx

nJ G(x, u(x), J7u(x), Adj J7u(x), Det Vu(x)dx

if 11=2.

if n=3.

M3" 3XM3" 3, Let G: .'/ -+ &l satisfy (H I)-(H4) and the following hypothesis:
'"
(H6) There exist N~runctions A,B with A sati.~ryingeither gA(t)dt<oo or both

'

Lll1da fUllction beL1(Q) such t/1c.I1 I

G(x, u, F, H)~b(x)+A(IF!)+

B(lH!)

for all (x, u, F, H)e.'l'.

Define
.cI= {ue WI LA(Q): Adj J7ueLB(Q), (J7u(x), Adj J7u(x)eW(x)
almost el'erywhere in Q}.

Let
1(11)= f G(x. U(X), I7I/(X), Adj Vu(x))dx.

We are now in a position to present our first existence theorem. which in;

J7ur(x) d;<

If ue.r:1 then it follows from results ofEKELANO'&'TEMAM [I, Prop. 1.1. p. 218]
that J(u) exists and is finiteor + 00,where

d;!

J C(Det

Theorem 7.4. Lel 11 3. In the dejlnitio/lS of W(x) (I/!Cf Y' replact! E by El

ifand only if u-veWJLA(Q)

J7u(x),

VUr(x)i)dx.

By Theorem 6.2(ii) H =Adj VUoand (5=Det J7uo' By Theorem 7.2,

n= I: .cI= (ueWI LA(Q): J7u(x)eW(x) almost eve~ywhere in Q},

n =2: ..cl= {ue WI LA(Q): Det J7ueLB(Q),(J7u(x), Det J7u(x)eW(x) almost every.
,
where in Q}"
'

= JG(x,u(x),
n ,

JB(lAdj

u).:..L..uo in WILA(Q),

otherwise,

then clearly GI is of Caratheodory type and satisfies (7.4)"for some N.function rp.

dd

'

TEMAM
[1, p. 223])that for a subsequence {u)}
we have.
.

(7.9)

if (x,u,a)e.9'

= + 00

d f

(7.10)

Uj-+"0 a~mosteverywhere,

GI(x,u,a)=G(x,u,a)-b(x)

J(u)';' JG(x,u(x).
n

J7u(x), Adj J711(X),Det Vu (x) dx.

are bounded independently of r. The Poincare inequality for WJ LA(Q) (GOSSEZ

[1, p.202]) implies (cf KRASNOSEL'SKII
& RUTICKII'[I, p.131], EKELAND..&,

If we define GI : Q x (Bl" x E) -+ 91 by

u-v

'

'

nJ A(Wur(x)l)dx,

the con.

ditions (6.12),A-A* and B-A*, and a function beLl (Q),such that

We definetheadmissibility
set .cIby

377

J(u) is bounded below on '(/. Let Ur be a minimizing sequence from f(i. By (7.9)

'

JI gA (t) d t < 00 or

J GI (x, u(x),

ror

. the quantities

, (7.8)

et>

~b(x)+A(lFi)+B(lHI)+C(c5)

'r,'

(7.7)

00

G(x,u,F,c5)~b(x)+A(lFl)+B(c5) for all (x,u,F,c5)e.9';

n = 3: there exist N -functions A, B, C satisfying either

Elasticicy

,Proof. We just give the proof for 11= 3. the other cases being easier. It is sufficient

n=2: there exist N-functions A,B, with A satisfying either JgA(t)dt<oo or

.
I .
both (6.12)and A-A*, and a function beV(Q), such that

'1.

.,.

(7.11)

'

dudes as a special case the displacement boundary-value problem of nonlinear

hyperelasticity.

= I. of course, the re~ult is well known.

For 11

Theorem 7.3. Lt!t G sati.~f.'v(HI )-(H~) above. Let '(j be a Dirichlet class in .cI, and
SL/{'{'ost!that thert! exists UIe'6' with J(UI)< 00. Then there exists uoe'li' that minimi:es J(u) in 'If.

Let '6' he a Dirichh,t

class in ...;, allCI SlCf7{'OSt!that the;'e t!:dsIS U I e'r,' with 1( UI) < co.

TIrenthere exists uoe't>'that millimi:t!s J(I/)

in (6'.

Remark. Similar modifications to Theorem 7.3 can be made when n = 2, and

when n = 3 and G is independent of both Hand (5:the details are ten to the reader.

,.,

.,.

'1\

J.M. BAI.!.

37!1,

Next we give an existence t)1eorem for the displacement boundary-value

problem in three dimensions for an incompressible hyperelastic body.
Theorem

7.5. Theorem 7.4 remains ,valid if d

is redefined by d

= {UE WI

cQ with OQI having positive measure. Let u: OQI-+ fJ1l"be measurable and let
tJ!E/!(iJQ2) with 0'~1. Let G: Y'-+Pll satisfy hypotheses (Hd-(H4) and (H7). 1f

LA(Q):

n=2, let 1'>1, 0'= ~' if 1'<2, (J> 1 ify=2,(J= 1 1(1'>2, K2 =0, and

Adj VuELB(Q), (Vu (x), Adj

VU(X)}EW(x) '"
almost everywhere in Q, Det Vu(x) =; 1.
,

d = {liE WI.Y(Q): Det VUE~B(Q),(Vu(x), Det VU(X)}E

W(x) almost
everywhere in Q, u=u almost everywhere in oQt'.}.

almost everywhere in Q}, provided that if J gA(t)dt = 00, we make the extra assumptionB~A*.
I
.
Proof. Let {u,} be a minimizing sequence from C(j.Then {Det Vu,} is bounded in
L"(Q) independently of r. As in the proof of Theorem 7.3 we may extract a sub-

~rn';' 3, let Y>t, !+!

I'
,

sequence {Uj};,<t&'with, among other properties,

.De.t VUj-2L..o

in

L"'(Q).

d
.

of this section we restrict .our attention

to the cases n= 2, 3

'.

(7.15)

Prpof. We give the proof just for n=3.

By the trace theorems (cf MORREY[2], NEtAS [I)
exists k >0 such that

iiELa(oQI) and there

IIullwl.>(nl~k IIullL~'(of1)for all UEWI.Y(Q).

(7.16)

Since oQI has positive measure, a result of MORREY[2, p. 82] implies that there
exists kl >0 such that

fluIYdx~k1[JWuIYdx+( fluldSF]

(7.12)
Kt>O,

G(x, u, F, H, <5)~b(x)+ Kt(IFIY+ IHI~)+K2Iul'+ C(o)

for all (x, u, F, H, (5)E.9'.

K2~0,

Jo(II)~

1'> I,

for all ueWI.Y(Q).

,Jf),

(7.17)

f h(x)dx + Kt af 117111)'
dx + Kt f IAdj Vul"dx + f C(Det Vu)dx
a
n

fJ

JII(X)' tR(x)dS

,1a,

~ Jh(x)dx +
a

(7.13)

(~.!.-I:)J
a

/17/11>dx

+ (2~t 1 -I;) a luIYdx-~.!.

(H8) n = 3: there exist constants Kt >0, K2 ~O, I'> I, Jl> 1, s~ 1, and a function
bEJJ(Q) such that
.

(P'

--G(x, /I, F, H)~h(x)+ Kt OFP'+rHlJl)+ Kllul'

for all (x, u, F, H)EY'.

. ...

oa,

By (H7), (7.16) and (7.17) we have for arbitrary IIE.cY

G(x, u, F, 0) ~ b(x) + K IIFIY+ K 2Iu" + B(o)

for all (x, u, F, O)EY'.
there exists an N-function C, real constants
J.1.
> I. oS~ I. and a function bE JJ (Q) such that

;..

J II(X)' tR(x)dS,

. Jo(u)=J(u)'-

where the integral is defined in the sense aftrace.

Suppose that there exists ulEd with Jo(Ut)<oo. 71len there exists UoE.#
that minimizes Jo(u) in d.

(H7) n=2: there exists an N-function B, real constants Kt >0, K2~0, )'> I, s~ I,
and a function bE JJ(Q) such that
.

n=3:

={uE WI, Y(Q):Adj VuEl!'(Q),Det VIIELdQ), (Vu(x),Adj Vu(x),Det Vu(x)}

'

and we impose growth hypotheses that are partly of polynomia, type. This will
enable us to work in Sobolev spaces, rather than in Orlicz-So,bolev spaces. It
would be possible to extend most of our results to an Orlicz-Sobo~ev space setting.
Such an extension would involve the use of trace theory for,' Orlicz-Sobolev
spaces (see DONALDSON
& TRUDINGER[1], FOUGERES
[I], LACROtx[1]). .
For ease of reference we now restate hypotheses (Hs) and (H6) 'in modified
form. Later we shall put extra restrictions on the constants appearing in these
hypotheses.
.

<1. 0'= 2;' if I' <3, er> 1 if 1'= 3,0'= 1 ify>3, K2 =0, and

Let

Remark. A variety of 'weakly closed' constraints can be tre~ted in this way.

Analogous results hold for n=2.

J.1.

