Abstract
We consider an initial boundary value problem of the multi-dimensional Boussinesq equations in the absence of thermal diffusion with velocity damping or velocity diffusion under the stress free boundary condition in horizontally periodic strip domain. We prove the global-in-time existence of classical solutions in high order Sobolev spaces satisfying high order compatibility conditions around the linearly stratified equilibrium, the convergence of the temperature to the asymptotic profile, and sharp decay rates of the velocity field and temperature fluctuation in all intermediate norms based on spectral analysis combined with energy estimates. To the best of our knowledge, our results provide first sharp decay rates for the temperature fluctuation and the vertical velocity to the linearly stratified Boussinesq equations in all intermediate norms.
Similar content being viewed by others
Avoid common mistakes on your manuscript.
1 Introduction
We consider the Boussinesq equations for buoyant fluids
where v, p, and \(\rho \) denote the fluid velocity field, scalar pressure and density (or temperature) respectively. The parameter \(\alpha \ge 0\) and \(\beta \ge 0\) represent the strength of dissipation and thermal diffusion, while the parameters \(\nu \ge 0 \) and \(\kappa \ge 0\) stand for the nonnegative constant fluid viscosity and thermal diffusivity, respectively. The d-dimensional vector \(e_d\) stands for \((0, \cdots , 0, 1)^T\).
The Boussinesq equations (1.1) arise in geophysical fluid dynamics to model and study atmospheric and oceanographic flows [36, 39] and describe interesting physical phenomena such as Rayleigh-Bénard convection [18, 22] and turbulence [10]. From a mathematical point of view, the Boussinesq equations are intimately tied to the Euler and Navier-Stokes equations and they share important features such as the vortex stretching. In fact, the two-dimensional inviscid Boussinesq equations can be viewed as the three-dimensional axisymmetric Euler equations for swirling flows [37]. Due to its physical and mathematical relevance, there have been a lot of works and progress made on the Boussinesq system in the past decades: for instance, see [1, 2, 4, 6, 7, 9, 11, 12, 20, 23, 24, 26, 27, 29,30,31, 33,34,35, 41,42,45] and references therein on the local, global well-posedness and regularity problem.
On the other hand, it is well-known that the system (1.1) has the exact solutions, called hydrostatic equilibrium, with the balance equation
In recent years, the stability around the linearly stratified state \((v_s, \rho _s, p_s):=(0, \cdots , 0, x_d, x_d^2/2)\) has been a subject of active research in the presence of dissipation where damping is understood as a limit of fractional diffusion. For \(d=2\), there exist many stability results (see [2, 3] and references therein), while less works are available for other space dimension. Among others, asymptotic stability with velocity damping was studied in \({\mathbb {R}}^3\) [15], and the stability result has been extended to \({\mathbb {R}}^d\) with more general initial data in [28].
In this paper, we focus on the domain with boundary, in particular \(\Omega =\mathbb T^{d-1} \times [-1,1]\). This type of domain with \(\rho = 1\) and \(\rho = -1\) fixed on the bottom boundary and top boundary has been used to demonstrate the Rayleigh-Bénard convection [18, 22], which leads to the instability of the solution by a continuously heated bottom fluid. On the contrary, the opposite case where \(\rho = -1\) and \(\rho = 1\) on the bottom and top boundary respectively stabilizes the system. We will show stabilizing aspects of the latter by analyzing the dynamics near linearly stratified hydrostatic equilibrium \((v_s, \rho _s, p_s)=(0, \cdots , 0, x_d, x_d^2/2)\). We consider two cases: \(\alpha = 0\) (velocity damping) and \(\alpha =1\) (velocity diffusion) without thermal diffusion \((\kappa = 0)\). When \(\alpha =0\), we take the no-penetration boundary condition \(v \cdot n = 0\) and when \(\alpha = 1\), we impose the stress free boundary condition, also known as the Lions boundary condition \(v \cdot n = 0\) and \({\text {curl}}v \times n = 0\), where the temperature is fixed at \(\rho _s = -1\) and \(\rho _s = 1\) on the each boundary. Here, n denotes the outward unit normal vector to \(\partial \Omega \). Let us set
Then, the perturbed system is given by
where the boundary conditions of the velocity field are preserved and \(\theta \) vanishes on \(\partial \Omega \) in each case \(\alpha = 0\) and \(\alpha = 1\) with \(\theta _0|_{\partial \Omega } =0\).
We now discuss some relevant prior works regarding (1.2) starting with the case \(\alpha = 0\). Castro, Córdoba, and Lear [5] showed the asymptotic stability of (1.2) for \(d=2\). In particular, the authors showed that high order compatibility conditions are satisfied for well-prepared data, and introduced proper solution spaces \(X^m(\Omega )\), \(Y^m(\Omega ) \subset H^m(\Omega )\) with orthonormal bases (see Sect. 2.2 for the definitions). For their main result, for \(m \in \mathbb N\) with \(m \ge 17\), the small data global existence with temporal decay estimate \((1+t)^{\frac{m-7}{8}} (\Vert v(t) \Vert _{H^4} + \Vert \bar{\theta } (t) \Vert _{H^4}) \le C\) was obtained, where \(\bar{\theta }(t) := \theta (t) - \int _{\mathbb T} \theta (t,x) \,\textrm{d}x_1\). It is worth pointing out that the temporal decay rates of \(H^4\)-norm increase as m gets larger, namely the solutions are more regular. Next we consider the case \(\alpha = 1\). In \(d=2\), long time behavior was first considered by Doering et al [13] for \(v\in H^2\) and \(\theta \in H^1\) and explicit decay rates were given in \({\mathbb {T}}^2\) by Tao et al [40] using the spectral analysis. Recently, Dong and Sun considered the asymptotic stability problem on the infinite flat strip \(\mathbb R^{d-1} \times (0,1)\) for \(d = 2\) and 3 in [16, 17] respectively, and Dong [14] obtained the stability result on \(\mathbb T\times (0,1)\).
In the aforementioned works, some explicit decay rates were obtained with high regularity index m or global existence (2D) with more general initial data was obtained without explicit decay rates. However, the convergence of the temperature fluctuation and its optimal equilibration rate has remained elusive. The goal of this paper is to establish the global existence in \(H^m\), \(m>2+\alpha + \frac{d}{2}\) satisfying high order compatibility conditions, the convergence of \(\theta \) to the asymptotic profile \(\sigma \), and sharp decay rates of \((v, \theta -\sigma )\) in \(H^s\) norms for all \(s\in [0,m]\). We now state the main results:
Theorem 1.1
Let \(d \in \mathbb {N}\) with \(d \ge 2\) and let \(m \in \mathbb {N}\) satisfying \(m > 3+\frac{d}{2}\). Then there exists a constant \(\delta > 0\) such that if initial data \((v_0,\,\theta _0) \in \mathbb X^m \times X^m(\Omega )\) with \({\text {div}} v_0 = 0\), \(\int _{\Omega } v_0 \,\textrm{d}x = 0\), and \(\Vert (v_0, \theta _0) \Vert _{H^m}^2 < \delta ^2\), then (1.2) with \(\alpha = 1\) possesses a unique global classical solution \((v,\,\theta )\) satisfying
with
Moreover, there exists a function
such that
for any \(s \in [0,m]\).
