Di Serta Tie
Di Serta Tie
Di Serta Tie
(t),
(t) X is a continuous mapping, we can dene an operator which has a unique xed
point
C([0, T]; X). Then, it will be proved that the unique solution of the variational system is the pair
(
) C([0, T]; X) C([0, T]; (g)). Nextly, we focus on continuous dependencies on the data f , u
0
and
g. As we shall see, the solution depends Lipschitz continuously on the initial data. In addition, if f is constant
in time, the solution depends Lipschitz continuously simultaneous on f and u
0
. Moreover, it will be proved that
the solution depends continuously on g, this being the most signicant result of the present paper.
The abstract results we deliver can be employed in the mathematical analysis of a class of frictional contact
problems for viscoelastic materials with short memory. To give an example, in the present paper we consider
an antiplane problem which models the frictional contact between a cylinder and a rigid foundation. The reason
for which we are placed in the antiplane framework is to simplify the writing; the corresponding 3D model
can be analyzed similarly into an appropriate functional setting. The proposed mechanical model leads to an
evolutionary mixed variational problem with friction bound dependent set of Lagrange multipliers. This weak
formulation is cast in the abstract variational problem studied previously. After proving the existence and the
uniqueness of the weak solution, we investigate the Lipschitz continuous dependence on the initial data and
on the densities of the volume forces and surface tractions. In addition, based on the boundedness of the weak
solution, the continuous dependence on the friction bound is also proved; herein, the difculty is generated by
the dependence of the Lagrange multipliers set on the friction bound.
To end this introductory part we recall an existence result which will be used in Section 2.
Theorem 1. Let X and Y be two Hilbert spaces. We assume that:
(i) A X and B Y are non-empty closed convex subsets;
(ii) for all B the mapping v L(v, ) is a convex and lower semicontinuous function;
(iii) for all v A the mapping L(v, ) is a concave and upper semicontinuous function.
Moreover, assume that A is bounded, or otherwise there exists
0
B such that lim
|v|
X
,vA
L(v,
0
) = , and
B is bounded, or otherwise lim
||
Y
,B
inf
vA
L(v, ) = .
Then L possesses a saddle point on A B.
For a proof of this theorem we refer the reader to [7, p. 176]. We recall that given two non-empty sets A and
B, a pair (u, ) A B is said to be a saddle point of a functional L : A B R if and only if
L(u, ) L(u, ) L(v, ) for all (v, ) A B.
It could also be helpful for the reader to consult [811] for some tools related to weak topologies or some
results in the theory of Sobolev spaces. Concerning contact mechanics, we refer the reader to, e.g., [1221].
The rest of the paper is structured as follows. In Section 2 we analyze an abstract problem. We study here
the existence, the uniqueness and the boundedness of the solution; also, the continuous dependence on the data
is discussed. In Section 3 we discuss a boundary value problem which models the frictional contact between
a viscoelastic cylinder and a rigid foundation, in the antiplane context. We deliver for this problem a weak
formulation in terms of dual Lagrange multipliers. This formulation allows us to analyze the well-posedness of
the model by using the abstract results we obtained. Section 4 contains some nal comments.
Matei 3
2. An abstract problem
Let T be a positive real number. In this section we are interested in studying the following evolutionary mixed
variational problem.
Problem 1 Given f : [0, T] X, g W and u
0
X, nd u : [0, T] X and : [0, T] (g) Y so that
for each t (0, T), we have
a( u(t), v) e(u(t), v) b(v, (t)) = (f (t), v)
X
for all v X, (1)
b( u(t), (t)) 0 for all (g), (2)
u(0) = u
0
. (3)
We shall study Problem 1 under the following assumptions.
Assumption 1 We assume that (X, (, )
X
, ||
X
), (Y, (, )
Y
, ||
Y
) and (W, (, )
W
, ||
W
) are three Hilbert spaces.
Assumption 2 We assume that a(, ) : X X R is a symmetric bilinear form so that:
(a
1
) there exists M
a
> 0 : [a(u, v)[ M
a
|u|
X
|v|
X
for all u, v X;
(a
2
) there exists m
a
> 0 : a(v, v) m
a
|v|
2
X
for all v X.