E W(x) almost everywhere in Q, u = u almost everywhere in CQI}'

Clearly o(x)= r almost everywhere.ByTheorem 6.2(ii),o=Det Vu,and the result

follows. 0

For the remainder

379

Existence Theorems in Non.Lincnt Elnsticity

)'

Kt

,a

J IAdj

Vul/dx

(,1al
JluldS)Y+i:kYllulli."'(MJJ

Y
I -)'.
IIuIIL~'(ml-- l)..lltRI11"(MI
Y{'

' I+

J,"C(DetVu)dx,

(7.14)

!'viixed displ(/cellle1Jttractioll hound;lrx-valur problems

ke<>t.< (.~"""'"
Theorem 7.6 (et: Section I A). Let Q}/sati4.i' a strong Lipschitz condition. Let
aQ=tiQtUuQ2,
let 7Q.(',iJQ2=1>, and let OQI and iJQ2 be measurable as subsets of

for 1:>0 and d>O. Choosing I: and cl small enough with d~k(I:I')I!Y we obtain
Jo(II)~C+(,o lIullh"(II1 + Kt S IAdj Vul/dx+ fC(Det Vu)dx,
{J

where
c and co>O are constants.
'-'-'-

In the sense or trace.

(7.18)

'f

,~.

380 . ,

oj\

1,

J.M. BAll.

Existence Theorems in Non-Linear

Let {ur} bea minimizingseq1Jence

for Jo' It follows from (7.18)that a sub-

Let
jo(u)=j(u)-C..

seq uence {u) satisfies

AdjVuj~AdjVllo

in E(Q),

DetVuJ---L.DetVu~
.

, '

in LdQ):
.

J given by (7.10) it follows that

(7.19)

J(uo)~ lim J(u),

J-<r>

while
~

Noting

Juo(;). tR(x)dS=
.

lim

~oo~

JIlj(X)'

(7.20)

tR(x)dS.

that Uo = ii almost everywhere in oQI we see, as in the proof of Theorem 7.3,

that UoE.Q{/.
The resultfollows. 0

where j is given by (7.11).

Suppose there exists uI Ed
minimizes jo(u) in d.

(7.21)

with jo(ul) < co. Then there exists uoEd

that

Pure traction boundary-value problems

Theorem 7.9 (cf Sect,ion 1, A 1). Let Q satisfy a strong Lipschitz condition. Let
tREIJ(iJQ), 0'~1. Let G:!/--+~ satisfy hypotheses (HI)-(H4) and (H7) with
K2>0. Let K=min(y,s),
4K'
.
n=2: Let 1'>3' 0'=2 if I~K<2, 0'>1 if K=2, u';'l if K>2. If s=1 let
IltRIIL~(om<koK2, where ko(QO is a certain constant, Let

d={UEWI'1(Q): Det VilELs(Q), (Vu(x), Det VU(X)EW(X)

almost everywhere in Q}.

Remark. In Theorem 7.6, and in the results below, the hypotheses on ii are
concealed in the assumption that d is non empty. .
I
Theorem 7.7. rfn=2let

Ju(x). tR(x)dS,

3D,

uJ~uo in WI'1(Q), urJ<~uo in'IJ(oQ),

uJ--+uo almost everywhere in Q and oQ,

For

381

Elasticity

y~2, 0'>1 (ft'=2, 0'=1 (h>2,

'

K2=0. {j"n=3Iet
y~2,
:

"

'
3
=0 .
3
2
3
-I +-I. ~ I.a= -21"
3 If ~}' < , 0'> I i/y = , 0'= I i/1'> , K 2

y p
..
Let the other hypotheses of Theorem 7.6 remain unchanged..7hen 71leorem 7.6
remains valid with Adj Vu, Det Vu replaced everywhere by adj Vu, det Vu respectively~
Proof. This is immediate from Lemma 6.1.

n=3:

.
1
1
Let y>!, -+-<-

2 K' ,

' 0'=-

,'

..

3 ifK<3, 0'> I if K=3, a= I ilK>3.

If s= I let IltRIIL~(3a)
< ko K 2' where ko(Q 0 is a certain constant. Let
d={UEWt'1(Q):

Adj VUEl!'(Q), Det VUELC<Q),

(Vu(x), Adj Vu(x), DetVu(x)E W(x) almost everywhere in Q}.

Let

Remark. In Theorem 7.7, and in those results below that concern the distributions det Vu, adj Vu it is only necessary for G to be defined on the set
{(x.u,a):xEQ,IlE~n,aECo{T(M3X3)(')W)}
(see Section 4 and the remark
after Example 6.1),
Next we give the analogue of Theorems 7.6 and 7.7 for incompressible materials.
The proof is similar to.that ofTheorem 7.5 and is omitted. An analogue of Theorem
7.4 may also be simply proved.
Theorem 7.8, Let 11=3. Let Q, ('QI' oQ2' ii, tR be as ill Theorem 7.6. In the definitions
of W(x) and!/ replace E by El' Let G: S" ~ satisfy hypotheses (HIHH4) and (Ha).
Either let }',tl. a, K2 be as in 71leorem7.6 alld let
.cI = {UE WI. )'(Q): Adj VUEl!'(Q), (l7u(x), Adj VIl(X))E W(x) almost everywhere
in Q. u = ii a/most e['erywhere in (iQI' Det Vu(x) = I almost eve/'ywhere
in Q}
.
or let ,',11, a, K 2 he clS in 711eorem 7.7 and let

.cV= {UE WI. 1(Q): adj VUEl!'(Q), (Vu(x), adj VU(.~)EW(x) almost everywhere
in Q, 11=ii cl/most everywhere in cQI' det Vu(x) 1 almost everywhere in Q},

Jo(U)=J(u)- S u(x).tR(x)dS.
oa

'(7.22)

Suppose that there exists UIEd such that JO(UI)< 00. Then there exists UoE.'"
that minimizes Jo(u) in d. If, in addition,
n=2: y~2

n=3: y~2, -+-~

y
Il

(7.23)

then the result. holds with Adj Vu, Det Vu replaced everywhere by adj Vu, det Vu
respectively.
Proof, Let n = 3. By using the hypothesis K 2> 0 instead of (7.17) we obtain the
a priori bound
Jo(u)~c+co IIVullb(al+K, SIAdj Vu IIIdx+ J C(Det Vu)dx+c, Slul" dx,
a
a
D
.
I
where Cl>0. If s= I then ko IS such that IIU"lVl.ltm~r-0 IlullL'(omfor all
UEWI.I(Q). By using the Poincare inequality (MORREY[2, p.82J) we can complete the proof in the same way as for Theorems 6.6, 6.7. 0

..
.'"

,r..

382

.y,

.ft

Theorem

!!.et 0 satisfy a strong Lipschitz condition. Let

(aO). Let G: fI' -+Bf satisfy hypotheses (Ht)-(H4)
and (H7) wirh K2 =0.

tRE~

7.10 (cf.Section

Mixed displacement
pressure.
boundary-value
problems

1,'A2).

If /1=2,
Clconstant
L'ector
and let 1'>1, 0'= ~ if 1'<2, fJ> 1 if 1'=2, 0'=1 if . 1'>2, e pe
'
.~={uEW1.Y(O):DetVuELB(n),.

(Vu(x), Det VU(X)}EW(x) almost everywhere

in 0, I u(x)

Theorem

1B).Let n = 3.

7.11 (cf. Section

Let aQ=aQlvl:,

.
dx

For these problems we restrict our attention to the case n=3.The

can be treated similarly.

aOlnl:=4>,

Let

satisfy

let y>~, ~+~<~,

0'=21'/ if 1'<3, 0'>1 if 1'=3, 0'=1 if "1>3, ebe a
I' j,( 3
3
.
constant vector, and

Lipschitz

condition.

with aQ, having positivemeasure as a subset of

and (H7). Let K2=0.

hypotheses (H1)-(H4)

'3
I
1
let "1>2' ~+-<JI

"I

and if~+~=1
I' p

4
3 ' If p(x)$constanl

If p(x)=constant

almost everywhere,

I
1
almost everywhere, let y>~, -+-<
"I

let Il17pIIL~(m~k3,where kJ=kJ(K.,

tain constant.Let

= {UE wl,y(O):

a strong

case n=2

eo #. Let jj:a Q I -+ Bf3 be measurable, and letpE W1. ex)

(0).Let G: fI'-+ Bf satisfy

=e}. .

ff'n=3,

d'

383

Existence. Theorems in Non-Linear Elasticity

J.M.BAll

K2,Y.JI,Q.aOI)
.

j,(

is a cer-

, ..

Adj VUEL:'(O), Det VUE Lc(Q),

.t1'={UEWI.Y(Q): Adj VUEL:'(Q),Det VueLc(Q),

(Vu(x), Adj Vu(x), Det Vu(x)e W(x) almost everywhere in Q,

(Vu(x). Adj l7u(x), Det VU(X)}EW(x) almost everywhere in Q, u(x) dx=e}.

.
a
Let PRElf(Q), let hoEBf"and let Jo be given by

=jj almost

everywhere

in tJQI}'

Let

JPR(X)

Jo(U)=J(U)-

ho

. u(x)

dx

- oa
J U . tR IS.

(7.24)

Suppose there exists Ul E,c1 with JO(UI)< 00. Then there exists uoEd' that minimizes

Jo(u) in ,<1.

where (c.f \.35)}

p(U)ddi,f

lj; in addition. we assume (7.23), then the result holds with Adj Vu, Det Vu
replaced everywhere by adj Vu, det Vu, resp,ectively.

Suppose thatthereexists

Proof. If UE.<1, then

minimizes JI (u) in d.
If in'addition

u(x)-~e

m(n)

dx=O.

I [p De~ Vu +tp,,(Adj

Poincare

y~2,
then the result
respectit'ely.

:..

IlulY dx~k3+k4Il17uIY

Ix

(7.25)

holds with

1
1
-+-~I,

Adj Vu, Det Vu replaced eperywhere by adj Vu, det J7u

JI(II)~(,+Co

Udx~ -

[-~. IIuIILY(D)+y.

dY' IIPR hollL'(n)

.