Remark 1.2
The assumption \(\int _{\Omega } v_{0}\,\textrm{d}x = 0\) is essential for the velocity field v decaying in t (see Lemma 2.1).
Remark 1.3
Indeed for any \(\epsilon >0\), there exists a constant \(C>0\) such that
for any \(s \in [0,m+1]\). See Proposition 6.6.
Theorem 1.4
Let \(d \in \mathbb {N}\) with \(d \ge 2\) and let \(m \in \mathbb {N}\) satisfying \(m > 2+\frac{d}{2}\). Then there exists a constant \(\delta > 0\) such that if initial data \((v_0,\,\theta _0) \in \mathbb X^m \times X^m(\Omega )\) with \({\text {div}} v_0 = 0\) and \(\Vert (v_0, \theta _0) \Vert _{H^m}^2 < \delta ^2\), then (1.2) with \(\alpha = 0\) possesses a unique global classical solution \((v,\,\theta )\) satisfying
with
Moreover, there exists a function \(\sigma (x_d)\) defined by (1.4) such that
for any \(s \in [0,m]\).
Remark 1.5
The decay rates for \(\theta \) and \(v_d\) in Theorem 1.1 and 1.4 are sharp (see Sect. 7).
To the best of our knowledge, our results provide the first sharp decay rates for the temperature fluctuation and the vertical velocity in all intermediate norms. In particular, they show the enhanced \(L^2\) decay rate for higher order initial data, while \(H^m\) decay rate doesn’t change for both velocity damping and velocity diffusion. This is in contrast to parabolic equations for which higher norms enjoy faster decay rates. The regularity index m required in our analysis is higher than the one required for the local existence, but it is still significantly smaller than the ones required in the previous results. Also our results demonstrate that the velocity damping leads to faster decay than the velocity diffusion in the presence of the slip boundary, despite having the Poincaré inequality for the velocity field in hand. This is because of coupling structure between the velocity field and the temperature fluctuation of Boussinesq equations: it causes the temperature to decay much slower than the velocity field and the velocity diffusion weakens the temperature damping in high frequency. Moreover, the method developed in this paper is robust and applicable to the periodic box \({\mathbb {T}}^d\), and to various partially dissipative PDEs including non-resistive MHD and IPM (cf. [25]).
The main difficulty comes from the non-decaying \(\theta \) and weak damping in \(\nabla _h\theta \), which makes the standard energy estimates alone hard to bootstrap the local theory to global theory and to capture precise decay rates. To establish the results, we employ the spectral analysis using the orthonormal basis associated to our domain with the slip boundary together with energy estimates, first to obtain the global existence and then to prove the decay rates by relying on the already established uniform bounds of the solutions. The relaxed condition for m comes from estimating the key quantities \(\int \Vert \nabla v(t) \Vert _{L^{\infty }} \,\textrm{d}t \) and \( \int \Vert \partial _d v_d(t) \Vert _{L^{\infty }} \,\textrm{d}t\) which appear in the energy estimates. The previous works on the stability problem of (1.2) (\(d=2\)) were devoted to obtaining the temporal decay estimate for \(\Vert u(t) \Vert _{H^4}\) or \(\Vert \partial _1 {\text {curl}} v(t) \Vert _{H^2}\), which obviously require stronger condition for m (see [5, 14]). Getting decay rates in bounded domains turns out to be more subtle than in the whole space, since \(\theta \) does not decay, while it decays in the whole space. To prove the sharp decay rates of \((v, \theta -\sigma )\) in \(H^s\) norms for all \(s\in [0,m]\) in our domain, we adapt Elgindi’s the splitting scheme of the density first used for the linearly stratified IPM equation in \(\mathbb T^2\) [19]. In particular, splitting the density into decaying part and non-decay part and using the boundedness of high norms obtained from the global existence part, the decay of low norms can be obtained through optimizing splitting scale of frequency in spirit of [19]. We refer to Lemma 3.1 for a clear view of the sharp decay estimates for the linearized system of (1.2), and Sect. 6 for controlling the nonlinear terms in (1.2) with the splitting scheme.
The rest of this paper proceeds as follows. In Sect. 2, we give some preliminary results used for the paper and introduce key function spaces \(X^m(\Omega ), Y^{m}(\Omega ), \mathbb X^m(\Omega )\) and their orthonormal bases. Section 3 is devoted to spectral analysis of (1.2) in frequency variables and the proof of linear decay estimates. In Sect. 4, we present the energy-dissipation inequalities for (1.2). In Sect. 5, we extend the local existence to global-in-time result by combining the energy estimates with spectral analysis to estimate key quantities (5.1) appearing in the energy estimates. Section 6 is devoted to the proof of temporal decay estimates based on the spectral analysis and the splitting scheme. In Sect. 7, we argue that the decay rates are sharp by showing that the linear decay rates can’t be algebraically improved.
2 Preliminaries
We first introduce some notations that will be used throughout this paper. Let \(\langle \cdot ,\cdot \rangle \) be the standard inner product on \(\mathbb C^d\) for any \(d \ge 2\). We use \(\gamma \) as a multi-index, and let \(v_h := (v_1, \cdots , v_{d-1})^T\), \(x_h := (x_1, \cdots , x_{d-1})^T\), and \(\nabla _h := (\partial _1, \cdots , \partial _{d-1})^T\). For any smooth function \(f:\Omega \rightarrow \mathbb R\), we use the notation
Next we investigate the average of the solution \((v,\theta )\) over time.
Lemma 2.1
Let \((v,\theta )\) be a smooth solution to (1.2) with \(\alpha \in \{0, 1\}\). Then, there hold
and
for all \(t \ge 0\). Moreover, if \(\alpha = 1\), then
and if \(\alpha = 0\),
Proof
By the divergence-free condition and the boundary condition \(v_d(x_h,-1) = 0\), we have
for all \(x_d \in [-1,1]\). From the \(v_h\) equation in (1.2), we have
Integration by parts and the boundary condition for \(v_d\) yield
thus,
This gives (2.4) when \(\alpha = 0\). In the case of \(\alpha = 1\), \(\partial _d v_h = 0\) on \(\partial \Omega \) implies (2.3). Similarly, we can obtain (2.2) by the use of (2.1). This completes the proof. \(\square \)
2.1 Boundary conditions
In the section, we briefly show in both cases \(\alpha = 0\) and \(\alpha =1\) the high compatibility conditions, whose statement is as follows: Let \((v,\theta )\) be a global-in-time smooth solution to (1.2) and suppose that there exists \(n \in \mathbb N\) such that \(\partial _d^{2k} \theta _0 = 0\) holds on the boundary for all \(0 \le k \le n\). Then, we have
for any \(1 \le k \le n\).