Assumption 3 We assume that e(, ) : X X R is a symmetric bilinear form so that:
(e
1
) there exists M
e
> 0 : [e(u, v)[ M
e
|u|
X
|v|
X
for all u, v X;
(e
2
) there exists m
e
> 0 : e(v, v) m
e
|v|
2
X
for all v X.
Assumption 4 We assume that b(, ) : X Y R is a bilinear form so that:
(b
1
) there exists M
b
> 0 : [b(v, )[ M
b
|v|
X
||
Y
for all v X, Y;
(b
2
) there exists > 0 : inf
Y,,=0
Y
sup
vX,v,=0
X
b(v,)
|v|
X
||
Y
.
Assumption 5 We assume that f C([0, T]; X).
Assumption 6 For each W, () is a closed convex subset of Y so that 0
Y
().
Assumption 7 If (
n
)
n
W and (w
n
)
n
X are two sequences so that
n
in W and w
n
w in X, as
n , then:
(k
1
) for each () Y, there exists a sequence (
n
)
n
Y so that
n
(
n
) for all n 1, and
limsup
n
b(w
n
,
n
) 0;
(k
2
) For each subsequence ((
n
/ ))
n
/ of the sequence ((
n
))
n
, if (
n
/ )
n
/ Y so that
n
/ (
n
/ ) and
n
/ in Y as n
/
, then ().
In order to prove the existence and the uniqueness of the solution of Problem 1 we need a few intermediate
results.
Let C([0, T]; X) be given and let us consider the following intermediate problem.
Problem 2 Given f : [0, T] X and g W nd w
: [0, T] X and
(t)) = (f (t), v)
X
for all v X, (4)
b(w
(t),
(t) X so that
(f
(t), v)
X
= (f (t), v)
X
e((t), v) for all v X. (7)
Let us associate with Problem 2 the following functional:
L
t
: X (g) R, L
t
(v, ) =
1
2
a(v, v) (f
(t), v)
X
b(v, ). (8)
By standard arguments, it can be proved that a pair (w
(t),
(t),
(w
(t), ) L
t
(w
(t),
(t)) L
t
(v,
(v, 0
Y
) = .
If (g) is bounded, then according to Theorem 1 the functional L
t
possesses a solution
(w
(t),
(t)) X (g).
If (g) is unbounded, in order to apply Theorem 1 we have to prove that
lim
||
Y
,(g)
inf
vX
L
t
(v, ) = . (10)
To this end, let (g) be xed and let u
(t), v) b(v, ) = (f
(t), v)
X
for all v X, (11)
guaranteed by the LaxMilgram lemma. Since a(, ) is symmetric, we can write
inf
vX
L
t
(v, ) =
1
2
a(u
(t), u
(t)) (f
(t), u
(t))
X
b(u
(t), ).
Setting v = u
(v, )
m
a
2
|u
(t)|
2
X
. (12)
Using the inf-sup property of the form b(, ), see (b
2
) in Assumption 4 and (a
1
) in Assumption 2, by (11) we
get
||
Y
|f
(t)|
X
M
a
|u
(t)|
X
.
Hence, there exists c > 0 so that
||
2
Y
c(|f
(t)|
2
X
|u
(t)|
2
X
). (13)
By (12) and (13) we deduce now (10).
Therefore, applying Theorem 1, the functional L
t
possesses a solution
(w
(t),
(t)) X (g).
Since the pair (w
(t),
(t)) X (g) is a solution of the problem 9 if and only if it veries (4) and (5),
we conclude that Problem 2 has at least one solution.
Matei 5
To prove the uniqueness of the solution, let us assume that (w
1
,
1
) and (u
2
,
2
(t) w
2
(t), w
1
(t) w
2
(t)) = b(w
1
(t),
2
(t)
1
(t)) b(w
2
(t),
1
(t)
2
(t)).
