Home Stability on 3D Boussinesq system with mixed partial dissipation
Article Open Access

Stability on 3D Boussinesq system with mixed partial dissipation

  • Hongxia Lin EMAIL logo , Sen Liu , Xiaochuan Guo and Ruiqi You
Published/Copyright: December 31, 2024

Abstract

In the article, we are concerned with the three-dimensional anisotropic Boussinesq equations with the velocity dissipation in x 2 and x 3 directions and the thermal diffusion in only x 3 direction. When the spatial domain is the whole space R 3 , the global well-posedness and stability problem for the partially dissipated Boussinesq system remain the extremely challenging open problems. Attention here focuses on the periodic domain Ω = R × T 2 . We aim at establishing the stability for the problem of perturbations near hydrostatic equilibrium and the large-time behavior of the perturbed solution. We first obtain the global existence of some symmetric fluids in H 2 ( Ω ) for small initial data. Then the exponential decay rates for the oscillations u ˜ and θ in H 1 ( Ω ) and the homogeneous Sobolev space H v 2 ˙ ( Ω ) are also shown. The proof is based on a key observation that we can decompose the velocity u into the average u ¯ on T 2 and the corresponding oscillation u ˜ . This enables us to establish the strong Poincaré-type inequalities on u ˜ , u 3 , θ and some anisotropic inequalities, which ensure the establishment of the closed priori estimates. In addition, we also prove the oscillations in one direction u ˜ ( 2 ) , u ˜ ( 3 ) in H 1 ( Ω ) decay to zero exponentially.

MSC 2010: 35A05; 35Q35; 76D03;

1 Introduction

The Boussinesq system describes the influence of the convection (or convection-diffusion) phenomenon in a viscous or inviscid fluid, such as atmospheric fronts and ocean circulations. It plays an important role in the study of turbulence and suitably modified in atmospheric and oceanographic situations where rotation and stratification play a dominant role.

We are concerned with the three-dimensional incompressible anisotropic Boussinesq equations in the space domain Ω = R × T 2 with T 2 being the 2D periodic box:

(1.1) t U + U U = ν Δ v U P + Θ e 3 , x Ω , t > 0 , t Θ + U Θ = η 33 Θ , U = 0 , U ( x , 0 ) = U 0 ( x ) , Θ 0 ( x , 0 ) = Θ 0 ( x ) ,

where U represents the velocity field of the fluid, Θ is the temperature, P is the pressure, and e 3 = ( 0 , 0 , 1 ) . ν > 0 denotes the kinematic viscosity and η > 0 denotes the thermal diffusivity. For notational convenience, we make the notation i i for the partial derivatives x i i with i = 1 , 2 , 3 . We also introduce the abbreviation: Δ v = 22 + 33 , v = ( 2 , 3 ) , u v = ( u 2 , u 3 ) , and the norm notation u H v 1 ( Ω ) 2 = u L 2 ( Ω ) 2 + v u L 2 ( Ω ) 2 .

The objective of this article is to comprehend the stability of perturbations near the hydrostatic balance or hydrostatic equilibrium for the 3D Boussinesq equations (1.1) and to present accurate predictions of the long-term behavior. The motivations are twofold. First, many geophysical fluids are in an important equilibrium known as hydrostatic balance in physics. The hydrostatic equilibrium occurs when the force of gravity is balanced by the force of the pressure gradient. Actually, between the downward-directed force of gravity and the upward-directed force of the pressure gradient, our atmosphere is typically in a state of hydrostatic equilibrium. Gaining insight into certain severe weather phenomena may be made easier by being aware of the stability of perturbations close to the hydrostatic equilibrium [14,27]. Second, the Boussinesq equations (1.1), which only take into account partial dissipation, are used to simulate anisotropic fluids. Even though significant progress has been made on the large-time behavior of fully dissipated partial differential equation (PDE) systems [2931], anisotropic PDE system behavior is still a hot topic in the field of partial differential equations.

Due to its physical background and mathematical features, the 2D or 3D Boussinesq equations with full or partial dissipation have attracted considerable attention from the community of mathematical fluid mechanics. In the past decades, the global regularity problem has been a topic of interest and significant progress were made. One can refer to [1,3,5,6,7,912,15,2022,25,32] for partial dissipations and to [13,18,19,26,35,36,38] for fractional dissipations. Recently, the stability problem for the Boussinesq equations near the equilibrium state has also been examined. The stability of the system close to the hydrostatic equilibrium is a significant subject [2,16,17,23,28,33,36,37]. Said et al. [28] solved the stability problem for a two-dimensional Boussinesq system with only vertical dissipation and horizontal thermal diffusion in R 2 . Some excellent works on the stability for the Boussinesq system in the periodic domain has also been made [2,8,37]. To our best knowledge, there is no work to examine the global well-posedness and stability for the Boussinesq with the velocity dissipations in x 2 , x 3 directions and vertical diffusion. It remains an extremely challenging open problem in the whole space R 3 .

This article investigates the Boussinesq equations (1.1) on the periodic domain Ω = R × T 2 . We aims at the stability for the perturbations of (1.1) near the hydrostatic equilibrium ( U h e , P h e , Θ h e )

U h e ( 0 , 0 , 0 ) , P h e = 1 2 x 3 2 , Θ h e = x 3 .

Consider the perturbations

u = U U h e , θ = Θ Θ h e , p = P P h e .

Then ( u , θ ) satisfies

(1.2) t u + u u + p = ν Δ v u + θ e 3 , x Ω , t > 0 t θ + u θ + u 3 = η 33 θ , u = 0 , u ( x , 0 ) = u 0 ( x ) , θ ( x , 0 ) = θ 0 ( x ) .

With loss of generality, let T 2 = [ 1 2 , 1 2 ] × [ 1 2 , 1 2 ] . The domain Ω allows us to decompose a physical quantity into its average and corresponding oscillation portion, which is the difference from the whole space R 3 and will make the stability feasible. We introduce the following average of f on T 2 , T , and their oscillations.

(1.3) f ¯ ( x 1 ) = T 2 f ( x 1 , x 2 , x 3 ) d x 2 d x 3 , f ˜ = f f ¯ ,

(1.4) f ¯ ( i ) = T f ( x 1 , x 2 , x 3 ) d x i , f ˜ ( i ) = f f ¯ ( i ) , i = 2 , 3 .

This article attempts to achieve two goals. The first is to prove the stability of the hydrostatic equilibrium. The second goal is to establish the large-time behavior of solutions to (1.2). We shall show that ( u ˜ , θ ) and ( u ˜ ( 2 ) , u ˜ ( 3 ) ) obey the exponential decay rate.

Before stating our results, we first make the symmetry assumptions for the initial data

(1.5) u 01 , u 02 are even in x 3 , and u 03 and θ 0 are odd in x 3 .

We obtain the first result as follows.

Theorem 1.1

Assume ( u 0 , θ 0 ) H 2 ( Ω ) satisfies u 0 = 0 and (1.5). Then there exists δ > 0 depending on ν , η > 0 such that, if

u 0 H 2 ( Ω ) + θ 0 H 2 ( Ω ) δ ,

then the system (1.2) possesses a unique global solution ( u , θ ) C ( 0 , ; H 2 ( Ω ) ) satisfying

(1.6) u ( t ) H 2 ( Ω ) 2 + θ ( t ) H 2 ( Ω ) 2 + ν 0 t v u H 2 ( Ω ) 2 d τ + η 0 t 3 θ H 2 ( Ω ) 2 d τ C δ 2 ,

(1.7) u 1 , u 2 , p are e v e n i n x 3 , and u 3 and θ are o d d i n x 3 .

As aforementioned, when the spatial domain is the whole space R 3 , the global regularity and stability remain open problems. When in R 3 , it is easy to handle the nonlinear term u u by using the anisotropic inequality related to the triple product (see [4])

R 3 f g h d x C f L 2 1 2 1 f L 2 1 2 g L 2 1 2 2 g L 2 1 2 h L 2 1 2 3 h L 2 1 2 .

However, due to lack of horizontal thermal diffusion, it is very challenging to control the growth of the nonlinear term involving u θ , namely,

Ω u θ θ d x = 1 u θ 1 θ d x + 2 u θ 2 θ d x + 3 u θ 3 θ d x .

The only vertical thermal diffusion prevents us from establishing the closed bound.

Therefore, we turn to consider the stability in the domain Ω = R × T 2 . The periodic domain will introduce new obstacles. Due to the boundary, the aforementioned anisotropic inequalities will include additional adverse terms. More precisely,

Ω f g h d x C f L 2 1 2 g L 2 1 2 ( g L 2 + 2 g L 2 ) 1 2 h L 2 1 2 ( h L 2 + 3 h L 2 ) 1 2 ,

which makes the bound for the integral term involving u u more difficult than that in R 3 . To overcome the obstacle, we make full use of the properties of the oscillation part. More precisely, we can construct the strong Poincaré-types inequality:

(1.8) f ˜ H k ( Ω ) C v f ˜ H k ( Ω ) .

With the fine inequality (1.8) at our proposal, several anisotropic inequalities associated with the oscillation are able to be established (Lemma 2.4). All the preparations together with the key property v f ˜ = v f are sufficient to ensure to close the term Δ ( u u ) Δ u d x . For example, by (2.13), we can derive

j = 2 3 1 j u j 1 u ˜ 1 2 u d x C j = 2 3 1 j u j L 2 1 2 1 2 j u j L 2 1 2 v 2 1 u ˜ L 2 1 2 u L 2 C u H 2 v u H 2 2 .

We point out that it remains impossible to bound all parts of the nonlinear integral terms with θ since the temperature equations possess the diffusion only in vertical direction. Thus, we impose assumptions (1.5) on the initial data so that the solution ( u , θ ) obeys the same symmetry (1.7), see Subsection 3.1 for the detailed proof. Thereby, another strong Poincaré-types inequality can be derived as follows:

θ L 2 C 3 θ L 2 .

And the difficult terms involving θ can be solved.

The procedure for the proof of Theorem 1.1 is standard. We define the energy functional

E ( t ) = sup 0 τ t ( u ( τ ) , θ ( τ ) ) H 2 2 + 2 ν v u H 2 2 d x + 2 η 3 θ H 2 2 d x .

Efforts are devoted to establishing the a priori estimate

(1.9) E ( t ) C 0 E ( 0 ) + C 1 E 3 2 ( t ) ,

where C 0 > 0 , C 1 > 0 are two pure constants independent of δ . Then we can employ the bootstrapping argument (see Tao’s book [34], p. 21) to (1.9) to obtain the stability of solution for sufficiently small initial data.

Our second result shows that oscillation part u ˜ , u ˜ ( 2 ) , u ˜ ( 3 ) , and θ decay exponentially in time.

Theorem 1.2

Assume that the initial data ( u 0 , θ 0 ) H 2 ( Ω ) satisfies the conditions in Theorem 1.1. Let ( u , θ ) be the corresponding solution of (1.2). Then we have the following decay rates for the ( u ˜ , θ ) :

(1.10) ( u ˜ , θ ) ( t ) H v 1 ( Ω ) + ( u ˜ ( 2 ) , u ˜ ( 3 ) ) ( t ) H v 1 ( Ω ) C δ e C t ,

where C > 0 is a constant independent of δ and t . If we further assume ( u 0 , θ 0 ) H 3 ( Ω ) , then we have

(1.11) ( 1 u ˜ , 1 θ ) ( t ) L 2 ( Ω ) + ( Δ v u ˜ , Δ v θ ) ( t ) L 2 ( Ω ) C δ e C t ,

(1.12) ( 1 u ˜ ( 2 ) , 1 u ˜ ( 3 ) ) ( t ) L 2 ( Ω ) C δ e C t .

We briefly outline the sketch of the proof of Theorem 1.2. The proof is split into three steps, which are devoted to the proof of (1.10), (1.11), and (1.12). First, we prove (1.10). Generally speaking, to obtain the exponential decay rates, we need to prove the following differential inequality:

(1.13) d d t ( u ˜ H v 1 2 + θ H v 1 2 ) + C ( u ˜ H v 1 2 + θ H v 1 2 ) 0

for some constant C > 0 . However, based on the strong poincaré-type inequalities

u ˜ L 2 C v u ˜ L 2 = C v u L 2 and θ L 2 C 3 θ L 2 ,

we attempt to prove the decay rates for ( v u , v θ ) , which will simplify the proof of (1.10). Indeed, by making use of the anisotropic inequalities derived in Lemma 2.4 together with the strong poincaré-type inequalities, we are able to obtain the self-closed differential inequality involving only v u and v θ . More precisely, we obtain

d d t ( v u , v θ ) L 2 2 + ( 2 ν C 2 ( u H 2 + θ H 2 ) ) Δ v u L 2 2 + ( 2 η C 2 ( u H 2 + θ H 2 ) ) 3 v θ L 2 2 0 .