When \(d=2\), Castro, Córdoba, and Lear [5] and Dong [14] showed (2.5) for \(\alpha =0\) and \(\alpha = 1\) respectively. It is not hard to extend it to the \(d \ge 3\) case. Here, we only give details for the case \(\alpha = 1\).
From our boundary conditions, we see that
By (2.6) and the incompressibility, it holds
Then from the \(v_d\) equation in (1.2), we can see
on the boundary. Next, we apply \(\partial _d\) to the \(v_h\) equation in (1.2) and have
The previous results imply that \(\partial _d^3 v_h = 0\) on the boundary. From the \(\theta \) equation in (1.2), we can see
hence,
Consider the flow map \(\Phi (t,x)\) with \(\partial _t \Phi (t,x) = (v_h(t,\Phi (t,x)),0)\). Then, it holds
By the use of Grönwall’s inequality, we have
Thus, \(\partial _d^2 \theta _0 = 0\) is conserved over time on the boundary, whenever \(\partial _d v_d \in L^1_t\). Thus, (2.5) with \(k=1\) is obtained. It is clear that
Repeating the above processes, we can deduce (2.5) for all \(1 \le k \le n\).
2.2 Functional spaces and orthonormal bases
To introduce our solution spaces, we define orthonormal sets \(\{ b_q \}_{q \in \mathbb N}\) and \(\{ c_q \}_{q \in \mathbb N\cup \{0\}}\) by
Note that each set is orthonormal basis for \(L^2([-1,1])\). Let
Then we have the following relations
Now, we consider the function spaces
where
Then, \(\{ \mathscr {B}_{n,q} \}_{(n,q) \in \mathbb Z^{d-1} \times \mathbb N}\) and \(\{ \mathscr {C}_{n,q} \}_{(n,q) \in \mathbb Z^{d-1} \times \mathbb N\cup \{0\}}\) become orthonormal bases of \(X^m(\Omega )\) and \(Y^m(\Omega )\) respectively. For the velocity field, we define a d-dimensional vector space \(\mathbb X^m(\Omega )\) by
We introduce series expansions of the elements in \(X^m(\Omega )\) and \(Y^m(\Omega )\). Let
for each \((n,q) \in \mathbb Z^{d-1} \times \mathbb N\) and \((n,q) \in \mathbb Z^{d-1} \times \mathbb N\cup \{ 0 \}\) respectively. Then for any \(f \in X^m(\Omega )\) and \(g \in Y^m(\Omega )\), we can write
We refer to [5, Lemma 3.1] for details.
We give two simple lemmas. The first one implies \(fg \in X^m\) when \(f \in X^m\) and \(g \in Y^m\), and the second one implies \(fg \in Y^m\) when \(f,g \in X^m\) or \(f,g \in Y^m\) for any given \(m \in \mathbb N\) with \(m > d/2\). Since the proofs are elementary, we omit them.
Lemma 2.2
Let \(q_e\) and \(q_o\) be even and odd number respectively. Then, there hold
Lemma 2.3
Let \(q-q'\) and \(q-q''\) be odd and even number respectively. Then, there hold
The next proposition provides convolution estimates similar to the Fourier expansion.
Proposition 2.4
Let \(f,f' \in X^m\) and \(g,g' \in Y^m\) for some \(m \in \mathbb N\) with \(m > \frac{d}{2}\). Then, there hold
Proof
We only show the first inequality because the others can be proved similarly. By the series expansions of \(f \in X^m(\Omega )\) and \(g \in Y^m(\Omega )\), it holds
By the use of Lemma 2.2, we can see for each \((n,q) \in \mathbb Z^{d-1} \times \mathbb N\) that
This estimate infers
This finishes the proof. \(\square \)
3 Spectral analysis
In this section, we give a different form of (1.2) via spectral analysis. Then, we provide temporal decay estimates for the linear operator of (1.2). From now on, we use the notations \(\tilde{n} := 2\pi n\) and \(\tilde{q} := \frac{\pi }{2} q\) for each \(n \in \mathbb Z^{d-1}\) and \(q \in \mathbb N\cup \{0\}\). We define two sets
We estimate the pressure term first. From the v equation in (1.2), we can see
Using the basis \(\mathscr {C}_{n,q}(x) = \mathscr {C}_{\eta }(x)\), we have for each \(\eta \in I \setminus \{0\}\) that
Since \(\nabla _h \mathscr {C}_{\eta } = i\tilde{n} \mathscr {C}_{\eta }\) and \(\partial _d \mathscr {C}_{\eta } = -\tilde{q} \mathscr {B}_{\eta }\), we can see
and
Thus, we obtain
where
and
From these formulas, we have
for \(\eta \in I \setminus \{0\}\), and
for \(\eta \in J\). Due to the linear structure of (3.2) and (3.3), we can observe a partially dissipative nature by writing the two equaitons at once with \(\textbf{u} := (v_d,\theta )^{T}\).
Let us define an operator \(\mathscr {F} : (L^2)^{d-1} \times L^2 \rightarrow \mathbb C^{d-1} \times \mathbb C\) by \(\mathscr {F} := (\mathscr {F}_c,\mathscr {F}_b)\). Then, it follows
where
For simplicity, we use the notation
Since the characteristic equation of \(M^{T}\) is given by
the two pair of eigenvalue and eigenvector \((\lambda _\pm (\eta ), \overline{\textbf{a}_\pm (\eta )})\) satisfy
where \(M^{T} \overline{\textbf{a}_\pm (\eta )} = \lambda (\eta )_\pm \overline{\textbf{a}_\pm (\eta )}\) holds. We note that there is no pair \(\eta \in J\) satisfying \(|\eta |^{4\alpha } - {4 |\tilde{n}|^2}/{|\eta |^2} = 0\). Since
satisfy \(BA = I\), it follows by Duhamel’s principle
However, using this formula directly can be problematic because of the unboundedness of \(|\textbf{b}_{\pm }|\) around the set \(\{ |\eta |^{4\alpha } = 4|\tilde{n}|^2/|\eta |^2 \}\). For this reason, we employ
which allows us to get rid of the singularity of \(\textbf{b}_{\pm }\). Here, we note some useful calculations when using (3.6). From the definition of \(\lambda _{\pm }\), \(\textbf{a}_{\pm }\), and \(\textbf{b}_{\pm }\), we have
Thus, it follows
Let us consider the three sets
with \(J = D_1 \cup D_2 \cup D_3\). Then, for any \(\textbf{f} \in \mathbb C^2\), there exists a constant \(C>0\) such that
For the first and second inequalities, we apply the mean value theorem so that
for any \(\eta \in D_1\) and
for any \(\eta \in D_2\). One can easily obtain the last inequality in (3.8) by the use of \(|\lambda _+ - \lambda _-| \ge \frac{1}{2}|\eta |^{2\alpha }\).