Using (5) and (a
2
) in Assumption 2, we obtain w
1
(t) = w
2
(t)
(t)) = 0 for all v X. Using now the inf-sup property of the form b(, ) we are led to
1
(t) =
2
(t).
Finally, let us justify (6). To this end, let t
1
, t
2
be in [0, T].
a(w
(t
1
) w
(t
2
), w
(t
1
) w
(t
2
)) = b(w
(t
1
) w
(t
2
),
(t
2
)
(t
1
))
(f
(t
1
) f
(t
2
), w
(t
1
) w
(t
2
))
X
.
Owing to (5), we have
b(w
(t
1
) w
(t
2
),
(t
2
)
(t
1
)) 0
and, thus,
a(w
(t
1
) w
(t
2
), w
(t
1
) w
(t
2
)) (f
(t
1
) f
(t
2
), w
(t
1
) w
(t
2
))
X
.
Taking into consideration (7) and Assumptions 23 it follows that
|w
(t
1
) w
(t
2
)|
X
M
e
m
a
|(t
1
) (t
2
)|
X
1
m
a
|f (t
1
) f (t
2
)|
X
.
As f C([0, T]; X) and C([0, T]; X), the previous inequality yields w
(t
1
)
(t
2
)|
Y
|f (t
1
) f (t
2
)|
X
M
e
|(t
1
) (t
2
)|
X
M
a
|w
(t
1
) w
(t
2
)|
X
.
Taking into account the regularity of f , and w
we conclude that
C(0, T; (g)).
Let us consider the operator T : C([0, T]; X) C([0, T]; X) dened as follows: for each C([0, T]; X),
T (t) =
_
t
0
w
(s) ds u
0
for all t [0, T]. (14)
Lemma 3. The operator T has a unique xed point
= sup
t[0,T]
| (t)|
X
e
t
,
it follows that
|T
1
(t) T
2
(t)|
X
e
t
M
e
m
a
|
1
2
|
M
e
m
a
|
1
2
|
C([0, T]; X) be the xed point of T . We consider the functions u : [0, T] X and : [0, T]
Y so that u(t) =
(t) for all t [0, T]. Since, for each t [0, T] we have
(t) = T
(t), then
it follows from (14) that u(0) = u
0
and u = w
we conclude that
the pair (u, ) is a solution of Problem 1. Next, by (6), we get C([0, T]; Y). Moreover, since u = w
and
w
M
e
m
a
_
t
0
_
M
e
m
a
|u
0
|
X
1
m
a
|f (s)|
X
_
e
Me
ma
(ts)
ds.
By (38), taking into account Assumption 5, we deduce that there exists c
1
> 0 so that
| u|
C([0,T];X)
c
1
.
10 Mathematics and Mechanics of Solids
Using now (37) we deduce that there exists c
2
> 0 so that
|u|
C([0,T];X)
c
2
.
Consequently (33) is fullled with K
1
= c
1
c
2
.
Furthermore, by using (1) and the inf-sup property of the form b(, ) it follows that
|(t)|
Y
|f (t)|
X
M
e
|u(t)|
X
M
a
| u(t)|
X
.
Therefore (34) is fullled for K
2
=
|f |
C([0,T];X)
M
e
c
2
M
a
c
1
.
Finally, let us prove the continuous dependence of the solution on the data g W.
Proposition 7. If Assumptions 17 hold true and (g
n
)
n
W is a sequence so that g
n
g in W as n , then
for all t [0, T],
u
n
(t) u(t) in X as n ; (39)
u
n
(t) u(t) in X as n ; (40)
n
(t) (t) in Y as n , (41)
where (u, ) and (u
n
,
n
) denote the solutions of Problem 1 associated with the data (f, g, u
0
) C([0, T]; X)
W X and (f, g
n
, u
0
) C([0, T]; X) W X for all n 1.
