On 2D Viscoelasticity with Small Strain

∗
Zhen Lei
arXiv:0904.1345v1 [math.AP] 8 Apr 2009

Abstract
An exact two-dimensional rotation-strain model describing the motion of
Hookean incompressible viscoelastic materials is constructed by the polar decom-
position of the deformation tensor. The global existence of classical solutions is
proved under the smallness assumptions only on the size of initial strain tensor.
The proof of global existence utilizes the weak dissipative mechanism of motion,
which is revealed by passing the partial dissipation to the whole system.

Keyword: Rotation-strain model, viscoelastic fluids, partial dissipation, complex flu-

ids.

1 Introduction
Viscoelastic materials include a wide range of fluids with elastic properties, as well
as solids with fluid properties. Let us start with the standard description of general
mechanical evolutions to introduce some notations and definitions. Any deformation
can be represented by a flow map (particle trajectory) x(t, X), 0 ≤ t < T , a time-
dependent family of orientation-preserving diffeomorphisms. X is the original labeling
(Lagrangian coordinate) of the particles. X is also referred to as the material/reference
coordinate. x is then the observer’s (Eulerian) coordinate. In general, the velocity field
u(t, x) is defined as the time derivative of the flow map, which is a vector field defined
on the Eulerian coordinate:


u(t, x) = xt t, X(t, x) .

The deformation tensor Fe in Lagrangian coordinate is defined by

∂xi
Feij (t, X) = (t, X).
∂Xj
When one uses the Eulerian description, the deformation gradient tensor F in spatial
coordinate is defined by

F (t, x) = Fe t, X(t, x) .
∗
School of Mathematical Sciences, Fudan University, Shanghai 200433, P. R. China. email:
leizhn@yahoo.com, zlei@fudan.edu.cn

1
 Recently, an extensively studied system modelling the motion of isotropic incom-
pressible viscoelastic materials of Oldroyd-B type takes:


 ∇ · u = 0,

ut + u · ∇u + ∇p = µ∆u + ∇ · ( ∂W∂F(F ) F T ), (1.1)


F + u · ∇F = ∇uF,
t

where p(t, x) is the pressure, µ is the viscosity and W (F ) is the elastic energy function.
The global existence of classical solutions near equilibrium to (1.1) in the 2D Hookean
case was established by Lin, Liu and Zhang [17] (see also Lei and Zhou [15] via the
incompressible limit method). Later on, Lei, Liu and Zhou [14] proved the global
existence of classical small solutions to system (1.1) in both 2D and 3D cases (see also
Chen and Zhang [3]). Very recently, an improved result was obtained by Kessenich [9]
by removing the dependence of the smallness of the initial data on the viscosity via
the hyperbolic energy method.
To study the different contributions of strain and rotation parts, Friedrich [5] ob-
served that the smallness of the strain in nonlinear elasticity can be realized through
the polar decomposition of the deformation tensor. John [6, 7] showed that no point-
wise estimate for rotations in terms of strains can exist even in the case of small strain.
In the work of Friedrich and John, no PDE is involved. From the PDE point of view,
Liu and Walkington [19] considered approximating systems resulting from the special
linearization of the original system with respect to the strain. Lei, Liu and Zhou [13]
constructed a 2D rotation-strain viscoelastic model and proved the global existence
their classical solutions with small strain. One shortcoming of the model in [13] is
that the dynamics of strain and rotation are not equivalent to that of the deformation
tensor.
One main concern of this paper is to construct an exact rotation-strain model for
motions of Hookean viscoelastic materials of Oldroyd-B type, which takes


 ∇ · u = 0,



 ut + u · ∇u + ∇p = µ∆u + ∇ · (V V T ) + 2∇ · V,




V + u · ∇V = D(u) + 1 (∇uV + V ∇uT )
t 2
1 (1.2)

 + ω 12 (u)(V A − AV ) − 21 γ(V A − AV ),

 2
!



 −∇ u · ∇θ

 ⊥ ⊥ 1 2
+ ∇⊥ γ.
∇ θt + u · ∇∇ θ = 2 ∆u −
∇1 u · ∇θ

Here V is the strain part of the deformation tensor, θ is the rotation angle and γ is a
nonlinear term. The derivation of this model will be presented in section 2. It seems
that the velocity field u is the only dissipative unknown. We will prove the global
existence of classical solutions to (1.2) for small initial strain V and initial gradient of
θ (θ may be arbitrarily large). The proof utilizes the weak dissipative mechanism of

2
system (1.2) by passing the dissipation of the velocity field u to the strain tensor V and
the gradient of the angle variable ∇θ. Note that the equations for u and V have already
formed an closed dynamical system. However, our proof needs the use of the equation
for θ to carry out the weak dissipative mechanism of the whole system. Without the
help of the equation for θ and the underlying physical relationship between V and θ
(see Lemma 3.3), the corresponding analysis issue is still open.
At last, let us also cite some related results on elastodynamics when the viscosity
µ is zero. The global existence of 3D compressible elastodynamics was established by
Agemi [1] and Sideris [20] independently under the null condition and small initial data.
An earlier almost global existence result can be founded in John and Klainerman [8] (see
also Klainerman and Sideris [11]). Hyperbolic energy method involving Klainerman’s
vector field developed by Klainerman [10] (see also Christodoulou [4]) plays an essential
role in [11, 20]. The incompressible case was later on studied by Sideris and Thomases
in [21, 22, 23]. A natural question to ask is that what happens for the solutions of
(1.1) when the viscosity vanishes, which is answered very recently by Kessenich [9] in
the case of 3D. Other results on Oldroyd-B model can be found in [2, 12, 16, 18].
The remaining part of this paper is organized as follows. In section 2 we present the
derivation of the rotation-strain model. Then the main result of this paper is stated.
Section 3 is devoted to exploring some of the intrinsic properties of the viscoelastic
system. The proof of global existence theorem is completed in section 4.

2 Derivation of the Strain-Rotation Model

Constitutive relations generally involve the deformation tensor F . In particular, the
polar decomposition of F is important. For any non-singular matrix F , it is easy to see
that F F T is positive definite symmetric. Thus, there exists a unique positive definite
symmetric matrix I +V such that (I +V )2 = F F T , where I is the 2×2 identity matrix.
Then the matrix R = (I + V )−1 F is orthogonal. Consequently the non-singular matrix
F can be decomposed uniquely in the form

F = (I + V )R, (2.1)

where R is orthogonal
RRT = I
and I + V positive definite symmetric

V = V T. (2.2)

Physically, this means that the deformation is decomposed into stretching and rotation.
Following a suggestion by K. O. Friedrichs [5], the tensor I + V is called the left stretch
tensor, V the strain matrix and R the rotation matrix.
Our first goal is to formulate the viscoelastic system in terms of the velocity u
and the strain V . We first look at the momentum equation, the second equation in

3
(1.1). We restrict our discussions below to a special elastic energy functional of the
Hookean linear elasticity W (F ) = 21 |F |2 , it does not reduce the essential difficulties for
mathematical analysis. Indeed, all the results we describe here can easily be generalized
to a more general class of elastic energy functionals of the deformation tensor F (see
[14]). In terms of the strain matrix V , the momentum equation takes the form of

ut + u · ∇u + ∇p = µ∆u + ∇ · (V V T ) + 2∇ · V (2.3)

where we used the equation (2.1) and (2.2). We point out here that through this paper,
∇ with no indices stands for the derivative with respect to the spatial coordinate x.
For a matrix-valued function A, (∇ · B)i means ∇j Bij , where summation over repeated
indices will always be well understood.
In order to get a dynamical equation for V , we plug (2.1) into the third equation
of system (1.1). Denote the material derivatives of a quantity Q by

Q̇ = Qt + u · ∇Q.