If we select the initial data sufficiently such that

C 2 ( u H 2 + θ H 2 ) C 3 δ min { 2 ν , 2 η } ,

then we assert that the decay rate holds

(1.14) ( v u ˜ , v θ ) ( t ) L 2 C δ e C t .

The rates ( u ˜ , θ ) ( t ) L 2 and ( u ˜ ( 2 ) , u ˜ ( 3 ) ) H v 1 further follow from (1.14), (2.4) (2.5) and (2.1), (2.2), (2.3), respectively.

Second, for the decay estimate in (1.11), based on a significant observation u ¯ 1 = 0 , that is, u 1 = u ˜ 1 , we can obtain the equations of ( u ˜ , θ ) as follows:

(1.15) t u ˜ 1 + u u ˜ 1 = 1 p ˜ + ν Δ v u ˜ 1 , t u ˜ 2 + 1 ( u 1 u 2 ) ˜ + 2 u 2 2 + 3 ( u 2 u 3 ) = 2 p + ν Δ v u ˜ 2 , t u ˜ 3 + u u ˜ 3 = 3 p + ν Δ v u ˜ 3 + θ , t θ + u θ + u ˜ 3 = η 33 θ , u ˜ = 0 ,

It is noted that when ( u 0 , θ 0 ) H 3 ( Ω ) , we can obtain the global bound of the solution in H 3 with a similar argument to Theorem 1.1. Then the application of a series of anisotropic inequalities in Lemma 2.4, 2.6 and strong poincaré-type inequalities along with the estimates on ( v u ˜ , v θ ) ( t ) L 2 yields the desired decay rates of ( 1 u ˜ , 1 θ ) and ( Δ v u , Δ v θ ) in L 2 . It is worth mentioning that we make some more delicate arguments in the proof of (1.11) to exhibit the integral term depending on the H 3 energy clearly. More precisely, we show that the two terms

1 u 2 2 θ 1 θ d x and v u 2 2 v θ Δ v θ d x

cannot be closed if ( u 0 , θ 0 ) H 2 .

Third, we present the proof of (1.12). That is the decay rates for 1 u ˜ ( 2 ) L 2 and 1 u ˜ ( 3 ) L 2 . Since from the equalities (2.1) and (2.2) we are able to drive

f ˜ ( 3 ) L 2 C f ˜ L 2 + C f ˜ ( 2 ) L 2 ,

or

f ˜ ( 2 ) L 2 C f ˜ L 2 + C f ˜ ( 3 ) L 2 ,

it suffices to deal with either of the two. With loss of generality, we consider the decay of 1 u ˜ ( 3 ) L 2 . It is easy to verify that ( u ˜ ( 3 ) , θ ) satisfies

(1.16) t u ˜ 1 ( 3 ) + 1 ( u 1 2 u 1 2 ¯ ( 3 ) ) + 2 ( u 1 u 2 u 1 u 2 ¯ ( 3 ) ) + 3 ( u ˜ 3 ( 3 ) u 1 ) = 1 p ˜ ( 3 ) + ν Δ v u ˜ 1 ( 3 ) , t u ˜ 2 ( 3 ) + 1 ( u 1 u 2 u 1 u 2 ¯ ( 3 ) ) + 2 ( u 2 2 u 2 2 ¯ ( 3 ) ) + 3 ( u ˜ 3 ( 3 ) u 2 ) = 2 p ˜ ( 3 ) + ν Δ v u ˜ 2 ( 3 ) , t u ˜ 3 ( 3 ) + u u ˜ 3 ( 3 ) = 3 p ˜ ( 3 ) + ν Δ v u ˜ 3 ( 3 ) + θ , t + u θ + u ˜ 3 ( 3 ) = η 33 θ , u ˜ ( 3 ) = 0 .

Compared to (1.15), the system (1.16) has more complicated forms. Therefore, to obtain a similar differential inequality to (1.13), we need more subtle estimates. A key step is to decompose the nonlinear terms in some items associated with u ˜ ( 3 ) . Take one of the nonlinear terms as instance,

u 1 2 u 1 2 ¯ ( 3 ) = 2 u ˜ 1 ( 3 ) u ¯ 1 ( 3 ) + ( u ˜ 1 ( 3 ) ) 2 ( u ˜ 1 ( 3 ) ) 2 ¯ = 2 u ˜ 1 ( 3 ) u ¯ 1 ( 3 ) + ( u ˜ 1 ( 3 ) ) 2 ˜ ( 3 ) .

Then by making full use of the strong poincaré inequality (2.3) and several anisotropic inequalities on the oscillation in x 3 -direction, we able to prove that 1 u ˜ ( 3 ) L 2 decays exponentially.

The rest of this article is divided into three sections. Section 2 is devoted to providing the strong poincaré-type inequalities and some anisotropic inequalities associated with oscillation, as well as the properties of the decomposition. Section 3 proves Theorem 1.1, while Section 4 presents the proof of Theorem 1.2.

2 Preliminary

Before proving Theorems 1.1 and 1.2, we shall make some preparations in this section. We first list some properties related to the decomposition. Then the strong Poincaré-type inequalities on the oscillations and u 3 , θ are provided, respectively. Since the system examined in this article has anisotropic dissipation, anisotropic inequalities appear to be necessary to deal with such partially dissipated systems. Therefore, we further present some crucial anisotropic inequalities for integrals involving triple products. The strong Poincaré-type inequalities and anisotropic inequalities associated with the oscillating functions are extremely powerful tools, which will be used frequently in our subsequent proof.

Lemma 2.1

Let f ¯ , f ¯ ( i ) and f ˜ , f ˜ ( i ) for i = 2 , 3 be defined as in (1.3).

(i) We have the following basic properties:

k f ¯ = k f ¯ , k f ¯ ( i ) = k f ¯ ( i ) , 2 f ¯ = 3 f ¯ = 0 , 2 f ¯ ( 2 ) = 3 f ¯ ( 3 ) = 0 , k f ˜ = k f ˜ , k f ˜ ( i ) = k f ˜ ( i ) , 2 f ˜ = 2 f ˜ ( 2 ) = 2 f , 3 f ˜ = 3 f ˜ ( 3 ) = 3 f .

(ii) If f is a divergence-free vector filed, namely, f = 0 , then f ¯ and f ˜ are also divergence-free,

f ¯ = f ˜ = f ˜ ( i ) = 0 .

(iii) Let f H s with s N . Then f ¯ and f ˜ are orthogonal in H s , namely,

( f ¯ , f ˜ ) H s Ω f ¯ f ˜ d x = 0 , f H s 2 = f ¯ H s 2 + f ˜ H s 2 .

Clearly,

f ¯ H s f H s , f ˜ H s f H s .

Also, f ¯ ( i ) and f ˜ ( i ) are orthogonal and obey

f ¯ ( i ) H s f H s , f ˜ ( i ) H s f H s .

(iv) f ˜ , f ˜ ( 2 ) , and f ˜ ( 3 ) satisfy the equalities

(2.1) f ˜ = f ˜ ( 3 ) + f ˜ ( 2 ) ¯ ( 3 ) ,

(2.2) f ˜ = f ˜ ( 2 ) + f ˜ ( 3 ) ¯ ( 2 ) .

Proof

For the properties in (i), (ii), and (iii), it suffices to prove that for f ¯ and f ˜ . The properties of f ¯ ( i ) and f ˜ ( i ) for i = 2 , 3 can be established similarly.

(i) The properties can be proved via the definitions of f ¯ and f ˜ .

k = 1 , 1 f ¯ = T 2 1 f ( x 1 , x 2 , x 3 ) d x 2 d x 3 = 1 T 2 f ( x 1 , x 2 , x 3 ) d x 2 d x 3 = 1 f ¯ , k = 2 , 2 f ¯ = T 2 2 f ( x 1 , x 2 , x 3 ) d x 2 d x 3 = 0 = 2 f ¯ ( x 1 ) .

Similarly, 3 f ¯ = 3 f ¯ = 0 . Also,

k = 1 , 1 f ˜ = 1 f 1 f ¯ = 1 ( f f ¯ ) = 1 f ˜ , k 1 , k f ˜ = k f k f ¯ = k f = k ( f f ¯ ) = k f ˜ = k f .

(ii) If u = 0 , then we have

u ¯ = 1 u ¯ 1 + 2 u ¯ 2 + 3 u ¯ 3 = 1 u 1 + 2 u 2 + 3 u 3 ¯ = u ¯ = 0 ,

and

u ˜ = u u ¯ = 0 .

(iii) Since

f ˜ ¯ = T 2 ( f ( x 1 , x 2 , x 3 ) f ¯ ( x 1 , x 2 , x 3 ) ) d x 2 d x 3 = f ¯ ( x 1 ) f ¯ ( x 1 ) = 0 ,

then, we have

( f ¯ , f ˜ ) = Ω f ¯ ( x 1 ) f ˜ ( x 1 , x 2 , x 3 ) d x 1 d x 2 d x 3 = R f ¯ ( x 1 ) T 2 f ˜ ( x 1 , x 2 , x 3 ) d x 2 d x 3 d x 1 = 0 .

For α = k , by k f ¯ = k f ¯ , we have

( D α f ¯ , D α f ˜ ) = Ω D α f ¯ ( x 1 ) D α f ˜ ( x 1 , x 2 , x 3 ) d x 1 d x 2 d x 3 = R D α f ¯ ( x 1 ) T 2 D α f ˜ ( x 1 , x 2 , x 3 ) d x 2 d x 3 d x 1 = R D α f ¯ ( x 1 ) D α f ˜ ¯ ( x 1 ) d x 1 = 0 .

Therefore, we obtain ( f ¯ , f ˜ ) H s = 0 .

Furthermore, by the orthogonality, we can obtain

α s D α f L 2 2 = α s Ω ( ( D α f ¯ ) 2 + 2 D α f ¯ D α f ˜ + ( D α f ˜ ) 2 ) d x = α s D α f ¯ L 2 2 + α s D α f ˜ L 2 2 .

That is, f H s 2 = f ¯ H s 2 + f ˜ H s 2 , which implies

f ¯ H s f H s , f ˜ H s f H s .

(iv) It follows from the definition of f ˜ and f ˜ ( i ) , i = 2 , 3 that

f ˜ = f T T f ( x 1 , x 2 , x 3 ) d x 2 d x 3 = f T f ¯ ( 2 ) ( x 1 , x 2 ) d x 3 = f T ( f f ˜ ( 2 ) ) d x 3 = f f ¯ ( 3 ) + T f ˜ ( 2 ) ( x 1 , x 3 ) d x 3 = f ˜ ( 3 ) + f ˜ ( 2 ) ¯ ( 3 ) .

Similarly, we have

f ˜ = f ˜ ( 2 ) + f ˜ ( 3 ) ¯ ( 2 ) .

This completes the proof of Lemma 2.1.□

Lemma 2.2

Let f ˜ be defined as in (1.3). If v f ˜ H s ( Ω ) , then we have

(2.3) f ˜ ( 2 ) L 2 C 2 f ˜ ( 2 ) L 2 , f ˜ ( 3 ) L 2 C 3 f ˜ ( 3 ) L 2 ,

(2.4) f ˜ H s ( Ω ) C v f ˜ H s ( Ω ) ,

where C > 0 is a constant depending on Ω and s only. In particular,

(2.5) u 3 H s ( Ω ) C 3 u 3 H s ( Ω ) , θ H s ( Ω ) C 3 θ H s ( Ω ) .

Proof

Obviously, we only take the case s = 0 into consideration. We may assume that f ˜ is smooth since every function in L 2 ( Ω ) can be approximated by smooth functions. With loss of generality, consider the case i = 2 . Due to f ˜ ( 2 ) ¯ ( 2 ) = 0 , there exists ψ 2 ( x 1 , x 3 ) such that

T 2 f ˜ ( 2 ) ( x 1 , x 2 , x 3 ) d x 2 = f ˜ ( 2 ) ( x 1 , ψ 2 , x 3 ) = 0 .