Now, we are ready to show temporal decay estimates of solutions to the linearized system of (3.4):
Lemma 3.1
Let \(d \in \mathbb {N}\) with \(d \ge 2\) and \(m \in \mathbb {N}\). Let \(\textbf{u}_0 \in X^m(\Omega )\). Then, there exists a unique smooth global smooth global solution \(\textbf{u}= (v_d,\theta )\) to (3.9) such that
and
for all \(s \in [0,m]\).
Proof
We recall
and prove (3.11) first. We can see
From (3.7) it is clear that
On the other hand, (3.8) gives
Since
and
it follows
Collecting the above estimates, we deduce (3.11).
It remains to show (3.10). We can see
Since \(\langle \textbf{b}_{-},e_1 \rangle = \frac{|\tilde{n}|^2}{|\lambda _+||\eta |^2}\langle \textbf{b}_{-},e_2 \rangle \), using \(\frac{|\tilde{n}|^2}{|\lambda _+||\eta |^2} \le C\) for \(\eta \in D_1 \cup D_2\) and \(\frac{|\tilde{n}|^2}{|\lambda _+||\eta |^2} \le 2 \frac{|\tilde{n}|^2}{|\eta |^{2+2\alpha }}\) for \(\eta \in D_3\), we have
Estimating as in (3.12), we deduce
from which we obtain (3.10). This completes the proof. \(\square \)
4 Energy estimates
In this section, we provide the energy estimates which specify the quantities that should be computed via the spectral analysis. We start with the following standard local existence result. For the proof, we refer to [5, 37].
Proposition 4.1
(Local well-posedness) Let \(d \in \mathbb {N}\) with \(d \ge 2\) and \(\alpha \in \{0,1\}\). Let \(m \in \mathbb N\) with \(m > 1+\frac{d}{2}-\alpha \) and an initial data \(\theta _0 \in X^m\) and \(v_0 \in \mathbb X^m\). Then there exists a \(T>0\) such that there exists a unique classical solution \((v, \theta )\) to the stratified Boussinesq equations (1.2) satisfying
Let \(T^* \in (0,\infty ]\) be the maximal time of existence. Moreover, if \(T^* < \infty \), then it holds
We will frequently use the following result on the product estimates
(see [21] for the proof).
Lemma 4.2
Let \(m \in \mathbb N\). Then for any subset \(D \subset \{ \gamma ; |\gamma | = m \}\), there exists a constant \(C=C(m)>0\) such that
for all \(f, g \in H^m(\Omega ) \cap L^{\infty }(\Omega )\). Moreover, if \(m > \frac{d}{2}\), then it holds
Let us use the notations for \(k \in \mathbb N\)
Note that Young’s inequality implies
Proposition 4.3
Let \(d \in \mathbb {N}\) with \(d \ge 2\) and \(m \in \mathbb {N}\) with \(m \ge 2+\frac{d}{2}\). Assume that \((v, \theta )\) is a smooth global solution to (1.2) with \(\alpha = 1\). Then there exists a constant \(C > 0\) such that
for all \(t>0\).
Proof
From the system (1.2), we have
We only consider the \(|\gamma | = m\) case because the others can be treated similarly. It is clear by the divergence-free condition and the boundary condition that
Thus, Lemma 4.2 implies
For estimating the remainder term
we use a simple formula \(\Delta v \cdot \nabla \theta = \Delta v_h \cdot \nabla _h \theta + \Delta v_d \partial _d \theta \). We can infer from Hölder’s inequality with Sobolev embeddings that
Since \(m \ge 2 + d/2\) implies
it follows
Otherwise, we have
Here, we need to estimate carefully with the boundary conditions. The first integral on the right-hand side is bounded by
Since \(v_d \in X^{m+1}(\Omega )\) and \(\theta \in X^m(\Omega )\) for a.e. \(t>0\), the boundary term vanishes. Lemma 4.2 implies
On the other hand, we write the second integral in (4.3) as
It can be shown by Hölder’s inequalities and Sobolev embeddings that
Since \(\partial _d v_h \partial _d \theta \in X^{m-1}\) and \(\theta \in X^m\), it holds
Then, we have
Combining the above estimates, we have
Now we claim that
We recall the \(v_d\) equation in (1.2)
We first take \(-\Delta \) on the both sides of (4.7). Since the definition of \(\mathbb {P}\) implies
and
it follows
Then, we have for \(|\gamma | = m-2\) that
On the other hand, we have from the \(\theta \) equation in (1.2)
Adding these two equalities gives
We have by Lemma 4.2 and the divergence-free condition that
From \(\partial _dv_d = -\nabla _h \cdot v_h\) and the cancellation property, we deduce
and
respectively. The above estimates yield
Similarly, we can repeat the above procedure for the lower order derivatives. Then, (4.6) is obtained.
We multiply (4.5) by 2
and recall (4.6)
Adding these two inequality, we obtain (4.2). This completes the proof. \(\square \)
Proposition 4.4
Let \(d \in \mathbb {N}\) with \(d \ge 2\) and \(m \in \mathbb {N}\) with \(m > 1+\frac{d}{2}\). Assume that \((v, \theta )\) is a smooth global solution to (1.2) with \(\alpha = 0\). Then there exists a constant \(C > 0\) such that
for all \(t>0\).
Proof
From (1.2), we can have
We only estimate \(|\gamma | = m\) case because the others can be treated similarly. The first integral on the right-hand side can be estimated by lemma 4.2 and the divergence-free condition that
To estimate the remainder term, we consider the case \(\partial ^{\gamma } \ne \partial _d^m\) first. As estimating the first one, we can see
In the case of \(\partial ^{\gamma } = \partial _d^m\), we have
We note that
By (4.4), we can deduce
thus,
Collecting the above estimates, we have
As estimating (4.6), we can show
We only consider the highest derivative case. We can see from the \(v_d\) equation in (1.2)
Thus, we have for \(|\gamma | = m-1\) that
From the \(\theta \) equation in (1.2), it follows
Combining the two above gives
The divergence-free condition and Lemma 4.2 imply
We also have with the divergence-free condition that
The cancellation property yields
Therefore, we deduce that
which implies (4.10).
Multiplying (4.9) by 2 and adding (4.10) gives (4.8). This completes the proof. \(\square \)
5 Global-in-time existence
In this section, we prove the global existence part of Theorem 1.1 and 1.4. It remains to estimate the key quantities in Proposition 4.3 and Proposition 4.4, namely,
respectively. For this purpose, we recall the notations introduced in Sect. 3. From now on, we use the notations
for \(f \in X^m(\Omega )\) and \(g \in Y^m(\Omega )\).