We have
V̇ R + (I + V )Ṙ = ∇u(I + V )R.
Noting that R is orthogonal, we deduce that

V̇ + (I + V )ṘRT = ∇u(I + V ). (2.4)

By transposing the above equality (2.4), it is rather easy to see

V̇ + RṘT (I + V ) = (I + V )∇uT . (2.5)

Now let us write the orthogonal matrix R as

 
cos θ − sin θ
R= .
sin θ cos θ

It is immediately verified that

    
T − sin θ − cos θ cos θ sin θ 0 −1
ṘR = θ̇ = θ̇.
cos θ − sin θ − sin θ cos θ 1 0

Define  
0 −1
A= , (2.6)
1 0
we have
ṘRT = Aθ̇, V A + AV = AtrV AT = A−1 = −A. (2.7)
Thus, it follows from (2.4), (2.5) and (2.7) that

(I + V )Aθ̇ − AT (I + V )θ̇ = ∇u(I + V ) − (I + V )∇uT

4
which is equivalent to
1
θ̇I = − A[∇u(I + V ) − (I + V )∇uT ]
2 + trV
1
= − [2ω12 (u) + ∇k u1 Vk2 − ∇k u2 Vk1]I
2 + trV
= −ω12 (u)I + γI,
where
1
ω(u) = (∇u − ∇uT ) (2.8)
2
is the virticity tensor and
1
γ= [trV ω12 (u) − (∇k u1 Vk2 − ∇k u2Vk1 )]. (2.9)
2 + trV
Consequently, we obtain that
θ̇ = −ω12 (u) + γ. (2.10)
Next, add up (2.4) and (2.5) and then insert (2.7) and (2.10) into the resulting
equation, we obtian
1
V̇ = [∇u(I + V ) + (I + V )∇uT ] (2.11)
2
1
− [(I + V )ṘRT + RṘT (I + V )]
2
1 1
= (∇u + ∇uT ) + (∇uV + V ∇uT )
2 2
1
+ [(I + V )A + AT (I + V )]ω12 (u)
2
1
− γ[(I + V )A + AT (I + V )]
2
1 1
= (∇u + ∇uT ) + (∇uV + V ∇uT )
2 2
1 1
+ ω12 (u)(V A − AV ) − γ(V A − AV ),
2 2
where in the last equality we used (2.7). If we denote the symmetric part of the gradient
of velocity by
1
D(u) = (∇u + ∇uT ), (2.12)
2
we can rewrite the above equation (2.11) as
1
Vt + u · ∇V = D(u) + (∇uV + V ∇uT ) (2.13)
2
1 1
+ ω12 (u)(V A − AV ) − γ(V A − AV ).
2 2
5
 In order to formulate an incompressible dynamical problem, the equations of mass
conservation ∇ · u = 0 and balance of momentum (2.3) supplemented by the transport
equation (2.13) have in fact made up of a closure system. However, to prove the global
existence, we must include an equation for ∇θ to carry out some intrinsic properties
of the rotation-strain viscoelastic system, which will be seen more clear in section 4.
Let  
⊥ −∇2
∇ = (2.14)
∇1
and then apply ∇⊥ to (2.10) to yield
 
⊥ ⊥1 −∇2 u · ∇θ
∇ θt + u · ∇∇ θ = ∆u − + ∇⊥ γ. (2.15)
2 ∇1 u · ∇θ

This equation will be used to explore the weak dissipation of ∇θ, which eventually
shows that V is also weakly dissipative. The equations (2.3), (2.13) and (2.15) form
the rotation-strain model (1.2) for viscoelastic fluids. For future use, we also need an
equation of ∇θ. For this purpose, we apply the gradient operator ∇ to (2.10) to get
 
∇1 u · ∇θ
∇θt + u · ∇∇θ = − − ∇ω12 (u) + ∇γ. (2.16)
∇2 u · ∇θ

We will consider the Cauchy problem or the periodic boundary-value problem of

system (1.2). The initial data takes of the form

u(0, x) = u0 (x), V (0, x) = V0 (x), ∇θ(0, x) = ∇θ0 (x), x ∈ Ω. (2.17)

where Ω ⊆ R2 is two dimensional torus or the entire space. The above initial data will
be imposed on the following constraints


∇ · u0 = 0,
det(I + V0 ) = 1, (2.18)


∇ · V0 = A(I + V0 )∇θ0 .

The first two are just the consequences of the incompressibility and the last one is
understood as the consistency condition for changing of variables (see Lemma 3.1,
Lemma 3.3, Remark 3.2 and Remark 3.4 in the next section).

Remark 2.1. In our earlier work [13], the rotation matrix R satisfies a spontaneous
but specified transport equation:

Rt + u · ∇R = −ARω12 (u) = ω(u)R.

Consequently, it follows that

θt + u · ∇θ = −ω12 .

6
This combining the third equation of system (1.1) gives a non-symmetric transport
equation for V . While in the model (1.2), noting (2.7) and with the aid of (2.10), it is
rather easy to find that

Rt + u · ∇R = ARθ̇ = −ARω12 (u) + γAR. (2.19)

The last term of the right side of the above equation or the one γ in (2.10) repre-
sents corrections of our previous model, which leads to that the strain matrix must
be symmetric. However, the corrections reflect the intrinsic properties of motions of
viscoelastic fluids and the underlying physical origins.

For the rotation-strain model (1.2), we will prove the following theorem:

Theorem 2.2. Consider the Cauchy problem or the periodic initial-boundary value
problem for the rotation-strain viscoelastic model (1.2) and (2.17) with the intrinsic
physical constraints (2.18) on the initial data. Then there exists a unique global classical
solution (u, V, ∇θ) which satisfies
Z ∞h
2 2 1 ⊥ 2
kukH 2 (Ω) + kV kH 2 (Ω) + k∆u + ∇ θk + µ k∇uk2H 2
µ 0
1 ⊥ 2 1 i µ4 (1 + µ4 )
2
+ k∇(∆u + ∇ θ)k + k∇(∆u + ∇ · V )k dt ≤ 2 ,
µ µ C (1 + µ10 )

if the initial data u0 , V0 ∈ H 2 (Ω) and ∇θ0 ∈ H 1 (Ω) and

µ8
ku0 k2H 2 (Ω) + kV0 k2H 2 (Ω) + k∇θ0 k2H 1 (Ω) < ,
M(1 + µ10 )

where Ω ⊆ R2 is two dimensional torus or the entire space, C and M (M > 2C 3 ) are
big enough constants independent of t and µ.