(7.26)

lIullwl,y,m+ ~I aJIAdj 17111"

dx+} aJ C(Det Vu)dx

Remark. We re-emphasise (e:f.Section I, A2 and Theorem 7.13)that for nonlinear elasticity, when G is independent of u. it is necessary to impose the extra
condition (1.3'1) in order to show that Uois in some sense a solution of the equilibrium equations.

(7.30)

for an UE,,",,where c. Co>0 are constants.

We use the Sobolev inequality

we again obtain the bound given for n = 3 by (7.18). The rest of the proof follows

the usual pattern.

(7.29)

J.l.

Proof. It suffices to establish the bound

for all-uE.cI.
Applying (7.25)and the simple estimate

- aJPRho'

(7.28)

inequality (MORREY[2, p. 83]) there are con-

stants k3,k4>0, such that'

a

Vu)~ Uk]Ix.

UIE.coIwith J1(UI)<(X)'Then there exists uoEd' that

"I

Thus by a version of the

(7.27)

JI (u)= J(u) + P(u),

lIuIIJI;"Y(m~k21111111..wl

for all U E WI. Y(Q),

where k2 >0 and v= 3)'/3 -"I if}' < 3, I < v < 00 if,' = 3. I' = <XJif}' > 3"
- .M

. In applicationswcshallhave1:= ,U., ,iQ, with

the no, as in Section lB.

(7.31 )

384

J.M. BALL

, iC".

'1

l4

.~.
Existence

For ued, we can use (7.17),(7.31) to obtain.

Jdu)?;

Jb(x)dx+

)aJWu'; dx+KdlAdj
'

.'

rul~ dx+ J C(D~d'u)dx

.n;.

k ~'

.'

i.

I
,

IIp,,IIL"'(m d~

11
(Adj
.

..

Vu)~llt"(m
.

I ~~I ;:;;;PI+ Cl [Iulf

If p is constant, then clearly (7.30) follows. If p is not constant and ..!..+..!..< 1,

.
. .
7 Jl
then y > )1' so that we obtain (7.30) by choosing &and d small enough. If p is not

constantand-.!..+-.!..=
1,thenY=Jl'and'(7.30)followssimilarly. 0

W(x)= {aeE: a>O}.,

,,
i
I

OG
. .'
of ;:;;;P2+C2[1ulf+IFIY+IHII'+loIY],

(7.38)

(7.39)

for all values of the arguments. If i' < 3 we assume further that!

(7.40)

_-Y .

= 337

Let f be given by
n=2: f(x,u,F)=G(x,/l,F,detF)
n=3: f(x,u, F)=G(x, u, F, adj F, det F)}

(7.41)

Two typical results are the following:

Theorem 6.1,2.In the hypotheses of Theorems 7.7, 7.9 replace (H7) by (H9). Suppose
also that (cl (7.23)

if n = 2: 7?; 2,

if n= 3: 7?; 2,
.

'; ,

..!..+..!..;:;;;1.
7 Jl

(7.42)

I,'

We replace(H7)by the followingstronger hypotheseson G:

(H9) n=2: G is continuously differentiable in u, F, a for all, xeD, and there

exist real constants KI >0, K2 ?;O, 81 >0. C,>O, PI?;O, 1<v;:;;;7,T?;O,s?; 1, and

a function bel!(Q) such that

:
,I.

G(x, u, F, a)?;b(x)+ KI IFIY

+ K2 lul' + 81 lalv,
,
.

(7.32)

;:;;;PI+CI[lulf+IFIY+laIV],

OG

of

;:;;;P2+

(7.37)

and (ii)when the material is incompressible.

.
,
We therefore consider the simpler situation when W(x)=E for all x, so that

+ IFIY+ IHI~+Ion,

1~~I;:;;;P4+~4[1UI;'
+1F17+IHI~-I+lol;'],

We now turn to the question of whether the minimizers whoseexistence we

have established satisfy the corresponding Euler-Lagrange equations; There is
at present no available regularity theory for our problems under acceptable
hypotheses, and we therefore confine our discussion to whether the minimizers
are weak solutions: Unfortunately t4ere are technical problems associated with
the two most important cases, namely (i) when the material is compressible and
.,
.
.
W(x) is given by

(7.36)

/;ZI;:;;;P34;C3[1uI7
+IFIY-I.+IHI7 +115'('7],

Solutions to the equilibrium equations

a;

7 Jl
.
The above proof is valid ifI>is taken to be zero, but the value of k3 so obtained
is then smaller. There is no difficulty in giving the analogous results to Theorem
7.11 for the pure pressure boundary-value problem and for incompressible
materials.

27

2 -)'

G(x,u, F, H, a)?;b(x)+KI(lFIY+IHI~)+ K2Iul'+81Ial',

where &>0, d>O.

aG

n=3: G is continuously differentiable in u,F,H,o for all xeD, and there

exist real cons~antsKI>O, K2?;O,81>0, CI>O,PI?;O,l<v~Jl~7, T?;O,s?;l.
and a function be1J(D)such that
,.

L,,'(a, '

thereisno localconstraint:

(7.35)

at;;:;;;P3+C3[1ulf+I,FIY-I+laIY'],

ll

11

Jl' d~

"

k. r-IJl

for all values of the arguments. If 7 < 2, we assume further that T=

+(2~1I -&)JluIYdx-~1
a
. (J
Ba, luldS)Y+ekUulltv(m
. ~lIpIlL"'(mJiDetVuldx-f

385

Elasticity

aG

','

~&

in Non-Linear
,

. '.

K
1
2

Theorems

C2[1ulf+IFIY+laIY],
'I-

(7.33)
(7.34)

Then the minimizing functions u ="0 satisfy the Euler-Lagrange equation

"

. ,'.

I.

afl Vl+ ",a~' v:.] d~= S lJ.tRdS

a [a u

vu,.

3a,

(7.43)

for alllJeCOo(ln) with 1J1311,

=0.
Proof. We give the proof for the case in which n = 3. Fix lJe COO(11t'3)with 1J13111
= O.
Since 1< v~ Jl;:;;;
7 it follows that Uo+ &lJe.w for any e. Also we clearly have

J (lIo+elJ).tRdS=
J
T e Ba,
Ba,lJ.tRdS.

(7.44)

;.('.
386

<t.

-A

J.M. BALL
387

Existence Theorems in Non-Linear Elaslicily

It thus suffices to show that

Proof. Proceeding as in Theorem 7.12 we obtain (7.50)for all t'E C'XJ(9f")satisfying

d.r

= nff(x,

/(11)

U(X), Ji'1I(X)dx

(7.45)

f vc!x=O.

(7.51)

n
is Gateaux dilTerentiable at uo, and that /'(uo)(v) is given by the left-hand side
of (7.43). By the dominated convergence theorem it is enough to establish the
estimate
.

Let WECXJ(9r") be arbitrary

and set

;:;- f w(x) dx.

vex)= w(x)(~ f(x, u(x)+ev(x), Ji'u(x)+eJi'v(x)I~e(x),

(7.46)

Then

for fixed ued with I(u)< 00, where eeLJ(Q) and is independent of ee(Q, I).
Carrying out the indicated dilTerentiation in (7.46), we find from (7.41) that

S[PRbo. w- ull..
~~

n.
.

Id~ f(x,u+ev,vu+r.vV)/
8G
OG
~c [/

a;; II+

OG

8G

8G

II

of + oH / Wul + / -g / Wul + 1-gladj

/

Vul~cP4ladj

Vul+cC4[1u(7

+Wulf-

.,, .

.I

~el(x),
for some e, eLI(Q), where we have used (7.36)and the facts that if 1';;;;3then
ueU'(Q). while if2~)'<3 then ueL.!(Q). 0
. ;
.
Theorem 7.13 (c/. Section 1, A2). In the hypotheses of Theorem 7.10 replace (H7)
by (H 9)' let (7.42) hold, and let f = iI/'(x,F) be jn~ependentof u so that

S iI"(x,

n
Suppose ill additioil that

Vu(x) dx - SPR(~')ho . u(x) dx - S U"R dS.

n..

~.

Existence under other hypotheses

cer:ainly has generalizations to integrands with x, u dependence.

~c[1 +lul.+Wul'+ladj Vul"+ldetVul'] .

Theorem 7.14. Let n =3. Let Uc;;;

M3x3 be sI/ch that Co T(U) is open. Let
Ul eWI"(Q),

adjVu 1 El.!'(Q), detVu,EL"(Q), where y~2, JI>I,.!..+.!..~I,

.
In the notation of Definition 4.3 let g: U -+f!I satisfy
sponding function G satisfy'
G(F, H, (5)~ b+ KI (1Ft' + tHI" + t<5!')

SPRhodx':"'O.
n

(7.48)

jJ '1ft"

for all ve C "',(:31'").

] dx+

fV"RdS=O

on

~.P at

v>1.

UI' Ll!t the corre,

for all (F, H, (5)eCo T( U),

(7.53)

G(F, H, <5)-+ 00

as'

(7.54)

(F, H, (5)-.. iJ(Co T( U))

Let

(7.49)

U-utEWd.1(Q), adj VUE!'(Q).

. det VUE.c(Q), I(u)ECoT(U)

Defilll!for Ue.~

Then the minimizillf{.li/llctiollU=Uo sati.~(iesthe Euler-Lagrange equations

f PRbo'v-~v~.01/..
{I [

I'

where band K I > 0 are constants, and

d={UEW"'(Q):
FRdS+
on

We next give a sample theorem under the hypothesis ~.~.. for the displacement boundary-value problem for a homogeneous material. This result almost

~c[ladj Vul+lut Y +WuIY+ladj Vu'''+ldet Vu I']

Jo(u)

Remarks. A similar result can be proved for the minimizer in Theorem 7.1 t.
We may also waive the assumption (7.42) at the expense of obtaining the EulerLagrange equations only in terms of the distributions Adj Vu, Det Vu and with
derivatives of G replacing the derivatives of f In Case (i) above an analogous
local result to Theorems 7.12, 7.13 may be proved under the a priori assumption
that det Ji'u~d>O locally. The details of these proofs are left to the reader.