Proof. Let (g
n
)
n
W be a sequence so that g
n
g in W as n and let t [0, T]. For every positive
integer n, we have
a( u
n
(t), v) e(u
n
(t), v) b(v,
n
(t)) = (f (t), v)
X
for all v X, (42)
b( u
n
(t),
n
(t)) 0 for all (g
n
), (43)
u
n
(0) = u
0
. (44)
Also, we have
a( u(t), v) e(u(t), v) b(v, (t)) = (f (t), v)
X
for all v X, (45)
b( u(t), (t)) 0 for all (g), (46)
u(0) = u
0
. (47)
Setting v = u
n
(t) u(t) and summing (42) and (45) we get
e(u
n
(t) u(t), u
n
(t) u(t)) = a( u
n
(t) u(t), u
n
(t) u(t))
b( u(t) u
n
(t),
n
(t) (t)).
Hence,
e(u
n
(t) u(t), u
n
(t) u(t)) b( u(t) u
n
(t),
n
(t) (t)).
Therefore,
1
2
d
dt
e(u
n
(t) u(t), u
n
(t) u(t)) b( u(t) u
n
(t),
n
(t) (t))
and from this,
m
e
|u
n
(t) u(t)|
2
X
2
_
t
0
b( u(s) u
n
(s),
n
(s) (s)) ds. (48)
Note that, according to Proposition 6, ( u
n
(t))
n
is a bounded sequence in X. Hence, there exists a subsequence,
still denoted by ( u
n
(t))
n
, to simplify the writing, which is weakly convergent to an element w(t) X.
Matei 11
By Assumption 7 (k
1
), as (t) (g), it follows that there exists a sequence (
n
(t))
n
Y, so that
n
(t)
(g
n
) for all n 1 and
limsup
n
b( u
n
(t), (t)
n
(t)) 0.
On the other hand, (
n
(t))
n
is a bounded sequence; see Proposition 6. Thus, there exists a subsequence, still
denoted by (
n
(t))
n
, to simplify the writing, so that
n
(t) (g
n
) and
n
(t)
(t) in Y as n . Hence,
according to Assumption 7 (k
2
), we have
(t) (g).
Owing to (43) we can write
b( u
n
(t),
n
(t)
n
(t)) 0
and due to (46)
b( u(t),
(t) (t)) 0.
We deduce that
b( u(t) u
n
(t),
n
(t) (t)) b( u(t),
n
(t)
(t)) b( u
n
(t), (t)
n
(t)).
Combining the previous inequality with (48) it follows that
m
e
|u
n
(t) u(t)|
2
X
2
_
t
0
[b( u
n
(s), (s)
n
(s)) b( u(s),
n
(s)
(s))] ds.
Passing to the superior limit as n , and taking into consideration (k
1
) in Assumption 7, Equation (39)
follows.
Using again (42) and (45) we can write
m
a
| u(t) u
n
(t)|
2
X
M
e
|u
n
(t) u(t)|
X
| u(t) u
n
(t)|
X
b( u
n
(t), (t)
n
(t))
b( u(t),
n
(t)
(t)).
Therefore,
| u(t) u
n
(t)|
2
X
M
2
e
2 m
2
a
|u
n
(t) u(t)|
2
X
| u
n
(t) u(t)|
2
X
2
(49)
1
m
a
(b( u
n
(t), (t)
n
(t)) b( u(t),
n
(t)
(t))).
Consequently,
1
2
| u(t) u
n
(t)|
2
X
M
2
e
2 m
2
a
|u
n
(t) u(t)|
2
X
1
m
a
(b( u
n
(t), (t)
n
(t)) b( u(t),
n
(t)
(t))).
Passing again to the superior limit we get (40).
Using now (42), (45) and (b
2
) in Assumption 4 it follows that
|
n
(t) (t)|
Y
M
a
| u(t) u
n
(t)|
X
M
e
|u
n
(t) u(t)|
X
and from this, taking into account (39) and (40), we get (41).
12 Mathematics and Mechanics of Solids
3. A frictional contact problem
In this section we shall indicate an example of frictional contact problem whose weak form can be cast in the
abstract setting studied before.