The proof of the above theorem relies on a local existence theorem and a priori
energy estimates. The proof of local existence is standard and can be similarly done
as in [14, 16], thus is omitted here. Below we will only present the a priori estimate
in Theorem 2.2.

Remark 2.3. In the case of Ω = T2 being a two-dimensional torus, we in fact recovered

the results in our earlier work [14]. To see this, we set
Z
1
θ0 = 2 θ0 dx.
|T | T2

By Poincaré’s inequality, we have

kθ0 − θ0 kL∞ ≤ Ckθ0 − θ0 kH 2 ≤ Ck∇θ0 kH 1 . (2.20)

7
Thus, if we define  
cos θ0 − sin θ0
R0 = ,
sin θ0 cos θ0
then  
F (0, x) = (I + V )R t=0 = R0 + E0
where  
E0 = R − R0 + V R t=0
is a small disturbance near a constant equilibrium R0 by (2.20), since k∇θ0 kH 1 is
assumed to be small in Theorem 2.2. In [14], we studied the general viscoelastic model
including an equation for the deformation tensor F and proved the global existence of
classical solutions near constant equilibrium in both two and three-dimensional spaces.
However, our results and proofs here are still of importance even in the periodic case
because they provide us a better understanding of the physical background of the
system.

3 Special Structures of the System

In this section, we will explore some of the intrinsic properties of the viscoelastic system
mentioned in section 2. These properties reflect the underlying physical origins of the
problem and in the meantime, and are also essential to the proof of the global existence
result here.
The following lemma will be used to explore the weak dissipation of V .
Lemma 3.1. Assume that the second equality of (2.18) is satisfied and (u, V ) is the
solution of system (1.2). Then the following is always true:

det(I + V ) = 1 (3.1)

for all the latter time t ≥ 0. In other words, we have

trV = − det V. (3.2)

P roof. Using the identity that for any non-singular matrix F ,

∂ det F
= det F F −T ,
∂F
we have

(det(I + V ))t + u · ∇(det(I + V )) (3.3)

= det(I + V )(I + V )−1
ji V̇ij
(   )
V22 −V12
= tr V̇ + V̇ .
−V12 V11

8
Return to (2.11), we in fact have that
1 1
V̇ = D(u) + (∇uV + V ∇uT ) − (V A − AV )θ̇. (3.4)
2 2
By (3.3), (3.4), we arrive at

(det(I + V ))t + u · ∇(det(I + V )) (3.5)

(   )
1 V22 −V12
= trD(u) − tr(V A − AV )θ̇ + tr ∇uV + D(u)
2 −V12 V11
(  ) (  )
V22 −V12 V22 −V12
+ tr ∇uV + tr (V A − AV ) .
−V12 V11 −V12 V11

We will compute the right side of the above equality term by term. First of all,
with the use of the first equation of (1.2), it is rather easy to see that

trD(u) = 0. (3.6)

Making use of (2.6), we deduce that

 
2V12 V22 − V11
tr(V A − AV ) = tr = 0. (3.7)
V22 − V11 −2V12

According to the expression of the Cauchy strain tensor (2.12), the third term of (3.5)
can be calculated as
(   )
V22 −V12
tr ∇uV + D(u) (3.8)
−V12 V11
= ∇1 u1 V11 + ∇2 u1 V12 + ∇1 u2 V12 + ∇2 u2 V22
+ V22 ∇1 u1 − V12 (∇1 u2 + ∇2 u1 ) + V11 ∇2 u2
= trV ∇ · u = 0.

Similarly, by a straightforward computation, we find that

(  )
V22 −V12
tr ∇uV (3.9)
−V12 V11
(  )
V22 −V12 ∇1 u1 V11 + ∇2 u1 V12 ∇1 u1 V12 + ∇2 u1 V22
= tr
−V12 V11 ∇1 u2 V11 + ∇2 u2 V12 ∇1 u2 V12 + ∇2 u2 V22
= V22 (∇1 u1 V11 + ∇2 u1 V12 ) + V11 (∇1 u2 V12 + ∇2 u2 V22 )
− V12 [(∇1 u2 V11 + ∇2 u2 V12 ) + (∇1 u1 V12 + ∇2 u1 V22 )]
= V11 V22 ∇ · u + V12 (V22 ∇2 u1 + V11 ∇1 u2 )
− V12 [∇1 u2 V11 + ∇2 u1 V22 + V12 ∇ · u]
= 0.

9
It remains to deal with the last term of (3.5). Recall (2.6) once more, it follows from
a simple calculation that
(  )
V22 −V12
tr (V A − AV ) (3.10)
−V12 V11
(  )
V22 −V12 2V12 V22 − V11
= tr
−V12 V11 V22 − V11 −2V12
= 2V22 V12 − 2V12 (V22 − V11 ) − 2V11 V12 = 0.
Finally, combining the above equalities (3.5) with (3.6)-(3.10), we arrive at
(det(I + V ))t + u · ∇(det(I + V )) = 0. (3.11)
This completes the proof Lemma 3.1.
Remark 3.2. The incompressibility can be exactly represented as
det F = 1. (3.12)
The usual incompressible condition ∇ · u = 0, the first equation in (1.2), is the direct
consequence of this identity. On the other hand, det(I + V ) = 1 is also a direct
conclusion of (3.12) since R is an orthogonal matrix. However, the above Lemma
illustrates the incompressible consistency of the the system (1.2).
To show the global existence of the classical small strain solutions to system (1.2),
we still need to show that the pressure is a high order term, which will be seen more
clear in the next section. To do so, we will use the following lemma to find out the
weak dissipation of ∇θ, which will play an important role for our proof below.
Lemma 3.3. Assume that the third equality of (2.18) is satisfied. Then any solution
(u, V ) of the system (1.2)-(2.17) will satisfy the following identity
∇ · V = A(I + V )∇θ (3.13)
for all the latter time t ≥ 0.
P roof. We rewrite the third equation of (1.2) as
1 1
Vt + u · ∇V = D(u) + (∇uV + V ∇uT ) − (V A − AV )θ̇, (3.14)
2 2
and then apply ∇· to the resulting equation (3.14) to yield
∇ · Vt + u · ∇∇ · V (3.15)
 
1 1 ∇j u · ∇V1j 1
= ∆u + AV ∇θ̇ − + ∇ · (∇uV + V ∇uT )
2 2 ∇j u · ∇V2j 2
1 1
− V A∇θ̇ − ∇ · (V A − AV )θ̇
2 2  
1 1 ∇j u · ∇V1j 1
= ∆u + AV ∇θ̇ − + ∇ · (∇uV + V ∇uT )
2 2 ∇ j u · ∇V 2j 2
1 1
− V A∇θ̇ − ∇ · (V A + AV )θ̇ + A∇ · V θ̇.
2 2
10
By the first equation of (1.2) and a straightforward computation, we obtain
 
