J. Korean Math. Soc. 57 (2020), No. 1, pp.

215–247
https://doi.org/10.4134/JKMS.j190028
pISSN: 0304-9914 / eISSN: 2234-3008

GLOBAL EXISTENCE AND ASYMPTOTIC BEHAVIOR IN

A THREE-DIMENSIONAL TWO-SPECIES
CHEMOTAXIS-STOKES SYSTEM WITH
TENSOR-VALUED SENSITIVITY

Bin Liu and Guoqiang Ren

Abstract. In this paper, we deal with a two-species chemotaxis-Stokes

system with Lotka-Volterra competitive kinetics under homogeneous Neu-
mann boundary conditions in a general three-dimensional bounded do-
main with smooth boundary. Under appropriate regularity assumptions
on the initial data, by some Lp -estimate techniques, we show that the
system possesses at least one global and bounded weak solution, in addi-
tion to discussing the asymptotic behavior of the solutions. Our results
generalizes and improves partial previously known ones.

1. Introduction
This paper is concerned with the following two-species chemotaxis-Stokes
system with Lotka-Volterra competitive kinetics:
(1.1)


 (n1 )t + u · ∇n1 = ∇ · (D1 (n1 )∇n1 ) − ∇ · (n1 S1 (x, n1 , v) · ∇v)

+µ1 n1 (1 − n1 − a1 n2 ) x ∈ Ω, t > 0,





(n2 )t + u · ∇n2 = ∇ · (D2 (n2 )∇n2 ) − ∇ · (n2 S2 (x, n2 , v) · ∇v)





 +µ2 n2 (1 − a2 n1 − n2 ) x ∈ Ω, t > 0,


 v t + u · ∇v = 4v − v + αn1 + βn2 , x ∈ Ω, t > 0,
ut + ∇P = 4u + (n1 + n2 )∇φ, ∇ · u = 0, x ∈ Ω, t > 0,



 ∂v
(Di (ni )∇ni − ni Si (x, ni , v) · ∇v) · ν = ∂ν = 0, u = 0, x ∈ ∂Ω, t > 0,





n1 (x, 0) = n10 (x), n2 (x, 0) = n20 (x), v(x, 0) = v0 (x), u(x, 0) = u0 (x), x ∈ Ω,


Received January 9, 2019; Revised July 10, 2019; Accepted August 14, 2019.
2010 Mathematics Subject Classification. 35D30, 35K46, 35A01, 35Q92, 35B35, 92C17.
Key words and phrases. Chemotaxis-Stokes, boundedness, asymptotic behavior, global
existence.
This work was partially supported by NNSF of China (Grant No. 11971185).

c 2020 Korean Mathematical Society

215
216 B. LIU AND G. REN

∂
where Ω ⊂ R3 is a bounded domain with smooth boundary ∂Ω and ∂ν denotes
the derivative with respect to the outer normal of ∂Ω. The system (1.1) is pro-
posed to describe the exercise of two species which impact on a single chemoat-
tractant in fluid, where n1 and n2 represent densities of species, v stands for
the chemical concentration, u shows the fluid velocity field and P denotes the
pressure of the fluid. µ1 , µ2 , α, β > 0 are constants, n10 , n20 , v0 , u0 , φ are known
functions satisfying

ϑ
n10 , n20 ∈ C (Ω) for certain ϑ > 0 with n10 , n20 ≥ 0 in Ω,

1,∞
(1.2) v0 ∈ W (Ω) satisfies v0 ≥ 0 in Ω,
u0 ∈ D(Ar ) for some  ∈ ( 34 , 1) and any r ∈ (1, ∞)

 

with Ar representing the Stokes operator with domain D(Ar ) := W 2,r (Ω) ∩
W01,r (Ω) ∩ Lrσ (Ω), where Lrσ (Ω) := {ϕ ∈ Lr (Ω)|∇ · ϕ = 0} for r ∈ (1, ∞).
As for the diffusion coefficient in (1.1), we suppose that D satisfies
θ
(1.3) D1 , D2 ∈ Cloc ([0, ∞)) for some θ > 0,
as well as
1 −1
(1.4) D1 (n1 ) ≥ CD1 nm
1 for all n1 > 0 with m1 > 1 and CD1 > 0
and
2 −1
(1.5) D2 (n2 ) ≥ CD2 nm
2 for all n2 > 0 with m2 > 1 and CD2 > 0.
Under the assumptions of Di (ni ), the first two equations of system (1.1) may
be degenerate at ni = 0, i = 1, 2.
Except for this, we assume that the tensor-valued sensitivity S1 , S2 satisfies
(1.6) S1 , S2 ∈ C 2 (Ω × [0, ∞)2 ; R3×3 )
as well as
(1.7) |S1 (x, n1 , v)| ≤ CS1 (1 + n1 )−α1 , |S2 (x, n2 , v)| ≤ CS2 (1 + n2 )−α2
for all (x, ni , v) ∈ Ω × [0, ∞)2 , i = 1, 2, with some CSi > 0 and αi > 0, i = 1, 2,
αi > 0 means that the magnitude of the chemotactic flux is weakened when
the bacterial density increases. The function φ is known and fulfills
(1.8) φ ∈ W 1,∞ (Ω).
Processes of directed movement of cells in response to a chemical stimulus,
referred to as chemotaxis, play an important role in the interaction of cells
with their environment. A typical model describing chemotaxis is the classical
Keller-Segel system derived by Keller and Segel in 1970s [13]. The mathe-
matical analysis of classical Keller-Segel system and the variant thereof mainly
concentrates on the boundedness and blow-up of the solutions [8,11,25]. As the
blow-up has not been observed in the real biological process, many mechanisms,
such as nonlinear porous medium diffusion, saturation effect, logistic source,
etc., are introduced to avoid the blow-up of solutions [12,15,29]. In the past few
decades, Keller-Segel system has attracted extensive attentions. For a helpful
 GLOBAL EXISTENCE AND ASYMPTOTIC BEHAVIOR 217

overview of many models arising out of this fundamental description we refer to

the survey [3]. To better understand the model (1.1), it is necessary to separate
study two-species chemotaxis system and chemotaxis-Stokes system with sin-
gle specie. For two-species chemotaxis model, this has been extensively studied
by many authors. When the two-species have influence in each other, namely,
the system has Lotka-Volterra competitive kinetics. In the two-dimensional
case, Bai and Winkler [1] obtained global existence of solution to the system
(1.1) with Di (ni ) = 1, Si (x, ni , v) = χi > 0, i = 1, 2. Moreover, they also
taken into account asymptotic behavior of solutions to the system (1.1), when
1−a1
a1 , a2 ∈ (0, 1), n1 (·, t) → 1−a 1 a2
1−a2
, n2 (·, t) → 1−a1 a2
, v(·, t) → α(1−a1−a
1 )+β(1−a2 )
1 a2
in
L∞ (Ω) as t → ∞; when a1 ≥ 1 > a2 > 0, n1 (·, t) → 0, n2 (·, t) → 1, v(·, t) → β
in L∞ (Ω) as t → ∞, Mizukami [22] improved this result. Recently, Mizukami
[23] further modified the result in [22], this is filling up the gap between [1]
and [22]. In the three-dimensional case, Lin and Mu [16] supposed that µ1
and µ2 are large enough to obtained a similar result. In the high dimensional
case, Lin et al. [17] and Zhang et al. [47] are obtained unique global classical
bounded solution, respectively. However, the asymptotic behavior of solution
is not involved. When the system has a logistic source, but the two-species
have not influence in each other, in other wards, the competitive kinetics term
µ1 n1 (1 − n1 − a1 n2 ) and µ2 n2 (1 − a2 n1 − n2 ) are replaced by µ1 n1 (1 − n1 )
and µ2 n2 (1 − n2 ), Negreanu et al. [27] and [26] separate discussed the system
has unique uniformly bounded solution with vt = ε4v + h(n1 , n2 , v), ε ∈ [0, 1),
Mizukami et al. [24] remove the restriction of ε ∈ [0, 1) to obtained a similar
result. When the third equation degenerate into elliptic equation, Stinner et
al. [32] and Lin et al. [18] are obtained global existence of solution to the system
(1.1) with Di (ni ) = 1, Si (x, ni , v) = χi > 0, i = 1, 2. Moreover, they also taken
into account asymptotic behavior of solutions to the system (1.1).
For chemotaxis-Stokes system with single specie, this model which was pro-
posed in [36] for the spatio-temporal evolution in populations of oxytactically
moving bacteria that interact with a surrounding fluid through transport and
buoyancy with the third equation of (1.1) is replaced by vt +u·∇v = 4v−nf (v),
the corresponding model is given by:


 nt + u · ∇n = 4n − ∇ · (nχ(n, v) · ∇v) x ∈ Ω, t > 0,

v + u · ∇v = 4v − nf (v), x ∈ Ω, t > 0,
t
(1.9)


 u t + κ(u · ∇)u = 4u + ∇P + n∇φ, x ∈ Ω, t > 0,
∇ · u = 0 x ∈ Ω, t > 0.


When 4n is replaced by 4nm , Winkler [43] recent analysis has revealed that
m > 67 for all reasonably regular initial data, a corresponding no-flux Neumann
initial-boundary value problem possesses a globally defined weak solution which
is bounded in three-dimensional bounded convex domains. Tao et al. [33] which
assured global solvability within the large range m > 78 , but only in a class
of weak solutions locally bounded in Ω × [0, ∞). Winkler [46] which allows
218 B. LIU AND G. REN

for the construction of global weak solution to an associated initial-boundary

value problem under the milder assumption that m > 89 . Moreover, the ob-
tainedR solutions are shown to approach the spatially homogeneous steady state
1
( |Ω| n , 0, 0) in the large time limit. As in the classical Keller-Segel model
Ω 0
where the chemoattractant is produced rather than consumed by bacteria, the
relevant Keller-Segel-fluid system of the form

nt + u · ∇n = 4n − ∇ · (nS(x, n, v) · ∇v) x ∈ Ω, t > 0,


v + u · ∇v = 4v − v + n, x ∈ Ω, t > 0,
t
(1.10)
ut + κ(u · ∇)u = 4u + ∇P + n∇φ, x ∈ Ω, t > 0,


∇ · u = 0 x ∈ Ω, t > 0.


Compare with (1.9), the mathematical analysis of (1.10) is quite few. When
S(x, n, v) is a tensor-valued sensitivity satisfying some dampening condition in
(1.7), Wang et al. [39] obtained global existence and boundedness in a Keller-
Segel-Stokes system (κ = 0) in two-dimensional smoothly bounded domains, to
the best of our knowledge, this is the first result on global existence and bound-
edness in a Keller-Segel-Stokes system with tensor-valued sensitivity. With the
same author [40], when α > 12 , they also obtained global classical solutions
which is uniformly bounded in three-dimensional smoothly bounded domains.
Parallel to the case of the corresponding Keller-Segel-Navier-Stokes system,
Wang [37] proved the system (1.10) possesses at least one global very weak
solution with α > 31 in three-dimensional smoothly bounded domains. Winkler
[45] shown that if α > 13 , the problem (1.10) with κ = 0 possesses a global
classical solution. When the system (1.10) has a logistic source rn − µn2 and
external force g in the fluid equation, Tao et al. [34] shown that under the
explicit condition µ ≥ 23 and suitable regularity assumptions on the initial
data, the corresponding initial-boundary problem possesses a global classical
solution which is bounded in three-dimensional smoothly bounded domains.
Apart from this, it is also proved that if r = 0, then both n(·, t) and v(·, t)
decay
R ∞ R to zero with respect to the norm in L∞ as t → ∞, and that if moreover
0 Ω
|g| < 0, then also u(·, t) → 0 in L∞ as t → ∞. In two-dimensional
2

smoothly bounded domains, Tao et al. [35] obtained the Keller-Segel-Navier-

Stokes possesses a global classical bounded solution when µ > 0, and also get
the same large time behavior. Jiu et al. [19] shown that under the conditions
m ≥ 13 and α > 56 −m, and proper regularity hypotheses on the initial data, the
corresponding initial-boundary problem possesses at least one global bounded
weak solution for the Keller-Segel-Stokes system with nonlinear diffusion and
logistic source in the three-dimensional bounded domains. When 4n is re-
placed by 4nm , S(x, n, v) ≡ 1, Black [2] proved that if m > 53 , the problem
(1.10) admits at least one global weak solution, in addition, if m > 43 , the
problem (1.10) admits at least one global very weak solution. For the latest
progress of other chemotaxis-fluid system, please refer to [4, 16, 21, 38, 44].
 GLOBAL EXISTENCE AND ASYMPTOTIC BEHAVIOR 219