Problem 3 Find a displacement eld u :
[0, T] R so that, for all t (0, T), we have
div ((x) u(x, t) (x) u(x, t)) f
0
(x, t) = 0 in , (50)
u(x, t) = 0 on
1
, (51)
(x)
u(x, t) (x)
u(x, t) = f
2
(x, t) on
2
, (52)
[(x)
u(x, t) (x)
u(x, t) (x)
u(x, t) = g(x)
u(x,t)
[ u(x,t)[
if u(x, t) ,= 0
_
on
3
, (53)
u(0) = u
0
in . (54)
Herein [0, T] is the time interval and R
2
is a bounded domain with Lipschitz continuous boundary.
The boundary will be denoted by and will be partitioned into three measurable parts
1
,
2
,
3
so that
the Lebesgue measure of
1
is positive. Problem 3 models the antiplane shear deformation of a viscoelastic,
isotropic, non-homogeneous cylindrical body in frictional contact on
3
with a rigid foundation. Referring the
body to a Cartesian coordinate system Ox
1
x
2
x
3
so that the generators of the cylinder are parallel with the axis
Ox
3
, the domain Ox
1
x
2
denotes the cross-section of the cylinder. The function = (x
1
, x
2
) :
R
is the viscoelastic coefcient, = (x
1
, x
2
) :
R denotes a coefcient of the material (one of Lams
coefcients), the functions f
0
= f
0
(x
1
, x
2
, t) : (0, T) R, f
2
= f
2
(x
1
, x
2
, t) :
2
(0, T) R are related
to the density of the volume forces and the density of the surface traction, respectively, and g :
3
R
is the
friction bound, a given function. Here = (
1
,
2
) (
i
=
i
(x
1
, x
2
), for each i {1, 2]), represents the outward
unit normal vector to the boundary and
u = u .
The unknown part of the problem is the function u = u(x
1
, x
2
, t) :
[0, T] R which represents the
third component of the displacement vector u. We recall that, in the antiplane physical setting, the displacement
vectorial eld has a particular form, u = (0, 0, u(x
1
, x
2
)). Once the eld u is determined, the stress tensor can
be computed:
=
0 0
u
x
1
0 0
u
x
2
u
x
1
u
x
2
0
.
In the study of Problem 3 we assume that the elasticity and the viscosity coefcients satisfy
L
a.e. x . (55)
L
a.e. x . (56)
The volume forces and surface tractions densities satisfy
f
0
C([0, T]; L
2
()), f
2
C([0, T]; L
2
(
2
)), (57)
and the friction bound g satises
g L
2
(
3
) so that g(x) 0 a.e. x
3
. (58)
Let us introduce the following Hilbert space
X = { v H
1
() [ v = 0 a.e. on
3
] (59)
Matei 13
endowed with the inner product
(u, v)
X
=
_
u v dx.
In (59), denotes the Sobolev trace operator. Finally, we made the following assumption for the initial
displacement:
u
0
X. (60)
If u is a regular enough function which satises (50)(54), then, using a Green formula, at each moment
t [0, T], we can write for all v X, the following identity
_
u(t) v dx
_
u(t) v dx =
_
f
0
(t) v dx
_
2
f
2
(t) v d (61)
3
(
u(t)
u(t)) v d.
We dene the bilinear forms a : X X R and e : X X R by equalities
a(u, v) =
_
f
0
(t) v dx
_
2
f
2
(t) v d for all v X. (64)
It follows from (57) that the integrals in (64) are well dened. Consequently, for each t (0, T) we can write
a( u(t), v) e(u(t), v) = (f (t), v)
X
_
3
(
u(t)
u(t)) v d. (65)
We consider the space
S = {v = v[
3
v X] (66)
endowed with the norm
|v|
3
=
_
_
3
_
3
(v(x) v(y))
2
|x y|
2
ds
x
ds
y
_
1/2
for all v S,
where | | denotes the Euclidian norm in R
2
.
We can dene now a second Hilbert space, the dual of the space S,
Y = S
/
. (67)
Also, we can dene a bilinear form b : X Y R as follows
b(v, ) = , v[
3
), (68)
where , ) denotes the duality pairing between the spaces Y and S.