∇j u · ∇V1j 1 1
− + ∇ · (∇uV + V ∇uT ) = ∇ · (∇uV − V ∇uT ). (3.16)
∇j u · ∇V2j 2 2
In view of (2.10), we can rewrite the last term of (3.15) as
A∇ · V θ̇ = [ω(u) + γA]∇ · V. (3.17)
Thus, it follows from (3.15), (3.16) and (3.17) that
∇ · Vt + u · ∇∇ · V (3.18)
1 1 1
= ∆u + AV ∇θ̇ + [ω(u) + γA]∇ · V − V A∇θ̇
2 2 2
1 1
+ ∇ · (∇uV − V ∇uT ) − ∇ · (V A + AV )θ̇.
2 2
On the other hand, combining (2.16) with (3.14), and splitting some terms of the
resulting expression, we conclude that
∂t [A(I + V )∇θ] + u · ∇[A(I + V )∇θ] (3.19)
1 1
= AD(u)∇θ + A(∇uV + V ∇uT )∇θ − (AV A + V )θ̇∇θ
2  2

∇1 u · ∇θ
+ A(I + V )∇θ̇ − A(I + V )
∇2 u · ∇θ
1 1 1
= A∇θ̇ + AV ∇θ̇ + AV ∇θ̇ + AD(u)∇θ + A(∇uV + V ∇uT )∇θ
2 2   2 
1 ∇1 u · ∇θ ∇1 u · ∇θ
− (AV A + V )θ̇∇θ − A − AV .
2 ∇2 u · ∇θ ∇2 u · ∇θ
In view of (2.10) and the first equation of system (1.2), we obtain, after a straightfor-
ward calculation,
1
A∇θ̇ = ∆u + A∇γ. (3.20)
2
It is rather easy to see that
 
∇1 u · ∇θ
AD(u)∇θ − A = ω(u)A∇θ (3.21)
∇2 u · ∇θ
and
 
1 ∇1 u · ∇θ 1
A(∇uV + V ∇uT )∇θ − AV = A(∇uV − V ∇uT )∇θ. (3.22)
2 ∇2 u · ∇θ 2
Using (2.7) and (2.10), we find that
1
− (AV A + V )θ̇∇θ (3.23)
2
1
= − A(V A + AV )θ̇∇θ − (−ω12 (u) + γ)V ∇θ
2
1
= − A(V A + AV )θ̇∇θ + Aω(u)V ∇θ − γV ∇θ.
2
11
Combining (3.19)-(3.23), we have
∂t [A(I + V )∇θ] + u · ∇[A(I + V )∇θ] (3.24)
1 1 1
= ∆u + AV ∇θ̇ + ω(u)A(I + V )∇θ − A(V A + AV )θ̇∇θ
2 2 2
1 1
+ A∇γ + A(∇uV − V ∇uT )∇θ + AV ∇θ̇ − γV ∇θ
2 2
Now, we subtract (3.24) from (3.18), and rearrange the right side terms of resulting
equation. The outcome of this straightforward calculations is
∂t [∇ · V − A(I + V )∇θ] + u · ∇[∇ · V − A(I + V )∇θ] (3.25)
1 1
= ω(u)[∇ · V − A(I + V )∇θ] − (V A + AV )∇θ̇ − ∇ · (V A + AV )θ̇
2 2
1 T
+ ∇ · (∇uV − V ∇u ) − A∇γ + γA∇ · V
2
1 1
+ [ A(V A + AV )θ̇ − A(∇uV − V ∇uT ) + γV ]∇θ
2 2
1
= ω(u)[∇ · V − A(I + V )∇θ] − ∇ · [(V A + AV )θ̇]
2
1 T
+ ∇ · (∇uV − V ∇u ) − A∇γ + γA∇ · V
2
1 1
+ [ A(V A + AV )θ̇ − A(∇uV − V ∇uT ) + γV ]∇θ.
2 2
Our objective is to show that the expression ∇·V −A(I+V )∇θ satisfies some differential
equation whose solution can be uniquely determined by the initial data. Recall (2.7)
and (2.10), we get
1 1
− ∇ · [(V A + AV )θ̇] + ∇ · (∇uV − V ∇uT ) − A∇γ + γA∇ · V
2 2
1
= − ∇ · [trV A(−ω12 (u) + γ)] − A∇γ + γA∇ · V
2
1
+ ∇ · (∇uV − V ∇uT )
2
1 1
= A∇[trV ω12 (u)] − A∇[(trV + 2)γ)] + γA∇ · V
2 2
1
− ∇ · [A(∇k u1 Vk2 − ∇k u2 Vk1)]
2
1
= A∇[trV ω12 (u) − (∇k u1 Vk2 − ∇k u2 Vk1)] + γA∇ · V
2
1
− A∇[(trV + 2)γ)].
2
Thus, by (2.9), we can easily deduce that
1 1
− ∇ · [(V A + AV )θ̇] + ∇ · (∇uV − V ∇uT ) (3.26)
2 2
− A∇γ + γA∇ · V = γA∇ · V.

12
On the other hand, it follows from (2.7), (2.9) and (2.10) that
1 1
A(V A + AV )θ̇ − A(∇uV − V ∇uT ) + γV (3.27)
2 2
1
= −γAA(I + V ) − γI − trV [−ω12 (u)
2
1
+ γ]I − A(∇uV − V ∇uT )
2
1
= −γAA(I + V ) − [−trV ω12 (u) + γ(trV + 2)]I
2
1
− A(∇uV − V ∇uT )
2
1
= −γAA(I + V ) − (∇k u1 Vk2 − ∇k u2 Vk1)I
2
1
+ A(∇uV + V ∇uT )
2
= −γAA(I + V ).

Hence we conclude from (3.25)-(3.27) that

∂t [∇ · V − A(I + V )∇θ] + u · ∇[∇ · V − A(I + V )∇θ] (3.28)

= [ω(u) + γA][∇ · V − A(I + V )∇θ].

which concludes the proof of the lemma, since the above quantity will remain zero all
the time with zero initial data.
Remark 3.4. In order to carry out the mechanical background of the above lemma,
we again go back to the deformation tensor F . By its definition, the following identity

∇ · FT = 0

is proved in [14, 15]. Insert (2.1) into the above equality, we find

∂i Rij + Rkj ∂i Vik + Vik ∂i Rkj = 0

which implies
∂i Vil + Rlj ∂i Rij + Rlj Vik ∂i Rkj = 0
since R is an orthogonal matrix. But then employing (2.2) a straightforward calculation
yields (3.13).