It is noted that

( f ˜ ( 2 ) ) 2 ( x 1 , x 2 , x 3 ) = ψ 2 x 2 2 ( f ˜ ( 2 ) ) 2 ( x 1 , y , x 3 ) d y = 2 ψ 2 x 2 f ˜ ( 2 ) 2 f ˜ ( 2 ) d y 2 f ˜ ( 2 ) L x 2 2 2 f ˜ ( 2 ) L x 2 2 .

Then integrating in space Ω and applying Hölder’s inequality, we can derive

f ˜ ( 2 ) L 2 2 f ˜ ( 2 ) L 2 .

(2.4) is thus established.

Now taking L 2 -norm on both sides of (2.1) and applying (2.3) yield

f ˜ L 2 C ( f ˜ ( 3 ) L 2 + f ˜ ( 2 ) L 2 ) C v f ˜ L 2 ,

where we have used the fact i f ˜ ( i ) = i f = i f ˜ for i = 2 , 3 .

Finally, by the symmetry condition (1.7), which will be proved in Section 3, the facts u 3 ˜ ( 3 ) = u 3 and θ ˜ ( 3 ) = θ hold. Consequently, we can derive (2.5) by (2.3). We complete the proof of Lemma 2.2.□

The lemma assesses the anisotropic inequalities in Ω involving in the general function and its oscillation part. Now we present some anisotropic inequalities. First, we introduce the 1D inequalities of L -norm as follows:

(2.6) f L ( R ) 2 f L 2 ( R ) 1 2 f L 2 ( R ) 1 2 ,

(2.7) f L ( T ) 2 f L 2 ( T ) 1 2 ( f L 2 ( T ) + f L 2 ( T ) ) 1 2 ,

which can be obtained by the direct estimates. One refers to a similar proof of Lemma 2.4 in [24].

Lemma 2.3

Assume that f H 1 ( Ω ) and g , v g , h , 3 h L 2 ( Ω ) . Then,

(2.8) Ω f g h d x C f L 2 1 2 1 f L 2 1 2 g L 2 1 2 ( g L 2 + 2 g L 2 ) 1 2 h L 2 1 2 ( h L 2 + 3 h L 2 ) 1 2 .

(2.9) Ω f g h d x C f L 2 1 4 1 f L 2 1 4 ( f L 2 + 2 f L 2 ) 1 4 ( 1 f L 2 + 1 2 f L 2 ) 1 4 g L 2 1 2 ( g L 2 + 3 g L 2 ) 1 2 h L 2 C f H 1 3 4 ( 1 f L 2 + 1 2 f L 2 ) 1 4 g L 2 1 2 ( g L 2 + 3 g L 2 ) 1 2 h L 2 .

Proof

By Hölder’s inequality, Minskowski’s inequality, (2.6), and (2.7), we can easily to obtain

Ω f g h d x f L x 1 L x 2 2 L x 3 2 g L x 1 2 L x 2 L x 3 2 h L x 1 2 L x 2 2 L x 3 C f L x 1 2 1 2 1 f L x 1 2 1 2 L x 2 2 L x 3 2 g L x 2 2 1 2 ( g L x 2 2 + 2 g L x 2 2 ) 1 2 L x 1 2 L x 3 2 × h L x 3 2 1 2 ( h L x 3 2 + 3 h L x 3 2 ) 1 2 L x 1 2 L x 2 2 C f L 2 1 2 1 f L 2 1 2 g L 2 1 2 ( g L 2 + 2 g L 2 ) 1 2 h L 2 1 2 ( h L 2 + 3 h L 2 ) 1 2 .

Similarly,

Ω f g h d x f L x 1 L x 2 L x 3 2 g L x 1 2 L x 2 2 L x 3 h L 2 C f L x 1 2 L x 3 2 L x 2 1 2 1 f L x 1 2 L x 3 2 L x 2 1 2 g L 2 1 2 ( g L 2 + 3 g L 2 ) 1 2 h L 2 C f L 2 1 4 ( f L 2 + 2 f L 2 ) 1 4 1 f L 2 1 4 ( 1 f L 2 + 1 2 f L 2 ) 1 4 × g L 2 1 2 ( g L 2 + 3 g L 2 ) 1 2 h L 2 .

This completes the proof of Lemma 2.4.□

If we replace g and h in Lemma 2.3 by the oscillation part g ˜ and h ˜ , we can obtain the following inequalities.

Lemma 2.4

Assume that f H 2 ( Ω ) and g , h H 1 ( Ω ) . Then we have

(2.10) Ω f g ˜ h ˜ d x C f L 2 1 2 1 f L 2 1 2 v g ˜ L 2 v h ˜ L 2 ,

(2.11) Ω f g h ˜ d x C f L 2 1 2 1 f L 2 1 2 g L 2 1 2 ( g L 2 + i g L 2 ) 1 2 v h ˜ L 2 , i = 2 , 3 ,

(2.12) Ω f ˜ g h d x C v f ˜ L 2 1 2 1 v f ˜ L 2 1 2 g L 2 1 2 ( g L 2 + 3 g L 2 ) 1 2 h L 2 ,

(2.13) Ω f ˜ g h d x C f ˜ L 2 1 4 v f ˜ L 2 1 2 v 2 f ˜ L 2 1 4 g L 2 1 2 1 g L 2 1 2 h L 2 C v 2 f ˜ L 2 g L 2 1 2 1 g L 2 1 2 h L 2 .

Proof

(2.10), (2.11), and (2.12) follow directly from (2.8), (2.9), and the Poincaré-type inequalities (2.4). For (2.13), a similar argument to (2.9) yields

f g h d x f L x 2 L x 3 L x 1 2 g L x 2 2 L x 3 2 L x 1 h L 2 C f L 2 1 4 ( f L 2 + 2 f L 2 ) 1 4 ( f L 2 + 3 f L 2 ) 1 4 ( f L 2 + 2 f L 2 + 3 f L 2 + 3 2 f L 2 ) 1 4 g L 2 1 2 1 g L 2 1 2 h L 2 .

Then replacing f by f ˜ and applying the Poincaré (2.4), we can show the desired inequality (2.13). This completes the proof of Lemma 2.4.□

When f ˜ is replaced by f ˜ ( 3 ) in Lemma 2.4, we can establish the similar anisotropic inequalities by Poincaré inequality (2.3).

Lemma 2.5

Assume that f H 2 ( Ω ) and g , h H 1 ( Ω ) . Then it holds

(2.14) Ω f g ˜ ( 3 ) h ˜ ( 3 ) d x C f L 2 1 2 1 f L 2 1 2 3 g ˜ ( 3 ) L 2 v h ˜ ( 3 ) L 2 ,

(2.15) Ω f g h ˜ ( 3 ) d x C f L 2 1 2 1 f L 2 1 2 g L 2 1 2 ( g L 2 + i g L 2 ) 1 2 v h ˜ ( 3 ) L 2 , i = 2 , 3 ,

(2.16) Ω f ˜ ( 3 ) g h d x C 3 f ˜ ( 3 ) L 2 1 2 1 3 f ˜ ( 3 ) L 2 1 2 g L 2 1 2 ( g L 2 + 2 g L 2 ) 1 2 h L 2 ,

(2.17) Ω f ˜ ( 3 ) g h d x C f ˜ ( 3 ) L 2 1 4 v f ˜ ( 3 ) L 2 1 2 v 2 f ˜ ( 3 ) L 2 1 4 g L 2 1 2 1 g L 2 1 2 h L 2 C v 2 f ˜ ( 3 ) L 2 g L 2 1 2 1 g L 2 1 2 h L 2 .

We now present the last lemma that assesses the anisotropic upper bound on f ˜ ( 3 ) g ˜ ( 3 ) L 2 .

Lemma 2.6

Assume that v f ˜ ( 3 ) , v 1 f ˜ ( 3 ) , v g ˜ ( 3 ) L 2 ( Ω ) . Then we have

(2.18) f ˜ ( 3 ) g ˜ ( 3 ) L 2 C v f ˜ ( 3 ) L 2 1 2 v 1 f ˜ ( 3 ) L 2 1 2 v g ˜ ( 3 ) L 2 .

Proof

First, the following Sobolev inequality in T 2 holds

(2.19) h ˜ ( 3 ) L v 4 C h ˜ ( 3 ) L v 2 1 2 v h ˜ ( 3 ) L v 2 1 2 C v h ˜ ( 3 ) L v 2

for h ˜ ( 3 ) H v 1 . In fact, due to h ˜ ( 3 ) ¯ ( 3 ) = 0 , there exists x 3 * T such that h ˜ ( x 1 , x 2 , x 3 * ) = 0 . By Hölder inequality, we have

( h ˜ ( 3 ) ) 2 = 2 x 3 * x 3 h ˜ ( 3 ) ( x 1 , x 2 , z ) 3 h ˜ ( 3 ) ( x 1 , x 2 , z ) d z 2 h ˜ ( 3 ) L x 3 2 3 h ˜ ( 3 ) L x 3 2 .

By multiplying aforementioned inequality on both sides by ( h ˜ ( 3 ) ) 2 , integrating in T 2 and using (2.7) and (2.3) yield

h ˜ ( 3 ) L v 4 4 2 T h ˜ ( 3 ) L x 3 2 3 3 h ˜ ( 3 ) L x 3 2 d x 2 2 h ˜ ( 3 ) L x 3 2 L x 2 2 h ˜ ( 3 ) L v 2 2 h ˜ ( 3 ) L v 2 C h ˜ ( 3 ) L v 2 ( h ˜ ( 3 ) L v 2 + 2 h ˜ ( 3 ) L v 2 ) h ˜ ( 3 ) L v 2 2 h ˜ ( 3 ) L v 2 C v h ˜ ( 3 ) L v 2 .

It then follows from (2.19), (2.7), and (2.3) that

f ˜ ( 3 ) g ˜ ( 3 ) L 2 2 f ˜ ( 3 ) L x 1 L v 4 2 g ˜ ( 3 ) L x 1 2 L v 4 2 C v f ˜ ( 3 ) L v 2 L x 1 2 v g ˜ ( 3 ) L 2 2 C v f ˜ ( 3 ) L 2 v 1 f ˜ ( 3 ) L 2 v g ˜ ( 3 ) L 2 2 .

This completes the proof of Lemma 2.6.□

3 Proof Theorem 1.1

This section is devoted to proving Theorem 1.1. To prove the global bound result (1.6), we need first show the the symmetry (1.7) for u , θ . Based on u 3 , θ , we are able to derive another strong Poincaré inequality, i.e.,

u 3 H s ( Ω ) 3 u 3 H s ( Ω ) , θ H s ( Ω ) 3 θ H s ( Ω ) .

Then by applying Poincaré inequalities and anisotropic inequalities, the a p r i o r estimate for E ( t ) are able to be established. Furthermore, we use the bootstrapping argument to yield the global well-posedenss.

3.1 Proof of the symmetry (1.7)

The symmetry (1.7) can be obtained via the uniqueness. First, we can use the standard energy method to obtain the local well-posedness of the solution. Thus, our attentions mainly focus on the uniqueness of the solution. It is simple to achieve it. For the convenience of the readers, here, we only give the sketch of the proof.

Proposition 3.1

Let ( u ( 1 ) , θ ( 1 ) ) and ( u ( 2 ) , θ ( 2 ) ) be two solutions of (1.2) satisfying

( u ( 1 ) , θ ( 1 ) ) , ( u ( 2 ) , θ ( 2 ) ) L ( 0 , T 2 ; H 2 ) .

Then, for any 0 < t T , we have

( u ( 1 ) , θ ( 1 ) ) = ( u ( 2 ) , θ ( 2 ) ) .

Proof of Proposition 3.1

Denote

ε u u ( 1 ) u ( 2 ) , ε θ θ ( 1 ) θ ( 2 ) , ε p p ( 1 ) p ( 2 ) .

It is easily to check

(3.1) t ε u + u ( 1 ) ε u + ε u u ( 2 ) = ε p + ν Δ v ε u + ε θ e 3 , t ε θ + u ( 1 ) ε θ + ε u θ ( 2 ) + ε u 3 = η 33 ε θ , ε u = 0 .

Taking the L 2 -inner product of (3.1) with ( ε u , ε θ ) , we have

(3.2) 1 2 d d t ( ε u L 2 2 + ε θ L 2 2 ) + ν v ε u L 2 2 + η 3 ε θ L 2 2 = ε u u ( 2 ) ε u d x ε u θ ( 2 ) ε θ d x I 1 + I 2 .