Next, for each L
2
(
3
) we dene () as follows:
() =
_
Y : , w[
3
)
_
3
[(x)[ [ w(x)[ d for all w X
_
. (69)
14 Mathematics and Mechanics of Solids
We dene a Lagrange multiplier , so that at each t [0, T], (t) Y and
(t), z) =
_
3
(
u(t)
3
(
u(t)
u(t)) u(t) d =
_
3
g[ u(t)[ d
and, taking into account (68) we obtain the following equality
b( u(t), (t)) =
_
3
g[ u(t)[ d for all t (0, T).
Moreover, keeping in mind (69), we have at each moment t (0, T),
b( u(t), )
_
3
g[ u(t)[ d for all (g).
Consequently, at each moment t (0, T), we get
b( u(t), (t)) 0 for all (g).
Thus, we deliver the following mixed formulation of Problem 3.
Problem 4 Find u : [0, T] X and : [0, T] (g) Y so that, for all t (0, T),
a( u(t), v) e(u(t), v) b(v, (t)) = (f (t), v)
X
for all v X,
b( u(t), (t)) 0 for all (g),
u(0) = u
0
.
Theorem 8. Assume that (55)(58) and (60) hold. Then, Problem 4 has a unique solution (u, ) with the
regularity
u C
1
([0, T]; X), C([0, T]; (g)).
Proof. Taking into account (56), the bilinear form a(, ), dened in (62), veries Assumption 3 with M
a
=
||
L
()
and m
a
=
. Also, by (55) the bilinear form e(, ) dened in (63) veries Assumption 3 with
M
e
= ||
L
()
and m
e
=
. To prove that the bilinear form b(, ) veries Assumption 4, arguments similar
to those used in [6, 24] can be used; for the convenience of the reader, we shall indicate below the justication.
First,
[b(v, )[ ||
Y
| v[
3
|
3
||
Y
| v|
where, herein, | |
denotes the norm on the space (X) = {v = v v X]. Since (X) is a closed subspace
of H
1/2
(), see [6], we deduce that
[b(v, )[ ||
Y
| v|
H
1/2
()
.
Owing to the fact that : H
1
() H
1/2
() is a linear and continuous operator and taking into account that
| |
X
and | |
H
1
()
are equivalent norms, we deduce that there exists M
b
> 0 so that (b
1
) in Assumption 4 holds
true.
Matei 15
We recall here that there exists a linear and continuous operator Z, the inverse to the right of the operator ,
so that
Z : H
1/2
() H
1
() (Z( )) = for all H
1/2
().
Note that,
(Z( w)) = w for all w X.
Since, for each w X, Z( w) has the same trace as w, we deduce that for each w X, we have Z( w) X.
Let us prove now (b
2
) in Assumption 4. As
||
Y
= sup
w[
3
S, w[
3
,=0
S
< , w[
3
>
| w[
3
|
3
c sup
w[
3
S, w[
3
,=0
S
b(Z( w), )
|Z( w)|
X
c sup
vX, v,=0
X
b(v, )
|v|
X
,
with a positive constant c, then we can take
=
1
c
. (71)
Also, we note that for each L
2
(
3
), () is a closed convex subset of the space Y which contains 0
Y
.
Hence, Assumption 6 is fullled. Consequently, Theorem 8 is a straightforward consequence of Theorem 4.
Based on Proposition 5 we are led to the following results related to our mechanical model.
Proposition 9. Assume that (55)(58) and (60) hold.
(i
1
) Given f
0
C([0, T]; L
2
()), f
2
C([0, T]; L
2
(
2
)), g L
2
(
3
) and two initial data u
1
0
, u
2
0
X, there exists
c
1
> 0 so that
|u
1
u
2
|
C
1
([0,T];X)
c
1
|u
1
0
u
2
0
|
X
(72)
where u
1
, u
2
are the corresponding solutions of Problem 4.