4 Proof of Theorem 2.2

In this section we prove Theorem 2.2. Weak dissipations on the strain V and the
gradient of rotation angle ∇θ are found by introducing auxiliary functions w and Θ
below. The way of defining such functions reveals the intrinsic dissipative nature of

13
the system. In what follows, k · k is used to denote the standard L2 (R2 ) norm, < ·, · >
and (·, ·) the Rd and L2 (R2 )d inner products with d = 1, 2 or 4. The proof is divided
into four steps.
Step 1. Standard Energy Estimates.
By taking the L2 inner product of the second and third equations of system (1.2)
with u and V , and then adding up the resulting equations together, we obtain
Z
1d
(|u|2 + |V |2 ) dx + µk∇uk2
2 dt R2
Z
|u|2 + |V |2
=− u·∇ + < u, ∇p > dx
R2 2
Z
1 1
+ < V, (∇uV + V ∇uT ) + ω12 (u)(V A − AV ) > dx
2 2 2
ZR
+ < u, 2∇ · V > + < V, D(u) > dx
R 2
Z Z
T 1
+ < u, ∇ · (V V ) > dx − < V, γ(V A − AV ) > dx.
R2 R2 2
On the other hand, the fourth equation of system (1.2) gives
Z Z Z
1d
2trV dx = − u · ∇trV dx + trD(u) dx
2 dt R2 R2 R 2
Z
1

+ tr ∇uV + (V A − AV )(ω12 (u) − γ) dx.
R2 2
Adding up the above two equation together, we have
Z
1d
(|u|2 + |V |2 + 2trV ) dx + µk∇uk2 (4.1)
2 dt R2
Z Z
|u|2 + |V |2

=− u·∇ + trV + < u, ∇p > dx + trD(u) dx
R 2 2 R2
Z
1 1
+ < V, (∇uV + V ∇uT ) + ω12 (u)(V A − AV ) > dx
2 2 2
ZR Z
+ < u, 2∇ · V > + < V, D(u) > dx + tr(∇uV ) dx
R 2 R 2
Z Z
T 1
+ < u, ∇ · (V V ) > dx − < V, γ(V A − AV ) > dx
2 R2 2
ZR
1

+ tr (V A − AV )(ω12 (u) − γ) dx.
R2 2
Noting that V is symmetric, and with the aid of the first equation of (1.2), it is
rather easy to see that
( R R
− R2 u · 21 (|u|2 + |V |2 + 2trV )+ < u, ∇p > dx + R2 trD(u) dx = 0,
R R (4.2)
R2
< u, 2∇ · V > + < V, D(u) > dx + R2 tr(∇uV ) dx = 0.

14
Noting the definition in (2.6) and (2.9), one has
Z
1 1
< V, (∇uV + V ∇uT ) + ω12 (u)(V A − AV ) > dx (4.3)
R2 Z 2 2
Z
T 1
+ < u, ∇ · (V V ) > dx − < V, γ(V A − AV ) > dx
2 R2 2
ZR
1

+ tr (V A − AV )(ω12 (u) − γ) dx
2 2
Z R
= < V, ∇uV > + < u, ∇ · (V V T ) > dx
R2Z
1

+ (ω12 (u) − γ) tr(V A − AV )+ < V, V A − AV > dx
R2 2
= 0.

Then, combining (4.1), (4.2) with (4.3), we have

Z
1d
(|u|2 + |V |2 + 2trV ) dx + µk∇uk2 = 0,
2 dt R2

Finally, by using (3.2), we arrive at

Z
1d
(|u|2 + |V11 − V22 |2 + |V12 + V21 |2 ) dx + µk∇uk2 = 0. (4.4)
2 dt R2

By taking the L2 inner product of the second and third equations of system (1.2)
with u and 2V , and then adding up the resulting equations together, we obtain
Z Z
1d 2 2 |u|2
|u| + 2|V | dx + u · ∇( + |V |2 )dx + (u, ∇p)
2 dt R2 R2 2
 
T
= V, ∇uV + V ∇u + ω12 (u)(V A − AV ) − γ(V A − AV )



+ u, ∇ · (V V T ) + (u, 2∇ · V ) + 2V, D(u) + µ(u, ∆u),

which gives that

Z
1d
|u|2 + 2|V |2 dx + µk∇uk2 (4.5)
2 dt R2
≤ k∇ukL∞ kV k2
≤ Ck∇V kk∆V k(k∆∇uk + k∇uk)
1

≤ Cµk∇V kL2 k∇uk2 + k∆∇uk2 + 2
k∆V k .
µ
In order to exploit the higher order energy estimates, we apply ∆ to the second
and third equations of system (1.2), and then take L2 inner product of the resulting

15
equations with ∆u and 2∆V , respectively, to yield
Z
1d
|∆u|2 + 2|∆V |2 dx + (∆u, ∇∆p) (4.6)
2 dt R2
Z
|∆u|2
+ u · ∇( + |∆V |2 )dx
2 2
 R 
= ∆V, ∆ ∇uV + V ∇uT + ω12 (u)(V A − AV )


− γ(V A − AV ) + ∆u, ∆∇ · (V V T ) + (∆u, 2∆∇ · V )

  
+ 2∆V, ∆D(u) + µ(∆u, ∆∆u) − ∆u, ∆(u · ∇u) − u · ∇∆u
  
− 2∆V, ∆(u · ∇V ) − u · ∇∆V .

Applying similar arguments as before, we conclude that

R |∆u|2
 2
 R2 u · ∇( 2 + |∆V | )dx + (∆u, ∇∆p)
− µ(∆u, ∆∆u) = µk∆∇uk2 , (4.7)



(∆u, 2∆∇ · V ) + 2∆V, ∆D(u) = 0.

To estimate the rest terms of the right side of (4.6), and also for later use, we
need the following Lemmas. Their proofs can found in many literatures, see [14], for
example.
Lemma 4.1. Assume v ∈ W k,2(R2 ), k ≥ 3. The following interpolation inequalities
hold.
1. For 1 ≤ s ≤ k,
1 1
kvkL4 ≤ Ckvk1− 2s k∇s vk 2s ,
3 3
k∇vkL4 ≤ Ckvk1− 2(s+1) k∇s ∇vk 2(s+1) ,
5 5
k∆vkL4 ≤ Ckvk1− 2(s+2) k∇s ∆vk 2(s+2) .

2. For 2 ≤ s ≤ k,
1 1
kvkL∞ ≤ Ckvk1− s k∇s vk s ,
2 2
k∇vkL∞ ≤ Ckvk1− s+1 k∇s ∇vk s+1 .

The following Lemma will also be used for several times below.
Lemma 4.2. Assume f, g ∈ H s (R2 ) for some s > 0. Then there exists a universal
constant C depending only on s such that

k∇s (f g)k ≤ C(kf kL∞ k∇s gk + kgkL∞ k∇s f k).