,
+Iadj Vul"-'+Idet Vufil'JladjVul
!4-

dx+ Sw' tR dS
on
1
\.
.
=Sw(y)dy,[SPRboclx,+
StRdS]=O.
Q
(
)
m
n
n
on
Wi..

(7.47)

Vul

for fixed ued with I(u)<oo, and for all ee(O,I). Here and below c denotes a
genericconstant. The estimate (7.46)now followsfrom (7.37)-(7.40)and HOlder's
inequality.For brevity we display the calculation only for the last term of (7.47).
We have that
cl~~lladj

(7.52)

m(.,) n

I(u)=

(7.50)

almost el'erywhere ill Q}.

J~(I7u) dx.

(7.55)

Then if I(Ut) < 00, there exists UoE.<JIthat millimizes I(u) ill .<JI.
Proof. Let {u,} be a minimizing

sequence.

By (7.53), {I(u,)}

is bounded

in the

reflexive;:Banach space !!A= LJ(Q)9 X!1(Q)9x L:(Q). By the Poincare inequality

.}

'<!,.

'

,~

Cl

J.M. BAL!.

388

Existence Theorems in Non.Linear

and our now standard arguments there exists a subsequence tU) such tha.t
Uj~UO
withuoesl.

in

W1.Y(!1),

I(u)~'.r(uo)

iri :A:NIq,fJ4,

Define G1: E ~~'by

For compressible materials, in the case when the local constraint set W(x) has
the form
(8.2)
W(x)= El X K(x),

I.

G1(F,H,b)=G(F,H,b)
if (F,.H,b)eCoT(U)
= 00 otherwise.

.1

Let
J(a) =

389

Elasticity

J G1 (a(x))

'n

(7.56) .

dx.

Then J: <6191
.Jf'NI-+ gj. Since g satisfies Ij./A." at u. it follows that J is convex.
Applying Fatou's lemma to the integrand
G1(F,H,b)-b-K1<lFIY+IHI/A+lbl")

with K(x)q,~ nonempty, open and convex, it follows from Theorem 4.3 that
Co U(x)= W(x), where U(x)={FeM3x3:detFeK(x)}. Necessaryand sufficient
conditions for if/(x, .) to be polyconvex on U(x) are given by Theorem 4.4. For
simplicity we shall assume in the compressible case that K(x) = {b> O}(continuity
condition), while in the incompressible case W(x) = El (no local constraint).
.
We consider a modification of a class of stored-energy functions introduced
by OGDEN[2, 3]. For a ~ 1 let
.p(a)

.1

we see that J is (strongly) lower semicontinuous. Hence (cf. EKELAND& TEMAM

[1. p. 33]) J is sequentially weakly lower semicontinuous. Thus

v~

x(a)=(v2

+ V2+ v; - 3,

V3)+(V3 VI)+(VI V2)-3.

Consider'lirst incompressible materials, and let

M

if/(x, F)= B(x)+

.

i-I

(8.3)

'1'1

aj(x) .p(lXi)+

I-t

(8.4)

Ci(X) X({Ji)'

J{2'(uo) ~ lim J{2'(u j)'

j-oo

and the proof is complete.

The most general integrands for which our methods establish the existence
of minimizers are given by the sum of polycorivex functions satisfying a suitable
subset of hypotheses (HI)-(H9), quasiconvex functions satisfying the hypotheses
of Theorem 6.1, and, wh'ere appropriate, functions satisfying condition Ij,/A."as
in the above theorem. By suitably combining the growth conditions of each of
the terms in this sum various existence theorems may be given. At present both
the scarcity of examples of quasiconvex functions that are not polyconvex and
the abundance of physically useful polyconvex functions make these theorems of
Iitt1e interest. We therefore leave the routine formulation of the results to the
reader.
8. Applications to Specific Models or Elastic Materials
Many forms of the stored-energy function have been proposed for nonlinear
elastic materials, particularly for various rubbers. An excellent review of the
literature can be found in the papers OfOGDEN[2, 3]. We now examine the extent
to which these models satisfy the hypotheses of our existence theorems. By concentrating on a certain class of models below we do not mean to imply that other
models are inferior for empirical reasons. Neither should the omission of a model
from the discussion be construed as suggesting that it fails to satisfy our existence
hypotheses. Our purpose is simply to indicate the flexibility of the hypotheses to
serve for a variety of stored-energy functions and also to discuss the position.
occupied with respect to these hypotheses by certain well known models.
We assume that n = 3 unless otherwise stated, and consider for simplicity
only isotropic materials, for which the stored-energy function has the form
(see(1.15)
.
if/(x, F)=cf>(x, vI' V2,'V3)'

. (8.1)

where al ~... ~IXM~ I, PI ~... ?'PN?' I, and where 8, ai' Ciare functions in l!(Q)
satisfying
.
(8.5)
for almost all xeQ
Cj(x)?, k > 0,
Qj(x)?,k>O,
for some constant k.
By Theorem 5.2 it/'(x, F) is polyconvex on U(x). Since v~+ V2+ v; is a continuous function of F* it follows that
(8.6)
V~+ v2 + v) ?, d(lX) IFI
for some constant

d(lX) > O. Similarly

(V2 t'3)"+(V31'1)+(VI v2)"?,e(IX)\adj FI',

(8.7)

e(lX)> O.

It fonows from (8.4)-(8.7) that 'ir considered as a function of x, F, adj F satisfies

hypotheses (H1HH4), and (Ha) with
Y=!XI'

J/=/11'

Thus if
.-

3
IX!>2'

1 1
_+_<4
a1 P1

K2=0.

(8.8)

}.

(8.9)

we obtain from Theorems 7.8 and the analogues of Theorems 7.9-7.11 for incompressible materials the existence of minimizers for the various boundaryvalue problems in terms of the distributions Vu, Adj Vu. These minimizers satisfy
the incompressibility

condition

Det Vu

=1

almost

everywhere.

Note

that

in

order to obtain existence for variable pressures p in the mixed displacement

pressure problem we require (Theorem 7.11) either
I
I
(8.10)
--+--<
I
IX!

PI

.--..-.-. Because.for example.it is a finite-valuedconvex function.

,}

i,;'

.1:
,

390

"1
391
Existence

J. M. BALL

.,

I'-

or

P.

"

-+-=1

IX.

If

I'

(8.11)

IIJ7PIlV"'tm<kJ(a/,C/,IX/,PI' n, an.).

'

'IXI~2,

1 I
~I,
-+IXI

It is clear that a widevariety of stored-energyfunctionshaving the form (5.11)

(with x.dependence if necessary) can be treated by our theory in an way analogous
to that for the models discussed above. We end this section by exhibiting such
a stored-energy function, which satisfies the hypotheses of Theorem 7.3 but not
those of Theorem 7.6, etc. (for iJQ2= 4, and thus requires the Orlicz-Sobolev
,space apparatus. Our example is ofa stored-energy funCtion with slow growth. For
functions of very fast growth the Orlicz-Sobolev space setting would also be
necessary for any proof that the Euler-Lagrange equations are satisfied. t We
need two lemmas:

then we have the stronger forms of the existence theorems with the incompressibil-

ity condition det J7u=I holdingalmosteverywhere.

'

As a special case we consider the inhomogeneous Mooney-Ril,lin material,

for which B=O, M=N= I,IX.=PI =2, so that

'

. '1r(x,F)=al(x)(I.-3)+cl(x)(lI.-3).
,(8.13)
Clearly (8.12) is satisfied so that the Mooney-Rivlin material is included in the
existence theory. Note that in the mixed displacement pressure problem the
critical case (8.11)applies. '
'
,
,
The incompressible Neo-Hookean material

with at satisfying

(8.5). To prove existence

by our methods

under

.. "

Lemma 8,1. Let C, D be N-functions. Then

--0
D-I(s)
(:"-1 (s)

'1r(x,F)=IXI(x)(l.-3)
,'(8.14)
is not covered by the theorems. To illustrate this point let us consider a single
term stored-energy function
'#"(x,F)=GI(X) !/I(IX)
(8.15)

Elasticity

the conditions (8.9)-(8.12).

(8.12)

pI

in Non-Linear

Then the modifiedstored-energyfJnction satisfies(HI)-(1/4) and(117)with y =lXIJi= PI' K 2=0. and thus satisfiesthe hypotheses of our existencetheorems under

Theorems

(8.18)

as s -+ co

if and only if D~C.

Proof. We just prove the' only if' part, the 'if' part being easier. Set s =C (),t)
for ),=0. Then by (~.18) ),t/D-I(C(),t)}-+O as t-+oo. Hence t~D-I(C(At)}
large enough. By the convexity of D we have for t large enough

the incom-

pressibility constraint Det Vu= I (det J7u= I) it is necessary that if u, -" u in

WI.Z(Q), then Det Vu, -" Det Vu in ~'(Q) (det J7u,-->detVu in ~'(Q)). We must
thus have ('I the methods of Section 6) IX>(IX~ 3). Note, however, that we do
get existence theorems for the Neo-Hookean material in two dimensions.
OGDENfitted a stored-energy function with three terms (M = 2, N = 1)of the
form (8.4) tq data of TRELOARfor homogeneous vulcanized rubbert. The values
of the various constants obtained were

D(t)
t
C(At);;;!D-I(C(),t))

I C-I(C(),t))
D-I(C(),t)} -->0

for t

as t-+ 00. 0

'

..,emma 8.2. Let g, k be non-negative continuous jimctions on [Jli.. such that

k(s)-->O as s-+ 00 and such that
00

Jg(t)dt=oo.