For two-species chemotaxis-Stokes system with Lotka-Volterra competitive

kinetics, the mathematical analysis of (1.1) is quite fragmentary. When Di (ni )
≡ 1, Si (x, ni , v) = χi , i = 1, 2, the third equation turned into vt + u · ∇v =
4v−(αn1 +βn2 )v, the fourth equation turn into Navier-Stokes equation, Hirata
et al. [9] obtained global classical and bounded solution which is unique in
two-dimensional smoothly bounded domains. Moreover, they also taken into
account asymptotic behavior of solutions to the system (1.1), when a1 , a2 ∈
1−a1 1−a2
(0, 1), n1 (·, t) → 1−a1 a2
, n2 (·, t) → 1−a1 a2
, v(·, t) → 0, u(·, t) → 0 in L∞ (Ω) as
t → ∞; when a1 ≥ 1 > a2 > 0, n1 (·, t) → 0, n2 (·, t) → 1, v(·, t) → 0, u(·, t) → 0
in L∞ (Ω) as t → ∞. To the best of our knowledge, this is the first result
on global existence and boundedness for two-species chemotaxis-Navier-Stokes
system with competitive kinetics. The same authors [10] shown that global
existence of weak solutions in three-dimensional smoothly bounded domains,
the system is the same as in [9]. When Di (ni ) ≡ 1, Si (x, ni , v) = χi , i = 1, 2,
the third equation turned into vt + u · ∇v = 4v − (αn1 + βn2 )v, the fourth
equation is the same as in (1.1), Cao et al. [5] obtained the system possesses
a classical solution which is unique in the sense that it allows up to addition
of spatially constants to the pressure P with χµ < ξ0 , where χ := max{χ1 , χ2 },
µ := min{µ1 , µ2 }, ξ0 > 0 is a constant in three-dimensional smoothly bounded
domains. Moreover, they obtained the similar results for asymptotic behavior
of solutions in [9]. When Di (ni ) ≡ 1, Si (x, ni , v) = χi , i = 1, 2, the third
equation and fourth equation is the same as in (1.1), Cao et al. [6] proved a
similar result with the different conditions.
Throughout above analysis, compared with two-species chemotaxis system
and chemotaxis-Stokes system with single specie, it is not so mature that the
two-species chemotaxis-Stokes system with Lotka-Volterra competitive kinetics.
Inspired by the arguments in previous studies [10, 20, 45], we mainly investi-
gate the global existence and boundedness in a three-dimensional two-species
chemotaxis-Stokes system with tensor-valued sensitivity to the system (1.1), in
addition to discuss the asymptotic behavior of the solutions. More precisely,
we have:

Theorem 3.1. Let a1 , a2 ≥ 0, α, β > 0, Ω ⊂ R3 be a bounded domain with

smooth boundary. Suppose that D(ni ) and Si , i = 1, 2, satisfy (1.3)-(1.7) with
mi ≥ 31 , i = 1, 2 and

23 23
m1 + α1 > , m2 + α2 > .
18 18
Then for any choice of the initial data n01 , n02 , v0 , u0 , φ fulfill (1.2) and (1.8),
system (1.1) possesses at least one non-negative global weak solution (n1 , n2 ,
v, u, P ) in the sense of Definition 2.1. Also, this solution is bounded in Ω ×
(0, ∞) in the sense that

kn1 (·, t)kL∞ (Ω) + kn2 (·, t)kL∞ (Ω) + kv(·, t)kW 1,∞ (Ω) + ku(·, t)kW 1,∞ (Ω) ≤ C
220 B. LIU AND G. REN

for all t > 0 with some constant C > 0. In addition, v and u are continuous in
Ω × (0, ∞), and n1 , n2 as an L∞ (Ω)-valued function is continuous on [0, ∞)
with respect to the weak-∗ topology, i.e.,
0
n1 , n2 ∈ Cw−∗ ([0, ∞); L∞ (Ω)).
Theorem 4.1. Let D1 (n1 ) = D2 (n2 ) ≡ 1, assume that the condition of Theo-
rem 3.1 holds. Then the solution of (1.1) has the following properties:
(i) Let a1 , a2 ∈ (0, 1), under the condition that there exists γ1 such that
4γ1 − (1 + γ1 )2 a1 a2 > 0
and
c1 CS21 (1 − a1 ) γ1 c2 CS22 (1 − a2 ) 4γ1 − (1 + γ1 )2 a1 a2
+ < ,
4a1 µ1 (1 − a1 a2 ) 4a2 µ2 (1 − a1 a2 ) a1 α γ1 + a2 β 2 − a1 a2 αβ(1 + γ1 )
2

then
n1 (·, t) → N1 , n2 (·, t) → N2 , v(·, t) → V1 , u(·, t) → 0 in L∞ as t → ∞,
where
1 − a1 1 − a2
N1 := , N2 := , V1 := αN1 + βN2
1 − a1 a2 1 − a1 a2
as well as
c1 = max{1, (1 + kn1 kL∞ [0,kn10 kL∞ (Ω) +1] )1−α1 }
and
c2 = max{1, (1 + kn2 kL∞ [0,kn20 kL∞ (Ω) +1] )1−α2 }.
(ii) Let a1 ≥ 1 > a2 . under the condition that there exist γ3 and a01 ∈ [1, a1 ]
such that
4γ3 − (1 + γ3 )2 a01 a2 > 0
and
C 2 γ3 c2 (α2 a01 γ3 + β 2 a2 − αβa01 a2 (1 + γ3 ))
µ2 > S2 ,
4a2 (4γ3 − a01 a2 (1 + γ3 )2 )
then
n1 (·, t) → 0, n2 (·, t) → 1, v(·, t) → β, u(·, t) → 0 in L∞ as t → ∞.
where
c2 = max{1, (1 + kn2 kL∞ [0,kn20 kL∞ (Ω) +1] )1−α2 }.
In this paper, we use symbols c and C as some generic positive constants.
Sometimes, in order to distinguish them, we use symbols Ci and ci (i = 1, 2, . . . )
which depend on m, CD1 , CD2 , CS1 , CS2 , p, Ω and the initial
R data only. More-
over,
R for simplicity, u(x, t) is written as u, the integral Ω
u(x)dx is written as
Ω
u(x).
The rest of this paper is organized as follows. In Section 2, we summarize
some basic definitions and some useful lemmas in order to prove the main result.
In Section 3, we show the main theorem, firstly, and give some fundamental
estimates for the solution to the system (2.1) to prove Theorem 3.1. In Section
 GLOBAL EXISTENCE AND ASYMPTOTIC BEHAVIOR 221

4, we start with stating the main theorem, and construct the function to prove
Theorem 4.1.

2. Preliminaries
Under the assumptions of Di (ni ), the first two equations of system (1.1) may
be degenerate at ni = 0, i = 1, 2. Therefore, system (1.1) does not allow for
classical solvability in general. We introduce the following definition of weak
solution.
Definition 2.1. Let T ∈ (0, ∞). A quadruple of nonnegative functions (n1 , n2 ,
v, u) defined in Ω×(0, T ) is called a weak solution of system (1.1) if the following
conditions are satisfied
ni ∈ L1loc (Ω × [0, T )), v ∈ L∞ 1
loc (Ω × [0, T )) ∩ Lloc ([0, T ); W
1,1
(Ω)),

u ∈ L1loc ([0, T ); W 1,1 (Ω)), G(ni ), ni |∇v|, ni |u| ∈ L1loc (Ω × [0, T ));
the integral equalities
Z TZ Z
− n1 ψt − n10 ψ(·, 0)
0 Ω Ω
Z T Z Z T Z
= − G1 (n1 ) 4 ψ + n1 (S1 (x, n1 , v) · ∇v) · ∇ψ
0 Ω 0 Ω
Z T Z Z T Z
+ n1 u · ∇ψ + µ1 n1 (1 − n1 − a1 n2 )ψ
0 Ω 0 Ω

for any ψ ∈ C0∞ (Ω × [0, T )) satisfying ∂ψ

∂ν = 0 on ∂Ω × (0, T ), as well as
Z TZ Z
− n2 ψt − n20 ψ(·, 0)
0 Ω Ω
Z T Z Z T Z
= − G2 (n2 ) 4 ψ + n2 (S2 (x, n2 , v) · ∇v) · ∇ψ
0 Ω 0 Ω
Z T Z Z T Z
+ n2 u · ∇ψ + µ2 n2 (1 − a2 n1 − n2 )ψ
0 Ω 0 Ω

for any ψ ∈ C0∞ (Ω × [0, T )) satisfying ∂ψ

∂ν = 0 on ∂Ω × (0, T ), and
Z TZ Z
− vψt − v0 ψ(·, 0)
0 Ω Ω
Z T Z Z T Z Z T Z
= − ∇v · ∇ψ − v·ψ+α n1 · ψ
0 Ω 0 Ω 0 Ω
Z T Z Z T Z
+β n2 · ψ + uv · ∇ψ
0 Ω 0 Ω
222 B. LIU AND G. REN

for any ψ ∈ C0∞ (Ω × [0, T )), as well as

Z TZ Z Z TZ Z T Z
− uψt − u0 ψ(·, 0) = − ∇u · ∇ψ + (n1 + n2 )∇φ · ψ
0 Ω Ω 0 Ω 0 Ω

for all ψ ∈ C0∞ (Ω × [0, T ); R3 ) fulfilling ∇ψ ≡ 0 in Ω × (0, T ), where we let

Z s
Gi (s) := Di (σ)dσ for s ≥ 0, i = 1, 2.
0
If (n1 , n2 , v, u) is a weak solution of system (1.1) in Ω×(0, T ) for all T ∈ (0, ∞),
then we call (n1 , n2 , v, u) a global weak solution.
We will construct solutions of (1.1) as limits of solutions to relevant regular-
ized approximated problems, and give some basic estimates for the solutions to
the regularized system. First of all, we approximated the diffusion coefficient
function in (1.1) by a family (Diε )ε∈(0,1) of functions
Diε ∈ C 2 ([0, ∞)) such that Diε (ni ) ≥ ε for all ni > 0 (i = 1, 2),
and
Di (ni ) ≤ Diε (ni ) ≤ Di (ni ) + 2ε for all ni > 0 (i = 1, 2).
Then, in order to obtain homogeneous Neumann boundary conditions for both
ni (i = 1, 2) and v, we let (ρε )ε∈(0,1) ⊂ C0∞ (Ω) be a family of standard cut-off
functions satisfying 0 ≤ ρε ≤ 1 in Ω and ρε → 1 in Ω as ε → 0, and define
Siε (x, ni , v) = ρε Si (x, ni , v), x ∈ Ω, ni ≥ 0, (i = 1, 2), v ≥ 0
for ε ∈ (0, 1) to approximate the sensitivity tensor Si , which implies that
Siε (x, ni , v) = 0 on ∂Ω for each fixed ε ∈ (0, 1), i = 1, 2. The initial data
ni0ε ∈ C ϑ (Ω) for some ϑ > 0 with ni0ε ≥ 0, i = 1, 2 in Ω, v0ε ∈ W 1,∞ (Ω)
fulfills v0ε ≥ 0 in Ω, and u0ε ∈ D(Ar ) for certain  ∈ ( 34 , 1) and
(
kn10ε kL∞ (Ω) ≤ kn10 kL∞ (Ω) + 1, kn20ε kL∞ (Ω) ≤ kn20 kL∞ (Ω) + 1,
kv0ε kW 1,∞ (Ω) ≤ kv0 kW 1,∞ (Ω) + 1, ku0ε kW 1,∞ (Ω) ≤ ku0 kW 1,∞ (Ω) + 1.
Therefore, for any such ε, the regularized problems
(2.1)


 (n1ε )t + uε · ∇n1ε = ∇ · (D1ε (n1ε )∇n1ε ) − ∇ · (n1ε S1ε (x, n1ε , vε ) · ∇vε )




 +µ1 n1ε (1 − n1ε − a1 n2ε ), x ∈ Ω, t > 0,
· ∇n ∇ · (D2ε (n2ε )∇n2ε ) − ∇ · (n2ε S2ε (x, n1ε , vε ) · ∇vε )




 (n )
2ε t + uε 2ε =
+µ2 n2ε (1 − a2 n1ε − n2ε ), x ∈ Ω, t > 0,




vεt + uε · ∇vε = 4vε − vε + αn1ε + βn2ε , x ∈ Ω, t > 0,

uεt + ∇Pε = 4uε + (n1ε + n2ε )∇φ, ∇ · uε = 0, x ∈ Ω, t > 0,




∂n1ε ∂n2ε ∂vε

∂ν = ∂ν = ∂ν = 0, uε = 0, x ∈ ∂Ω, t > 0,








 n1ε (x, 0) = n10ε (x), n2ε (x, 0) = n20ε (x),
vε (x, 0) = v0ε (x), uε (x, 0) = u0ε (x), x ∈ Ω.