By using the anisotropic inequality (2.8), L 1 can be estimated as follows:

I 1 C ε u L 2 1 2 ( ε u L 2 + 2 ε u L 2 ) 1 2 ε u L 2 1 2 × ( ε u L 2 + 3 ε u L 2 ) 1 2 u ( 2 ) L 2 1 2 1 u ( 2 ) L 2 1 2 C u ( 2 ) H 2 ε u L 2 ε u L 2 1 2 + 2 ε u L 2 1 2 ε u L 2 1 2 + 3 ε u L 2 1 2 C ε u L 2 2 ( 1 + u ( 2 ) H 2 2 ) + 1 2 ν 3 ε u L 2 2 + 1 4 ν 2 ε u L 2 2 .

Also, we have

I 2 C ( 1 + θ ( 2 ) H 2 2 ) ( ε u L 2 2 + ε θ L 2 2 ) + 1 4 ν 2 ε u L 2 2 + 1 2 η 3 ε θ L 2 2 .

Therefore, we obtain

(3.3) d d t ( ε u L 2 2 + ε θ L 2 2 ) + ν v ε u L 2 2 + η 3 ε θ L 2 2 C ( 1 + ( u ( 2 ) ( t ) , θ ( 2 ) ( t ) ) H 2 2 ) ( ε u L 2 2 + ε θ L 2 2 ) .

Then by Gronwall’s inequality, for any 0 < t T , we obtain

( ε u ( t ) , ε θ ( t ) ) L 2 2 ( ε u ( 0 ) , ε θ ( 0 ) ) L 2 2 e C 0 T ( 1 + ( u ( 2 ) , θ ( 2 ) ) H 2 2 ) d τ ,

which together the local well-posedness of the solution on ( 0 , T ) implies the uniqueness, i.e. ε u ( t ) = ε θ ( t ) = 0 . This completes the proof of proposition 3.1.□

Proposition 3.2

Suppose that ( u 0 , θ 0 ) H 2 ( Ω ) satisfies the symmetries in (1.5) and u 0 = 0 . Then the symmetries (1.5) can persist for any 0 < t T .

Proof of Proposition 3.2

We will utilize the uniqueness to verify proposition 3.2. Let ( u , θ ) = ( u 1 , u 2 , u 3 , θ ) be the solution of the system (1.2). We define U , P , Θ with

U 1 = u 1 ( x 1 , x 2 , x 3 , t ) , U 2 = u 2 ( x 1 , x 2 , x 3 , t ) , U 3 = u 3 ( x 1 , x 2 , x 3 , t ) , Θ = θ ( x 1 , x 2 , x 3 , t ) , P = p ( x 1 , x 2 , x 3 , t ) .

To prove the symmetries (1.7), it is only to verify that ( U , Θ , P ) also satisfies the Boussinesq equations (1.2). Indeed, through a simple calculation,

U 1 1 U 1 = u 1 ( x 1 , x 2 , x 3 , t ) 1 u 1 ( x 1 , x 2 , x 3 , t ) = u 1 1 u 1 , U 2 2 U 1 = u 2 ( x 1 , x 2 , x 3 , t ) 2 u 1 ( x 1 , x 2 , x 3 , t ) = u 2 2 u 1 , U 3 3 U 1 = u 3 ( x 1 , x 2 , x 3 , t ) ( 1 ) 3 u 1 ( x 1 , x 2 , x 3 , t ) = u 3 3 u 1 .

We derive U U 1 = u u 1 . Similarly

1 P ( x 1 , x 2 , x 3 , t ) = 1 p ( x 1 , x 2 , x 3 , t ) , Δ v U 1 = Δ v u 1 .

Therefore, U 1 satisfies

(3.4) t U 1 + U U 1 + 1 P = Δ v U 1 .

For U 2 , U 3 , and Θ , we can obtain with a similar argument

t U 2 + U U 2 + 2 P = Δ v U 2 , t U 3 + U U 3 + 3 P = Δ v U 3 + Θ , t Θ + U Θ = 3 2 Θ ,

which together with (3.4) asserts that ( U , Θ ) is still the solution of the Boussinesq equations (1.2). Then by the uniqueness stated in Proposition 3.1, we find that ( u , θ , p ) admits the desired symmetry. We complete the proof of Proposition 3.2.□

3.2 Proof of the global well-posedness (1.6)

To prove the global uniform bound of the solution, as aforementioned in Section 1, we first establish the a priori estimate for E ( t ) as follows.

Proposition 3.3

Let the initial datum ( u 0 , θ 0 ) satisfy conditions in Theorem 1.1. Then, for some constant C 0 > 0 , C 1 > 0 , we have

(3.5) E ( t ) C 0 E ( 0 ) + C 1 E 3 2 ( t ) .

With the energy inequality at our proposal, we are able to prove (1.6) in Theorem 1.1.

Proof of (1.1)

We will use the bootstrapping argument to obtain the global existence of the solution and its stability. A rigorous statement of the abstract bootstrap principle can be found in Tao’s book (see [34] p. 21). More precisely, once (3.5) holds, the bootstrapping argument will imply E ( t ) admits the global uniform bound (1.6) provided that the initial data is sufficiently small, say

(3.6) E ( 0 ) = u 0 H 2 2 + θ 0 H 2 2 δ 1 4 C 0 C 1 .

The argument starts with the ansatz that, for t T ,

E ( t ) M 1 4 C 1 2 .

Then it follows from (3.5) and initial data assumption (3.6) that

E ( t ) C 0 δ 2 + 1 2 E ( t )

or

E ( t ) 2 C 0 δ 2 2 C 0 1 16 C 0 C 1 2 = 1 8 C 1 2 = M 2 .

That asserts that E ( t ) actually admits a smaller upper bound. The bootstrapping argument then shows that E ( t ) is bounded uniformly for all time, i.e., E ( t ) C δ 2 . This completes the proof of the stability result (1.6) in Theorem 1.1.□

In what follows, we present the proof of Proposition 3.3.

Proof of Proposition 3.3

Due to the equivalence of the norms

( u , θ ) H 2 2 ( u , θ ) L 2 2 + ( u , θ ) H ˙ 2 2 .

We only need to bound ( u , θ ) L 2 2 and i = 1 3 ( i 2 u , i 2 θ ) L 2 2 . It is obviously for the global L 2 -bound.

(3.7) ( u , θ ) ( t ) L 2 2 + 2 ν 0 t v u L 2 2 d τ + 2 η 0 t 3 θ L 2 2 d τ ( u ( 0 ) , θ ( 0 ) ) L 2 2 ,

where we have used the fact that

( θ e 3 u u 3 θ ) d x = 0 .

Next we focus on the second-order derivatives estimates. We first apply the operator i 2 to (1.2) and then take the L 2 inner product of the resulted equations with ( i 2 u , i 2 θ ) to obtain

(3.8) 1 2 d d t i = 1 3 ( i 2 u , i 2 θ ) L 2 2 + ν i = 1 3 v i 2 u L 2 2 + η i = 1 3 3 i 2 θ L 2 2 = i = 1 3 i 2 ( u u ) i 2 u d x i = 1 3 i 2 ( u θ ) i 2 θ d x M 1 + M 2 .

Since u possesses the anisotropic dissipation in x 2 , x 3 directions, we need to separate the x 1 -derivative from x 2 , x 3 -derivatives. That is,

M 1 = i = 2 3 i 2 ( u u ) i 2 u d x j = 2 3 1 2 ( u j j u ) 1 2 u d x 1 2 ( u 1 1 u ) 1 2 u d x M 11 + M 12 + M 13 .

Since all of the derivatives in M 11 are in x 2 , x 3 direction, by Sobolev inequality v L 4 C v H 1 , it is easily to derive

M 11 = i = 2 3 m = 1 2 C 2 m i m u i 2 m u i 2 u d x i = 2 3 m = 1 2 C 2 m i m u L 4 i 2 m u L 2 i 2 u L 4 C u H 2 v u H 2 2 .

By means of Lemma 2.1 and integration by parts, we first split M 12 into two parts.

M 12 = j = 2 3 m = 1 2 C 2 m 1 m u j 1 2 m j u ˜ 1 2 u d x = j = 2 3 m = 1 2 C 2 m 1 m j u j 1 2 m u ˜ 1 2 u d x + j = 2 3 m = 1 2 C 2 m 1 m u j 1 2 m u ˜ j 1 2 u d x M 12 , 1 + M 12 , 2 .

Then by Lemmas 2.1 and 2.3, invoking Hölder’s inequality and (2.4) yields

M 12 , 1 C j = 2 3 1 j u j L 2 1 2 1 2 j u j L 2 1 2 1 2 u L 2 1 2 ( 1 2 u L 2 + 2 1 2 u L 2 ) 1 2 v 1 u ˜ L 2 + C j = 2 3 1 2 j u j L 2 u ˜ L 1 2 u L 2 C u H 2 v u H 2 2 ,

and

M 12 , 2 C j = 2 3 1 u j L 4 1 u ˜ j L 4 1 2 j u L 2 + C j = 2 3 1 2 u j L 2 u ˜ j L 1 2 j u L 2 C j = 2 3 1 u j H 1 1 v u ˜ j H 1 v u H 2 2 + C j = 2 3 u H 2 v u ˜ H 2 v u H 2 C u H 2 v u H 2 2 .

To deal with M 13 , we make full use of the incompressible condition 1 u 1 = v u v and integrate by parts to rewrite it as follows:

M 13 = 1 2 u 1 1 u 1 2 u d x 2 1 u 1 1 2 u 1 2 u d x = 1 u ˜ v 1 v u 1 2 u d x 1 u ˜ v 1 2 v u 1 u d x 4 u ˜ v v 1 2 u 1 2 u d x ,

where we have used v u = v u ˜ . Then it follows from Hölder’s inequality, Sobolev inequality, and u ˜ L 2 C v u ˜ L 2 that

M 13 1 u ˜ v L 4 ( 1 v u L 4 1 2 u L 2 + 1 u L 4 1 2 v u L 2 ) + 4 u ˜ v L 1 2 u L 2 1 2 v u L 2 C u H 2 v u H 2 2 .

By collecting the estimates for M 1 , we obtain

(3.9) M 1 C u H 2 v u H 2 2 .

The bound for M 2 is complicated. Due to the only vertical dissipation for θ , we distinguish the horizontal derivatives of θ from the vertical derivative and divide it into three parts.

M 2 = i = 2 3 j = 1 2 i 2 ( u j j θ ) i 2 θ d x i = 2 3 i 2 ( u 3 3 θ ) i 2 θ d x j = 1 2 1 2 ( u j j θ ) 1 2 θ d x 1 2 ( u 3 3 θ ) 1 2 θ d x = i = 2 3 m = 1 2 C 2 m i m u i 2 m θ i 2 θ d x j = 1 2 1 2 u j j θ 1 2 θ d x 2 j = 1 2 1 u j 1 j θ 1 2 θ d x m = 1 2 1 m u 3 1 2 m 3 θ 1 2 θ d x M 21 + M 22 + M 23 + M 24 ,

where we have used θ ˜ * = θ and v u = v u ˜ .

By applying the anisotropic inequality (2.11) and Hölder’s inequality along with θ L 2 C 3 θ L 2 , we have

M 21 C i = 2 3 i 2 u L 2 1 2 1 i 2 u L 2 1 2 θ L 2 1 2 ( θ L 2 + 2 j θ L 2 ) 1 2 × i 2 θ L 2 1 2 ( i 2 θ L 2 + 3 i 2 θ L 2 ) 1 2 + C i = 2 3 i u L i θ L 2 i 2 θ L 2 C v u H 2 θ H 2 3 θ H 2 + C v u j H 2 3 θ H 2 θ H 2 C ( θ H 2 + u H 2 ) ( v u H 2 2 + 3 θ H 2 2 ) .

Similarly,

M 22 C j = 1 2 1 2 u j L 2 1 2 ( 1 2 u j L 2 + 2 1 2 u j L 2 ) 1 2 j θ L 2 1 2 1 j θ L 2 1 2 1 2 θ L 2 1 2 × ( 1 2 θ L 2 + 3 1 2 θ L 2 ) 1 2 C ( u , θ ) H 2 ( v u H 2 2 + 3 θ H 2 2 ) .