(i
2
) Given f
1
0
, f
2
0
L
2
(), f
1
2
, f
2
2
L
2
(
2
), g L
2
(
3
) and two initial data u
1
0
, u
2
0
X, there exists c
2
> 0 so
that
|u
1
u
2
|
C
1
([0,T];X)
c
2
(|f
1
0
f
2
0
|
L
2
()
|f
1
2
f
2
2
|
L
2
(
2
)
|u
1
0
u
2
0
|
X
) (73)
where u
1
, u
2
are the corresponding solutions of Problem 4.
Proof. Assuming that (55)(58) and (60) hold true, then Assumptions 1-6 are fullled. Proposition 9 is a
straightforward consequence of Proposition 5.
In addition, we have the following boundedness result.
Proposition 10. Assume that (55)(58) and (60) hold. The solution (u, ) of Problem 4 is bounded.
Proof. The proof is a straightforward consequence of Proposition 6.
Let us investigate now the continuous dependence of the weak solution on the friction bound.
Proposition 11. Assume that (55)(58) and (60) hold. If (g
n
)
n
L
2
(
3
) is a sequence of friction bounds so that
g
n
0 a.e. on
3
, for all n 1, and g
n
g in L
2
(
3
) as n , then, for all t [0, T],
u
n
(t) u(t) in X as n ;
u
n
(t) u(t) in X as n ;
n
(t) (t) in Y as n ,
16 Mathematics and Mechanics of Solids
where (u, ) and (u
n
,
n
) are solutions of Problem 4 associated with the friction bounds g and g
n
, for all n 1.
Proof. The idea is to apply Proposition 7. A crucial point is to verify Assumption 7, the other assumptions
being already veried. Precisely, we shall prove below that Assumption 7 is fullled with W = L
2
(
3
), X
dened by (59), Y dened by (67), the set of Lagrange multipliers dened by (69) and the form b(, ) dened
by (68).
Let us verify (k
1
) in Assumption 7. To this end, let (
n
)
n
L
2
(
3
) and (w
n
)
n
X be two sequences, so that
n
in L
2
(
3
) and w
n
w in X, as n . In addition, let (). We dene
n
Y as follows
n
, v[
3
) =
_
3
[
n
(x)[ sgn w
n
(x) v(x) d (74)
3
[(x)[ [ w
n
(x)[ d
, w
n
[
3
) for all v[
3
S,
where, as usual,
sgn w
n
(x) =
_
1 if w
n
(x) > 0;
0 if w
n
(x) = 0;
1 if w
n
(x) < 0.
Since (), we have
3
[(x)[ [ w
n
(x)[ d , w
n
[
3
) 0.
Therefore, by (69), for each positive integer n, we deduce that
n
(
n
). Setting v = w
n
in (74) we can write
n
, w
n
[
3
) =
_
3
([
n
(x)[ [(x)[)[ w
n
(x)[ d. (75)
Since : H
1
() L
2
() is a compact operator and (w
n
)
n
is a weakly convergent sequence in X, then ( w
n
)
n
is a bounded sequence in X. Passing to the superior limit as n in (75), it follows that
limsup
n
b(w
n
,
n
) = limsup
n
([
n
[ [[, [ w
n
[)
L
2
(
3
)
= 0.
Finally, taking into account (69) we get (k
2
) in Assumption 7.
In order to conclude Proposition 10, we have to apply Proposition 7.
4. Conclusions and comments
The study in this paper provides abstract results for a new mixed variational problem. Then, a frictional contact
model for viscoelastic materials of type KelvinVoigt is revisited, employing in its analysis the abstract results
we have obtained. The new variational approach allows to prove the existence and the uniqueness of the weak
solution, the boundedness of the weak solution, the Lipschitz continuous dependence on the initial data and
on the densities of the volume forces and surface tractions. In addition, it is worth mentioning the continuous
dependence of the weak solution on the friction bound. Since the Lagrange multipliers set depends on the fric-
tion bound, signicant mathematical difculties arise. The present study was motivated by numerical reasons,
such kind of weak formulations being appropriate to efcient approximations of the weak solutions for a class
of frictional contact problems.
Funding
This work was supported by the Romanian National Authority for Scientic Research, CNCS-UEFISCDI (project number PN-II-ID-
PCE-2011-3-0257).