 GLOBAL EXISTENCE AND ASYMPTOTIC BEHAVIOR 223

According to the well-established fixed point arguments, the local solvability

of (2.1) can be obtained, the proof is similar to that in [42], so here we omit
the proof.
Lemma 2.1. Let Ω ⊂ R3 be a bounded domain with smooth boundary. Assume
that (1.2)-(1.8) holds. Then for each ε ∈ (0, 1), q > 3, there exist Tmax ∈ (0, ∞)
and a classical solution (n1ε , n2ε , vε , uε , Pε ) such that
 0 2,1
n1ε , n2ε ∈ C (Ω × [0, Tmax )) ∩ C (Ω × (0, Tmax )),

v ∈ C 0 (Ω × [0, T 2,1
(Ω × (0, Tmax )) ∩ L∞ 1,q

ε max )) ∩ C loc ([0, Tmax ); W (Ω)),
0 2,1 ∞
uε ∈ C (Ω × [0, Tmax )) ∩ C (Ω × (0, Tmax )) ∩ Lloc ([0, Tmax ); D(A )), 


Pε ∈ C 1,0 (Ω) × (0, Tmax )),


where Tmax denotes the maximal existence time. Also, the above solution is
unique up to addition of spatially constants to the pressure Pε . Moreover, we
have niε > 0 i = 1, 2 and vε > 0 in Ω × (0, Tmax ), and if Tmax < +∞, then
(2.2)
kn1ε (·, t)kL∞ (Ω) + kn2ε (·, t)kL∞ (Ω) + kvε (·, t)kW 1,q (Ω) + kAr uε (·, t)kL2 (Ω) → ∞
as t → Tmax , where , r are taken from (1.2).
Lemma 2.2. For each ε ∈ (0, 1), there exists a constant C := C(τ ) > 0, the
solutions of (2.1) satisfies
Z
(2.3) niε (·, t) ≤ C for all t ∈ (0, Tmax ) i = 1, 2
Ω
as well as
Z t+τ Z
(2.4) n2iε ≤ C for all t ∈ (0, Tmax − τ ) i = 1, 2
t Ω
and
Z nZ o
(2.5) vε (·, t) ≤ max v0 , (α + β)C for all t ∈ (0, Tmax ).
Ω Ω

Proof. The proof is similar to [28, Lemma 2.2], so we omitted it. 

Lemma 2.3 ([37, Lemma 3.4]). Let T > 0 and y ∈ C 0 ([0, T )) ∩ C 1 (0, T ) be
such that
y 0 (t) + ay(t) ≤ g(t) for all t ∈ (0, T ),
where g ∈ L1loc (R) has the property that
1 t+τ
Z
g(s)ds ≤ b for all t ∈ (0, T )
τ t
with some τ > 0 and b > 0. Then
bτ
y(t) ≤ y(0) + for all t ∈ [0, T ).
1 − e−aτ
224 B. LIU AND G. REN

Lemma 2.4 (Gagliardo-Nirenberg interpolation inequality, [30, Lemma 2.2]).

Let 0 < θ ≤ p ≤ N2N
−2 . There exists a positive constant CGN such that for all
1,2 θ
uε ∈ W (Ω) ∩ L (Ω),
(2.6) kuε kLp (Ω) ≤ CGN (k∇uε kaL2 (Ω) kuε k1−a
Lθ (Ω)
+ kuε kLθ (Ω) )
N
θ −N
p
is valid with a = 1− N N ∈ (0, 1).
2 + θ

Lemma 2.5 ([43, Corollary 3.4]). Let p ∈ [1, ∞) and r ∈ [1, ∞] be such that
(
3p
r < 3−p if p ≤ 3,
(2.7)
r ≤ ∞ if p > 3.
Then for all K > 0 exists C = C(p, r, K) such that if for some ε ∈ (0, 1) and
Tmax > 0 we have
(2.8) kniε (·, t)kLp (Ω) ≤ K for all t ∈ (0, Tmax ),
then
(2.9) kDuε (·, t)kLr (Ω) ≤ C for all t ∈ (0, Tmax ).
Lemma 2.6 ([19, Lemma 3.3]). Suppose that Ω ⊂ R3 be a bounded domain with
smooth boundary. Let q > 1 and γ ∈ [2, 6q]. Then there exists C = C(q, γ) > 0
such that for all ω ∈ C 2 (Ω) the following interpolation inequality
 3γ−6 6q−γ 
(2.10) k∇ωkγLγ (Ω) ≤ C k∇|∇ω|q kL3q−1 3q−1 γ
2 (Ω) k∇ωkL2 (Ω) + k∇ωkL2 (Ω)

holds.
Lemma 2.7 ([41, Lemma 1.3]). Let Ω ⊂ RN (N ∈ N) be a bounded domain
with smooth boundary and let (et4 )t≥0 be the Neumann heat semigroup in Ω.
Then there exist constants C, λ1 > 0 depending only on Ω such that if 1 ≤ q ≤
p ≤ ∞, then
1 N 1 1
k∇et4 ψkLp (Ω) ≤ C(1 + t− 2 − 2 ( q − p ) )e−λ1 t kψkLq (Ω)
holds for all t > 0 and each ψ ∈ Lp (Ω).

3. Existence of global and bounded weak solution

In this section, we start with expounding the main theorem, and then es-
tablish some a priori estimates for solutions to the approximated system (2.1)
with non-degenerate diffusion, it is crucial ingredient for the proof of our main
results. As a first step towards this, we show the main theorem.
Theorem 3.1. Let a1 , a2 ≥ 0, α, β > 0, Ω ⊂ R3 be a bounded domain with
smooth boundary. Suppose that D(ni ) and Si , i = 1, 2, satisfy (1.3)-(1.7) with
mi ≥ 31 , i = 1, 2 and
23 23
m1 + α1 > , m2 + α2 > .
18 18
 GLOBAL EXISTENCE AND ASYMPTOTIC BEHAVIOR 225

Then for any choice of the initial data n01 , n02 , v0 , u0 , φ fulfill (1.2) and (1.8),
system (1.1) possesses at least one non-negative global weak solution (n1 , n2 ,
v, u, P ) in the sense of Definition 2.1. Also, this solution is bounded in Ω ×
(0, ∞) in the sense that
kn1 (·, t)kL∞ (Ω) + kn2 (·, t)kL∞ (Ω) + kv(·, t)kW 1,∞ (Ω) + ku(·, t)kW 1,∞ (Ω) ≤ C
for all t > 0 with some constant C > 0. In addition, v and u are continuous in
Ω × (0, ∞), and n1 , n2 as an L∞ (Ω)-valued function is continuous on [0, ∞)
with respect to the weak-∗ topology, i.e.,
0
n1 , n2 ∈ Cw−∗ ([0, ∞); L∞ (Ω)).
Remark 3.1. Theorem 3.1 show that the system (1.1) admit a global bounded
weak solution nothing to do with the size of χµ and the dampening intensity of
logistic source, this is very important in the proof of [5], [6] and [10]. We now
have to leave an open question on existence of global bounded weak solution
when mi ≥ 13 , i = 1, 2, m1 + α1 > 23 23 23
18 , m2 + α2 ≤ 18 or m1 + α1 ≤ 18 , m2 + α2 >
23 1 23 23
18 and mi < 3 , i = 1, 2, m1 + α1 > 18 , m2 + α2 > 18 .

Then, we will give some priori estimates and prove the main theorem.
Lemma 3.1. There exists a constant C > 0 depending on ε ∈ (0, 1) such that
Z
(3.1) |∇uε |2 ≤ C for all t ∈ (0, Tmax )
Ω
and
Z t+τ Z
(3.2) |Auε |2 ≤ C for all t ∈ (0, Tmax − τ ),
t Ω

where τ = min{1, 21 Tmax }.

Proof. The proof is similar to that in [34, Lemma 2.4], so we omitted it. 
Lemma 3.2. There exists C > 0 depending on ε ∈ (0, 1) such that
Z Z
2
(3.3) vε + |∇vε |2 ≤ C for all t ∈ (0, Tmax ).
Ω Ω

Proof. The proof boundedness of k∇vε kL2 (Ω) is similar to that in [34, Lemma
2.6], so we omitted it, by the Poincaré inequality, the boundedness of kvε kL2 (Ω)
is obtained. This completes the proof. 
Lemma 3.3. Let Ω ⊂ R3 be a bounded domain with smooth boundary, p > 1,
suppose that the assumptions in Lemma 2.1 holds. Then for all ε ∈ (0, 1), we
can find a constant C > 0 independent of ε and obtain the following inequality
(3.4)
Z Z
d p

n1ε + np2ε
dt Ω Ω
226 B. LIU AND G. REN

p(p − 1) 
Z Z 
+ CD1 nm1ε
1 +p−3
|∇n 1ε |2
+ C D2 nm
2ε
2 +p−3
|∇n2ε |2
2 Ω Ω
p(p − 1)  CS21 CS22
Z Z 
p+1−m1 −2α1 2 p+1−m2 −2α2
≤ n1ε |∇vε | + n2ε |∇vε |2 +C
2 CD1 Ω CD2 Ω
for all t ∈ (0, Tmax ), where CD1 , CD2 , CS1 and CS2 are as in (1.4), (1.5) and
(1.7), respectively.

Proof. We multiply the first two equation in (2.1) by pnp−1 1ε and use the fact
S1ε (x, n1ε , vε ) = 0 on ∂Ω, we have
Z Z
d p
n + p(p − 1) np−2
1ε D1ε (n1ε )|∇n1ε |
2
dt Ω 1ε Ω
Z
(3.5) = p(p − 1) np−1
1ε ∇n1ε · (S1ε (x, n1ε , vε ) · ∇vε )
Ω
Z Z Z
p p+1
+ µ1 p n1ε − µ1 p n1ε − a1 µ1 p np1ε n2ε
Ω Ω Ω
Z
≤ p(p − 1) np−1
1ε ∇n1ε · (S1ε (x, n1ε , vε ) · ∇vε )
Z Ω Z
+ µ1 p np1ε − µ1 p np+1
1ε
Ω Ω

for all t ∈ (0, Tmax ). Here we use the definition of Dε and (1.4), (1.5) to get
that
Z Z
(3.6) p(p − 1) np−2
1ε D (n
ε 1ε )|∇n 1ε |2
≥ CD1 p(p − 1) m1 +p−3
n1ε |∇n1ε |2
Ω Ω

for all t ∈ (0, Tmax ). Then, due to (1.7), we obtain

Z
(3.7) p(p − 1) np−1
1ε ∇n1ε · (S1ε (x, n1ε , vε ) · ∇vε )
Ω
Z
1 −1
≤ CS1 p(p − 1) np−α
1ε |∇n1ε ||∇vε |
Ω
CD1 p(p − 1) C 2 p(p − 1)
Z Z
p+1−m1 −2α1
≤ nm
1ε
1 +p−3
|∇n1ε |2 + S1 n1ε |∇vε |2
2 Ω 2CD1 Ω

for all t ∈ (0, Tmax ), here we use the Young’s inequality. Dealing with the last
two items, we also use the Young’s inequality to see that
Z Z
µ1 p+1 p p
(3.8) µ1 p np1ε − µ1 p np+1
1ε ≤ p( ) ( ) =: C1
Ω Ω p +1 µ1
for all t ∈ (0, Tmax ), where C1 is a positive constant. Similar to n1ε , n2ε has
similar inequality as above. Thus, we can obtain (3.4). This completes the
proof. 
 GLOBAL EXISTENCE AND ASYMPTOTIC BEHAVIOR 227

Lemma 3.4. Let q > 1, it holds for all ε ∈ (0, 1) that

2(q − 1)
Z Z Z
d 2q q 2
(3.9) |∇vε | + |∇|∇vε | | + q |∇vε |2q−2 |D2 vε |2
dt Ω q Ω Ω
√ 2Z 2 2 2 2 2q−2
Z
≤ 2q(2q−2+ 3) (α n1ε +β n2ε )|∇vε | + 2q |∇vε |2q |Duε | + C
Ω Ω

on (0, Tmax ) with some positive constant C determined by q.