'

IXI

= 5.0,

al=2.4x
B=Q

'1X2

= 1.3, PI=2,

IO-J,

'a2=4.8,

Cl=;,0.05kgcm-2,

Jk(t)g(t)dt

i.

,,'

Clearly (8.12) is satisfied. Furthermore, since IXI> 3 the minimizers Uo of the

various boundary-value problems belong to C(Q); indeed by the results of
MORREY[2] (see also FRtEDMAN[I] and NEAS[I]) we have in this case "oE
CO.*(Q), the space of Holder continuous functions on Q with exponent
Note
that the imbedding theorems do not imply that "oE C(Q) for the Mooney-Rivlin

material.

Let

(8.16)

'

'

J~(t)dt

0
.'

Then (}(.~)-+0 as s -+ 00.

Proof. This followsimmediatelyfrom L'Hopital's rule. 0

For compressible materials OGDEN [3] considered the eITect of adding a

term f(det F) to (8.4) #. Suppose that
f(t)~ C(t)
for all 1>0.
where C is an'N-function.
(8.17)
risconvexon(O.oo),
f(t)-->oo
as t-+O+.
}
--- n. ...
'

. For related experimental work see TRELOAR[I].

OOI)ENretained the values oflhe conslllnts B.IX,.1X2:1/,.1/2' "" but for incompressible materials
replaced the lerm (', ;((2) by", [", 2+ ('2' 2+ ",.2 - 3]. These terms are identical if 1'1.1'2DJ..I, and
since 1',1'1'" is in practice very close to I the alteration is insignificant for experimental correlations.
,

O(.~)

Now let A be an N-function with principal part

A(t)=ti logt.

(8.19)

Let
---"-.Y"(F)=aA(vl
+V2 +vJ)+CA(V2VJ + (lJI'1 + 1'1(12)+ f(1'1 V21'J)
(8.20)
,t In order 10givea comprehensivedescription or ~lRisotropyund inhomogeneityin rod theories
ANn-tAN[H] uses more general Orlic7.-Sobolevspuces than are IIsed in this work. The methods described here probably extend to Ihese spaces. with a consequenl brQadcningin Ihe applicabilityof
the existenceIheorems presented here.

. ,

. ~.

1.-\
t

()

J.M. BALL

392

wherer isasin (8.17)anda,c arepositiveconstants.SinceA*ti+c

where
dor
k(I)= ..

. d' ->0

(8.21)

"

where ii: Q -> ~3 is given by

00

- 'XI

ii(x)=().

'if"(F) = a(!.

- 3)+ crI111-3),

K2=0,

W(x)=Et=M3x3XM3x3,

(9.6)

00

d=(UEWI.2(Q):

A*-I.(S)! k(t) gB(t) dl

B*-I(S)

adjVuEL2(Q), u=ii almost everywherein OQl'

det Vu = 1 almost everywhere in Q},

fj

(8.22)

f gB(t) dl

By Lemmas 8.1 and 8.2 we deduce B*~A*

Vii =diag(..1.

as required.

- t,..1.- t, A),

In this section we establish nonuniquenessfor the mixed displacementzero

traction boundary-value problem corresponding to buckling ora rod of a homogeneous,incompressible,Mooney-Rivlin material having a unifornl cross-section.
We do this by exhibiting an admissibledisplacementfield with total energylower
than that of the trivial solution, and then applying Theorem 7.8 to ensure the
existenceof a nontrivial minimizer for the total energy.Under suitable conditions
a similar but more complex analysis can be carried out for incompressiblerods
consisting of material not of Mooney-Rivlin type. The extensionto compressible
materials,however,isnot soeasybecausean explicit trivial solution is not available.
In the stress-freereferenceconfiguration the rod occupiesthe region
/>0,

(9.1)

.
.
where the cross-section D is a nonempty bounded open set in ~2 satisfying a
strong Lipschitz condition. We suppose that the density in the reference configuration isa constant fiR > 0, and that the plane X2=Ocontains the line ofcentroids
of the rod in the reference configuration. so that
.

Jx2dS=0,
.

dS=dxl

iJQ2=iJDx(O,/).

we consider the equilibrium

dX2'

(9.2)

Let A=SdS be the area of D. For IbO

D
.
mixed boundary-value problem with boundary

U= U
tR=O

on

j)Q2'

(9.8)

(9.3) .

b(2..1. +..1.-2 -3)J

A I< <Xi.

(9.9)

It is easily shown that ii satisfies the equilibrium equations and boundary conditions for a suitable hydrostatic pressure. By Theorem 7.8 (with UI=.ii) there
existsuoEd
that minimizes j(u) in d.
It thus remains to construct a function UEd with.
(9.10)

j(u)<J(ii).

We perform this construction in a manner reminiscent of derivations of rod

theories in engineering. Let Yo, 00 be real valued functions with YoE C2([0, /]),
80EC2([O,/]) and
.
Yo(O)= yo(1)=0,

.00(0)= o(l)= O~(l)=.f}~(l)=0.

(9.11)

Let 8>0, y=r.Yo, O=r.o, and define

u.(X)=(XI g(X2' X3)' y(X3)+..1.-, X2 COSO(X3)'A.x3-). -, X2 sin O(X3). (9.12)
where g is a function to be chosen. u, represents a deformation in which points
(0,0, X3) are mapped to (0, Y(X3)' ..1.X3)'and in which cross-sections normal to the
x3-axis in the reference configuration stay plane and are so inclined that their
normals remain parallel to the X2X3 plane and make an angle 0(X3) with the
x3-axis (Fig. 3).
From (9.12)we obtain

~rI
on (..1'

det Vii= I,

and

of Nonuniqueness; Buckling of a Rod

Q=D x (0,I),

(9.7)

From (9.4) we obtain

J(ii) = [a(2..1. -2 +..1.2 -3)+

9. An Example

.. .

j(u)= f'iY{Vu(x)dx.

conditions

(9.5)

where a>O, c>O are constants. In the notation of Theorem 7.8 we set

as t->oo.

Also gB(I)=-r, and g;.(t)dl=

gB(t)dt=<Xi. Therefore
I
0
0
.

Let DQ1=Dx(0./}.

(9.4)

,..1.-, X2' ..1.X3)'

The stored-energy function has the form (cf. 8.11)

Y=J.l=2,
. 1

393

Existence Theorems in Non-Linear Elasticity

for any 8>0

Theorem 7.6 does not apply. To show that (Hs) is satisfied, so that Theorem 7.3 .
may be applied, we must prove that A~A*. Let B(t)=ti. Then B-<.A so that
A -<.8'" t3'" B*. It is therefore sufficient to prove that .B*~A*. We have that
gA(t)= A~;(t) = kt~) ,

~~

Xl g.2

.._------

pu,~G

xlg.3

).- t cos ()
y' - X. t x2 sin (). ()' .
-..1.- i sinO A-A.'! X2cos(}. ()')

(9.13)

. Nole Ihal we do nol have 10 satisfy the zero traction condition on ,1(12because our existence
theorem in:orporates this as a natural boundary condition.

"" ,
~'

'rf\

394

\,)

,'~I

J.M, BAll

Existence

-'

1t.2

1<:2

395 '"''

Ela~ticity

do=A).-2(a).+c),
dl =a[(3k2 +klP -4+k2A -I] +c[(kl +k2)'- 2+3k2)' -S],
d2=k3).-5(a).+c),
d3=2A).-3(a).+c),
d4=A).-I(a).+c),

/~~~
I

le.

(9.21)

and

x'3

kl = f xl dS,
(i) Reference configuration

(ii) Deformed configuration

Fig. 3

Therefore

The Euler-Lagrange equations corresponding to tpe quadratic part of (9.20) are

'

'

g= [).i cos O+J.- 1y' sin (}-). -I X20']-I,

(9.16)

for all xe!2.

(9.17)

B: =). -I +2d - t X20~+,,2 [A-I O~-2), - 20oY~+(3x~+xf).-40~2+). -4 xf x~0~2]

+0(62)"

(9.24)

(9.25)

0~"-(X(O~+(X200=0,
where
dl
(XI=-d '

(X2=

(9.23)

where y is a constant. Setting y=O and combining (9.23), (9.24) we obtain

It follows from (9.11)-(9.16) that u.e.s:1 for" small enough. Let

B=Vu.y~.
.
A routine but tedious calculation shows that

d2 o~"-dl ()~+doOo- ~3Yo=0,

.. (9.15)

whence

'f'?

(9.14)

Since ),>0 the expression in brackets in (9.14) is positive for all xe!2 ife is small
enough. Thus we may choose

2d4Yo-d3Oo=Y'

det Vu,=g[A.t cos 0+), - 1y' sin 0-), -I X20'].

(9.22)

k3= f x~x~dS.

k2 = f x~ dS,
D

det Vu.= 1

in Non-Linear

where

Theorems

- A().3- I) .

do-d;/4d4
d2

k3

(9.26)

With the above as motivation we seek a solution, antisymmetric about X3:': ~ '
to the equation
(9.27)
O~" -(XI ()~ +a200 =0
subject to boundary conditions.