5.1 Proof of Theorem 1.1:Global-in-time existence part
Here, we fix \(\alpha = 1\). We show the two propositions that provide proper upper-bound of the key quantity. Then combining with Proposition 4.3, we finish the proof.
Proposition 5.1
Let \(d \in \mathbb {N}\) with \(d \ge 2\) and \(m \in \mathbb {N}\) satisfying \(m > 1+\frac{d}{2}\). Assume that \((v, \theta )\) is a smooth global solution to (1.2) with \(\alpha = 1\). Then there exists a constant \(C > 0\) such that
for all \(T > 0\).
Proof
We first note that the divergence-free condition implies
Thus, it suffices to show
We only estimate the first term on the left-hand side because the other can be treated similarly. Applying Duhamel’s principle to (3.1), we can have
where
We have used that \(|\mathscr {F} \mathbb {P} f| \le |\mathscr {F} f|\) for \(I_2\) and \(I_3\). It is clear that
We can see by Fubini’s theorem that
Similarly,
Using \((v \cdot \nabla )v_h = (v_h \cdot \nabla _h)v_h + v_d \partial _d v_h\) and Proposition 2.4, we have
In a similar way with the above, we can have the same upper-bound for \(I_3\). Collecting the estimates for \(I_1\), \(I_2\), \(I_3\), and \(I_4\), we obtain the claim. This completes the proof. \(\square \)
Proposition 5.2
Let \(d \in \mathbb {N}\) with \(d \ge 2\) and \(m \in \mathbb {N}\) satisfying \(m > 3+\frac{d}{2}\). Assume that \((v, \theta )\) is a smooth global solution to (1.2) with \(\alpha = 1\), and \(\int _{\Omega } v_0 \,\textrm{d}x\) be satisfied. Then there exists a constant \(C > 0\) such that
for all \(T>0\).
Proof
We recall (3.6) and have
where
We estimate \(I_6\) and \(I_8\) first. By (3.7) we have
With Fubini’s theorem, we also have
Due to (2.1) and (2.3), Poincaré inequality implies
Now, we estimate \(I_5\) and \(I_7\). To apply (3.8), we consider \(\eta \in D_2\) and \(\eta \in D_3\) separately. Note that \(D_1 = \emptyset \) when \(\alpha = 1\). In the former case, as the previous estimates, we have
and
In the latter case,
On the other hand, we similarly have
Let \(\tilde{n}' + \tilde{n}'' = \tilde{n}\) and \(\tilde{q}' + \tilde{q}'' = \tilde{q}\). Then, it holds
Similarly, for \(\tilde{n}' + \tilde{n}'' = \tilde{n}\) and \(|\tilde{q}' - \tilde{q}''| = \tilde{q}\), we can see
They imply
On the other hand, with \(\mathscr {F}_b(v \cdot \nabla )\theta = \mathscr {F}_b(v_h \cdot \nabla _h)\theta + \mathscr {F}_b[v_d \partial _d \theta ]\) it follows for \(m > 3+d/2\)
Since
and
we have
Collecting the estimates for \(I_5\), \(I_6\), \(I_7\), and \(I_8\), we complete the proof. \(\square \)
Now, we are ready to prove the global existence part of Theorem 1.1. Let \(T^*>0\) and \((v,\theta )\) be the maximal time of existence and the local solution given in Proposition 4.1 respectively. We define
Then, from (5.2) and (5.4), we have
for some \(C_1>0\). For a while, we assume that \(C_1B_m(T) \le \frac{1}{2}\) for all \(T \in (0,T^*)\). Then, we have
On the other hand, we recall (4.2) and integrate it on the interval [0, T]. Then by (4.1), we have
Since
it holds
for some \(C_2>0\). If we assume \(C_2 \Vert (v_0,\theta _0) \Vert _{H^m} \le \frac{1}{16}\) and \(C_2 B_m(T) \le \frac{1}{16}\), then
Here, we take \(\delta >0\) such that \(C_1 (2\delta ) < \frac{1}{2}\) and \(C_2 (2\delta ) < \frac{1}{16}\). By the above estimates, we can deduce that (5.6) holds for all \(T \in (0,T^*)\), hence, \(T^* = \infty \). Thus, (1.3) is obtained. This completes the proof.
5.2 Proof of Theorem 1.4:Global-in-time existence part
In this subsection, we fix \(\alpha = 0\). We only provide two propositions counterparts of Proposition 5.1 and 5.2, because the rest of the proof is similar with that of theorem 1.1.
Proposition 5.3
Let \(d \in \mathbb {N}\) with \(d \ge 2\) and \(m \in \mathbb {N}\) satisfying \(m > 2+\frac{d}{2}\). Assume that \((v, \theta )\) is a smooth global solution to (1.2) with \(\alpha = 0\). Then there exists a constant \(C > 0\) such that
for all \(T > 0\).
Proof
From (3.2) and Duhamel’s principle, we can have
where
We can easily show
and
Fubini’s theorem and Propostion 2.4 gives
Similarly, we can estimate \(J_3\) and have
From the estimates for \(J_1\), \(J_2\), \(J_3\), and \(J_4\), we deduce (5.7). This completes the proof. \(\square \)
Proposition 5.4
Let \(d \in \mathbb {N}\) with \(d \ge 2\) and \(m \in \mathbb {N}\) satisfying \(m > 2+\frac{d}{2}\). Assume that \((v, \theta )\) is a smooth global solution to (1.2) with \(\alpha = 1\). Then there exists a constant \(C > 0\) such that
for all \(T>0\).
Proof
We recall (3.6) and have
where
It is clear by (3.7)
Similarly, we have with Proposition 2.4 that
To estimate \(J_5\) and \(J_7\) with (3.8), we consider \(\eta \in D_1 \cup D_2\) and \(\eta \in D_3\) separately. We note that
and
On the other hand,
We can see
As estimating \(I_7\) on the set \(D_3\), we can deduce
and
for \(m > 2+d/2\). Collecting the estimates for \(J_5\), \(J_6\), \(J_7\), and \(J_8\), we obtain (5.8). This completes the proof. \(\square \)
6 Proof of temporal decay estimates
In this section, let \((v,\theta )\) be a smooth global-in-time solution to (1.2). In addition, we assume that (1.6) or (1.3) holds in each case with
for sufficiently small \(\delta >0\). The next three propositions are for the temporal decay estimates of \(\Vert \bar{\theta }(t) \Vert _{L^2}\), \(\Vert v(t) \Vert _{L^2}\), and \(\Vert v_d(t) \Vert _{L^2}\) in both cases \(\alpha =0\) and \(\alpha = 1\). After that, we prove (1.7) and (1.5) combining with the temporal decay estimates for \(\Vert v(t) \Vert _{{\dot{H}}^m}\) and \(\Vert v_d(t) \Vert _{{\dot{H}}^m}\).