Matei 17
References
[1] Heber, S, and Wohlmuth, BI. An optimal a priori error estimate for nonlinear multibody contact problems. SIAM J Numer Anal
2005; 43: 156173.
[2] Heber, S, Matei, A, and Wohlmuth, B. Efcient algorithms for problems with friction. SIAM J Scientic Computat 2007; 29:
70-92.
[3] Hild, P, and Renard, Y. A stabilized Lagrange multiplier method for the nite element approximation of contact problems in
elastostatics. Numer Math 2010; 115: 101129.
[4] Amdouni, S, Hild, P, Lleras, V, Moakher, M, and Renard, Y. A stabilized Lagrange multiplier method for the enriched nite-
element approximation of contact problems of cracked elastic bodies. ESAIM: M2AN Math Modell Numer Anal 2012; 46: 813
839.
[5] Sofonea, M, and Matei, A. A mixed variational formulation for the Signorini frictionless problem in viscoplasticity. Ann Univ
Ovidius Constanta 2004; 12: 157170.
[6] Matei, A, and Ciurcea, R. Weak solutions for contact problems involving viscoelastic materials with long memory. Math Mech
Solids 2011; 16: 393405.
[7] Ekeland, I, and Temam, R. Convex Analysis and Variational Problems (Studies in Mathematics and its Applications, vol. 1).
Amsterdam: North-Holland, 1976.
[8] Adams, RA. Sobolev Spaces. New York: Academic Press, 1975.
[9] Brzis, H. Analyse fonctionnelle, Thorie et applications. Paris: Masson, 1987.
[10] Lions, J-L, and Magenes, E. Problmes aux limites non homognes. Paris: Dunod, 1968.
[11] Grisvard, P. Elliptic Problems in Nonsmooth Domains. London: Pitman Advanced Publishing Program 1985.
[12] Duvaut, G, and Lions, JL. Inequalities in Mechanics and Physics (A Series of Comprehensive Studies in Mathematics, vol. 219).
Berlin: Springer, 1976.
[13] Han, W, and Sofonea, M. Quasistatic Contact Problems in Viscoelasticity and Viscoplasticity (Studies in Advanced Mathematics).
Providence, RI: American Mathematical Society, International Press, 2002.
[14] Johnson, KL. Contact Mechanics. Cambridge: Cambridge University Press, 1987.
[15] Kikuchi, N, and Oden, JT. Contact Problems in Elasticity: A Study of Variational Inequalities and Finite Element Methods.
Philadelphia, PA: SIAM, 1988.
[16] Laursen, TA. Computational Contact and Impact Mechanics. Berlin: Springer, 2002.
[17] Shillor, M, Sofonea, M, and Telega, J. Models and Variational Analysis of Quasistatic Contact. Berlin: Springer, 2004.
[18] Sofonea, M, and Matei, A. Variational Inequalities with Applications. A Study for Antiplane Contact Problem. Berlin: Springer,
2009.
[19] Wriggers, P. Computational Contact Mechanics. Chichester: Wiley, 2002.
[20] Wriggers, P, and Nackenhorst, U (eds). Proceedings of the IUTAM Symposium, Hannover, Germany, 58 November 2006. Berlin:
Springer, 2007.
[21] Sofonea, M, and Matei, A. Mathematical Models in Contact Mechanics (London Mathematical Society Lecture Note Series, vol.
398). Cambridge: Cambridge University Press, 2012.
[22] Ekeland, I, and Temam, R. Convex Analysis and Variational Problems (Studies in Mathematics and its Applications, vol. 1).
Amsterdam: North-Holland, 1976.
[23] Haslinger, J, Hlav cek, I, and Ne cas, J. Numerical methods for unilateral problems in solid mechanics. In Lions, J-L and Ciarlet,
P (eds), Handbook of Numerical Analysis, vol. IV. Amsterdam: North-Holland, 1996, pp. 313485.
[24] Matei, A. On the solvability of mixed variational problems with solution-dependent sets of Lagrange multipliers. Proc R Soc
Edinburgh A, in press.