' ,

00=00=0

B~ =). -1+,,2 CY02- ).-1 OJ]+0(82),

B~ =).2 -2df X2O~+,,2 [).-I O~+),-I X~0~2]+0(62),
B1 = Bf =d -hi 00 +0(8},
B~= B: =d -I XIx2o~+0(,,),
B~ = B~=e[Ayo -). - (00] +0(6),

(9.18)

for some a2 > (X2'If 00 is such a solution and

dcr d3 "
YO(X3)

(9.28) .

atx3=0,1,
.

'to.

"

= U 4 0f

o(s)ds

(9.29),

then Yo satisfies (9.11) and we have from (9.20), (9.26)-(9.29) that

and hence that

III= 2A- (+).2 + 2CX20~(J..

-1- ).f)+ e2[).-I ()~+((3x~ +xf). - 4+). - Ixn002
+X2X2).-4(}"2_2).-2 y ' +y '2 ] +0 (62)
I 2

liB

0'

..

A-

J(u.)=J(U)+I:

(9 19)

= ). - 2 + 2), + 2eX200().- J -A. - 1)+ ,,2[).- 20~+(x~ + xf). - 2+ 3x~~ - S}002 .

+xf x~).- S(}~2- 2),- 300YO+).-I y;?]+ 0(82),
,

d2((X2 -(X2)

f () ()dX3
0

2
+0(1:

(9.30)

).

It follows that for f. small enough J(u,)<J(ii) as required.

To solve (9.27), (9.28) we first note that for all ).e[O, I] we have

"

where 0(8) denotes a function of x such that 0 (e)/8 -> 0 uniformly

From (9.2), (9.5), (9.7) we 'obtain (with the standard meaning
I

J (u.) = J(ii)

+ C2 f
0

in Q as e -> O.
for: 0(e2)}

'>(XI-

3ck
2
0
k 3(a+c ) >,

Set

'

[do O~ +d( (}02+ d 2 (}~2 -d3

oYo +d4Y02]

dX3 + a(,,2),

(9.20)

(X2<
=0.

a2=(X2+'t,

0<'t<1

3ck2
k3(a+c)'

(9.31)
2

).

(9.32)

',,'.

396
.. .

...,

\-/1

I. ,. ,. . .,'"

J.M. BALL

..;I

(#

397

Exislence Theorems in Non-Lineni' Elasticity

I
I
I
1
I
I
I
I
I.

\;

2~

It is.not hard

to show that K and Jl are continuous functions of), in [0, I], and
that K is not constant in [0, 1]. Therefore there exists an interval in the range of

K: [0, 1] -+ 9i'+ with length ;b O.If KO* KIthen we may take

(9.38)

15=IKo-Kd + r(f),

where r(f) -+ 0 as t -+ O.
From Figure 4 it is thus clear that if
151
T>21t,

Tt

"~I
I
I
I
I.
I
I
I
.I

(9.39)

then there exists 0 <). < 1 such that (9.36) is satisfied. The corresponding 00 is
the function required~ To satisfy (9.39) one need only choose I large enough.,
Thus sufficiently long rodsof arbitrary cross-sectionwill exhibit 'n~nuniqueness
for
some ).e(O,1).(An obvious refinement of this argument shows that if 0<...1.0<1
then for [large enough there will be nonuniqueness for some). with ...1.0
<A.< 1.)
..' Usually Ko=FKtso that by choosing f>O small enough we get nonuniqueness
for some ).e(O, 1) whenever
.
41t

(9.40)

IKo-Kd>[.

The indicial equation for (9.27)is

(9.33)

m4 -IXI m2+&2 =0

..This is a condition expressed entirely in terms of I and the cross-sectional parameters kl,k2,k3 and A.
Example. Let D be the disc x~ + x~ < a2. Then

with roots m = :!::K, :!::iJl, where.

,,2

f1.

~-:-4&2-IXI

'

. ..

0(X3)= Alsinh K (X3

-+) +A2 ~inJl (X3 -+)

willsatisfy(9.27),(9.28)provided that
2
f1.1 2
-tan-=-tanh-.
Jll
2 "I

3k

"0=

(2/

9k

4k~+k
3
3..

"I=

~J

so that

4k +k

'

(9.42)

V,)V.

v~0.09.

(9.43)

(9.36)

at ,1,=0, I, respectively,

A 1

=-.136
a

Condition (9.40) therefore becomes

21

KI

(9.41)

(9.35)

Also 00 given by (9.35)is antisymmetricabout ,'1(3

=: ~ '. The grap~s of the functions ,=~ tan '1, ,=.!. tanh '1 are sketched in Fig.4. It is easy to prove the
'1
'1
IX +VIX2-41X

. l'
Ko=-a V9+.1f65
~IV,""

!:.<v,
[

,d
. 2.

indicated monotonicity properties.

So that

A =1ta2

k3=24""'

k1=k2=T'

(9.34)

and ">f1.>0. (K,J.lare real by (9.31),(9.32).)Hence

Let Ko, "I be the values of

1ta6

1ta4

IXI+~-4&2

).

(9.37)

The condition (9.40) is somewhat crude; indeed it is possible that no such condition is necessary. Improved estimates, and lower bounds on the sup remum of
0 <). < 1 for which nonuniqueness occurs, may be obtained by more detailed
calculations based on Figure 4. I have not included these results since they are
messy and since my method is severely limited in scope due to the type of trial
deformation considered in (9.12). Even in 'situations where we envisage nonuniqueness occurring by Euler buckling. the deformation of cross-sections implied

by (9.12)is unrealistic.

There are numerous formal stability calcult\tions in the literature for rods in
tension or compression, and for other problems in three dimensional non linear
elasticity. For the most part these calculations are based on the theory of small
deformations superposed upon large: the status of this theory with respect to

v.l
~

'78.-.

l)

~.)

.J,M.B...ll

.)

Existe?ce

noniinear elasticity has yet to be established. The reader is referred for details
and references to ,FOSDICK& SHIELD[I]. HOLDENtl], KNOPS& WILKES[I].
SENSENIG
[1]. WESOLOWSKI
[I], WILKES[1]. Nonuniqueness for the pure traction
boundary-value problem of a rectangular block of Neo-Hookean material loaded
uniformly on each face has been established explicitly by RIVLlN[I. 2. 3]. For
various one and two-dimensional 'rod and shell theories rigorous proofs of nonuniqueness have been given by ANTMAN[2, 3, 6].
. .
,

10. Concluding

Remarks

in Non.Linear

Elasticity

399.

it is violat'ed for nearly incompressible materials such as rubber (see HILL [2].
. OGDEN[I, 3]. RIVLlN[2]. SIDOROI'F[1]).
'
.

. For

bodies that are not homeomorphic

to an open ball the various minimizers

whose existence we have established may represent deformations topologically

isolated from those desired on physical grounds t. One might require. for example"
.

all admissible deformations to be accessible from a given deformation by a

homotopy of globatIy invertible configurations. In this article we have not studied
,such global constraints (although they may be of a weakly closed type), but have
concentrated on local constraints such as the local invertibility condition
det Vu>O. Local invertibility is a relatively weak requirement; indeed a hollow
sphere may be everted without violation of the condition in any intermediate

'.

The main implication of this work for constitutive inequalities is that the
quasiconvexity'condition (and in particular the Legendre.Hadamard condition)
is consistent with realistic models of hyperelastic solids. In the 'one-dimensional
case. when convexityand quasiconvexity of "If"(x.') are the same. Theorem3.2.'
shows that the existenceof Ct (.0)minimizers for various homogeneousdisplacement boundary-value problems implies that'll/' is quasiconvex.while the same
result holds in three dimensions if Q is a cube. If "If"is not quasiconvex then
minimizers may exist that are not Ct. Some examples in one dimension are
discussed by ERICKSEN[3]. It should be noted that we have not proved that
Cl (.Q) minimizers exist in general for dispJacement boundary-value problems"
when oQ is suitably regular under any reasonable hypotheses on "If".
.
The existence theorems proved in this article take the form that. existence 'is
established for a given material for qll suitable boundary data. In general such
unqualified existence is not to be expected for real materials. since Irupture will
occur under extreme conditions. of deformation. We may also not be interested
in solutions having at some points deformation gradients that lie: outside the
range in which the material behaves elastically. One way of partiially circumventing these difficulties is to choose the local constraint set W(x) ~o as to prohibit such behaviour. and then' to check a posteriori whether the rrinimizer Uo
is such that Vuo(x)eo W(x) for any x. One would then like a priori conditions
on the size of the boundary data to prevent this happening. The derivation of
any such conditions would require delioate estimates. The reader is referred to
the papers by ERICKSEN
[4] and KNOWLES
& STERNBERG
[1] for further discussion
of some of these points.
,
.
In genera,l weak lower semicontinuity will not hold if the quasiconvexity or
polyconvexity hypotheses are replaced by a hypothesis of convexity of the function '1/' restricted to positive definite symmetric tensors U. It is nevertheless
instructive to see how an attempt to establish lower semicontinuity In this case
breaks down. The difficultyis that if u.~u in the Sobolevspace Wt. Y(Q),then
th~ weak limit in LI(Q) of the sequence U.=yVu; Vu.. will not necessarily be
VVUT VU. and indeed may not arise from any displacement. This is because
yVu'f Vu is not of the form (4.2) and hence not sequentially weakly continuous.
The difficulty is also connected with the nonlinearity of the Riemann-ChristoITel
tensor based on C. Similar mathematical problems arise from attempts to estab.
lish existence under the COLEMAN& NOLLcondition [I]. or HILL'Sinequalities
[2. 3]. These conditions do not imply the Legendre-Hadamard condition. The
. COLEMAN
& NOLLcondition cannot apply to all hyperelastic materials because

Theorems

deformation

(SMALE[I])."