Proof. The proof is similar to that in [19, Lemma 3.2], so we omitted it. 
23
Lemma 3.5. Let mi > 18 − αi (i = 1, 2) and suppose that p > 1 and q ≥ 2
satisfies
1
max{3 − 3mi − 2αi , mi + 2αi − } ≤ p
3
5 4
(3.10) < (mi + αi − )(3q − 1) + − mi ,
6 3
i = 1, 2. Then for all η > 0 there exists C = C(p, q, η) > 0 such that for all
ε ∈ (0, 1)
Z Z Z
i −2αi
(3.11) np+1−m
iε |∇v ε | 2
≤ η n mi +p−3
iε |∇n iε | 2
+ η |∇|∇vε |q |2 + C
Ω Ω Ω

for all t ∈ (0, Tmax ), i = 1, 2.

Proof. By the Hölder inequality, we have
Z Z 3
 23 Z 1
i −2αi 2 (p+1−mi −2αi ) 6 3
np+1−m
iε |∇v ε |2
≤ n iε |∇v ε |
Ω Ω Ω
p+mi −1 2(p+1−mi −2αi )

k∇vε k2L6 (Ω)

p+mi −1
(3.12) = kniε 2
k 3(p+1−mi −2αi )
L p+mi −1 (Ω)

i −2αi )
for all t ∈ (0, Tmax ). Due to p ≥ 3 − 3mi − 2αi , we have 3(p+1−m
p+mi −1 ≤ 6,
i = 1, 2. Thus, by the Gagliardo-Nirenberg interpolation inequality there exist
some positive constants c1 , c2 may be determined by p such that
p+mi −1 2(p+1−mi −2αi )
p+mi −1
kniε 2
k 3(p+1−mi −2αi )
L p+mi −1 (Ω)
p+mi −1 2(p+1−mi −2αi ) p+mi −1 2(p+1−mi −2αi )
p+mi −1 σ1 p+mi −1 (1−σ1 )
(3.13) ≤ c1 k∇niε 2
kL2 (Ω) kniε 2
k 2
L p+m i −1 (Ω)
p+mi −1 2(p+1−mi −2αi )
p+mi −1
+ kniε 2
k 2
L p+mi −1 (Ω)
Z  3(p+1−m i −2αi )−2
mi +p−3 3(p+m −1)−1
≤ c2 niε |∇niε |2 +1 i

Ω
228 B. LIU AND G. REN

for all t ∈ (0, Tmax ) with

3(p+mi −1) p+mi −1
2 − p+1−m i −2αi
σ1 = ∈ [0, 1]
i −1)
1 − 32 + 3(p+m2

due to p > 3 − 3mi − 2αi and p > mi + 2αi − 13 , i = 1, 2. Due to 6 ∈ [2, 6q],
from Lemma 2.6, there exist c3 , c4 > 0 such that
4 2q−2

k∇vε k2L6 (Ω) ≤ c3 k∇|∇vε |q kL3q−1 3q−1 2

2 (Ω) k∇vε kL2 (Ω) + c3 k∇vε kL2 (Ω)

Z 2
 3q−1
(3.14) ≤ c4 |∇|∇vε |q |2 + 1 for all t ∈ (0, Tmax ).
Ω
5 4
Since p < (mi + αi − 6 )(3q − 1) + 3 − mi , i = 1, 2, we have
3(p + 1 − mi − 2αi ) − 2 2
+ < 1.
3(p + mi − 1) − 1 3q − 1
Combining with (3.12)-(3.14) and using Young’s inequality, the desired results
are obtained. This completes the proof. 
23
Lemma 3.6. Let mi > 18 − αi (i = 1, 2) and suppose that p > 1 and q ≥ 2
satisfies
4 4
(3.15) (3q − 1) + − mi }, i = 1, 2.
p > max{2 − mi ,
9 3
Then for all η > 0 there exists C = C(p, q, η) > 0 such that for all ε ∈ (0, 1)
Z Z Z Z
i −3
(3.16) n2iε |∇vε |2q−2 + n2iε ≤ η np+m
iε |∇n iε |2
+ η |∇|∇vε |q |2 + C
Ω Ω Ω Ω
for all t ∈ (0, Tmax ), i = 1, 2.
Proof. By the Hölder inequality, we have
Z Z  32 Z  13
n2iε |∇vε |2q−2 ≤ n3iε |∇vε |6q−6
Ω Ω Ω
p+mi −1 4
p+mi −1 2(q−1)
(3.17) = kniε 2
k 6 k∇vε kL6(q−1)(Ω)
L p+mi −1 (Ω)

for all t ∈ (0, Tmax ), i = 1, 2. Since p > 2 − mi , i = 1, 2, we have p+m6i −1 ≤ 6,

by the Gagliardo-Nirenberg interpolation inequality there exist some positive
constants c1 , c2 may be determined by p such that
p+mi −1 4 p+mi −1 4 p+mi −1 4
p+mi −1 p+m −1 σ2 p+mi −1 (1−σ2 )
kniε 2
k 6 ≤ c1 k∇niε 2
kL2 (Ω)
i
kniε2
k 2
L p+mi −1 (Ω) L p+mi −1 (Ω)
p+mi −1 4
p+mi −1
(3.18) + c1 kniε 2
k 2
L p+mi −1 (Ω)
Z  3(p+m4−1)−1
i −3
≤ c2 np+m
iε |∇niε |2 + 1 i

Ω
 GLOBAL EXISTENCE AND ASYMPTOTIC BEHAVIOR 229

for all t ∈ (0, Tmax ), with

3(p+mi −1) i −1)
2 − 3(p+m 6
σ2 = ∈ [0, 1]
i −1)
1 − 32 + 3(p+m
2
due to p > 2 − mi . Since 6(q − 1) ∈ [2, 6q], from Lemma 2.6, we have
2[3(q−1)−1] 2
2(q−1) 2(q−1)
k∇vε kL6(q−1) (Ω) ≤ c3 k∇|∇vε |q kL2 (Ω)
3q−1
k∇vε kL3q−1
2 (Ω) + c3 k∇vε kL2 (Ω)

 Z  3(q−1)−1
3q−1
(3.19) ≤ c4 |∇|∇vε |q |2 + 1 for all t ∈ (0, Tmax ).
Ω
4 4
Since p > 9 (3q − 1) + 3 − mi , we have
4 3(q − 1) − 1
+ < 1.
3(p + mi − 1) − 1 3q − 1
By the Young’s inequality, there exists a positive constant c5 such that
Z Z Z
η p+mi −3 η
(3.20) n2iε |∇vε |2q−2 ≤ niε |∇niε |2 + |∇|∇vε |q |2 + c5
Ω 2 Ω 2 Ω
for all t ∈ (0, Tmax ). Similarly, by the Gagliardo-Nirenberg interpolation in-
equality there exists a positive constant c6 such that
Z Z
η i −3
(3.21) 2
niε ≤ np+m
iε |∇niε |2 + c6
Ω 2 Ω
for all t ∈ (0, Tmax ). Combining with (3.20) and (3.21), the desired results are
obtained. This completes the proof. 
Lemma 3.7. For any q > 1, then for all η > 0, there exists a positive constant
C such that
Z Z
(3.22) |∇vε |2q |Duε | ≤ η |∇|∇vε |q |2 + C for all t ∈ (0, Tmax ),
Ω Ω
where C determined by q and λ4 .
Proof. We invoke the Hölder inequality with same exponents 2 to see that
Z Z  1 Z 1 Z  12
2q 4q 2 2 2
(3.23) |∇vε | |Duε | ≤ |∇vε | |Duε | ≤M |∇vε |4q
Ω Ω Ω Ω
for all t ∈ (0, Tmax ), here we used the result of Lemma 3.1. Owing to 2 ≤ 4q ≤
6q satisfying the condition of Lemma 2.6, thus exists a positive constant c1
determined by q such that
Z  21
|∇vε |4q = k∇vε k2q
L4q (Ω)
Ω
 3(2q−1) q 
2q
≤ c1 k∇|∇vε |q kL23q−1 3q−1
(Ω) k∇vε kL2 (Ω) + k∇vε kL2 (Ω)
Z
η C
(3.24) ≤ |∇|∇vε |q |2 +
M Ω M
230 B. LIU AND G. REN

for all t ∈ (0, Tmax ), here we use the Young’s inequality, combine with (3.23)
and (3.24), the desired result can be obtained. This completes the proof. 
Lemma 3.8. Let mi > 1 and mi > 2318 − αi , (i = 1, 2). Then for sufficiently
large p > 1 and q ≥ 2 in Lemmas 3.5 and 3.6, there exists a constant C =
C(p, q) > 0 such that
(3.25) kn1ε kLp (Ω) + kn2ε kLp (Ω) + k∇vε kL2q (Ω) ≤ C for all t ∈ (0, Tmax ).
23 5
Proof. Since mi > 18 − αi , we have mi + αi − 6 > 94 , so
4 4 5 4
(3q − 1) + − mi < (mi + αi − )(3q − 1) + − mi , i = 1, 2
9 3 6 3
for any q > 1, there exist sufficiently large p ≥ 1 such that
4 4 5 4
(3q − 1) + − mi < p < (mi + αi − )(3q − 1) + − mi , i = 1, 2.
9 3 6 3
Combining with Lemmas 3.5-3.7, choosing properly small η > 0, there exist
some constant c1 = c1 (p, q), c2 = c2 (p, q) > 0 such that
Z Z Z
d 
n1ε + n2ε + |∇vε |2q
dt Ω Ω Ω
Z p+m1 −1
Z p+mi −1
Z 
2
(3.26) + c1 |∇n1ε 2
| + |∇n2ε 2 |2 + |∇|∇vε |q |2 ≤ c2
Ω Ω Ω
for all t ∈ (0, Tmax ). By the Gagliardo-Nirenberg interpolation inequality, there
exist constants c1 , c2 > 0 may be determined by p such that
p+mi −1 2p

kniε kpLp (Ω) = kniε

p+mi −1
2
k 2p
L p+mi −1 (Ω)
p+mi −1 2p p+mi −1 2p
p+mi −1 %1 p+mi −1 (1−%1 )
(3.27) ≤ c1 k∇niε 2
kL2 (Ω) kniε2
k 2
L p+mi −1 (Ω)
p+mi −1 2p
p+m −1
+ c1 kniε2
kL2 (Ω)
i

p+mi −1 2p
%2
≤ c2 k∇niε 2
k p+mi −1 + c9
for all t ∈ (0, Tmax ), i = 1, 2, where
1 1
2 − 2p
%1 = p+mi −1 (p + mi − 1) ∈ [0, 1]
2 − 61
and
2p p−1
%1 = p+mi −1 1
≤ 2,
p + mi − 1 2 − 6
here we use the fact that mi ≥ 31 . Using the Young’s inequality to (3.27), there
exists a constant c3 > 0 independent of ε ∈ (0, 1) such that
Z Z p+mi −1
p
(3.28) niε ≤ c3 |∇niε 2 |2 + c3 , i = 1, 2.
Ω Ω
 GLOBAL EXISTENCE AND ASYMPTOTIC BEHAVIOR 231