For M 23 , we use another anisotropic inequality (2.9) and combine with the Poincaré inequality (2.5) to obtain

M 23 C 1 u j H 1 3 4 ( 1 2 u j L 2 + 1 2 2 u j L 2 ) 1 4 1 j θ L 2 1 2 ( 1 j θ L 2 + 3 1 j θ L 2 ) 1 2 2 2 θ L 2 C ( u , θ ) H 2 ( v u H 2 2 + 3 θ H 2 2 ) .

Due to u 3 and θ in M 24 , it is simple to obtain

M 24 C m = 1 2 1 m u 3 L 2 1 2 ( 1 m u 3 L 2 + 2 1 m u 3 L 2 ) 1 2 × 1 2 m 3 θ L 2 1 2 1 3 m 3 θ L 2 1 2 1 2 θ L 2 1 2 ( 1 2 θ L 2 + 3 1 2 θ L 2 ) 1 2 C u H 2 1 2 v u ˜ H 2 1 2 θ H 2 1 2 3 θ H 2 3 2 C ( θ H 2 + u H 2 ) ( v u H 2 2 + 3 θ H 2 2 ) .

Thus, we obtain the following bound:

(3.10) M 2 C ( θ H 2 + u H 2 ) ( v u H 2 2 + 3 θ H 2 2 ) .

By substituting (3.9) and (3.10) into (3.8), integrating over [ 0 , t ] , and invoking (3.7)

E ( t ) C 0 E ( 0 ) + C 0 t ( θ H 2 + u H 2 ) ( v u H 2 2 + 3 θ H 2 2 ) d τ C 0 E ( 0 ) + C 1 E 3 2 ( t ) ,

which is the desired inequality (3.5). We complete the proof of Proposition 3.3.□

4 Proof of Theorem 1.2

This section focuses on the proof of the decay estimates in Theorem 1.2. We will apply the inequalities obtained in Section 2 and use the global existence result (1.6) as well as the symmetry (1.7) to achieve this proof. Before the proof, we first present the stability in H 3 . This can be obtained with a similar argument to Theorem 1.1. To avoid repetitions, we omit the proof here.

Proposition 4.1

Assume the initial datum ( u 0 , θ 0 ) H 3 satisfies the symmetry conditions (1.5). Then there exists δ = δ ( ν , η ) > 0 with ν , η > 0 such that, if

u 0 H 3 + θ 0 H 3 δ ( ν , η ) ,

then (1.2) has a unique global solution ( u , θ ) L ( 0 , ; H 3 ) satisfying

(4.1) u ( t ) H 3 2 + θ ( t ) H 3 2 + ν 0 t v u H 3 2 d τ + η 0 t 3 θ H 3 2 d τ C δ 2 ,

(4.2) u 1 , u 2 , p are e v e n i n x 3 , and u 3 and θ are o d d i n x 3 .

Proof of Theorem 1.2

The proof is split into three steps which is devoted to the proof of (1.10), (1.11), and (1.12), respectively.

Step 1: the proof of (1.10)

It is noted that v u ˜ = v u . Thus, we only need to establish the decay for ( v u , v θ ) L 2 that makes the proof much easier than that for ( v u ˜ , v θ ) . Taking the L 2 -inner product of (1.2) with ( Δ v u , Δ v θ ) and using integration by parts, we have

1 2 d d t ( v u , v θ ) L 2 2 + ν Δ v u L 2 2 + η 3 v θ L 2 2 = v u u v u d x v u θ v θ d x .

The anisotropic inequality (2.10) together with v u ˜ = v u leads to

v u u v u d x = v u ˜ u v u ˜ d x C v 2 u ˜ L 2 2 u L 2 1 2 1 u L 2 1 2 C u H 2 v 2 u L 2 2 .

Applying (2.11) for i = 3 with g = v θ and the poincaré inequality (2.5) yields

v u θ v θ d x = v u ˜ θ v θ d x C v 2 u ˜ L 2 θ L 2 1 2 1 θ L 2 1 2 v θ L 2 1 2 ( v θ L 2 + 3 v θ L 2 ) 1 2 C θ H 2 ( v 2 u θ L 2 2 + 3 v θ L 2 2 ) .

Then we derive, for some constant C 2 > 0 ,

(4.3) d d t ( v u , v θ ) L 2 2 + 2 ν Δ v u L 2 2 + 2 η 3 v θ L 2 2 C ( u H 2 + θ H 2 ) Δ v u L 2 v 3 θ L 2 C 2 δ ( Δ v u L 2 2 + v 3 θ L 2 2 ) .

Now we choose δ small enough such that C 2 δ < min { 2 ν , 2 η } . Then there exists a constant C 3 ( ν , η ) = min { 2 ν C 2 δ , 2 η C 2 δ } such that

d d t ( v u , v θ ) L 2 2 + C 3 ( Δ v u L 2 2 + 3 v θ L 2 2 ) 0 .

Then employing the Poincaré inequality (2.3) together with v θ = v θ ˜ ( 3 ) to bound the dissipative terms yields

v u L 2 + v θ L 2 δ e C t .

Furthermore, by (2.1), (2.2), (2.3), and (2.4), we infer

u ˜ L 2 + θ L 2 δ C e C t , u ˜ ( 2 ) L 2 + u ˜ ( 3 ) L 2 C 2 u ˜ ( 2 ) L 2 + C 3 u ˜ ( 3 ) L 2 C v u C δ e C t , 3 u ˜ ( 2 ) L 2 C ( 3 u ˜ L 2 + 3 u ˜ ( 3 ) L 2 ) C δ e C t , 2 u ˜ ( 3 ) L 2 C ( 2 u ˜ L 2 + 2 u ˜ ( 2 ) L 2 ) C δ e C t ,

where we have used i u ˜ ( i ) = i u = i u ˜ for i = 2 , 3 .

The proof of (1.10) is now completed.

Step 2: the proof of (1.11)

With (4.1) and (4.2) at our proposals, we are able to present the exponential decay of ( v 2 u , v 2 θ ) and ( 1 u ˜ , 1 θ ) in L 2 . Different from Step 1, we first need to establish the equations of ( u ˜ , θ ) . By u ¯ 3 = 0 , θ ¯ = 0 and divergence free for u , it is easily to check that u ¯ obeys

(4.4) u ¯ 1 = 0 , 1 u 1 2 ¯ + 1 p ¯ = 0 , t u ¯ 2 + 1 u 1 u 2 ¯ = 0 .

Then by subtracting (4.4) from (1.2) yields

(4.5) t u ˜ 1 + u u ˜ 1 = 1 p + ν Δ v u ˜ 1 , t u ˜ 2 + 1 ( u 1 u 2 ) ˜ + 2 u 2 2 + 3 ( u 2 u 3 ) = 2 p + ν Δ v u ˜ 2 , t u ˜ 3 + u u ˜ 3 = 3 p + ν Δ v u ˜ 3 + θ , t θ + u θ + u ˜ 3 = η 33 θ , u ˜ = 0 ,

where we have used the fact k f = k f ˜ for k = 2 , 3 .

By multiplying (4.5) by ( 1 2 u ˜ , 1 2 θ ) and integrating in space, we infer

1 2 d d t ( 1 u ˜ , 1 θ ) L 2 2 + ν 1 v u ˜ L 2 2 + η 3 1 θ L 2 2 J 1 + + J 6 ,

where

J 1 = 1 2 u ˜ 1 ( u u ˜ 1 ) d x , J 2 = 1 2 u ˜ 2 1 ( u 1 u 2 ) ˜ d x , J 3 = 1 2 u ˜ 2 2 u 2 2 d x , J 4 = 1 2 u ˜ 2 3 ( u 2 u 3 ) d x , J 5 = 1 2 u ˜ 3 ( u u ˜ 3 ) d x , J 6 = 1 2 θ ( u θ ) d x .

Next we bound J 1 to J 6 one by one. By the integration parts, the anisotropic inequality (2.10), J 1 and J 5 can be bounded as follows:

J 1 + J 5 = 1 u u 1 ˜ 1 u ˜ 1 d x 1 u u 3 ˜ 1 u ˜ 3 d x C v u ˜ 1 L 2 v 1 u ˜ 1 L 2 1 u L 2 1 2 1 2 u L 2 1 2 + C v u ˜ 3 L 2 v 1 u ˜ 3 L 2 1 u L 2 1 2 1 2 u L 2 1 2 C u H 2 v u ˜ L 2 2 .

To bound J 2 , we first decompose 1 ( u 1 u 2 ) ˜ into some terms associated with the oscillations and use the integration by parts to derive

J 2 = 1 2 u ˜ 2 1 ( u 1 u 2 ) ˜ d x = 1 2 u ˜ 2 ( 1 u ˜ 1 u 2 ˜ + u ˜ 1 1 u 2 ˜ ) d x = 1 u ˜ 2 1 ( 1 u ˜ 1 u 2 ˜ + u ˜ 1 1 u 2 ˜ ) d x .

Then, by applying Sobolev’s inequality, Poincaré inequality (2.4) and f ˜ = 0 , we obtain

J 2 C 1 u ˜ 2 L 2 ( 1 ( 1 u ˜ 1 u 2 ) L 2 + 1 ( u ˜ 1 1 u 2 ) L 2 ) C 1 u ˜ 2 L 2 ( 1 2 u ˜ 1 L 2 u 2 L + 1 u ˜ 1 L 4 1 u 2 L 4 + u ˜ 1 L 1 2 u 2 L 2 ) C 1 v u ˜ 2 L 2 ( v u ˜ L 2 u H 2 + 1 u ˜ 1 H 1 1 u 2 H 1 + u ˜ 1 H 2 1 2 u 2 L 2 ) C u H 2 v u ˜ L 2 2 ,

where we have used

u ˜ 1 H 2 C ( u ˜ 1 L 2 + 2 u ˜ 1 L 2 ) C ( v 2 u ˜ 1 L 2 + v 2 u ˜ 1 L 2 + 1 u ˜ 1 L 2 ) C v u ˜ 1 L 2 .

For J 3 , by means of Poincaré’s inequality (2.4), anisotropic inequality (2.10), and Hölder inequality, we obtain

J 3 = 1 u ˜ 2 1 2 u 2 2 d x = 2 1 u ˜ 2 1 2 u 2 u 2 + 1 u ˜ 2 2 u ˜ 2 1 u 2 d x C 1 u ˜ 2 L 2 1 2 u 2 L 2 u 2 L + C 1 u 2 L 2 1 2 1 2 u 2 L 2 1 2 v 1 u ˜ 2 L 2 v 2 u ˜ 2 L 2 C v u ˜ L 2 2 u H 2 .

Similarly, for J 4 , we have

J 4 = 1 u ˜ 2 1 3 u ˜ 2 u ˜ 3 d x 1 u ˜ 2 3 u ˜ 2 1 u 3 d x 1 u ˜ 2 1 u 2 3 u ˜ 3 d x 1 u ˜ 2 u 2 1 3 u ˜ 3 d x C 1 3 u ˜ 2 L 2 1 u 2 ˜ L 2 u ˜ 3 L + C 1 u 3 L 2 1 2 1 2 u 3 L 2 1 2 1 v u ˜ 2 L 2 3 v u ˜ 2 L 2 + C 1 u 2 L 2 1 2 1 2 u 2 L 2 1 2 1 v u ˜ 2 L 2 3 v u ˜ 3 L 2 + C 1 u ˜ 2 L 2 u 2 L 1 3 u ˜ 3 L 2 C u H 2 v u L 2 2 .

We turn to bound the last terms. By integration parts, applying the anisotropic inequality (2.11) yields

J 6 = 1 u 1 1 θ 1 θ d x 1 u 2 2 θ 1 θ d x 1 u 3 3 θ 1 θ d x C v 1 u ˜ L 2 θ L 2 1 2 ( θ L 2 + 3 θ L 2 ) 1 2 1 θ L 2 1 2 1 2 θ L 2 1 2 + 1 u 2 L 2 θ L 2 1 θ L 2 C ( u H 3 + θ H 2 ) Δ v u ˜ 1 L 2 v 3 θ L 2 ,

where we have used 1 u 1 = 1 u ˜ 1 , 1 u 3 = 1 u ˜ 3 , and θ L 2 C 3 θ L 2 .