Proposition 6.1
Let \(d \in \mathbb {N}\) with \(d \ge 2\) and \(\alpha \in \{0,1\}\). Let \(m \in \mathbb {N}\) with \(m > 1+\frac{d}{2} +\alpha \) and \((v, \theta )\) be a smooth global solution to (1.2) with (1.3) or (1.6). Suppose that (6.1) be satisfied. Then, there exists a constant \(C > 0\) such that
Proof
From the v equations in (1.2), we have
On the other hand, we have from (2.1) and the \(\theta \) equation in (1.2) that
where
Hence,
We can deduce from (3.2), (3.3), and (2.1) that
Combining the above, we have
To estimate the integral on the right-hand side, we note
We consider \(\alpha = 0\) case first. The right-hand side is bounded by
Using \(v_d = -(\Delta _h)(-\Delta )^{-1} v_d + \partial _d \nabla _h (-\Delta )^{-1} v_h\), we have
For \(\alpha = 1\), we can see
and
where \(v_d = -(\Delta _h)(-\Delta )^{-1} v_d + \partial _d \nabla _h (-\Delta )^{-1} v_h\) also used here. Hence, we can deduce
in both cases. Combining the above and using (6.1), we can have
Let \(M\ge 1\) which will be specified later. Since
and
it holds
Thus,
Taking \(M = 1+\frac{t}{8\frac{m}{1+\alpha }}\) and multiplying both terms by \(2M^{\frac{m}{1+\alpha }}\), we obtain by (6.3) that
Using Grönwall’s inequality, we obtain (6.2). This completes the proof. \(\square \)
Proposition 6.2
Let \(d \in \mathbb {N}\) with \(d \ge 2\) and \(\alpha \in \{0,1\}\). Let \(m \in \mathbb {N}\) with \(m > 1+\frac{d}{2} +\alpha \) and \((v, \theta )\) be a smooth global solution to (1.2) with (1.3) or (1.6). Suppose that (6.1) be satisfied. Then, there exists a constant \(C > 0\) such that
Proof
From the v equations in (1.2), we have
Since Duhamel’s principle implies
we have
On the other hand, from (3.2) and (3.3), we have
and
respectively. Thus,
Moreover, we can deduce from (3.2) and (3.3) that
Combining the above, we have
By \((v \cdot \nabla )\theta = (v_h \cdot \nabla _h) \theta + v_d \partial _d \theta \), we deduce
Thus, by \(W^{1,\infty }(\Omega ) \hookrightarrow H^{m-\alpha }(\Omega )\), (6.1), and Young’s inequality, we have
Let \(M\ge 1\) which will be specified later. Since
together with
we can have as estimating (6.4) that
Thus,
We take \(M = 1+\frac{t}{8(1+\frac{m}{1+\alpha })}\). Then, we can have with (6.6) that
We integrate it over time and use (6.7) with
Then, for
it holds
Applying Grönwall’s ineqaulity, we obtain
With (6.6), we deduce (6.5). This completes the proof. \(\square \)
Proposition 6.3
Let \(d \in \mathbb {N}\) with \(d \ge 2\) and \(\alpha \in \{0,1\}\). Let \(m \in \mathbb {N}\) with \(m > 2+\frac{d}{2} +\alpha \) and \((v, \theta )\) be a smooth global solution to (1.2) with (1.3) or (1.6). Suppose that (6.1) be satisfied. Then, there exists a constant \(C > 0\) such that
Proof
Recalling the definition of \(\textbf{b}_{\pm }\), we can verify that
We note that
Together with
we have
We show
where \(\delta \) is a small constant in (6.1). Then by taking \(\delta \) small enough, we obtain
We recall (3.5) and have
Since \(|e^{-\lambda _+ t}| \le e^{-|\eta |^{2\alpha } \frac{t}{2}}\) for \(\eta \in J\), it follows by the Minkowski inequality
From the simple fact \(|\textbf{a}_{+}|^2 = |\lambda _{+}|^2 + \frac{|\tilde{n}|^4}{|\eta |^4} \le C|\eta |^{4\alpha }\) with (6.1), we have
We note that
Thus, it holds
We have used
in the last inequality. We note by \(H^{m-2-\alpha } \hookrightarrow L^{\infty }\)
and
Combining (6.5), (6.2) and our assumptions, we can see
where \(\delta \) is a small constant in (6.1). Hence,
Collecting the above estimates, we obtain (6.10) and (6.11).
Now, we show
Since we have from (3.2) and (3.3),
and
respectively, it holds
We can infer from (3.2) and (3.3) that
Combining the above, we have
We estimate the last integral with \((v \cdot \nabla )\theta = (v_h \cdot \nabla _h) \theta + v_d \partial _d \theta \). Hölder’s inequality implies
where \(\frac{1}{p} + \frac{1}{q} = \frac{1}{2}\). We take \(\frac{1}{q} = \frac{1}{2} - \frac{\alpha }{d}\). Then for \(\epsilon \in (0,1)\) with \(m>2+ \frac{d}{2} + \alpha +2\epsilon \), we can see
and
together with \(\Vert R_h^4 \Lambda ^{-6\alpha } v_d \Vert _{L^q} + \Vert R_h^{3} \Lambda ^{-4\alpha }\theta \Vert _{L^q} \le C\Vert R_h \Lambda ^{-\alpha } v_d \Vert _{L^2} + C\Vert R_h^{3} \Lambda ^{-3\alpha }\theta \Vert _{L^2}\), we have
On the other hand, it holds
The second integral on the right-hand side is bounded by
We note
When \(\alpha = 0\), with the integration by parts, it holds
For \(\alpha = 1\), we have
Therefore,
With Young’s inequality and (6.1) we can have
Let \(M\ge 1\) which will be specified later. Since
and
we have
Thus,
We take \(M = 1+\frac{t}{8(2+\frac{m}{1+\alpha })}\) and multiply \(2M^{2+\frac{m}{1+\alpha }}\) both sides. Then,
Since the interpolation inequality implies
we have from (6.5)
By (6.11), it holds
Then, we deduce that
Since we can verify
using Grönwall’s inequality and (6.14), (6.13) is obtained. Then, (6.8) follows from (6.11). This completes the proof. \(\square \)
6.1 Proof of Theorem 1.4:Temporal decay part
In this section, we completes the proof of Theorem 1.4. For this purpose, we suppose (6.15) and (6.16) hold true, which will be proved in the following proposition. From (6.5) and (6.15), we obtain
On the other hand, (6.8) and (6.16) imply
Hence, it suffices to prove
because of (1.6). We recall (1.4) and use (6.2) to have
Since
by (6.5) and (6.8), we can deduce
This completes the proof.