". Finally I remark on the implications of the results of S~ctjon 6 for' theories
of elasticity incorporating pointwise constraints on the deformation gradient F.
These resuItssuggest strongly t!1'atthe only non trivial homogeneous constraints

..
.
~

giving rise to a well posed theory ha~e the form (see (4.2)
4>(F)=A + BjF; + Cf(adjF)~+Ddet F=O.

(to.1)

where A. Bj, q. D are constants.' It is not hard to show that the only objective
constraints of this form (i.e.. cPsatisfying cP(QF)=cP(F) for all orthogonal Q) are
. those with Bj= q=O, so that detF is specified. In particular, as we have seen,
the incompressibilitycondition detF = 1 gives rise to a well posed theory. Note,
h9wever. that the constraint of inextensibility (TRuESDELL& NOLL [I. p.72J) is
not in~luded. It seems possible. therefore, that solutions do not in general exist for
boundary-value problems of inextensible elasticity, and that a higher order
.

Acknowledgement

I
"

l-

theory is required ,to make such constraints well behaved.

"

The research reported here was begun in the summer of 1974 when I held part of a Science
Research Council research fellowship at the Lefschetz Center for Dynamical Systems. Brown University. I am greatly indebted to CONSTANTINEDAFERMOS,without whose enthusiastic interest
and consistently excellent advice this article would not have been wriUen. I would also like to thank
STUARTANTMANfor several useful suggestions. and for his careful criticism of the manuscript. Finally.
I am glad to acknowledge helpful discussions at various stages of the project with ZVI ARTSTEIN.
KENBROWN.DAVIDEDMUNDs.J.L. ERICKSEN.ROBINKNops'and A.C. PIPKIN.'

References
A. R. AMIR-MoEZ

A. R. AMIR-MoEZ

[I)

Extreme properties .of eigenvalues of a Hermilian transformation

.and singular values of sum and product of linear transformat.ions.
Duke Math. J.. 23 (1956). 463-476. .
[I) Singular values of a matrix. Amer. Math. Monthly (1958). 742-748.

& A. HORN
S.S.

ANTMAN

u"-,,

- ,,'0

[I)

E4uilibrilim stales of nonlinearly elllstic rods. J. Mllth. Anal. Appl.

23 (I 96R).459-470.

t For bodies th!\! arc homeomorphic to an opcn balllhc samc may occur for mixed problems: the
situation for displacement bounuary-value problems is unclcar.'
A visualization of the eversion due 10 SIIAPIROcan be found in I'll III,II'S[I).

..,

....

");:}
,~' 4':X'

$.M. BALL

. ,>,/\
Existence

t'
[2] Existence QCsolutionsoCthe equilibrium equations for nonlinearly
elastic rings and arches, Indiana Univ. Math. J..20 (1970) 281-302.
[3] Existence and nonuniqueness ofaxisymmetric equilibrium states of
nonlinearly elastic shells. Arch. Rational Mech. Anal. 40 (1971),'

329-372.

"

[4] "The theory of rods", .in' Handbuch

der Physik, VoIVIa/2,

C. TkUESDELL,
Springer,Berlin,1972.

'

ed.

"

. . .

[5] Monotonidty and invertibility conditions in one-dimensional non.

. linear elasticity, in "Nonlinear Elasticily". ed. R. W. DICKEY,Aca-

demicPress,NewYork.1973.
[6] Nonuniqueness

. '

Anal.Appl.44(1973),333-349.

Soc.Edin(A)72(1973/4).275-280.

I. BEJU

L. CESARI

[I]

[3]
[I]

B. D. COlEMAN

[I]
& W. NOll
T. K. DONAlDSON
[I]
& N.S. TRUDINGER
P. DUHEM
[I]

Scl. Lincei Rend.. Cl: Sci. Fis. Mat. Na!. Ser 8, J (1946),

'

'.

Closure theorems for orientor fields and weak convergence, Arch.

Rational Mech. Anal. 55 (1974), 332-356.
"
L,
.
Lower semicontinuity and lower closure theorems Without seminor. .
mality conditions. Annali Mat. Pura. Appl. 98 (1974), 381-397.
A necessary and sufficient condition for lower sem/continuity.Bull.
i
Amer. Math. Soc. 80 (1974),467-472.,
Ober die zweite Variation vielfacher Integrale, l Reine Angew.
Math., 56 (1859),122-149.
i,
On the thermostatics of continuous media, Arch. :Rationai Mech.
Anal.. 4 (1959), 97-128.,
.
Orlicz.Sobolev spaces and imbedding theorems. j. Funct. Anal.,
8 (1971), 52-75.
'.
Recherches sur I'elasticite. troisieme partie. La stabilite des milieux
eJastiques, Ann. Ecole Norm., 22 (1905), 143-217. Reprinted Paris.

Gauthier-Villars 1906.
N. DIINFORD
& J. T. SnlWARTZ
D.G.B. Elm.EN

[2] "Non local variations and local invariance of fields", Modern

analytic and computatiomil methods in science and engineering
No. 19, Elsevier, New York, 1969.
.
I. EKELAND & R. TEMAM
[I] "Analyse convexe et probl~mes variationnels", Dunod, GauthierVillars, Paris, 1974.
.
J. L. ERICKSEN
(I] Nilpotent energies in liquid crystal theory, Arch. Rational Mech.
Anal., 10 (1962), 189-196.
.
.
(2] Loading devices and stability of equilibrium, in "Nonlinear Elasticity", ed. R. W. DICKEY,Academic Press, New York 1973.
[3] Equilibrium of bars, J. of Elasticity, 5 (1975~ 191-20/.
[4] Special topics in elastostatics. to appear.
:G. FICHERA
[I] Existence theorems in elasticity, in Handbuch dcr Ph'ysik, ed. C.
TRUESDEll, Vol. Vla/2, Springer, Berlin,.1972.
R. L. FOSDICK
[1] Smalt bending of a circular bar superposed on finite extension or
& R. T. SHIELD
compression, Arch. Rational Mecti. Anal.. 12 (1963), 223-248.
.A. FOUGERES
[I] Thesis, Besan~on, 1972.. ,
A. FRIEDMAN
[I] "Partial differential equations", Holt Rinehart and Winston, New
York, 1969.
.
,
.
J.P. GOSSEZ
[I] Nonlinear elliptic boundary yaluc problems for equalions with
rapidly ,(or slowly) increasing coefficients, Trans. Amer. Math. Soc.,
190 (1974), 163-204.
.
'
L.M.GRAVES
[I] The Weierstrass condition for multiplc integral variation problems,
Duke Math. J., 5 (1939), 656-660.
.
A.E.GREEN& W.ZERNA . [I] "Theoretical elasticity", 2nd edition, Oxford Univ. Press, 1968. .
J. HADAMAkD
Sur une question de calcul des variations, Bull. Soc. Math. France,
., [I] 30
(1902), 253-256.
.
,
. [2] "Le~ons sur la propagation des ondes", Paris, Hermann, 1903.
[3] Sur quelques questions de calcul des variations, Bull. Soc. Math de
France, 33 (1905), 73-80.
'
R.HILL
[I] On uniqueness and stability in the theory of finite elastic strain, J.
Mech. Phys. Solids 5 (1957). 229-241.
.f
[2] On constitutive inequalities for simple materials. I, J. Mech. Phys.'
Solids, 16 (1968), 229-242.
'
l ~
[3] Constitutive, inequalities for isotropic elastic solids under finite
strain. Proc. Roy..Soc. London A3J4 (1970). 457-472
J.T.HoLDEN
[I] Estimation of critical loads in elastic stability theory, Arch. Rational
. Mech. Anal., 17 (1964),171-183.
R.J.KNOPs'&
[I] "Theory of elastic stability", in Handbuch der Physik, Vol: Vla/3,
E. W. WllKES
cd. C. TRUESDEll,Springer, Berlin, 1973.
J.K.KNOWlES&
[I] On the ellipticity of the equations of nonlinear elastostatics for a
.
E.STERN8ERG
special material. J. of Elasticity. 5 (1975). 341-361.
'M'-A. KRASNOSEl'SKII
[I] "Convex functions and O"rlicz spaces", trans. L. F. BORON,NordhoIT.
& Y A. B. RurlcKII
Groningen, 1961.
I
M.T.LAcRolx
[I] Espaces de traces des espaces de Sobolev-Orlicz. J. de Math. Pure!
Appl., 53 (1974), 439-458.
E.J.MCSHANE
[I] On the necessary condition of Weierstrass in the multiple integral
problem of the calculus of variations, Annals of Math. Series 2, 32

I:'
f
I'
I.

(1931),578-590.

"Linear operators",

[I]

The null set oC the Euler-Lagrange operator, Arch. Rational Mech.

Anal.. 11 (1962), 1t7-12/.

[I]

L. MIRSKY

[I] On the trace of matrix products, Math. Nach. 20 (1959), 171-174.

[2] A trace inequality of John von Neumann, Monat. fUr Math., 79
(1975).303-306.

Pt. I., Interscience, New York 1958.

J.J.

MOREAU

N.G.MEYERS

'

[t]

401

Elasticity

Convexity on nonconvex sets. Proc. Col/. on Convexity, Copen.

hagcn. Univ. Math. Inst., Copenhagen, (1965),20-33.
:
Su un teoremafondamentale nella teoria delle onde di discontinuitli,
Aui. Accad.
66-72.