Finally, collecting all estimates for J 1 through J 8 and combining (4.3) and the uniform bound (4.1), we have for some positive pure constant C independent of δ ,

d d t ( u ˜ , θ ) L 2 2 + 2 ν v u ˜ L 2 2 + 2 η 3 θ L 2 2 C ( u H 3 + θ H 2 ) v u ˜ L 2 3 θ L 2 C δ ( v u ˜ L 2 2 + 3 θ L 2 2 ) .

If the initial data δ is sufficiently small, we are able to derive

(4.6) d d t ( u ˜ , θ ) L 2 2 + C 4 ( v u ˜ L 2 2 + 3 θ L 2 2 ) 0 ,

where C 4 > 0 is a pure constant independent of δ and t . Then by (4.6) we conclude

u ˜ L 2 + θ L 2 δ e C t .

That means

1 u ˜ L 2 + 1 θ L 2 δ e C t .

In what follows, we show the proof of the decay on ( v 2 u ˜ , v 2 θ ) L 2 . Similar to Step 1, we consider ( v 2 u , v 2 θ ) L 2 . By applying the operator Δ v to (1.2), multiplying the resulted equations by ( Δ v u , Δ v θ ) , and integrating in space, we obtain

1 2 d d t ( Δ v u , Δ v θ ) L 2 2 + ν v Δ v u L 2 2 + η 3 Δ v θ L 2 2 = Δ v ( u u ) Δ v u d x Δ v ( u θ ) Δ v θ d x K 1 + K 2 .

Similar to ( 1 u ˜ , 1 θ ) L 2 , to reveal the term which need H 3 -energy, we give a more precise estimate. K 1 can be closed by the energy u H 2 . Indeed, by integration by parts, the poincaré inequality (2.4) and the anisotropic inequalities (2.10) and (2.12), we can obtain

K 1 = ( Δ v u u Δ v u + 2 v u 1 1 v u Δ v u + 2 v u v v 2 u Δ v u ) d x C Δ v v u L 2 2 u L 2 1 2 1 u L 2 1 2 + C v 2 u 1 L 2 1 2 1 v 2 u 1 L 2 1 2 1 v u L 2 Δ v u L 2 1 2 ( Δ v u L 2 + 3 Δ v u L 2 ) 1 2 C Δ v v u L 2 2 u H 2 ,

where we have used v u = v u ˜ and the incompressible condition for u .

Also, we first split K 2 into four parts

K 2 = Δ v u θ Δ v θ d x v u 1 1 v θ Δ v θ d x v u 2 2 v θ Δ v θ d x v u 3 3 v θ Δ v θ d x K 21 + K 22 + K 23 + K 24 .

Applying (2.11) and (2.12) again, except for K 23 the other terms can be bounded by ( u , θ ) H 2 .

K 21 + K 22 + K 24 C Δ v v u L 2 Δ v θ L 2 1 2 ( Δ v θ L 2 + Δ v 3 θ L 2 ) 1 2 θ L 2 1 2 1 θ L 2 1 2 + C v 2 u 1 L 2 1 2 1 v 2 u 1 L 2 1 2 1 v θ L 2 Δ v θ L 2 1 2 ( Δ v θ L 2 + Δ v 3 θ L 2 ) 1 2 + C v u 3 L 2 1 2 1 v u 3 L 2 1 2 3 v 2 θ L 2 Δ v θ L 2 1 2 ( Δ v θ L 2 + Δ v 3 θ L 2 ) 1 2 C ( Δ v v u L 2 2 + Δ v 3 θ L 2 2 ) ( u , θ ) H 2 ,

where we use 3 v θ = 3 v θ ˜ . K 23 can only be estimated as follows:

K 23 C v 2 u 2 L 2 Δ v θ L 2 1 2 ( Δ v θ L 2 + Δ v 3 θ L 2 ) 1 2 2 v θ L 2 1 2 1 2 v θ L 2 1 2 C ( Δ v v u L 2 2 + Δ v 3 θ L 2 2 ) θ H 3 .

In summary,

(4.7) d d t ( Δ v u , Δ v θ ) L 2 2 + 2 ν v Δ v u L 2 2 + 2 η 3 Δ v θ L 2 2 C ( u H 2 + θ H 3 ) Δ v v u L 2 Δ v 3 θ L 2 C δ ( Δ v v u L 2 2 + Δ v 3 θ L 2 2 ) ,

which implies

Δ v u L 2 + Δ v θ L 2 δ e C t ,

provided that δ is sufficiently small. This completes the proof of (1.11).

Step 3: the proof of (1.12)

We first establish the equations of ( u ˜ ( 3 ) , θ ) . By u ¯ 3 ( 3 ) = 0 , θ ¯ ( 3 ) = 0 , we can see

(4.8) t u ¯ 1 ( 3 ) + 1 u 1 2 ¯ ( 3 ) + 2 u 1 u 2 ¯ ( 3 ) = 1 p ¯ ( 3 ) + ν Δ v u ¯ 1 ( 3 ) , t u ¯ 2 ( 3 ) + 1 u 1 u 2 ¯ ( 3 ) + 2 u 2 2 ¯ ( 3 ) = 2 p ¯ ( 3 ) + ν Δ v u ¯ 2 ( 3 ) , 1 u ¯ 1 ( 3 ) + 2 u ¯ 2 ( 3 ) = 0 .

Then from (4.8) and (1.2), we can deduce

(4.9) t u ˜ 1 ( 3 ) + 1 ( u 1 2 u 1 2 ¯ ( 3 ) ) + 2 ( u 1 u 2 u 1 u 2 ¯ ( 3 ) ) + 3 ( u ˜ 3 ( 3 ) u 1 ) = 1 p ˜ ( 3 ) + ν Δ v u ˜ 1 ( 3 ) , t u ˜ 2 ( 3 ) + 1 ( u 1 u 2 u 1 u 2 ¯ ( 3 ) ) + 2 ( u 2 2 u 2 2 ¯ ( 3 ) ) + 3 ( u ˜ 3 ( 3 ) u 2 ) = 2 p ˜ ( 3 ) + ν Δ v u ˜ 2 ( 3 ) , t u ˜ 3 ( 3 ) + u u ˜ 3 ( 3 ) = 3 p ˜ ( 3 ) + ν Δ v u ˜ 3 ( 3 ) + θ , t + u θ + u ˜ 3 ( 3 ) = η 33 θ , u ˜ ( 3 ) = 0 .

By taking the L 2 inner product of (4.9) with ( 1 2 u ˜ ( 3 ) , 1 2 θ ) and integrating in space, we infer

1 2 d d t ( 1 u ˜ ( 3 ) , 1 θ ) L 2 2 + ν 1 v u ˜ ( 3 ) L 2 2 + η 3 1 θ L 2 2 L 1 + + L 8 ,

where

L 1 = 1 2 u ˜ 1 ( 3 ) 1 ( u 1 2 u 1 2 ¯ ( 3 ) ) d x , L 2 = 1 2 u ˜ 1 ( 3 ) 2 ( u 1 u 2 u 1 u 2 ¯ ( 3 ) ) d x , L 3 = 1 2 u ˜ 1 ( 3 ) 3 ( u ˜ 3 ( 3 ) u 1 ) d x , L 4 = 1 2 u ˜ 2 ( 3 ) 1 ( u 1 u 2 u 1 u 2 ¯ ( 3 ) ) d x , L 5 = 1 2 u ˜ 2 ( 3 ) 2 ( u 2 2 u 2 2 ¯ ( 3 ) ) d x , L 6 = 1 2 u ˜ 2 ( 3 ) 3 ( u ˜ 3 ( 3 ) u 2 ) d x , L 7 = 1 2 u ˜ 3 ( 3 ) ( u u ˜ 3 ( 3 ) ) d x , L 8 = 1 2 θ ( u θ ) d x .

Now we estimate L 1 through L 8 one by one. To bound L 1 and L 2 , we fist divide u 1 2 u 1 2 ¯ ( 3 ) and u 1 u 2 u 1 u 2 ¯ ( 3 ) into the terms as follows:

(4.10) u 1 2 u 1 2 ¯ ( 3 ) = 2 u ˜ 1 ( 3 ) u ¯ 1 ( 3 ) + ( u ˜ 1 ( 3 ) ) 2 ( u ˜ 1 ( 3 ) ) 2 ¯ ( 3 ) = 2 u ˜ 1 ( 3 ) u ¯ 1 ( 3 ) + ( u ˜ 1 ( 3 ) ) 2 ˜ ( 3 ) ,

(4.11) u 1 u 2 u 1 u 2 ¯ ( 3 ) = u ˜ 1 ( 3 ) u ¯ 2 ( 3 ) + u ˜ 2 ( 3 ) u ¯ 1 ( 3 ) + u ˜ 1 ( 3 ) u ˜ 2 ( 3 ) ˜ ( 3 ) .

Then by Hölder’s inequality, the Poincaré inequality (2.3), and the anisotropic inequality (2.17), we can derive

L 1 = 1 2 u ˜ 1 ( 3 ) ( 2 1 u ˜ 1 ( 3 ) u ¯ 1 ( 3 ) + 2 u ˜ 1 ( 3 ) 1 u ¯ 1 ( 3 ) + 1 ( u ˜ 1 ( 3 ) ) 2 ˜ ( 3 ) ) d x C 1 2 u ˜ 1 ( 3 ) L 2 ( 1 u ˜ 1 ( 3 ) L 2 u ¯ 1 ( 3 ) L + v 2 u ˜ 1 ( 3 ) L 2 1 u ¯ 1 ( 3 ) L 2 1 2 1 2 u ¯ 1 ( 3 ) L 2 1 2 + 1 ( u ˜ 1 ( 3 ) ) 2 L 2 ) C v u ˜ ( 3 ) L 2 2 u H 2 + 2 1 2 u ˜ 1 ( 3 ) L 2 1 u ˜ 1 ( 3 ) L 2 u ˜ 1 ( 3 ) L C v u ˜ ( 3 ) L 2 2 u H 2 .

Similarly, L 2 and L 3 can be bounded as follows:

L 2 = 1 2 u ˜ 1 ( 3 ) ( 2 u ˜ 1 ( 3 ) u ¯ 2 ( 3 ) + u ˜ 1 ( 3 ) 2 u ¯ 1 ( 3 ) + 2 u ˜ 2 ( 3 ) u ¯ 1 ( 3 ) + u ˜ 2 ( 3 ) 2 u ¯ 1 ( 3 ) + 2 ( u ˜ 1 ( 3 ) u ˜ 2 ( 3 ) ) ˜ ) d x C 1 2 u ˜ 1 ( 3 ) L 2 ( 2 u ˜ ( 3 ) L 2 u ¯ ( 3 ) L + v 2 u ˜ ( 3 ) L 2 2 u ¯ ( 3 ) L 2 1 2 1 2 u ¯ ( 3 ) L 2 1 2 + 2 ( u ˜ 1 ( 3 ) u ˜ 2 ( 3 ) ) L 2 ) C v u ˜ ( 3 ) L 2 2 u H 2 + 2 1 2 u ˜ 1 ( 3 ) L 2 2 u ˜ ( 3 ) L 2 u ˜ ( 3 ) L C v u ˜ ( 3 ) L 2 2 u H 2 , L 3 = 1 2 u ˜ 1 ( 3 ) ( 3 u 3 ˜ ( 3 ) u 1 + u 3 ˜ ( 3 ) 3 u 1 ) d x 1 2 u ˜ 1 ( 3 ) L 2 ( 3 u 3 ˜ ( 3 ) L 2 u L + v 2 u 3 ˜ ( 3 ) L 2 3 u 1 L 2 1 2 3 1 u 1 L 2 1 2 ) v u ˜ ( 3 ) L 2 2 u H 2 .

For L 4 , we first use the integration by parts to divide it into three parts

L 4 = ( 1 u ˜ 2 ( 3 ) 1 2 ( u ˜ 1 ( 3 ) u ¯ 2 ( 3 ) ) 1 u ˜ 2 ( 3 ) 1 2 ( u ˜ 2 ( 3 ) u ¯ 1 ( 3 ) ) + 1 2 u ˜ 2 ( 3 ) 1 ( u ˜ 1 ( 3 ) u ˜ 2 ( 3 ) ˜ ( 3 ) ) ) d x L 41 + L 42 + L 43 .