Proposition 6.4
Let \(d \in \mathbb {N}\) with \(d \ge 2\) and \(\alpha = 0\). Let \(m \in \mathbb {N}\) with \(m > 2+\frac{d}{2}\) and \((v, \theta )\) be a smooth global solution to (1.2) with (1.6). Suppose that (6.1) be satisfied. Then, there exists a constant \(C > 0\) such that
and
Proof
From the v equations in (1.2), it follows
Using (5.3) gives
Then, applying Duhamel’s principle shows
thus,
Since (5.3) implies \(\Vert v_d (t) \Vert _{{\dot{H}}^m}^2 \le C\Vert R_h v(t) \Vert _{{\dot{H}}^m}^2\), we can similarly obtain
We omit the details.
Now, we show that
Since this implies
(6.15) and (6.16) follow by (6.17) and (6.18) respectively. Since \(H^m(\Omega )\) is a Banach algebra, we can show from (3.2) that
From (3.3),
Thus,
We note that
where
The integration by parts and \((v \cdot \nabla ) \theta = (v_h \cdot \nabla _h) \theta + v_d \partial _d \theta \) show
Again using the integration by parts with the continuous embedding \(L^{\infty }(\Omega ) \hookrightarrow H^{m-1}(\Omega )\), we obtain
On the other hand, due to the cancellation property, we can have
The divergence-free condition and the integration by parts imply
Thus,
From the above estimates, we deduce
On the other hand, using (3.2) and
we have
Since (3.3) yields
we have
We note that
where
We can see
and
Due to the cancellation property, we have
Collecting the above estimates gives
Thus, we arrived at
By Young’s inequality and (6.1), it follows
We consider \(M \ge 1\) which will be specified later. Since
and
it holds
Hence,
Here, we take \(M = 1+\frac{t}{16}\). Then multiplying the both sides by \(2M^2\) and using (6.18), we have
We integrate it over time and use (6.21) and
Then, for
it holds
By Grönwall’s ineqaulity, we obtain (6.19). This completes the proof. \(\square \)
6.2 Proof of Theorem 1.1:Temporal decay part
Now, we finish the proof of Theorem 1.1 assuming (6.22) and (6.23), which are given in Proposition 6.5. We also provide Proposition 6.6 for improved temporal estimates. From (6.5) and (6.22), we obtain
On the other hand, (6.8) and (6.23) imply
It suffices to prove
due to (1.3). Recalling (1.4), we can estimate from (6.2) that
Since
This completes the proof.
Proposition 6.5
Let \(d \in \mathbb {N}\) with \(d \ge 2\) and \(\alpha =1\). Let \(m \in \mathbb {N}\) with \(m > 3+\frac{d}{2}\) and \((v, \theta )\) be a smooth global solution to (1.2) with (1.3). Suppose that (6.1) be satisfied. Then, there exists a constant \(C > 0\) such that
and
Proof
From the v equations in (1.2), it follows
Using (5.3) gives
Then, applying Duhamel’s principle shows
Thus, from
it follows
Recalling from the v equations in (1.2) that
and using (6.1), we can see
By Duhamel’s principle,
Now, we show that
We can show from (3.2) that
From (3.3),
Thus,
Using integration by parts and Hölder’s inequality, we have
We note that
where
The integration by parts and \((v \cdot \nabla ) \theta = (v_h \cdot \nabla _h) \theta + v_d \partial _d \theta \) show
By the use of the integration by parts, we can estimate the second integral on the right-hand side as
Similarly, the first one is bounded by
Thus,
Due to the cancellation property, we have for \(|\gamma '|=1\) that
By integration by parts and the calculus inequality, it can be shown that
The first term on the right-hand side is bounded by
where \(\frac{1}{p} + \frac{1}{q} = \frac{1}{2}\) and \(\frac{1}{p} = \frac{2}{d} + (\frac{1}{2} - \frac{m-2}{d})\frac{2}{m-2}\).On the other hand, the integration by parts yields
Hence,
By the above estimates, we deduce
On the other hand, using (3.2) and (6.20), we have
Since (3.3) yields
we have
It is clear
Since \(H^{m-3}(\Omega )\) is banach algebra, we can see
Collecting the above estimates gives
Now, we arrived at
We have used Young’s inequality and (6.1) in the last inequality. We consider \(M \ge 1\) which will be specified later. Since
and
it holds
Hence,
Here, we take \(M = 1+\frac{t}{16}\). Then multiplying the both sides by \(2M^2\) and using (6.18), we have
From (6.25), it is clear
Using (6.5), (6.24) and the interpolation inequality gives
Therefore,
We integrate it over time and use (6.27) with
Then, for
it holds
By Grönwall’s ineqaulity, we obtain (6.26).
Now, we prove that
As showing (6.9), we can have
Thus, it suffices to show
We can see from (3.5) that
Due to \(|e^{-\lambda _+ t}| \le e^{-\frac{|\eta |^{2}}{2} t}\) for \(\eta \in J\), it follows by the Minkowski inequality
From the simple fact \(|\textbf{a}_{+}|^2 = |\lambda _{+}|^2 + \frac{|\tilde{n}|^4}{|\eta |^4} \le C|\eta |^{4}\) with (6.1), we have
By (6.12) we have
We note
and
Combining (6.5) and (6.22), we can see
Thus,
Collecting the above estimates, we obtain (6.29), which implies (6.28). This completes the proof. \(\square \)
Proposition 6.6
Let \(d \in \mathbb {N}\) with \(d \ge 2\) and \(\alpha =1\). Let \(m \in \mathbb {N}\) with \(m > 3+\frac{d}{2}\) and \((v, \theta )\) be a smooth global solution to (1.2) with (1.3). Suppose that (6.1) be satisfied. Then, for any \(\epsilon \in (0,1)\), there exists a constant \(C > 0\) such that
and
Proof
Since (6.8) implies \(\Vert \Lambda ^{-\epsilon } v_d(t) \Vert _{L^2} \le C t^{-(\frac{3}{4} + \frac{m}{4})}\), it suffices to show that
Applying Duhamel’s principle to (3.1) with (2.3), we obtain
for any \(\epsilon \in (0,1)\). We clearly have by (6.5) and (6.22) that
On the other hand, we can see
Thus, we have
We can infer from (6.5) and (6.8) that
Since this implies
combining the above estimates gives (6.30).