By Lemma 2.6 and Young’s inequality, there exist two constants c4 , c5 > 0 may
be determined by p, such that
 6(q−1) 4q 
k∇vε k2q q 3q−1 3q−1 2q
L2q (Ω) ≤ c4 k∇|∇vε | kL2 (Ω) k∇vε kL2 (Ω) + k∇vε kL2 (Ω)
Z
(3.29) ≤ c5 |∇|∇vε |q |2 + c5
Ω
for all t ∈ (0, Tmax ). Therefore, together with (3.28) and (3.29), we have
Z Z Z
(3.30) np1ε + np2ε + |∇vε |2q
Ω Ω Ω
Z p+m1 −1
Z p+m2 −1
Z
2
≤ c6 |∇n1ε 2
| + |∇n2ε 2 |2 + |∇|∇vε |q |2
Ω Ω Ω
for all t ∈ (0, Tmax ), with c6 = c3 + c5 . Now, let y(t) := Ω np1ε + Ω np2ε +
R R
p+m1 −1 p+m2 −1
|∇vε |2q , h(t) := Ω |∇n1ε 2 |2 + Ω |∇n2ε 2 |2 + Ω |∇|∇vε |q |2 , from
R R R R
Ω
(3.28) and (3.32), there exist some positive constants c7 , c8 > 0 such that
y 0 (t) + c7 y(t) + c7 h(t) ≤ c8 for all t ∈ (0, Tmax ).
Following from a comparison argument, we have
c8
y(t) ≤ c9 := max{y(0), } for all t ∈ (0, Tmax ).
c7
This completes the proof. 
23 23
Corollary 3.1. Let m1 > 1, m2 > 1 and m1 > − α1 , m2 >
18 − α2 . Then
18
for any p > 1, q > 1 and r ≥ 1, there exists some constat C > 0 such that for
any ε ∈ (0, 1)
(3.31)
kn1ε kLp (Ω) +kn2ε kLp (Ω) +k∇vε kLq (Ω) +kDuε (·, t)kLr (Ω) ≤ C for all t ∈ (0, Tmax ).
Proof. This proof is almost similar to that of [40, Corollary 3.2], to avoid rep-
etition, so we omitted it. 
Proposition 3.1. Suppose the assumptions of Lemma 2.1 hold with m1 > 1,
23
m2 > 1 and m1 > 18 − α1 , m2 > 23 18 − α2 . Then system (2.10) admits a
global classical solution (n1ε , n2ε , vε , uε , Pε ), which is uniformly bounded for all
ε ∈ (0, 1),
(3.32)
kn1ε (·, t)kL∞ (Ω) + kn2ε (·, t)kL∞ (Ω) + kvε (·, t)kW 1,∞ (Ω) + kuε (·, t)kW 1,∞ (Ω) ≤ C
for all t ∈ (0, ∞), with some constant C > 0. Moreover, we also have
(3.33) kA uε (·, t)kL2 (Ω) ≤ C for all t ∈ (0, ∞).
Proof. This proof is relying on the properties for the Neumann heat semigroup
and Stokes semigroup, we can find in [7,31,41]. In accordance with (3.27) with
p > 3, we can apply Lemma 2.5 to obtain (2.9) with r = ∞, and therefore
(3.34) kuε (·, t)kW 1,∞ (Ω) ≤ C for all t ∈ (0, Tmax )
232 B. LIU AND G. REN

is valid. Taking the results of Corollary 3.1 with appropriately large p and q as
a initial point, we make use of Moser-type iteration to the first two equations
in (2.1) and then obtain
(3.35) kniε (·, t)kL∞ (Ω) ≤ C (i = 1, 2) for all t ∈ (0, Tmax ).
On account of (3.34) and (3.35), we apply the parabolic regularity theory to
the third equation in (2.1) to get
(3.36) kvε (·, t)kW 1,∞ (Ω) ≤ C for all t ∈ (0, Tmax ).
Finally, we prove (3.33), let  ∈ ( 34 , 1), we apply the Helmholtz projection and
the fractional power Ar to the fourth equation in (2.1), it follows from the
variation-of-constants formula that
Z t
uε (·, t) = e−tAr u0ε + e−(t−s)Ar P[(n1ε + n2ε )∇φ](·, s)ds, t ∈ (0, Tmax ),
0
we can obtain
kAr uε (·, t)kL2 (Ω) ≤ kAr e−tAr u0ε kL2 (Ω)
Z t
+ kAr e−(t−s)Ar P[(n1ε + n2ε )∇φ](·, s)kL2 (Ω) ds
0
≤ c1 t− ku0ε kL2 (Ω) +c18 (kn1ε kL∞ (Ω) +kn2ε kL∞ (Ω) )k∇φkL2 (Ω)
Z t
× (t − s)− e−λ(t−s) ds
0
≤ c2
for all t ∈ (0, Tmax ), with some constants c1 , c2 , λ > 0. Along with (3.34)-(3.36)
and blow-up criterion (2.2), we infer that Tmax =∞ and prove the proposition.
This completes the proof. 
Lemma 3.9. Suppose the assumptions of Lemma 2.1 hold with m1 > 1, m2 > 1
23 23
and m1 > 18 − α1 , m2 > 18 − α2 . Then one can find θ ∈ (0, 1) such that for
some C > 0
(3.37) kvε k θ ≤ C for all t ∈ (0, ∞)
C θ, 2 (Ω×[t,t+1])

as well as
(3.38) kuε k θ ≤ C for all t ∈ (0, ∞)
C θ, 2 (Ω×[t,t+1])

and such that for any ς > 0 there exists C(ς) > 0 fulfilling
(3.39) k∇vε k θ ≤ C for all t ∈ [ς, ∞).
C θ, 2 (Ω×[t,t+1])

Proof. This proof can find in [19, Lemma 3.12], see also [14]. so we omitted
it. 
Finally, we prove the main theorem.
 GLOBAL EXISTENCE AND ASYMPTOTIC BEHAVIOR 233

The proof of Theorem 3.1. Firstly, with the help of Proposition 3.1 we derive
that for each ε ∈ (0, 1), system (2.1) admits a classical solution (n1ε , n2ε , vε ,
uε , Pε ) which is defined for all t > 0. Let ϕ ∈ W03,2 (Ω), it is known by the
embedding theorem, we see that W03,2 (Ω) ,→ W 1,∞ (Ω) with N = 3. Thus,
ϕ ∈ L∞ (Ω) and kϕkW 1,∞ (Ω) ≤ c1 kϕkW 3,2 (Ω) with some positive constant c1 .
0
Let κ > m1 satisfy κ ≥ 2(m1 − 1). Multiplying both sides of the first equation
in (2.10) by κnκ−11ε ϕ and integrating by parts on Ω, we obtain
Z 
1 ∂ κ
n ϕ
κ Ω ∂t 1ε
Z h   i
= nκ−1
1ε ∇ · D1ε (n1ε )∇n1ε − n1ε S1ε (x, n1ε , vε ) · ∇vε −uε ∇n1ε ·ϕ
Ω
Z
+ µ1 nκ1ε (1 − n1ε − a1 n2ε )ϕ
Ω
Z Z
(3.40) = − (κ − 1) nκ−2
1ε D 1ε (n 1ε )|∇n1ε |2
ϕ − κ−1
n1ε D1ε (n1ε )∇n1ε · ∇ϕ
Ω Ω
Z  
+ (κ − 1) nκ−1
1ε ∇n1ε · S1ε (x, n1ε , vε ) · ∇vε ϕ
Ω
Z Z
κ
  1
+ n1ε S1ε (x, n1ε , vε ) · ∇vε ·∇ϕ + nκ1ε uε · ∇ϕ
Ω κ Ω
Z
κ
+ µ1 n1ε (1 − n1ε − a1 n2ε )ϕ
Ω

for all t ∈ (0, ∞). From Lemma 3.9, we can fix positive constants c2 , c3 and c4
such that
(3.41) |niε | ≤ c2 , |∇vε | ≤ c3 and |uε | ≤ c4 in Ω × (0, ∞) for all ε ∈ (0, 1),
i = 1, 2, on account of the fact that D1ε < D1 + 2ε in (0, ∞) for all ε ∈ (0, 1),
we have
(3.42) D1ε (n1ε ) ≤ c5 := kD1 kL∞ (0,c2 ) + 2 in Ω × (0, ∞) for all ε ∈ (0, 1).
Let p := κ − m1 + 1 satisfies p > 1 and p ≥ m − 1, by (3.26), yield that
Z ∞Z Z ∞Z
κ−2 2 p+m1 −3
(3.43) n1ε |∇n1ε | = n1ε |∇n1ε |2 ≤ c6
0 Ω 0 Ω

for all t ∈ (0, ∞) with certain constant c6 > 0. Using (3.41), (3.42) and Young’s
inequality, we have
(3.44) Z Z 
κ−2
−(κ − 1) 2
n1ε D1ε (n1ε )|∇n1ε | ϕ ≤ (κ − 1)c5 · nκ−2 2
1ε |∇n1ε | ·kϕkL∞ (Ω)
Ω Ω
and
Z
κ−1
− n1ε D1ε (n1ε )∇n1ε · ∇ϕ
Ω
234 B. LIU AND G. REN

Z 
≤ c5 · nκ−1
1ε |∇n 1ε ·k∇ϕkL∞ (Ω)
|
Ω
 Z Z 
κ−2 2
(3.45) ≤ c5 · n1ε |∇n1ε | + nκ1ε ·k∇ϕkL∞ (Ω)
Ω Ω
 Z 
κ−2
≤ c5 n1ε |∇n1ε | + c24 cκ21 |Ω| ·k∇ϕkL∞ (Ω)
2
Ω
as well as
Z  
(κ − 1) nκ−1
1ε ∇n1ε · S1ε (x, n1ε , vε ) · ∇vε ϕ
Ω
Z 
≤ (κ − 1) · nκ−1
1ε |∇n 1ε ·CS1 c3 kϕkL∞ (Ω)
|
Ω
Z 
κ−2
(3.46) ≤ (κ − 1)CS1 c3 · n1ε |∇n1ε |2 + cκ2 |Ω| ·kϕkL∞ (Ω)
Ω
and
Z  
(3.47) nκ1ε S1ε (x, n1ε , vε ) · ∇vε ·∇ϕ ≤ cκ2 c3 CS1 |Ω|k∇ϕkL∞ (Ω)
Ω
as well as
Z
1 1 κ
(3.48) nκ1ε uε · ∇ϕ ≤ c c4 |Ω|k∇ϕkL∞ (Ω)
κ Ω κ 2
and
Z
(3.49) µ1 nκ1ε (1 − n1ε − a1 n2ε )ϕ ≤ µ1 (cκ2 + cκ+1
2 )kϕkL∞ (Ω) .
Ω

Due to W03,2 (Ω) ,→ W 1,∞ (Ω), combining with (3.44)-(3.49), there exists a
constant c7 > 0 such that
∂ Z 
k nκ1ε (·, t)k(W 3,2 (Ω))∗ ≤ c7 · nκ−2
1ε |∇n 1ε |2
+ 1
∂t 0
Ω
for all t ∈ (0, ∞) and any ε ∈ (0, 1). According to (3.43), for each T > 0 we
have
Z T
∂
(3.50) k nκ1ε (·, t)k(W 3,2 (Ω))∗ dt ≤ c6 c7 + c7 T
0 ∂t 0

for all ε ∈ (0, 1). A similar argument we have

Z T
∂
(3.51) k nκ2ε (·, t)k(W 3,2 (Ω))∗ dt ≤ c8 T + c8
0 ∂t 0

for all ε ∈ (0, 1), with some constant c8 > 0. Multiplying both sides of the first
equation in (2.10) by ψ and integrating by parts on Ω, we obtain
Z 
∂ 
(3.52) n1ε ψ
∂t
ZΩ h   i
= ∇ · D1ε (n1ε )∇n1ε − n1ε S1ε (x, n1ε , vε ) · ∇vε −uε ∇n1ε ·ψ
Ω
 GLOBAL EXISTENCE AND ASYMPTOTIC BEHAVIOR 235

Z
+ µ1 n1ε (1 − n1ε − a1 n2ε )ψ
Ω
Z Z   Z
= G1 (n1ε )4ψ + n1ε S1ε (x, n1ε , vε ) · ∇vε ∇ψ + n1ε uε · ∇ψ
Ω Ω Ω
Z
+ µ1 n1ε (1 − n1ε − a1 n2ε )ψ
Ω
Rs
for all t ∈ (0, ∞), where we have set G1 (s) := 0 D1 (σ)dσ for s ≥ 0. Recalling
that D1ε < D1 + 2ε in (0, ∞) we can estimate
 
(3.53) G1 (n1ε ) ≤ c9 := c2 · kD1 kL∞ (0,c21 ) + 2 in Ω × (0, ∞) for all ε ∈ (0, 1).