16
 Now we estimate the second line of (4.6). Invoking the definition of γ in (2.9), and
with the aid of Lemma 4.2, we deduce that
  
 T
 ∆V, ∆ ∇uV + V ∇u + ω12 (u)(V A − AV ) − γ(V A − AV )  (4.8)

≤ Ck∆V k k∆V kk∇ukL∞ + k∇∆ukkV kL∞ + k∇∆ukkV k2L∞
 1 
+ k∇ukL∞ kV kL∞ k∆V k + kV k2L∞ k∆ k
2 + trV
1
≤ CµkV kH 2 (k∇uk2 + k∇∆uk2 + 2 k∆V k2 )
µ
provided that
kV kL∞ < 1, (4.9)
which will be verified below.
Similarly, it is easy to show that


 ∆u, ∆∇ · (V V T )  (4.10)
≤ kV kL∞ k∆V kk∇∆uk
1
≤ CµkV kH 2 (k∇∆uk2 + 2
k∆V k2 ).
µ
At last, it remains to estimate the last line of (4.6). Again, by Sobolev imbedding
theorem and Lemma 4.2, it leads to
      

 − ∆u, ∆(u · ∇u) − u · ∇∆u − 2∆V, ∆(u · ∇V ) − u · ∇∆V  (4.11)


≤ Ck∆uk k∇ukL∞ k∆uk + k∇∇ukkukL∞


+ Ck∆V k k∇ukL∞ k∆V k + k∇∇ukkV kL∞
1
≤ C(1 + µ)(kukH 2 + kV kH 2 )(k∇uk2 + k∇∆uk2 + 2 k∆V k2 ).
µ
Combining all these estimates (4.7)-(4.11) with (4.5), and then noting (4.6), we can
finally conclude that
Z
1d


|u|2 + |∆u|2 + 2|V |2 + 2|∆V |2 dx + µ k∇uk2 + k∇∆uk2 (4.12)
2 dt R2
1

≤ C(1 + µ)(kukH 2 + kV kH 2 ) k∇uk2 + k∇∆uk2 + 2 k∆V k2
µ
provided (4.9) holds.

Step 2. Weak Dissipation for The Strain Matrix V .

To get dissipative energy estimates, by (4.12), it is clear that we should explore the
dissipation of V .

17
 Let
2
w = ∆u + ∇ · V. (4.13)
µ
Now we apply ∆ to the second equation of system (1.2), and respectively, µ1 ∇· to
the third one, and then adding up the outcomes to get a new equation. In terms of w
introduced in (4.13), this new equation can be written as
1
wt + u · ∇w + ∇∆p = µ∆w + ∆u + f. (4.14)
µ
where
  2
f = − ∆(u · ∇u) − u · ∇∆u − ∇ · (u · ∇V ) (4.15)
µ
 1 
− u · ∇(∇ · V ) + ∆∇ · (V V T ) + ∇ · (∇uV + V ∇uT )
µ

+ ω12 (u)(V A − AV ) − γ(V A − AV ) .

Take the inner product of (4.14) with w in (L2 (R2 ))2 , and use the similar arguments
as in (4.2), we have
Z
1d 1
|w|2dx + µk∇wk2 = (∆p, ∇ · w) + (∆u, w) + (f, w). (4.16)
2 dt R2 µ

As before, we should estimate the right side of (4.16) term by term. Firstly, it is
obvious that
1 µ 1
|(∆u, w)| ≤ k∇wk2 + 3 k∇uk2. (4.17)
µ 4 µ
On the other hand, noting that ∇ · u = 0, we can use integration by parts and
Lemma 4.2 to estimate the last term of the right side of (4.16) as follows:
   

 
|(f, w)| ≤  ∆(ui uj ) − ui ∆uj , ∇i wj  +  ∆(V V T ), ∇w  (4.18)
1  
+  (∇uV + V ∇uT ) + ω12 (u)(V A − AV ) − γ(V A − AV ) ,
µ
 2   
  
∇w  +  ∇k · (ui Vjk ) − ui (∇k Vjk ) , ∇i wj 
µ
h

≤ Ck∇wk kukL∞ k∆uk + k∇ukL∞ k∇uk
1 1 i
2
+ kV kL k∇uk + kV kL k∆V k + kV kL∞ k∇uk
∞ ∞
µ µ
1
≤ C(µ + )(kukH 2 + kV kH 2 )
µ
1

× k∇uk2 + k∆∇uk2 + k∇wk2 + 2 k∆V k2
µ

18
provided kV kL∞ < 1.
It also remains to estimate the first term of the right side of (4.16). For this purpose,
we apply the divergence operator for the momentum equation of system (1.2) to get

∆p = −tr(∇u∇u) + ∇ · [∇ · (V V T )] + 2∇ · (∇ · V ). (4.19)

Thus, by (4.19), Lemma 4.1, Lemma 4.2 and Sobolev imbedding theorem, we can easily
deduce that


|(∆p, ∇w)| ≤  tr(∇u∇u), ∇w  (4.20)
  

 
+  ∇ · [∇ · (V V T )], ∇w  + 2 ∇ · (∇ · V ), ∇w 
 
2
≤ Ck∇wk k∇ukL4 + kV kL k∆V k∞

µ 1
+ k∇wk2 + k∇ · (∇ · V )k2
4 µ
≤ C(1 + µ)(kukH 2 + kV kH 2 ) k∇uk2 + k∇∆uk2
1
µ 1
+ k∇wk2 + 2 k∆V k2 + k∇wk2 + k∇ · (∇ · V )k2 .
µ 4 µ
We must deal with the last term of the above estimates. Now Lemma 3.3 enters the
argument in this stage. Noting the definition of ∇⊥ , it is easy to see that ∇ · ∇⊥ = 0.
Thus, by Lemma 4.1 and a straightforward calculation, we have
2 2
k∇ · (∇ · V )k2 = k∇ · (AV ∇θ)k2 (4.21)
µ µ
C
2
2
≤ k∇V kL4 k∇θkL4 + kV kL∞ k∇2 θk
µ
C C
≤ k∇V kk∇θkk∆V kk∆θk + kV k2L∞ k∆θk2
µ µ
1 1
≤ C(1 + µ)(kV kH 2 + kukH 2 )(k∇∆uk2 + 2 k∆θk2 + 2 k∆V k2 )
µ µ

provided (4.9) is satisfied and

k∇θk ≤ 1 (4.22)
holds.
Finally, the combination of (4.16)-(4.21) leads to
Z
d
|w|2 dx + µk∇wk2 (4.23)
dt R2
1
≤ C(µ + )(kukH 2 + kV kH 2 ) k∇uk2 + k∆∇uk2
µ
1 1
2
+ k∇wk2 + 2 k∆V k2 + 2 k∆θk2 + 3 k∇uk2
µ µ µ

19
provided (4.22) holds.

Step 3. Weak Dissipation for ∇θ.