Then, applying Poincaré inequality (2.4) and the anisotropic inequality (2.14), (2.16) yields

L 41 = 1 u ˜ 2 ( 3 ) ( 1 2 u ˜ 1 ( 3 ) u ¯ 2 ( 3 ) + 2 1 u ˜ 1 ( 3 ) 1 u ¯ 2 ( 3 ) + u ˜ 1 ( 3 ) 1 2 u ¯ 2 ( 3 ) ) d x 1 u ˜ 2 ( 3 ) L 2 1 2 u ˜ 1 ( 3 ) L 2 u ¯ 2 ( 3 ) L + C 1 v u ˜ 2 ( 3 ) L 2 1 v u ˜ 1 ( 3 ) L 2 1 u ¯ 2 ( 3 ) L 2 1 2 1 2 u ¯ 2 ( 3 ) L 2 1 2 + C 1 u ˜ 2 ( 3 ) L 2 1 2 ( 1 u ˜ 2 ( 3 ) L 2 + 2 1 u ˜ 2 ( 3 ) L 2 ) 1 2 3 u ˜ 1 ( 3 ) L 2 1 2 1 3 u ˜ 1 ( 3 ) L 2 1 2 1 2 u ¯ ( 3 ) L 2 C v u ˜ ( 3 ) L 2 2 u H 2 .

Similarly,

L 42 = 1 u ˜ 2 ( 3 ) ( 1 2 u ˜ 2 u ¯ 1 ( 3 ) + 2 1 u ˜ 2 ( 3 ) 1 u ¯ 1 ( 3 ) + u ˜ 2 ( 3 ) 1 2 u ¯ 1 ( 3 ) ) d x C v u ˜ ( 3 ) L 2 2 u H 2 .

It follows from (2.18) that

L 43 1 2 u ˜ 2 ( 3 ) L 2 1 ( u ˜ 1 ( 3 ) u ˜ 2 ( 3 ) ) L 2 1 2 u ˜ 2 ( 3 ) L 2 v u ˜ ( 3 ) L 2 1 2 v 1 u ˜ ( 3 ) L 2 1 2 1 v u ˜ ( 3 ) L 2 C v u ˜ ( 3 ) L 2 2 u H 2 .

Therefore,

L 4 C v u ˜ ( 3 ) L 2 2 u H 2 .

A similar argument to L 4 yields

L 5 = ( 2 1 u ˜ 2 ( 3 ) 1 2 ( u ˜ 2 ( 3 ) u 2 ¯ ( 3 ) ) + 1 2 u ˜ 2 ( 3 ) 2 ( u ˜ 2 ( 3 ) ) 2 ˜ ( 3 ) ) d x C v u ˜ ( 3 ) L 2 2 u H 2 .

For L 6 , by the integration by parts and (2.4) and (2.17), we have

L 6 = 1 3 u ˜ 2 ( 3 ) ( 1 u 3 ˜ ( 3 ) u 2 + u 3 ˜ ( 3 ) 1 u 2 ) d x 1 3 u ˜ ( 3 ) L 2 ( 1 u 3 ˜ ( 3 ) L 2 u 2 L + v 2 u 3 ˜ ( 3 ) L 2 1 u 2 L 2 1 2 1 2 u 2 L 2 1 2 ) v u ˜ ( 3 ) L 2 2 u H 2 .

We turn to bound the last two terms. By integration parts, applying the anisotropic inequality (2.14) yields

L 7 + L 8 = 1 u u 3 ˜ ( 3 ) 1 u ˜ 3 ( 3 ) d x 1 u 1 1 θ 1 θ d x 1 u 2 2 θ 1 θ d x 1 u 3 3 θ 1 θ d x C 3 u ˜ 3 ( 3 ) L 2 v 1 u ˜ 3 ( 3 ) L 2 1 u L 2 1 2 1 2 u L 2 1 2 + 1 u L θ L 2 1 θ L 2 + C v 1 u ˜ ( 3 ) L 2 3 2 θ L 2 1 θ L 2 1 2 1 2 θ L 2 1 2 C ( u H 3 + θ H 2 ) v u ˜ ( 3 ) L 2 3 θ L 2 ,

where we have used 1 u 3 = 1 u ˜ 3 ( 3 ) , 3 = 3 θ ˜ 3 ( 3 ) , and θ L 2 C 3 θ L 2 .

Therefore, invoking the uniform bound (4.1), we have for some positive pure constant C independent of δ ,

d d t ( 1 u ˜ ( 3 ) , 1 θ ) L 2 2 + 2 ν 1 v u ˜ ( 3 ) L 2 2 + 2 η 3 1 θ L 2 2 C δ ( v u ˜ ( 3 ) L 2 2 + 3 θ L 2 2 ) ,

which together with (4.7) leads to

d d t ( ( 1 u ˜ ( 3 ) , 1 θ ) L 2 2 + ( Δ v u , Δ v θ ) L 2 2 ) + 2 ν ( 1 v u ˜ ( 3 ) L 2 2 + v Δ v u L 2 2 ) + 2 η ( 3 1 θ L 2 2 + 3 Δ v θ L 2 2 ) C δ ( v u ˜ ( 3 ) L 2 2 + 3 θ L 2 2 + Δ v v u L 2 2 + Δ v 3 θ L 2 2 ) .

Since by poincaré inequality (2.3), we have

v 2 u ˜ ( 3 ) L 2 2 C 3 v 2 u ˜ ( 3 ) L 2 2 = C 3 v 2 u L 2 2 , v 3 θ L 2 2 C 3 v 2 θ L 2 2 ,

then we can derive for C 5 > 0

d d t ( ( 1 u ˜ ( 3 ) , 1 θ ) L 2 2 + ( Δ v u , Δ v θ ) L 2 2 ) + 2 ν ( 1 v u ˜ ( 3 ) L 2 2 + v Δ v u L 2 2 ) + 2 η ( 3 1 θ L 2 2 + 3 Δ v θ L 2 2 ) C 5 δ ( 1 v u ˜ ( 3 ) L 2 2 + 1 3 θ L 2 2 + Δ v v u L 2 2 + Δ v 3 θ L 2 2 ) ,

which implies that if the initial data δ is sufficiently small, then the decay holds

1 u ˜ ( 3 ) L 2 + 1 θ L 2 δ e C t .

For 1 u ˜ ( 2 ) , it is not necessary to repeat the proof of 1 u ˜ ( 3 ) since the proof is more complicated. In fact, it follows from (2.2) that

f ˜ ( 2 ) L 2 C ( f ˜ L 2 + f ˜ ( 3 ) L 2 ) .

Then by the decay for 1 u ˜ ( 3 ) and 1 u ˜ , we obtain

1 u ˜ ( 2 ) L 2 C ( 1 u ˜ L 2 + 1 u ˜ ( 3 ) L 2 ) δ e C t .

This completes the proof of Theorem 1.2.□

Acknowledgements

The authors are grateful for the reviewer’s valuable comments that improved the manuscript. We also express our gratitude to the editors for efficiently handling our manuscript.

  1. Funding information: Lin was partially supported by the National Natural Science Foundation of China (NNSFC) under Grant 11701049 and the Natural Science Foundation of Sichuan Province under Grant 2023NSFSC0056. Our manuscript has no associated data.

  2. Author contributions: All authors have accepted responsibility for the entire content of this manuscript and consented to its submission to the journal, reviewed all the results, and approved the final version of the manuscript. Lin and Liu made formal analysis and the derivation of the proof. Liu and Guo prepared the original manuscript with contributions from all co-authors. You performed the review and editing.

  3. Conflict of interest: The authors state no conflict of interest.

References

[1] D. Adhikari, C. Cao, H. Shang, J. Wu, X. Xu, and Z. Ye, Global regularity results for the 2D Boussinesq equations with partial dissipation. J. Differ. Equ. 260 (2016), 1893–1917. 10.1016/j.jde.2015.09.049Search in Google Scholar

[2] D. Adhikari, O. Ben Said, U. R. Pandey, and J. Wu, Stability and large-time behavior for the 2D Boussineq system with horizontal dissipation and vertical thermal diffusion, Nonl. Differ. Equ. Appl. NoDEA 29 (2022), no. 4, 1–43. 10.1007/s00030-022-00773-4Search in Google Scholar

[3] D. Chae, Global regularity for the 2D Boussinesq equations with partial viscosity terms, Adv. Math. 203 (2006), 497–513. 10.1016/j.aim.2005.05.001Search in Google Scholar

[4] C. Cao and J. Wu, Global regularity for the 2D MHD equations with mixed partial dissipation and magnetic diffusion, Adv. Math. 226 (2011), 1803–1822. 10.1016/j.aim.2010.08.017Search in Google Scholar

[5] C. Cao and J. Wu, Global regularity for the 2D anisotropic Boussinesq equations with vertical dissipation, Arch. Ration. Mech. Anal. 208 (2013), 985–1004. 10.1007/s00205-013-0610-3Search in Google Scholar

[6] R. Danchin and M. Paicu, Global existence results for the anisotropic Boussinesq system in dimension two, Math. Models Methods Appl. Sci. 21 (2011), 421–457. 10.1142/S0218202511005106Search in Google Scholar

[7] C. R. Doering, J. Wu, K. Zhao, and X. Zheng, Long time behavior of the two-dimensional Boussinesq equations without buoyancy diffusion. Phys. D 376 (2018), no. 377, 144–159. 10.1016/j.physd.2017.12.013Search in Google Scholar

[8] B. Dong, J. Wu, and X. Xu, Stability and exponential decay for the 2D anisotropic Boussinesq equations with horizontal dissipation, Calc. Var. 60 (2021), 116. 10.1007/s00526-021-01976-wSearch in Google Scholar

[9] D. Fang, W. Le, and T. Zhang, Global solutions of 3D axisymmetric Boussinesq equations with nonzero swirl, Nonlinear Anal. 166 (2018), 48–86. 10.1016/j.na.2017.10.008Search in Google Scholar

[10] T. Hou and C. Li, Global well-posedness of the viscous Boussinesq equations, Discrete Cont. Dyn. Syst. 12 (2005), 1–12. 10.3934/dcds.2005.12.1Search in Google Scholar

[11] T. Hmidi, S. Keraani, and F. Rousset, Global well-posedness for Euler-Boussinesq system with critical dissipation, Commun. Partial Differ. Equ. 36 (2011), 420–445. 10.1080/03605302.2010.518657Search in Google Scholar

[12] T. Hmidi, S. Keraani, and F. Rousset, Global well-posedness for a Boussinesq-Navier-Stokes system with critical dissipation, J. Differ. Equ. 249 (2010), 2147–2174. 10.1016/j.jde.2010.07.008Search in Google Scholar

[13] W. Hu, Y. Wang, J. Wu, B. Xiao, and J. Yuan, Partially dissipated 2D Boussinesq equations with Navier type boundary conditions. Phys. D 376 (2018), no. 377, 39–48. 10.1016/j.physd.2017.07.003Search in Google Scholar

[14] J. R. Holton, An Introduction to Dynamic Meteorology, Academic Press, New York, 1972. Search in Google Scholar

[15] Q. Jiu, C. Miao, J. Wu, and Z. Zhang, The 2D incompressible Boussinesq equations with general critical dissipation, SIAM J. Math. Anal. 46 (2014), 3426–3454. 10.1137/140958256Search in Google Scholar

[16] R. Ji, L. Yan, and J. Wu, Optimal decay for the 3D anisotropic Boussinesq equations near the hydrostatic balance, Calc. Var. 61 (2014), 136. 10.1007/s00526-022-02242-3Search in Google Scholar

[17] R. Ji, D. Li, Y. Wei, and J. Wu, Stability of hydrostatic equilibrium to the 2D Boussinesq systems with partial dissipation, Appl. Math. Lett. 98 (2019), 392–397. 10.1016/j.aml.2019.06.019Search in Google Scholar

[18] Q. Jiu, J. Wu, and W. Yang, Eventual regularity of the two-dimensional Boussinesq equations with supercritical dissipation, J. Nonl. Sci. 25 (2015), 37–58. 10.1007/s00332-014-9220-ySearch in Google Scholar

[19] A. Kiselev and C. Tan, Finite time blow up in the hyperbolic Boussinesq system, Adv. Math. 325 (2018), 34–55. 10.1016/j.aim.2017.11.019Search in Google Scholar

[20] J. Li, H. Shang, J. Wu, X. Xu, and Z. Yeee, Regularity criteria for the 2D Boussinesq equations with supercritical dissipation, Commun. Math. Sci. 14 (2016), 1999–2022. 10.4310/CMS.2016.v14.n7.a10Search in Google Scholar