Similar to Lemma 3.23 in [43], there exist constants c10 , c11 > 0 such that
Z T
∂
(3.54) k n1ε (·, t)k(W 2,2 (Ω))∗ dt ≤ c10
0 ∂t 0

and
Z T
∂
(3.55) k n2ε (·, t)k(W 2,2 (Ω))∗ dt ≤ c11
0 ∂t 0

for all ε ∈ (0, 1). In accordance with Lemma 3.9, the Arzelá-Ascoli theorem
along with a standard extraction procedure yields a sequence (εj )j∈N ⊂ (0, 1)
with εj → 0 as j → ∞ such that
0
(3.56) vε → v in Cloc (Ω × [0, ∞)),

0
(3.57) ∇vε → ∇v in Cloc (Ω × (0, ∞)),

0
(3.58) uε → u in Cloc (Ω × [0, ∞)),
hold with some limit functions v and u belonging to the indicated spaces.
Passing to a subsequence if necessary, by means of Proposition 3.1 we can
achieve that for some ni ∈ L∞ (Ω × (0, ∞)) we moreover have
∗
(3.59) niε * ni in L∞ (Ω × (0, ∞)), i = 1, 2,

∗
(3.60) ∇vε * ∇v in L∞ (Ω × (0, ∞)),

∗
(3.61) Duε * Du in L∞ (Ω × (0, ∞)).
We fix κ > m1 satisfy κ ≥ 2(m1 − 1) and combine (3.51) with (3.44) for
p := 2κ − m1 + 1 to see 
that
for each T > 0, (εκ )ε∈(0,1) is bounded in
L2 ((0, T ); W 1,2 (Ω)) with ∂t
∂ κ
n1ε being bounded in L1 ((0, T ); (W03,2 (Ω))? ).
ε∈(0,1)
Therefore, an Aubin-Lions lemma applies to yield strong precompactness of
(nκ1ε )ε∈(0,1) in L2 (Ω × (0, T )), whence along a suitable subsequence we have
236 B. LIU AND G. REN

nκ1ε → bκ and hence n1ε → b a.e. in Ω × (0, ∞) for some nonnegative mea-
surable b : Ω × (0, ∞) → R. By Egorov’s theorem, we know that necessarily
n1 = b, thus, we have
(3.62) niε → ni a.e. in Ω × (0, ∞), i = 1, 2.
Finally, as the embedding L∞ (Ω) ,→ (W02,2 (Ω))? is compact, the Arzelà-Ascoli
once more applies to say that the equicontinuity property (3.54) together with
the boundedness of (n1ε )ε∈(0,1) in C 0 ([0, ∞); L∞ (Ω) ensures that
 
(3.63) 0
niε → ni in Cloc [0, ∞); (W02,2 (Ω))∗ , i = 1, 2,

holds after a further extraction of an adequate subsequence. The additional

regularity property
0 ∞ 0 ∞
(3.64) n1 ∈ C w −∗ ([0, ∞); L (Ω)), n2 ∈ Cw −∗ ([0, ∞); L (Ω)).
thereafter is a consequence of (3.63) and the fact that Ci := kni kL∞ (Ω×(0,∞)) ,
i = 1, 2 is finite: first, from the latter property it follows that there exists a null
set N ⊂ [0, ∞) such that for all t ∈ [0, ∞)\N we have ni (·, t) ∈ L∞ (Ω) with
kni (·, t)kL∞ (Ω) ≤ Ci , i = 1, 2. As [0, ∞)\N is dense in [0, ∞), for an arbitrary
t0 ∈ [0, ∞) we can find (tj )j∈N ⊂ [0, ∞)\N such that tj → t0 as j → ∞,
∗
and extracting a subsequence if necessary we can obtain that ni (·, t) * n̄i in
L∞ (Ω) as j → ∞, with some n̄i ∈ L∞ (Ω) satisfying kn̄i kL∞ (Ω) ≤ Ci , i = 1, 2.
Since (3.61) assert that moreover ni (·, tj ) → ni (·, t0 ) in (W02,2 (Ω))? as j → ∞,
this allows us to identify n̄i = ni (·, t0 ) and to conclude that thus actually
ni (·, t) ∈ L∞ (Ω) for all t ∈ [0, ∞), with kni (·, t)kL∞ (Ω) ≤ Ci for all t ≥ 0.
The property (3.62) can now be verified by partially repeating this argument:
given any t0 ≥ 0 and (tj )j∈N ⊂ [0, ∞)\N such that tj → t0 as j → ∞ we
∞ ∞
R is bounded in L (Ω), and that for all ϕ ∈ C0 (Ω)
know Rthat (ni (·, tj )))j∈N we
have Ω ni (·, tj ))ϕ → Ω ni (·, t0 )ϕ as j → ∞ by (3.61). By density of C0∞ (Ω)
∗
in L1 (Ω), this proves that indeed ni (·, tj ) * ni (·, t0 ) in L∞ (Ω) as j → ∞.
Now the verification of the claimed weak solution property of (n1 , n2 , v, u)
is straightforward: whereas the nonnegativity of ni , i = 1, 2 and v and the
integrability requirements in conditions of Definition 2.1 are immediate from
(3.56)-(3.59) and (3.62), the integral identities in Definition 2.1 can be derived
by standard arguments from the corresponding weak formulations in the ap-
proximate system (2.1) upon letting ε = εj → 0 and using (3.59) and (3.62) as
well as (3.56)-(3.58), (3.60) and (3.61). This completes the proof. 

4. Asymptotic behavior of the solution

In this section, we consider a special case with D1 (n1 ) = D2 (n2 ) ≡ 1, then
the weak solution becomes classical solution, we begin with stating the main
theorem, and construct the function to prove it. As a first step towards this,
we show the main theorem.
 GLOBAL EXISTENCE AND ASYMPTOTIC BEHAVIOR 237

Theorem 4.1. Let D1 (n1 ) = D2 (n2 ) ≡ 1, assume that the condition of Theo-
rem 3.1 holds. Then the solution of (1.1) has the following properties:
(i) Let a1 , a2 ∈ (0, 1), under the condition that there exists γ1 such that
4γ1 − (1 + γ1 )2 a1 a2 > 0
and
c1 CS21 (1 − a1 ) γ1 c2 CS22 (1 − a2 ) 4γ1 − (1 + γ1 )2 a1 a2
+ < ,
4a1 µ1 (1 − a1 a2 ) 4a2 µ2 (1 − a1 a2 ) a1 α2 γ1 + a2 β 2 − a1 a2 αβ(1 + γ1 )
then
n1 (·, t) → N1 , n2 (·, t) → N2 , v(·, t) → V1 , u(·, t) → 0 in L∞ as t → ∞,
where
1 − a1 1 − a2
N1 := , N2 := , V1 := αN1 + βN2
1 − a1 a2 1 − a1 a2
as well as
c1 = max{1, (1 + kn1 kL∞ [0,kn10 kL∞ (Ω) +1] )1−α1 }
and
c2 = max{1, (1 + kn2 kL∞ [0,kn20 kL∞ (Ω) +1] )1−α2 }.
(ii) Let a1 ≥ 1 > a2 under the condition that there exist γ3 and a01 ∈ [1, a1 ]
such that
4γ3 − (1 + γ3 )2 a01 a2 > 0
and
CS22 γ3 c2 (α2 a01 γ3 + β 2 a2 − αβa01 a2 (1 + γ3 ))
µ2 > ,
4a2 (4γ3 − a01 a2 (1 + γ3 )2 )
then
n1 (·, t) → 0, n2 (·, t) → 1, v(·, t) → β, u(·, t) → 0 in L∞ as t → ∞,
where
c2 = max{1, (1 + kn2 kL∞ [0,kn20 kL∞ (Ω) +1] )1−α2 }.
We will give the following lemma which will give stabilization in (1.1).
Lemma 4.1 ([9, Lemma 4.6]). Let n ∈ C 0 (Ω × [0, ∞)) satisfying that there
exist C1 > 0 and θ ∈ (0, 1) such that
knk θ ≤ C for all t ≥ 1.
C θ, 2 (Ω×[t,t+1])

Assume that Z ∞ Z
(n − N ∗ )2 < ∞
1 Ω
with some constant N ∗ . Then
n(·, t) → N ∗ as t → ∞.
238 B. LIU AND G. REN

Lemma 4.2 ([23, Lemma 2.1]). Let a, b, c, d, e, f ∈ R. Suppose that

b2 bce c2 d b2 f ae2
a > 0, ad − > 0, adf + − − − > 0.
4 4 4 4 4
Then there exists ε > 0 such that
ay12 + by1 y2 + cy1 y3 + dy22 + ey2 y3 + f y32 ≥ ε(y12 + y22 + y32 )
holds for all y1 , y2 , y3 ∈ R.
4.1. Case 1: a1 , a2 ∈ (0, 1)
In this subsection, we will analysis convergence of (n1 , n2 , v, u) as t → 0.
To get the desire results, we give the following key estimate for stabilization in
(1.1) in the case a1 , a2 ∈ (0, 1).
Lemma 4.3. Let a1 , a2 ∈ (0, 1). Under the assumption of Theorem 4.1(i), the
solution of (1.1) has the property that there exist γ1 , γ2 > 0 and ε1 > 0 such
that the nonnegative function E1 and F1 defined by
Z Z Z
n1 a1 µ1 n2 γ2
E1 := (n1 − N1 − N1 ln ) + γ1 (n2 − N2 − N2 ln )+ v2
Ω N 1 a µ
2 2 Ω N 2 2 Ω
and Z Z Z
F1 := (n1 − N1 )2 + (n2 − N2 )2 + (v − V1 )2
Ω Ω Ω
satisfy
(4.1) E1 (t) ≥ 0 for all t > 0
and
d
(4.2) E1 (t) ≤ −ε1 F1 (t) for all t > 0,
dt
1−a1 1−a2
where N1 = 1−a 1 a2
, N2 = 1−a 1 a2
and V1 = αN1 + βN2 .
Proof. Let A1 (t), B1 (t) and C1 (t) defined as
Z
n1
A1 (t) := (n1 − N1 − N1 ln ),
Ω N 1
Z
n2
B1 (t) := (n2 − N2 − N2 ln )
Ω N 2
and Z
1
C1 (t) := v2 ,
2 Ω
and we write
a1 µ1
E1 (t) = A1 (t) + γ1 B1 (t) + γ2 C1 (t).
a2 µ2
Taking the similar procedure in [9, Lemma 4.1], we know (4.1) holds. By the
straightforward calculation we have
d d a1 µ1 d d
E1 (t) = A1 (t) + γ1 B1 (t) + γ2 C1 (t)
dt dt a2 µ2 dt dt
 GLOBAL EXISTENCE AND ASYMPTOTIC BEHAVIOR 239

Z Z
(4.3) ≤ − µ1 (n1 − N1 )2 − a1 µ1 (1 + γ1 ) (n1 − N1 )(n2 − N2 )
Ω Ω
|∇n1 |2 ∇n1 · ∇v
Z Z
− N1 + CS1 N1
Ω n21 Ω (1 + n1 )
α1 +1

|∇n2 |2
Z Z
γ1 a1 µ1 γ1 a1 µ1 N2
− (n2 − N2 )2 −
a2 Ω a2 µ2 Ω n22
∇n2 · ∇v
Z Z Z
γ1 a1 µ1 N2 CS2 2
+ α2 +1
− γ 2 |∇v| − γ 2 (v − V1 )2
a2 µ2 Ω (1 + n2 ) Ω Ω
Z Z
+ γ2 α (n1 − N1 )(v − V1 ) + γ2 β (n2 − N2 )(v − V1 ).
Ω Ω
Using the (2.5) and Young’s inequality we have
∇n1 · ∇v |∇n1 |2 N1 CS21 |∇v|2
Z Z Z
(4.4) N1 CS1 α +1
≤ N1 3 + α1 −1
,
Ω (1 + n1 ) n1 4 Ω (1 + n1 )
1
Ω
due to
(4.5) (
(1 + kn1 kL∞ [0,kn01 kL∞ (Ω) ] )1−α1 |∇v|2 , if α1 < 1,
R
|∇v|2
Z
≤ Ω
(1 + n1 )α1 −2 |∇v|2 , if α1 ≥ 1,
R
Ω Ω

combining with (3.30) and (4.5), there exists a positive constant c1 such that
|∇v|2
Z Z
(4.6) α1 −1
≤ c 1 |∇v|2 ,
Ω (1 + n1 ) Ω
substituting (4.6) into (4.4), we have
∇n1 · ∇v |∇n1 |2 c1 N1 CS21
Z Z Z
(4.7) N1 CS1 ≤ N1 + |∇v|2 .
Ω (1 + n1 )
α1 +1
Ω n21 4 Ω
Similar to (4.7), there exists a constant c2 > 0 such that
∇n2 · ∇v
Z
γ1 a1 µ1 N2 CS2
a2 µ2 Ω (1 + n2 )α2 +1
2 Z
γ1 a1 µ1 N2 CS2 |∇v|2 |∇n2 |2
Z
γ1 a1 µ1 N2
(4.8) ≤ +
4a2 µ2 Ω (1 + n2 )
α2 −1 a2 µ2 Ω n22
2
|∇n2 |2 c2 γ1 a1 µ1 N2 CS2
Z Z
γ1 a1 µ1 N2
≤ 2 + |∇v|2 .
a2 µ2 Ω n2 4a2 µ2 Ω
Combining with (4.3)-(4.8), we obtain
Z Z
d
E1 (t) ≤ − µ1 (n1 − N1 )2 − a1 µ1 (1 + γ1 ) (n1 − N1 )(n2 − N2 )
dt Ω Ω
Z Z
γ1 a1 µ1
+ γ2 α (n1 − N1 )(v − V1 ) − (n2 − N2 )2
Ω a2 Ω
Z Z
+ γ2 β (n2 − N2 )(v − V1 ) − γ2 (v − V1 )2
Ω Ω
240 B. LIU AND G. REN

c1 N1 CS21 c2 γ1 a1 µ1 N2 CS22 
 Z
− γ2 − − |∇v|2 .
4 4a2 µ2 Ω

By the condition of Theorem 4.1(i), taking γ2 > 0 satisfies

c1 CS21 N1 γ1 a1 µ1 c2 CS22 N2 a1 µ1 [4γ1 − (1 + γ1 )2 a1 a2 ]