At this stage, to extract the dissipative nature of the system, it is clear that we
need also explore some kind of weak dissipation of ∆θ. For this purpose, we introduce
the auxiliary variable Θ:
2
Θ = ∆u + ∇⊥ θ. (4.24)
µ
Rewrite the momentum equation as follows:

ut + u · ∇u + ∇p = µ∆u + 2∇⊥ θ + ∇ · (V V T ) + 2AV ∇θ. (4.25)

Then, apply ∆ to (4.25) and µ2 ∇· to the last equation of (1.2), and then add up the
resulting equations together, we obtain a new equation in terms of Θ:
1
Θt + u · ∇Θ + ∇∆p = ∆Θ + ∆u + g. (4.26)
µ
where
2
g = ∆∇ · (V V T ) + 2∆(AV ∇θ) + ∇⊥ γ (4.27)
µ
 
  2 −∇2 u · ∇θ
− ∆(u · ∇u) − u · ∇∆u − .
µ ∇1 u · ∇θ

Similarly, we take the L2 inner product of (4.26) with Θ to get

Z
1d 1
|Θ|2dx + µk∇Θk2 = (∆p, ∇ · Θ) + (∆u, Θ) + (g, Θ). (4.28)
2 dt R2 µ

By the expression of Θ in (4.24) and the first equation of system (1.2), it is easy to see
that
(∆p, ∇ · Θ) = 0. (4.29)
A straightforward computation implies
1   µ 1
(∆u, Θ) ≤ k∇Θk2 + 3 k∇uk2. (4.30)
µ 4 µ

To estimate the last term in (4.28), we insert (4.27) into (g, Θ) and then use inte-
gration by parts to yield
   2  −∇ θ∇ u 
   2 2 
|(g, Θ)| ≤  ∆(ui uj ) − ui ∆uj , ∇i Θj  +  , Θ  (4.31)
µ ∇1 θ∇1 u

 2 
 

+  ∆(V V T ), ∇Θ  +  γ, ∇⊥ · Θ  + 2 ∆(AV ∇θ), Θ .
µ

20
By a similar argument as before, we have
   
 

 
 ∆(ui uj ) − ui ∆uj , ∇i Θj  +  ∆(V V T ), ∇Θ  + 2 ∆(AV ∇θ), Θ  (4.32)
h
i
≤ Ck∇Θk kukL∞ k∆uk + k∇ukL∞ k∇uk + kV kL∞ k∆V k


+ Ck∇Θk kV kL∞ k∆θk + k∇V kL4 k∇θkL4
3 1
≤ C(1 + µ 2 )(kukH 2 + kV kH 2 + k∇θk)
µ
1 1

× k∇uk2 + k∆∇uk2 + k∇Θk2 + 2 k∆V k2 + 2 k∆θk2 ,
µ µ

where we used Lemma 4.1 and Lemma 4.2. On the other hand, by (2.9), it follows that
  
2  −∇2 θ∇2 u  2

 , Θ  +  γ, ∇⊥ · Θ  (4.33)
µ ∇1 θ∇1 u µ
C

≤ k∇ukk∇θkL4 kΘkL4 + kV kL∞ k∇uk∇Θk
µ
C 1 1 1 1

≤ k∇ukk∇θk 2 kΘk 2 k∆θk 2 k∇Θk 2 + kV kL∞ k∇uk∇Θk
µ
1 1
≤ C(1 + )(kΘk + kV kH 2 + k∇θk)
µ µ
1

× k∇uk2 + k∇Θk2 + 2 k∆θk2 .
µ

The combination of (4.31)-(4.33) implies that

1 1
|(g, Θ)| ≤ C(µ2 +)(kukH 2 + kV kH 2 + kΘk + k∇θk) (4.34)
µ µ
1 1

× k∇uk2 + k∆∇uk2 + k∇Θk2 + 2 k∆θk2 + 2 k∆V k2 .
µ µ

Finally, combining (4.28)-(4.30) with (4.34), and noting the definition of Θ in (4.24),
we arrive at
Z
d
|Θ|2dx + µk∇Θk2
dt R2
C 1
≤ 3 k∇uk2 + C(µ2 + )(kukH 2 + kV kH 2 + kΘk)
µ µ
1

× k∇uk2 + k∆∇uk2 + k∇Θk2 + 2 k∆V k2 .
µ
Step 4. A P riori Weakly Dissipative Energy Estimates.

21
 This step ties everything together. Firstly, by Hodge’s decomposition, we calculate
∆V = ∇∇ · V − ∇ × ∇ × V (4.35)
 
∇2 (∇1 V12 − ∇2 V11 ) −∇1 (∇1 V12 − ∇2 V11 )
= ∇∇ · V −
∇2 (∇1 V22 − ∇2 V21 ) −∇1 (∇1 V22 − ∇2 V21 )
= ∇∇ · V + (∇⊥ )2 trV − ∇⊥ (A∇ · V ).
Thus, noting (3.2), and by (4.13) and Lemma 4.2, we obtain
k∆V k ≤ Cµ(k∆∇uk + k∇wk + k∆trV k) (4.36)
≤ Cµ(k∆∇uk + k∇wk) + Cµk∆V kkV kL∞
which implies that
k∆V k ≤ Cµ(k∆∇uk + k∇wk) (4.37)
provided that
1
kV kL∞ ≤ . (4.38)
2µC
On the other hand, by (4.24),
(
k∇θk ≤ Cµ(k∆uk + kΘk),
(4.39)
k∆θk ≤ Cµ(k∆∇uk + k∇Θk).
Now we can proceed with the dissipative energy method. Plug (4.39) into (4.12),
we find that
Z
1d


|u|2 + |∆u|2 + 2|V |2 + 2|∆V |2 dx + µ k∇uk2 + k∇∆uk2 (4.40)
2 dt R2


≤ C(1 + µ)(kukH 2 + kV kH 2 ) k∇uk2 + k∇∆uk2 + k∇wk2 .
Similarly, insert (4.37) and (4.39) into (4.23) and (4.35), and then add the resulting
inequalities together, we obtain that
Z
d


|w|2 + |Θ|2 dx + µ k∇wk2 k + ∇Θk2 (4.41)
dt R2
1
≤ C(µ2 + )(kukH 2 + kV kH 2 + kΘk) k∇uk2 + k∆∇uk2
µ
C
+ k∇wk2 + k∇Θk2 ) + 3 k∇uk2.
µ
Combining (4.40) with (4.41), we finally arrive at
Z h i
d

|u|2 + |∆u|2 + 2|V |2 + 2|∆V |2 + |w|2 + |Θ|2 dx (4.42)
dt R2


+ µ k∇uk2 + k∇∆uk2 + k∇wk2 + k∇Θk2
C 1
≤ 3 k∇uk2 + C(µ2 + )(kukH 2 + kV kH 2 + kΘk)
µ µ


× k∇uk + k∇∆uk + k∇wk2 + k∇Θk2 .
2 2

22
 Therefore, if the initial data is sufficiently small, we can find some positive time T ,
such that
µ2
kukH 2 + kV kH 2 + kΘk ≤ (4.43)
2C(1 + µ3 )
for all 0 ≤ t ≤ T . Moreover, in this case, (4.42) implies that
Z h i
d 2 2 2 2