[21] A. Larios, E. Lunasin, and E. S. Titi, Global well-posedness for the 2D Boussinesq system with anisotropic viscosity and without heat diffusion, J. Differ. Equ. 255 (2013), 2636–2654. 10.1016/j.jde.2013.07.011Search in Google Scholar

[22] M. Lai, R. Pan, and K. Zhao, Initial boundary value problem for two-dimensional viscous Boussinesq equations, Arch. Ration. Mech. Anal. 199 (2011), 739–760. 10.1007/s00205-010-0357-zSearch in Google Scholar

[23] S. Lai, J. Wu, and Y. Zhong, Stability and large-time behavior of the 2D Boussinesq equations with partial dissipation, J. Differ. Equ. 271 (2021), 764–796. 10.1016/j.jde.2020.09.022Search in Google Scholar

[24] H. Lin, Q. Sun, S. Liu, and H. Zhang, The stability and decay for the 2D incompressible Euler-like equations, J. Math. Fluid Mech. 25 (2023), 85. 10.1007/s00021-023-00824-5Search in Google Scholar

[25] C. Miao and L. Xue, On the global well-posedness of a class of Boussinesq-Navier-Stokes systems, NoDEA Nonl. Differ. Equ. Appl. 18 (2011), 707–735. 10.1007/s00030-011-0114-5Search in Google Scholar

[26] M. Paicu and N. Zhu, On the striated regularity for the 2D anisotropic Boussinesq system, J. Nonl. Sci. 30 (2020), 1115–1164. 10.1007/s00332-019-09603-xSearch in Google Scholar

[27] J. Pedlosky, Geophysical Fluid Dynamics, 2nd edn. Springer, Berlin, 1987. 10.1007/978-1-4612-4650-3Search in Google Scholar

[28] O. B. Said, U. Pandey, and J. Wu, The stabilizing effect of the temperature on buoyancy-driven fluids, Indiana University Math. J. 71 (2020), 2605–2646. 10.1512/iumj.2022.71.9070Search in Google Scholar

[29] M. Schonbek, L2 decay for weak solutions of the Navier-Stokes equations, Arch. Ration. Mech. Anal. 88 (1985), 209–222. 10.1007/BF00752111Search in Google Scholar

[30] M. Schonbek and T. Schonbek, Moment sand lower bounds in the far-field of solutions to quasi-geostrophic flows, Discret. Contin. Dyn. Syst. 13 (2005), 1277–1304. 10.3934/dcds.2005.13.1277Search in Google Scholar

[31] M. Schonbek and M. Wiegner, On the decay of higher-order norms of the solutions of Navier-Stokes equations, Proc. R. Soc. Edinb. Sect. A 126 (1996), 677–685. 10.1017/S0308210500022976Search in Google Scholar

[32] S. Sokrani, On the global well-posedness of 3D Boussinesq system with partial viscosity and axisymmetric data, Discrete Contin. Dyn. Syst. 39 (2019), 1613–1650. 10.3934/dcds.2019072Search in Google Scholar

[33] L. Tao, J. Wu, K. Zhao, and X. Zheng, Stability near hydrostatic equilibrium to the 2D Boussinesq equations without thermal diffusion, Arch. Ration. Mech. Anal. 237 (2020), 585–630. 10.1007/s00205-020-01515-5Search in Google Scholar

[34] T. Tao, Nonlinear Dispersive Equations: Local and Global Analysis, CBMS Regional Conference series in Mathematics, American Mathematical Society, Providence, RI, 2006.10.1090/cbms/106Search in Google Scholar

[35] J. Wu, X. Xu, and Z. Yeee, The 2D Boussinesq equations with fractional horizontal dissipation and thermal diffusion, J. Math. Pures et Appl. 115 (2018), no. 9, 187–217. 10.1016/j.matpur.2018.01.006Search in Google Scholar

[36] J. Wu, X. Xu, and N. Zhu, Stability and decay rates for a variant of the 2D Boussinesq-Bénard system, Commun. Math. Sci. 17 (2019), no. 8, 2325–2352. 10.4310/CMS.2019.v17.n8.a11Search in Google Scholar

[37] J. Wu and Q. Zhang, Stability and optimal decay for a system of 3D anisotropic Boussinesq equations, Nonlinearity 34 (2021), 5456. 10.1088/1361-6544/ac08e9Search in Google Scholar

[38] W. Yang, Q. Jiu, and J. Wu, The 3D incompressible Boussinesq equations with fractional partial dissipation, Commun. Math. Sci. 16 (2018), no. 3, 617–633. 10.4310/CMS.2018.v16.n3.a2Search in Google Scholar

Received: 2024-07-28
Accepted: 2024-11-22
Published Online: 2024-12-31

© 2024 the author(s), published by De Gruyter

This work is licensed under the Creative Commons Attribution 4.0 International License.

Articles in the same Issue

  1. Regular Articles
  2. Existence of normalized peak solutions for a coupled nonlinear Schrödinger system
  3. Orbital stability of the trains of peaked solitary waves for the modified Camassa-Holm-Novikov equation
  4. Global boundedness in a two-dimensional chemotaxis system with nonlinear diffusion and singular sensitivity
  5. Existence and uniqueness of solution for a singular elliptic differential equation
  6. The bounded variation capacity and Sobolev-type inequalities on Dirichlet spaces
  7. Vanishing and blow-up solutions to a class of nonlinear complex differential equations near the singular point
  8. Existence, uniqueness, localization and minimization property of positive solutions for non-local problems involving discontinuous Kirchhoff functions
  9. Concentration phenomena for a fractional relativistic Schrödinger equation with critical growth
  10. Beyond the classical strong maximum principle: Sign-changing forcing term and flat solutions
  11. Monotonicity of solutions for parabolic equations involving nonlocal Monge-Ampère operator
  12. On a nonlinear Robin problem with an absorption term on the boundary and L1 data
  13. Multiple solutions for the quasilinear Choquard equation with Berestycki-Lions-type nonlinearities
  14. Gradient estimates for a class of higher-order elliptic equations of p-growth over a nonsmooth domain
  15. Global existence and decay estimates of the classical solution to the compressible Navier-Stokes-Smoluchowski equations in ℝ3
  16. Global existence and finite-time blowup for a mixed pseudo-parabolic r(x)-Laplacian equation
  17. Multiple positive solutions for a class of concave-convex Schrödinger-Poisson-Slater equations with critical exponent
  18. A minimization problem with free boundary for p-Laplacian weakly coupled system
  19. Multiplicity of semiclassical solutions for a class of nonlinear Hamiltonian elliptic system
  20. k-convex solutions for multiparameter Dirichlet systems with k-Hessian operator and Lane-Emden type nonlinearities
  21. Infinitely many solutions for Hamiltonian system with critical growth
  22. On optimal control in a nonlinear interface problem described by hemivariational inequalities
  23. Variational–hemivariational system for contaminant convection–reaction–diffusion model of recovered fracturing fluid
  24. Potential and monotone homeomorphisms in Banach spaces
  25. Critical fractional Schrödinger-Poisson systems with lower perturbations: the existence and concentration behavior of ground state solutions
  26. Supercritical Hénon-type equation with a forcing term
  27. Concentration of blow-up solutions for the Gross-Pitaveskii equation
  28. Mountain-pass-type solutions for Schrödinger equations in R2 with unbounded or vanishing potentials and critical exponential growth nonlinearities
  29. Regularity for critical fractional Choquard equation with singular potential and its applications
  30. Double phase anisotropic variational problems involving critical growth
  31. Normalized solutions of NLS equations with mixed nonlocal nonlinearities
  32. Nontrivial solutions for resonance quasilinear elliptic systems
  33. Standing waves for Choquard equation with noncritical rotation
  34. Low regularity conservation laws for Fokas-Lenells equation and Camassa-Holm equation
  35. Modified quasilinear equations with strongly singular and critical exponential nonlinearity
  36. Limit profiles and the existence of bound-states in exterior domains for fractional Choquard equations with critical exponent
  37. Nests of limit cycles in quadratic systems
  38. Regularity of minimizers for double phase functionals of borderline case with variable exponents
  39. Boundedness, existence and uniqueness results for coupled gradient dependent elliptic systems with nonlinear boundary condition
  40. Ground state solutions for magnetic Schrödinger equations with polynomial growth
  41. Nonoccurrence of Lavrentiev gap for a class of functionals with nonstandard growth
  42. Dynamics for wave equations connected in parallel with nonlinear localized damping
  43. Boussinesq's equation for water waves: Asymptotics in Sector I
  44. Optimal global second-order regularity and improved integrability for parabolic equations with variable growth
  45. Normalized solutions of Schrödinger equations involving Moser-Trudinger critical growth
  46. Normalized solutions for the double-phase problem with nonlocal reaction
  47. Normalized solutions for Sobolev critical fractional Schrödinger equation
  48. Well-posedness of Cauchy problem of fractional drift diffusion system in non-critical spaces with power-law nonlinearity
  49. Improved results on planar Klein-Gordon-Maxwell system with critical exponential growth
  50. Existence and multiplicity of solutions for a new p(x)-Kirchhoff equation
  51. Quasiconvex bulk and surface energies: C1,α regularity
  52. Time decay estimates of solutions to a two-phase flow model in the whole space
  53. Bounds for the sum of the first k-eigenvalues of Dirichlet problem with logarithmic order of Klein-Gordon operators
  54. The properties of a new fractional g-Laplacian Monge-Ampère operator and its applications
  55. A fractional profile decomposition and its application to Kirchhoff-type fractional problems with prescribed mass
  56. Cauchy problem for a non-Newtonian filtration equation with slowly decaying volumetric moisture content
  57. Multiplicity of normalized semi-classical states for a class of nonlinear Choquard equations
  58. The second Yamabe invariant with boundary
  59. Peaked solitary waves and shock waves of the Degasperis-Procesi-Kadomtsev-Petviashvili equation
  60. Normalized solutions for the Choquard equations with critical nonlinearities
  61. Boundedness and long-time behavior in a parabolic-elliptic system arising from biological transport networks
  62. Initial boundary value problem and exponential stability for the planar magnetohydrodynamics equations with temperature-dependent viscosity
  63. Nonexistence of mean curvature flow solitons with polynomial volume growth immersed in certain semi-Riemannian warped products
  64. Analysis of a vector-borne disease model with vector-bias mechanism in advective heterogeneous environment
  65. Normalized solutions for the Kirchhoff equation with combined nonlinearities in ℝ4
  66. Nonexistence of the compressible Euler equations with space-dependent damping in high dimensions
  67. Multiplicity of solutions for a nonhomogeneous quasilinear elliptic equation with concave-convex nonlinearities
  68. On semilinear inequalities involving the Dunkl Laplacian and an inverse-square potential outside a ball
  69. Besov regularity for the elliptic p-harmonic equations in the non-quadratic case
  70. Uniqueness and nondegeneracy of ground states for Δ u + u = ( I α u 2 ) u in R 3 when α is close to 2
  71. Existence of solutions for a class of quasilinear Schrödinger equations with Choquard-type nonlinearity
  72. Weighted Hardy-Adams inequality on unit ball of any even dimension
  73. Existence and regularity for a p-Laplacian problem in ℝN with singular, convective, and critical reaction
  74. Temporal periodic solutions of non-isentropic compressible Euler equations with geometric effects
  75. Nodal solutions for a zero-mass Chern-Simons-Schrödinger equation
  76. The modified quasi-boundary-value method for an ill-posed generalized elliptic problem
  77. Nonlocal heat equations with generalized fractional Laplacian
  78. Choquard equations with recurrent potentials
  79. Local and global well-posedness of the Maxwell-Bloch system of equations with inhomogeneous broadening
  80. The Riemann problem for two-layer shallow water equations with bottom topography
  81. Global stability and asymptotic profiles of a partially degenerate reaction diffusion Cholera model with asymptomatic individuals
  82. On Kirchhoff-Schrödinger-Poisson-type systems with singular and critical nonlinearity
  83. Energy-variational solutions for viscoelastic fluid models
  84. Stability on 3D Boussinesq system with mixed partial dissipation
  85. Existence and multiplicity results for non-autonomous second-order Hamiltonian systems
  86. Erratum
  87. Erratum to “Normalized solutions for the Kirchhoff equation with combined nonlinearities in ℝ4
Downloaded on 7.9.2025 from https://www.degruyterbrill.com/document/doi/10.1515/anona-2024-0060/html
Scroll to top button