+ < γ2 < ,
4 4a2 µ2 a1 α2 γ1 + a2 β 2 − a1 a2 αβ(1 + γ1 )
it is easy to know that
c1 N1 CS21 c2 γ1 a1 µ1 N2 CS22
γ2 − − > 0.
4 4a2 µ2
Letting
γ1 a1 µ1
a = µ1 , b = a1 µ1 (1 + γ1 ), c = −γ2 α, d = , e = −γ2 β, f = γ2 .
a2
By the straightforward calculation we have
b2 a1 µ21 (4γ1 − (1 + γ1 )2 a1 a2 )
a = µ1 > 0, ad − = >0
4 4a2
and
bce c2 d b2 f ae2
adf + − − −
4 4 4 4
µ1 γ2
= [a1 µ1 (4γ1 −(1+γ1 )2 a1 a2 ) − (a1 α2 γ1 + a2 β 2 −a1 a2 αβ(1 + γ1 ))γ2 ] > 0.
4a2
From Lemma 4.2, there exists a constant ε1 > 0 such that
d Z Z Z 
E1 (t) ≤ −ε1 (n1 − N1 ) + (n2 − N2 ) + (v − V1 )2 for all t > 0.
2 2
dt Ω Ω Ω

This completes the proof. 

Lemma 4.4. Let a1 , a2 ∈ (0, 1). Under the assumption of Theorem 4.1(i), the
solution of (1.1) satisfies that there exists a constant C2 > 0 such that
Z ∞Z Z ∞Z Z ∞Z
(4.9) (n1 − N1 )2 + (n2 − N2 )2 + (v − V1 )2 ≤ C2 .
1 Ω 1 Ω 1 Ω

Proof. Integrating (4.2) over (1, t), we infer

Z t
(4.10) E1 (t) + ε1 F1 (s)ds ≤ E1 (0).
1

Therefore, combination of (4.10) with (4.2) implies (4.9). This completes the
proof. 
 GLOBAL EXISTENCE AND ASYMPTOTIC BEHAVIOR 241

4.2. Case 2: a1 ≥ 1 > a2

In this subsection, we will analysis convergence of (n1 , n2 , v, u) as t → 0.
To get the desire results, we give the following key estimate for stabilization in
(1.1) in the case a1 ≥ 1 > a2 .
Lemma 4.5. Let a1 ≥ 1 > a2 . Under the assumption of Theorem 4.1(ii), the
solution of (1.1) has the property that there exist γ3 , γ4 > 0 and ε2 > 0 such
that the nonnegative function E2 and F2 defined by
a0 µ1
Z Z Z
γ4
E2 := n1 + γ3 1 (n2 − 1 − ln n2 ) + (v − β)2
Ω a2 µ2 Ω 2 Ω
and Z Z Z
F2 := n21 + (n2 − 1)2 + (v − β)2
Ω Ω Ω
satisfy
(4.11) E2 (t) ≥ 0 for all t > 0
and
d
(4.12) E2 (t) ≤ −ε2 F2 (t) for all t > 0.
dt
Proof. Letting
a01 µ1
Z Z Z
γ4
E2 (t) = n1 + γ3 (n2 − 1 − ln n2 ) + (v − β)2 .
Ω a2 µ2 Ω 2 Ω

Taking the similar procedure in [9, Lemma 4.1], we know (4.11) holds. By the
straightforward calculation we have
Z Z Z
d 2 0 0
E2 (t) ≤ − µ1 n1 − a1 µ1 (1 + γ3 ) n1 (n2 − 1) − µ1 (a1 − 1) n1
dt Ω Ω Ω
γ3 a01 µ1 γ3 a01 µ1 |∇n2 |2
Z Z
− (n2 − N2 )2 −
a2 Ω a2 µ2 Ω n22
γ3 a01 µ1 CS2 ∇n2 · ∇v
Z Z
(4.13) + α2 +1
− γ4 |∇v|2
a2 µ2 Ω (1 + n2 ) Ω
Z Z Z
− γ4 (v − β)2 + γ4 α n1 (v − β) + γ4 β (n2 − 1)(v − β).
Ω Ω Ω

Similar to (4.8), there exists a constant c2 > 0 such that

(4.14)
γ3 a01 µ1 CS2 ∇n2 · ∇v γ3 a01 µ1 |∇n2 |2 c2 γ3 a01 µ1 CS22
Z Z Z
≤ + |∇v|2 .
a2 µ2 Ω (1 + n2 )
α2 +1 a2 µ2 Ω n22 4a2 µ2 Ω

Combining (4.13) with (4.14), we obtain

Z Z Z
d 2 0
E2 (t) ≤ − µ1 n1 − a1 µ1 (1 + γ3 ) n1 (n2 − 1) + γ4 α n1 (v − β)
dt Ω Ω Ω
242 B. LIU AND G. REN

γ3 a01 µ1
Z Z
(4.15) − (n2 − 1)2 + γ4 β (n2 − 1)(v − β)
a2 Ω Ω
c2 γ3 a01 µ1 CS22 
Z  Z
2
− γ4 (v − β) − γ4 − |∇v|2 .
Ω 4a µ
2 2 Ω
By the condition of Theorem 4.1(ii), taking γ4 > 0 satisfies
γ3 a01 µ1 c2 CS22 a0 µ1 [4γ3 − (1 + γ3 )2 a01 a2 ]
< γ2 < 0 2 1 ,
4a2 µ2 a1 α γ3 + a2 β 2 − a01 a2 αβ(1 + γ3 )
it is easy to know that
c2 γ3 a01 µ1 CS22
γ4 − > 0.
4a2 µ2
Letting
γ3 a01 µ1
a = µ1 , b = a01 µ1 (1 + γ3 ), c = −γ4 α, d = , e = −γ4 β, f = γ4 .
a2
By the straightforward calculation we have
b2 a0 µ2 (4γ3 − (1 + γ3 )2 a01 a2 )
a = µ1 > 0, ad − = 1 1 >0
4 4a2
and
bce c2 d b2 f ae2
adf + − − −
4 4 4 4
µ1 γ4 0
= [a µ1 (4γ3 −(1 + γ3 )2 a01 a2 ) − (a01 α2 γ3 + a2 β 2 −a01 a2 αβ(1 + γ3 ))γ4 ] > 0.
4a2 1
From Lemma 4.2, there exists a constant ε2 > 0 such that
d Z Z Z 
E2 (t) ≤ −ε2 (n1 − N1 )2 + (n2 − N2 )2 + (v − V1 )2 for all t > 0.
dt Ω Ω Ω
This completes the proof. 
Lemma 4.6. Let a1 ≥ 1 > a2 . Under the assumption of Theorem 4.1(ii), the
solution of (1.1) satisfies that there exists a constant C3 > 0 such that
Z ∞Z Z ∞Z Z ∞Z
2 2
(4.16) n1 + (n2 − 1) + (v − β)2 ≤ C3 .
1 Ω 1 Ω 1 Ω

Proof. The proof is similar to Lemma 4.4, so we omitted it. 

Lemma 4.7. Under the assumptions of Theorem 2.1 and Theorem 2.2, the
solution of (1.1) has the following property:
(4.17) ku(·, t)kL∞ (Ω) → 0 as t → ∞.
Proof. Noting
R ∞ R from Lemma 4.3 and Lemma 4.5 that exists a constant T > 0
such that T Ω (αn1 +βn2 )2 v 2 < ∞ and using Lemma 4.5 in [5], we can obtain
that Z ∞Z
v 2 < ∞,
T Ω
 GLOBAL EXISTENCE AND ASYMPTOTIC BEHAVIOR 243

which entails kvkL∞ (Ω) → 0. Next we will show that kukL∞ (Ω) → 0, similar to
[5, Lemma 4.6], there exist some positive constants c1 , c2 , c3 such that
kukL∞ (Ω) ≤ c1 kAθ ukL2 (Ω) ≤ c2 kA ukaL2 (Ω) kAθ uk1−a 1−a
L2 (Ω) ≤ c3 kukL2 (Ω) ,

where θ ∈ ( 43 , ) and a = θ ∈ (0, 1), which means that it is sufficient to show

that
ku(·, t)kL2 (Ω) → 0 as t → ∞.
From the Poincaré inequality that there exists a constant c4 > 0 such that
ku(·, t)k2L2 (Ω) ≤ c4 k∇u(·, t)k2L2 (Ω)
for all t ∈ (0, ∞). Put (n̄1 , n̄2 ) := (N1 , N2 ) if a1 , a2 ∈ (0, 1) or (n̄1 , n̄2 ) := (0, 1)
if a1 ≥ 1 > 0. We infer from the fourth equation in (2.10) and Young’s
inequality that
Z Z
1 d
u2 + |∇u|2
2 dt Ω Ω
Z Z
= (n1 − n̄1 + n2 − n̄2 )∇φ · u + (n̄1 + n̄2 ) ∇φ · u
Ω Ω
Z Z Z Z
1 2

2 2

≤ u + c5 (n1 − n̄1 ) + (n2 − n̄2 ) +(n̄1 + n̄2 ) ∇φ · u
4c4 Ω Ω Ω Ω

for all t ∈ (0, ∞) with some constant c5 > 0. Letting

Z Z Z 
y(t) := u2 and h(t) := 2c5 (n1 − n̄1 )2 + (n2 − n̄2 )2
Ω Ω Ω

satisfy
y 0 (t) + c6 y(t) ≤ h(t)
with some c6 > 0. Hence it holds that
Z t
−c6 t
y(t) ≤ y(0)e + e−c6 (t−s) h(s)ds
0
t
Z 2
Z t
(4.18) ≤ y(0)e−c6 t + e−c6 (t−s) h(s)ds + e−c6 (t−s) h(s)ds.
t
0 2

From Proposition 3.1 that there exists a constant c7 > 0 such that h(s) ≤ c7
for all s > 0 and hence we have
Z 2t Z 2t
c6
−c6 (t−s) −c6 t
(4.19) e h(s)ds ≤ c7 e ec6 s ds ≤ c8 e− 2 t
0 0

with
R ∞ some c7 > 0. On the other hand, noting from (4.9) and (4.13) that
0
h(s)ds < ∞, we have
Z t Z t
(4.20) 0≤ e−c6 (t−s) h(s)ds ≤ h(s)ds → 0 as t → ∞.
t t
2 2
244 B. LIU AND G. REN

Thus, combination of (4.18) with (4.19) and (4.20) leads to

ku(·, t)kL2 (Ω) = y(t) → 0 as t → ∞.
(4.17) follows from the dominated convergence theorem. This completes the
proof. 
Finally, we prove the main theorem.
The proof of Theorem 4.1. A combination of Lemmata 4.1-4.7 directly leads to
Theorem 4.1. 
Acknowledgment. The authors express their gratitude to the anonymous
reviewers and editors for their valuable comments and suggestions which led to
the improvement of the original manuscript.