2 2
|u| + |∆u| + 2|V | + 2|∆V | + |w| + |Θ| dx
dt R2

C
+ µ k∇uk2 + k∇∆uk2 + k∇wk2 + k∇Θk2 ≤ 3 k∇uk2.
µ
Thus, noting (4.13), (4.24), and by integrating the above inequality from 0 to t, 0 ≤
t ≤ T , with respect to t, we obtain


kuk2H 2 + kV k2H 2 + kwk2 + kΘk2 (4.44)
Z ∞


+µ k∇uk2H 2 + k∇wk2 + k∇Θk2 dt
0
Z
2 2 2 2 C ∞
≤ C(ku0kH 2 + kV0 kH 2 + kw0 k + kΘ0 k ) + 3 k∇uk2 dt
µ 0
1
≤ C(1 + 4 )(ku0k2H 2 + kV0 k2H 2 + k∇θ0 k2H 1 ),
µ
where in the last inequality, we used the basic energy law (4.4). Then, it follows from
(4.44) that if the initial data satisfies
µ8
ku0 k2H 2 + kV0 k2H 2 + k∇θk2H 1 ≤ (4.45)
M(1 + µ10 )
for a big enough constant M (M ≥ 2C 3 ) independent of µ, then (4.43) will still hold
for all the latter time t ≥ T with a uniform constant C independent of t and µ.
Furthermore, when (4.45) holds, we can deduce from (4.44) that


kuk2H 2 + kV k2H 2 + kwk2 + kΘk2 (4.46)
Z ∞


+µ k∇uk2H 2 + k∇wk2 + k∇Θk2 dt
0
Z
2 2 2 2 C ∞
≤ C(ku0kH 2 + kV0 kH 2 + kw0 k + kΘ0 k ) + 3 k∇uk2 dt
µ 0
4 4
µ (1 + µ )
≤ 2
C (1 + µ10 )
provided C is big enough.
At last, we should show (4.9), (4.22) and (4.38). By (4.43) and Sobolev imbedding
theorem, we have
cµ2
kV kL∞ ≤ ckV kH 2 ≤
2C(µ3 + 1)

23
for appropriate constants c and C. Thus, we have
(
1
≤ 1, for µ ≥ 1,
kV kL∞ ≤ µ 1 (4.47)
1 ≤ µ , for µ ≤ 1,

which means that (4.9) and (4.38) hold. To show (4.22), we deduce from (4.43) and
(4.24) that
µ3
k∇θk ≤ 2µ(kΘk + kukH 2 ) ≤ ≤ 1.
C(1 + µ3 )
The proof of Theorem 2.2 is completed.

Acknowledgments
The author would like to thank professor Chun Liu and professor Yi Zhou for their
constructive suggestions and helpful discussions. The author was partially supported
by the NSFC grant 10801029, the PSFC grants 20070410160 and 200801175.

References
[1] R. Agemi, Global existence of nonlinear elastic waves, Invent. Math., 142(2)
(2000), 225–250.

[2] J. Y. Chemin and N. Masmoudi, About lifespan of regular solutions of equations

related to viscoelastic fluids, SIAM J. Math. Anal., 33(1) (2001), 84–112.

[3] Y. Chen and P. Zhang, The global existence of small solutions to the incom-
pressible viscoelastic fluid system in 2 and 3 space dimensions. Comm. Partial
Differential Equations 31 (2006), no. 10-12, 1793–1810.

[4] D. Christodoulou, Global existence of nonlinear hyperbolic equations for small

data, Comm. Pure. Appl. Math., 39 (1986), 267–286.

[5] K. O. Friedrichs, On the boundary-value problems of the theory of Elasticity and

Korn’s inequality, Ann. Math., 48 (1947), 441–471.

[6] F. John, Rotation and strain, Comm. Pure Appl. Math., 14 (1961), 391–413.

[7] F. John, Distance changes in deformations with small strain. 1970 Studies and
Essays (Presented to Yu-why Chen on his 60th Birthday, April 1, 1970) pp. 1–15
Math. Res. Center, Nat. Taiwan Univ., Taipei

[8] F. John and S. Klainerman, Almost global existence to nonlinear wave equations
in three space dimensions, Comm. Pure Appl. Math., 37 (1984), 443–455.

24
 [9] P. Kessenich, Global existence with small initial data for three-dimensional
incompressible isotropic viscoelastic materials, preprint. available at
http://arxiv.org/abs/0903.2824v1
[10] S. Klainerman , The null condition and global existence to nonlinear wave equa-
tions, Lect. in Appl. Math., 23 (1986), 293–326.
[11] S. Klainerman and T. C. Sideris, On almost global existence for nonrelativistic
wave equations in 3D, Comm. Pure Appl. Math. 49 (1996), 307–322.
[12] Z. Lei, Global existence of classical solutions for some Oldroyd-B model via the
incompressible limit, Chin. Ann. Math. Ser. B, 27 (2006), no. 5, 565–580.
[13] Z. Lei, C. Liu and Y. Zhou, Global existence for a 2D incompressible viscoelastic
model with small strain. Comm. Math. Sci., 5 (2007), no. 3, 595–616.
[14] Z. Lei, C. Liu and Y. Zhou, Global solutions for incompressible viscoelastic fluids.
Arch. Ration. Mech. Anal., 188 (2008), no. 3, 371–398.
[15] Z. Lei and Y. Zhou, Global existence of classical solutions for 2D Oldroyd model
via the incompressible limit, SIAM J. Math. Anal., 37 (2005), no. 3, 797–814.
[16] F. H. Lin and C. Liu, Nonparabolic dissipative systems modelling the flow of liquid
crystals, Comm. Pure Appl. Math., 48(5) (1995), 501–537.
[17] F. H. Lin, C. Liu and P. Zhang, On hydrodynamics of viscoelastic fluids, Comm.
Pure Appl. Math., 58(11) (2005), 1437–1471.
[18] P. L. Lions and N. Masmoudi, Global solutions for some Oldroyd models of non-
Newtonian flows, Chinese Ann. Math. Ser. B 21(1) (2000), 131–146.
[19] C. Liu and N. J. Walkington, An Eulerian description of fluids containing visco-
hyperelastic particles, Arch. Rat. Mech Ana. 159 (2001), 229–252.
[20] T. C. Sideris, Nonresonance and global existence of prestressed nonlinear elastic
waves, Ann. of Math., 151 (2000), 849–874.
[21] T. C. Sideris and B. Thomases, Global existence for three-dimensional incom-
pressible isotropic elastodynamics via the incompressible limit. Comm. Pure Appl.
Math. 58 (2005), no. 6, 750–788.
[22] T. C. Sideris and B. Thomases, Local energy decay for solutions of multi-
dimensional isotropic symmetric hyperbolic systems. J. Hyperbolic Differ. Equ. 3
(2006), no. 4, 673–690.
[23] T. C. Sideris and B. Thomases, Global existence for three-dimensional incom-
pressible isotropic elastodynamics. Comm. Pure Appl. Math. 60 (2007), no. 12,
1707–1730.

25