[I]
[2]

A. ClEBSCH

in Non.Lrii~a;

'

[2] On the calculus of variations and sequentially weakly continuous

maps. Proc. Dundee Conference on' Ordinary and Partial DiITerential .
Equations 1976. Springer Lecture Notes in Mathematics, to appear. .
[I] Stability of hypere1astic bodies subject to hydrostatic loading, Nonlinear Mech. 5 (1970). 367-383..
. '
[I] Theorems on existence. uniqueness and stability of the solution of
the place boundary.value problem, in statics, for hyperelastic tnate. rials, Arch. Rational Mech. Anal., 42 (1971),1-23. .
.'
[2] The place boundary.value problem in hypei'elastostatics,' I. Dif. .
ferential properties of the operator of finite elastostatics, Bull. Math.
Soc. Sci. Math. R.S.Roumanie
16 (1972). 132-149, 11. Existence,
uniqueness and stability of the solution, Ibid. 283-313.
.
[I] Convex bodies and convexity on Grassman cones, Parts I-IV, Math.
Ann., 151 (1963), 1-41..
,".,'
.

H. BUSEMANN. G. EWALD
& G.c. SIIEPHARD
H. BUSEMANN
. [I]
& G.c. SHEPHARD
C. CATTANEO

'.'

[7] Ordinary diITerential equations of non linear elasticity I: Foundations

, of the theories of non linearly elastic rods and shells. Arch: Rational
Mech. Anal. 61 (i9761. 307-351.
""
[8] Ordinary diITerential equations of nonlinear elasticity 11: Existence
and regularity theory for conservative boundary 'value problems,
Arch. Rational Mech. Anal. 61 (1976). 353-393.
.[1] Weak continuity properties of mappings and semigroups,Proc. Roy"

J. M. BALL

M. F. BEATTY

oT equilibrium states for bars in tension. J. Math.

Theorems

C.B.MoRREy,Jr.

. [I]

Quasi-convexity and lower semicontinuity of multiple variational

integrals of any order. Trans. Amer. Math. Soc., 119 (l965~ 125-149.

Fonctionneiles con vexes, Semina ire sur les equations aux derivees
partielles, Collegc de France. 1966-1967.
[I] Quasi-convexity and the lower semicontlnuity of multiple integrals.
Pacific J. Math. 2 (19Sn 25-53.
'

,",.",.
'y,

'--

..,/
"'-.(
i'~
I

..

...

' .

_..,--

J.M.BALL"

c;'

I~. :

."

[2] "Multiple Integrals"'inthe Calculus of Variations", Springer, Berlin,

1966. .
'.
[I] "Les methodesdirectesen theorie des equationselliptiques",Masson,
Paris. 1967.

J. NEtAS

[I]

A molecular study of the ,mechanical

R.C.TIIOMI'SON &
L-J. FREEUP.

Lllrge

deformation

Isotropic

elasticity

and experiment (or'incompressibJe

London A326 (972), 565-584.

[3]

A. PHILLlPS

R.S. RIVLIN

[I]
[I]

[I]

London 240(1948),491-508.
"

1/

R. T. ROCKAFELLAR
H.RUN\)'

L.SCHWARTZ
C. B. SENSENIG

. M.l.SEWELL
F.SIDOROFF

- on

the

corrclation

-,

Elusticity

On the eigenvalucs of sums of Hermitian matrices, Linear Algeb.

and Appl., 4 (1971),369-376.

"

[2]

Some

restrictions

on constitutive

'

equations,

W.NOLL
W. VAN BUREN

'to VAN HOVE

"

",'

on the

[I]

"The Hamilton-Jacobi theory in the calculus of l'variations", Van

Nostrand. London, 1966.
"
[2] Integral formulae associated with the Euler.Lagninge operators of
multiple integral problems in the calculus of variations, Aequationes
Math.. 11 (1974),212-229.
.
'

[I] "Theorie des distributions", Hermann. Paris, 1966.

'

hyperelastiljue,C. R.Acad. Sc. Paris A, 279 (1974).379-382.

E. SII. VERMAN

[1] Strong quasi-convexity.

S.SMALE

[I] A'c!nssificationof immersionsof the two-sphere.Trans. Amer. Math.

F.STOPPEI.LI

Soc., 90, (1959), 281-290.

'
[I] Un teorema di esistenza e di unicita relativo alle equazioni dell'elastostatica isoterma per deformazioni finite, Ricerche Matematica, 3

R. TliMAM

[I] On the theory and numerical analysis of the Navier-Stokes equations.

F.l. TERPSTRA

(I]

Die Darstellung biquadratischer Formen als Summen van Quadraten

mit Anwendung auf die Variationsrechnung, Math. Ann. 116 (1938).
166-180.

C. M. TIIEOnAW

[I]

An ine<juality for the trace of the product of two symmetric

Math. Proc. Camb. Phi!. Soc., 77 (1975), 265-268.

R. C. THOMPSON

[I] Singular value inequalities for matrix sums and minors, Linear

Pacific J. Math.. 46 (1973).549-554.

(1954).247-267.

Lecture notes in Mathematie~No. 9, Universityof Maryland.

. Algebra

and Appl..

11 (1975).251-269.

'

(I] "The non-linear field theories of mechanics", in Handbuch der

Physik Vol. 111/3.ed. S.FLUOGE,Springer, Berlin, 1965.
[11 "On the existence and unjquenes~ of solutions to boundary value
problems in finite elasticity", Thesis, Department oC Mathematics.
Carnegie-Mellon
University, 1968.' R,esearch Report 68-1D 7.
MEK
MA-RI, Westinghnuse
Research Lahorutorics. Pittsburgh,
. Pa.
J96R.
"

J. VON NF,IIMANN

c.-Co WANe; &

Co TRtlEsDELl
Z. WESOU)WSKI

Sur I'extension de IIIcondition de Legendre du ca/cui des variations

aux integrales muhi'ples 1\ plu"ieurs fonctions ineonnues. Proe.
Koninkl. Ned. Aknd. Wetenschap SO(1947). IH..23.
,
[:!] Sur le signc de la variatinn secundc des intcgrules multiples 1\plusieurs
fonctions ineonnucs:KoninkJ. Bclg. Acad., Kfolsscder Wctenschappcn.
Vcrhandelingen. 24 (1949).
,

[I]

Some matri~-inc4ualities and mclrillllinn oC matric-spllce, Tomsk

Univ. Rev. 1 (1937),2116-300. Rcprinted in Cnlleetcd Works Vol. IV
Pergnmon, Oxfnrd. 1962.

[I]

"Introdition

[I]

"Zagadnienia dynamiczne nicliniowej leorii sprezysto~ci".

Akad. Nauk. IPPT, Warsaw, 1974.

[I]

On the ~tability of a circular tube under end thrust, Quart. J. Mech.

Appl. Math. 8 (1955). 88-100.

E. W. WllKES

to rational elnsticity," Noordhorr, Groningcn, 1973.

Polska

[1] Instability of thick elastic solids. Comm. Pure Appl. Math., 17 (1964).
451-491.
'.
'
[I] On configuration-dependent
loading. Arch. Rational Mech. Anal.,
23(1967),327-351.'
,
[I] Sur les restrictions a imposer a I'energie de deformation d'un materiau

Oxford, 1975.

[I] The main' open problem in the finite theory of elasticity (1955),
reprinted in "Foundations oC Elasticity Theory", Intl. Sci. Rev. Ser.
New York: Gordon and Breach 1965.

[I]

Foundations of Continuum Thermodynamics, Buss~co, 1973.

.
'[3] Stability of pure homogeneous deformations of an el~stic cube under r
dead loading, Quart. Appl. Math. 32 (1974), 265-271;[I] "Convex analysis", Princeton University Press, ,'Princeton, New
,
Jersey, 1970.'
.
! '
.

[1] "The physics of rubber eJa~ticity~, 3rd edition, Oxford Univ.Press,

, C. TRUEkDELL

C. TRUESDELl &

Proc. 1nl. Symp.

[3] Eigenvalues of sums of Hermitian matrices I1I, J. Research Nat. Bur.

Standards B, 7SB (1971).115-120.

L: R. G. TRELOAR

of theory

rubberlike solids, Proe. Roy. Soc.

Large .deformation isotropic elasticity: on the cor'relation of theory

and experiment for compressible rubberlike SOlids,.Proc. Roy. Soc.
London A 328 (1972). 567-583.
.:
.. '"
.
'.
Turning a surface inside out; Scientific American, May 1966.
On the stability of con formal mappings in multidimensional spaces,
Sibirskii Math. 8 (1967), 91-114.
-.oN
"
Stability theorems for mappings with bounded excursion, Sibirskii
Math. 9 (1968), 667-684.
'
"
Large elastic deformations of isotropic materials. It. So'me uniqueness,
theorems for pure homogeneous deformation, Phil. l)rans. Roy. Soc.

.[2]

'I

in Nnn-Lincnr

[2] On
thc 5eigcnvalucs
of 5.
sums of lIermitian matrices 11, Aeljuationes
Math.,
(1970). 103-11

Approximations and numerical analysis of finite deformations of

elastic solids, in "Nonlinear Elasticity" ed. R. W.DICKEY,Academic
Press, New York, 1973..
.
.
.
.
. [I] Compressible isotropic elnstic solids under finite strain-constitutive
inequalities, Quart, J. 'Mcch. Appl. Math.. 23 (1970). 457-468.
.
[2]

RESHETNYAK

Theorems

[I]

R. W.OODEN

V.G.

[I]

of arterial wall

vessels,Z.A.M.P.. 25 (1974).565-578.'

J.T.()l)IiN

properties

.~, 'r' r',1

Existcnee

"

P.NII:I>ERER

."')

"

'

matrices,

Department of Mathematics
Heriot-Wait University
Edinburgh EH 144AS
(R/'C'/'Ir'eclAIII(II.vtI5, 1976)

'