References
[1] X. Bai and M. Winkler, Equilibration in a fully parabolic two-species chemotaxis system
with competitive kinetics, Indiana Univ. Math. J. 65 (2016), no. 2, 553–583. https:
//doi.org/10.1512/iumj.2016.65.5776
[2] T. Black, Global very weak solutions to a chemotaxis-fluid system with nonlinear dif-
fusion, SIAM J. Math. Anal. 50 (2018), no. 4, 4087–4116. https://doi.org/10.1137/
17M1159488
[3] N. Bellomo, A. Bellouquid, Y. Tao, and M. Winkler, Toward a mathematical theory of
Keller-Segel models of pattern formation in biological tissues, Math. Models Methods
Appl. Sci. 25 (2015), no. 9, 1663–1763. https://doi.org/10.1142/S021820251550044X
[4] X. Cao, Global classical solutions in chemotaxis(-Navier)-Stokes system with rotational
flux term, J. Differential Equations 261 (2016), no. 12, 6883–6914. https://doi.org/
10.1016/j.jde.2016.09.007
[5] X. Cao, S. Kurima, and M. Mizukami, Global existence an asymptotic behavior of clas-
sical solutions for a 3D two-species chemotaxis-Stokes system with competitive kinetics,
Math. Methods Appl. Sci. 41 (2018), no. 8, 3138–3154. https://doi.org/10.1002/mma.
4807
[6] , Global existence and asymptotic behavior of classical solutions for a 3D two-
species Keller–Segel-Stokes system with competitive kinetics, preprint.
[7] D. Henry, Geometric theory of semilinear parabolic equations, Lecture Notes in Mathe-
matics, 840, Springer-Verlag, Berlin, 1981.
[8] M. A. Herrero and J. J. L. Velázquez, A blow-up mechanism for a chemotaxis model,
Ann. Scuola Norm. Sup. Pisa Cl. Sci. (4) 24 (1997), no. 4, 633–683.
[9] M. Hirata, S. Kurima, M. Mizukami, and T. Yokota, Boundedness and stabilization in a
two-dimensional two-species chemotaxis-Navier-Stokes system with competitive kinetics,
J. Differential Equations 263 (2017), no. 1, 470–490. https://doi.org/10.1016/j.jde.
2017.02.045
[10] , Boundedness and stabilization in a three-dimensional two-species chemotaxis-
Navier-Stokes system with competitive kinetics, Proceedings of EQUADIFF 2017 Con-
ference, (2017) 11–20.
[11] D. Horstmann and M. Winkler, Boundedness vs. blow-up in a chemotaxis system, J. Dif-
ferential Equations 215 (2005), no. 1, 52–107. https://doi.org/10.1016/j.jde.2004.
10.022
[12] S. Ishida, K. Seki, and T. Yokota, Boundedness in quasilinear Keller-Segel systems of
parabolic-parabolic type on non-convex bounded domains, J. Differential Equations 256
(2014), no. 8, 2993–3010. https://doi.org/10.1016/j.jde.2014.01.028
 GLOBAL EXISTENCE AND ASYMPTOTIC BEHAVIOR 245

[13] E. F. Keller and L. A. Segel, Initiation of slime mold aggregation viewed as an instability,
J. Theoret. Biol. 26 (1970), no. 3, 399–415. https://doi.org/10.1016/0022-5193(70)
90092-5
[14] O. A. Ladyženskaja, V. A. Solonnikov, and N. N. Ural’ceva, Linear and quasilinear equa-
tions of parabolic type (Russian), Translated from the Russian by S. Smith. Translations
of Mathematical Monographs, Vol. 23, American Mathematical Society, Providence, RI,
1968.
[15] H. Li and Y. Tao, Boundedness in a chemotaxis system with indirect signal production
and generalized logistic source, Appl. Math. Lett. 77 (2018), 108–113. https://doi.
org/10.1016/j.aml.2017.10.006
[16] X. Li, Global classical solutions in a Keller-Segel(-Navier)-Stokes system modeling coral
fertilization, J. Differential Equations, In press.
[17] K. Lin, C. Mu, and L. Wang, Boundedness in a two-species chemotaxis system, Math.
Methods Appl. Sci. 38 (2015), no. 18, 5085–5096. https://doi.org/10.1002/mma.3429
[18] K. Lin, C. Mu, and H. Zhong, A new approach toward stabilization in a two-species
chemotaxis model with logistic source, Comput. Math. Appl. 75 (2018), no. 3, 837–849.
https://doi.org/10.1016/j.camwa.2017.10.007
[19] J. Liu and Y. Wang, Boundedness and decay property in a three-dimensional Keller-
Segel-Stokes system involving tensor-valued sensitivity with saturation, J. Differential
Equations 261 (2016), no. 2, 967–999. https://doi.org/10.1016/j.jde.2016.03.030
[20] , Global existence and boundedness in a Keller-Segel-(Navier-)Stokes system with
signal-dependent sensitivity, J. Math. Anal. Appl. 447 (2017), no. 1, 499–528. https:
//doi.org/10.1016/j.jmaa.2016.10.028
[21] , Global weak solutions in a three-dimensional Keller-Segel-Navier-Stokes sys-
tem involving a tensor-valued sensitivity with saturation, J. Differential Equations 262
(2017), no. 10, 5271–5305. https://doi.org/10.1016/j.jde.2017.01.024
[22] M. Mizukami, Boundedness and asymptotic stability in a two-species chemotaxis-
competition model with signal-dependent sensitivity, Discrete Contin. Dyn. Syst. Ser.
B 22 (2017), no. 6, 2301–2319. https://doi.org/10.3934/dcdsb.2017097
[23] , Boundedness and asymptotic stability in a two-species chemotaxis-competition
model with signal-dependent sensitivity, Discrete Contin. Dyn. Syst. Ser. B 22 (2017),
no. 6, 2301–2319. https://doi.org/10.3934/dcdsb.2017097
[24] M. Mizukami and T. Yokota, Global existence and asymptotic stability of solutions to
a two-species chemotaxis system with any chemical diffusion, J. Differential Equations
261 (2016), no. 5, 2650–2669. https://doi.org/10.1016/j.jde.2016.05.008
[25] T. Nagai, T. Senba, and K. Yoshida, Application of the Trudinger-Moser inequality to
a parabolic system of chemotaxis, Funkcial. Ekvac. 40 (1997), no. 3, 411–433.
[26] M. Negreanu and J. I. Tello, On a two species chemotaxis model with slow chemical
diffusion, SIAM J. Math. Anal. 46 (2014), no. 6, 3761–3781. https://doi.org/10.
1137/140971853
[27] , Asymptotic stability of a two species chemotaxis system with non-diffusive
chemoattractant, J. Differential Equations 258 (2015), no. 5, 1592–1617. https://doi.
org/10.1016/j.jde.2014.11.009
[28] G. Ren and B. Liu, Global boundedness and asymptotic behavior in a two-species
chemotaxis-competition system with two signals, Nonlinear Anal. Real World Appl. 48
(2019), 288–325. https://doi.org/10.1016/j.nonrwa.2019.01.017
[29] , Global existence of bounded solutions for a quasilinear chemotaxis system with
logistic source, Nonlinear Anal. Real World Appl. 46 (2019), 545–582. https://doi.
org/10.1016/j.nonrwa.2018.09.020
[30] , Boundedness of solutions for a quasilinear chemotaxis-haptotaxis model,
Hakkaido Mathematical Journal, Accepted.
246 B. LIU AND G. REN

[31] H. Sohr, The Navier-Stokes equations, Birkhäuser Advanced Texts: Basler Lehrbücher.,
Birkhäuser Verlag, Basel, 2001.
[32] C. Stinner, J. I. Tello, and M. Winkler, Competitive exclusion in a two-species chemo-
taxis model, J. Math. Biol. 68 (2014), no. 7, 1607–1626. https://doi.org/10.1007/
s00285-013-0681-7
[33] Y. Tao and M. Winkler, Locally bounded global solutions in a three-dimensional
chemotaxis-Stokes system with nonlinear diffusion, Ann. Inst. H. Poincaré Anal. Non
Linéaire 30 (2013), no. 1, 157–178. https://doi.org/10.1016/j.anihpc.2012.07.002
[34] , Boundedness and decay enforced by quadratic degradation in a three-dimen-
sional chemotaxis-fluid system, Z. Angew. Math. Phys. 66 (2015), no. 5, 2555–2573.
https://doi.org/10.1007/s00033-015-0541-y
[35] , Blow-up prevention by quadratic degradation in a two-dimensional Keller-Segel-
Navier-Stokes system, Z. Angew. Math. Phys. 67 (2016), no. 6, Art. 138, 23 pp. https:
//doi.org/10.1007/s00033-016-0732-1
[36] I. Tuval, L. Cisneros, C. Dombrowski, C. W. Wolgemuth, J. O. Kessler, and R. E.
Goldstein, Bacterial swimming and oxygen transport near contact lines, Proc. Natl.
Acad. Sci. 102 (2005), 2277–2282.
[37] Y. Wang, Global weak solutions in a three-dimensional Keller-Segel-Navier-Stokes sys-
tem with subcritical sensitivity, Math. Models Methods Appl. Sci. 27 (2017), no. 14,
2745–2780. https://doi.org/10.1142/S0218202517500579
[38] Y. Wang, M. Winkler, and Z. Xiang, Global classical solutions in a two-dimensional
chemotaxis-Navier-Stokes system with subcritical sensitivity, Ann. Sc. Norm. Super.
Pisa Cl. Sci. (5) 18 (2018), no. 2, 421–466.
[39] Y. Wang and Z. Xiang, Global existence and boundedness in a Keller-Segel-Stokes sys-
tem involving a tensor-valued sensitivity with saturation, J. Differential Equations 259
(2015), no. 12, 7578–7609. https://doi.org/10.1016/j.jde.2015.08.027
[40] , Global existence and boundedness in a Keller-Segel-Stokes system involving a
tensor-valued sensitivity with saturation: the 3D case, J. Differential Equations 261
(2016), no. 9, 4944–4973. https://doi.org/10.1016/j.jde.2016.07.010
[41] M. Winkler, Aggregation vs. global diffusive behavior in the higher-dimensional Keller-
Segel model, J. Differential Equations 248 (2010), no. 12, 2889–2905. https://doi.org/
10.1016/j.jde.2010.02.008
[42] , Global large-data solutions in a chemotaxis-(Navier-)Stokes system modeling
cellular swimming in fluid drops, Comm. Partial Differential Equations 37 (2012), no. 2,
319–351. https://doi.org/10.1080/03605302.2011.591865
[43] , Boundedness and large time behavior in a three-dimensional chemotaxis-Stokes
system with nonlinear diffusion and general sensitivity, Calc. Var. Partial Differential
Equations 54 (2015), no. 4, 3789–3828. https://doi.org/10.1007/s00526-015-0922-2
[44] , How far do chemotaxis-driven forces influence regularity in the Navier-Stokes
system?, Trans. Amer. Math. Soc. 369 (2017), no. 5, 3067–3125. https://doi.org/10.
1090/tran/6733
[45] , Does fluid interaction affect regularity in the three-dimensional Keller-Segel
system with saturated sensitivity?, J. Math. Fluid Mech. 20 (2018), no. 4, 1889–1909.
https://doi.org/10.1007/s00021-018-0395-0
[46] , Global existence and stabilization in a degenerate chemotaxis-Stokes system
with mildly strong diffusion enhancement, J. Differential Equations 264 (2018), no. 10,
6109–6151. https://doi.org/10.1016/j.jde.2018.01.027
[47] Q. Zhang and Y. Li, Global boundedness of solutions to a two-species chemotaxis system,
Z. Angew. Math. Phys. 66 (2015), no. 1, 83–93. https://doi.org/10.1007/s00033-013-
0383-4
 GLOBAL EXISTENCE AND ASYMPTOTIC BEHAVIOR 247

Bin Liu
School of Mathematics and Statistics
Huazhong University of Science and Technology
Wuhan 430074, Hubei, P. R. China
and
Hubei Key Laboratory of Engineering Modeling and Scientific Computing
Huazhong University of Science and Technology
Wuhan 430074, Hubei, P. R. China
Email address: binliu@mail.hust.edu.cn

Guoqiang Ren
School of Mathematics and Statistics
Huazhong University of Science and Technology
Wuhan 430074, Hubei, P. R. China
and
Hubei Key Laboratory of Engineering Modeling and Scientific Computing
Huazhong University of Science and Technology
Wuhan 430074, Hubei, P. R. China
and
Institute of Artificial Intelligence
Huazhong University of Science and
Technology, Wuhan 430074, Hubei, P. R. China
Email address: 597746385@qq.com