Using (2.3) and (2.1), we can infer from (3.1) and (3.2) that
for any \(\epsilon \in (0,1)\). We can see
Since
and
we have by (6.22) that
Collecting the above estimates gives (6.31). This completes the proof. \(\square \)
7 Sharpness of decay rates
In this section, we prove that the decay rates in Theorem 1.1 and 1.4 are sharp in the following sense. We recall the linearized system of (3.4):
where \(\textbf{u} = (v_d,\theta )^{T}\). The eigenvalues and eigenvectors of the linear operator are previously given by
and it holds
where
Note that If we consider \(\textbf{u}_0\) such that \(\mathscr {F}_b \textbf{u}_0 = 0\) for \(\eta \not \in D_3\), then \(|\textbf{a}_{\pm }||\textbf{b}_{\pm }| \le C\) for some \(C>0\) not depending on \(\eta \). Now, we are ready to provide the sharpness of the decay rates.
Proposition 7.1
Let \(m \in \mathbb N\). Then for any \(\epsilon > 0\), there exists an initial data \(\textbf{u}_0 \in X^m(\Omega )\) such that the solution \(\textbf{u}=(v_d,\theta )\) to (7.1) satisfies
and
for any \(s \in [0,m]\) and \(t \ge C\).
Proof
Let \(\epsilon > 0\) and
where \( \mathscr {F}_b \theta _0(\eta ) := |q|^{-(m+\frac{1}{2} + \epsilon )}\) for \(\eta \in D_3 \cap \{|n| = 1\}\), \(\mathscr {F}_b \theta _0(\eta ) = 0\) for \(\eta \not \in D_3 \cap \{|n| = 1\}\). For simplicity, we use the notation \(A:= D_3 \cap \{|n| = 1\}\). We show (7.3) first. From (7.2), we can see
Due to \(|e^{-\lambda _{+}t}| \le e^{-|\eta |^{2\alpha } \frac{t}{2}} \le e^{-\frac{t}{2}}\) it is clear that
On the other hand, we have
Thus,
for some \(C_1>0\). Let
Then, we can verify that there exists \(C_2 \ge C_1\) not depending on t such that \(f(\tau )\) is decreasing on the interval \((C_2 t^{\frac{1}{2(1+\alpha )}},\infty )\). Thus, it holds for \(t \ge 1\) that
By the change of variable \(\tilde{\tau } = \tau t^{-\frac{1}{2(1+\alpha )}}\), we can see
for some \(C>0\). Combining the above yields
Therefore, (7.3) is obtained.
The proof of (7.4) is similar with the previous one. By (7.2) and (7.5), it holds
Note that
Using that \(\frac{|\tilde{n}|^2}{|\eta |^{2(1+\alpha )}} \ge \frac{C}{|q|^{2(1+\alpha )}}\) for all \(\eta \in A\), repeating the above procedures, we obtain (7.4). This completes the proof. \(\square \)
Data Availability
Data sharing not applicable to this article as no datasets were generated or analysed during the current study.
References
Abidi, H., Hmidi, T.: On the global well-posedness for Boussinesq system. J. Differ. Equ. 233(1), 199–220 (2007)
Bianchini, R., Natalini, R.: Asymptotic behavior of 2D stably stratified fluids with a damping term in the velocity equation. ESAIM Control Optim. Calc. Var. 27(43), 16 (2021)
Bianchini, R., Crin-Barat, T., Paicu, M.: Relaxation approximation and asymptotic stability of stratified solutions to the IPM equation. arXiv:2210.02118
Cao, C., Jiahong, W.: Global regularity for the two-dimensional anisotropic Boussinesq equations with vertical dissipation. Arch. Ration. Mech. Anal. 208(3), 985–1004 (2013)
Castro, Á., Córdoba, D., Lear, D.: On the asymptotic stability of stratified solutions for the 2D Boussinesq equations with a velocity damping term. Math. Models Methods Appl. Sci. 29(7), 1227–1277 (2019)
Chae, D.: Global regularity for the 2D Boussinesq equations with partial viscosity terms. Adv. Math. 203(2), 497–513 (2006)
Chae, D., Nam, H.-S.: Local existence and blow-up criterion for the Boussinesq equations. Proc. R. Soc. Edinb. Sect. A 127(5), 935–946 (1997)
Chae, D., Kim, S.-K., Nam, H.-S.: Local existence and blow-up criterion of Hölder continuous solutions of the Boussinesq equations. Nagoya Math. J. 155, 55–80 (1999)
Chen, J., Hou, T.Y.: Stable nearly self-similar blowup of the 2D Boussinesq and 3D Euler equations with smooth data. arXiv:2210.07191
Constantin, P., Doering, C.R.: Heat transfer in convective turbulence. Nonlinearity 9(4), 1049–1060 (1996)
Danchin, R.: Remarks on the lifespan of the solutions to some models of incompressible fluid mechanics. Proc. Am. Math. Soc. 141(6), 1979–1993 (2013)
Desjardins, B., Lannes, D., Saut, J.-C.: Normal mode decomposition and dispersive and nonlinear mixing in stratified fluids. Water Waves 3(1), 153–192 (2021)
Doering, C.R., Wu, J., Zhao, K., Zheng, X.: Long time behavior of the two-dimensional Boussinesq equations without buoyancy diffusion. Phys. D 376(377), 144–159 (2018)
Dong, L.: On stability of Boussinesq equations without thermal conduction. Z. Angew. Math. Phys. 72, 128 (2021)
Dong, L.: On Asymptotic Stability of the 3D Boussinesq equations with a velocity damping term. J. Math. Fluid Mech. 24(23), 25 (2022)
Dong, L., Sun, Y.: Asymptotic stability of the 2D Boussinesq equations without thermal conduction. J. Differ. Equ. 337, 507–540 (2022)
Dong, L., Sun, Y.: On asymptotic stability of the 3D Boussinesq equations without thermal conduction. arXiv:2107.10082
Weinan, E., Shu, C.-W.: Small-scale structures in Boussinesq convection. Phys. Fluids 6(1), 49–58 (1994)
Elgindi, T.M.: On the asymptotic stability of stationary solutions of the inviscid incompressible porous medium equation. Arch. Ration. Mech. Anal. 225(2), 573–599 (2017)
Elgindi, T.M., Widmayer, K.: Sharp decay estimates for an anisotropic linear semigroup and applications to the surface quasi-geostrophic and inviscid Boussinesq systems. SIAM J. Math. Anal. 47(6), 4672–4684 (2015)
Ferrari, A.B.: On the blow-up of solutions of the 3-D Euler equations in a bounded domain. Commun. Math. Phys. 155(2), 277–294 (1993)
Getling, A. V.: Rayleigh-Bénard convection, Structures and dynamics. Advanced Series in Nonlinear Dynamics, 11. World Scientific Publishing Co., Inc., River Edge (1998)
Hou, T.Y., Li, C.: Global well-posedness of the viscous Boussinesq equations. Discrete Contin. Dyn. Syst. 12(1), 1–12 (2005)
Hu, Weiwei, Kukavica, I., Ziane, M.: On the regularity for the Boussinesq equations in a bounded domain. J. Math. Phys. 54(8), 081507 (2013)
Jo, M.J., Kim, J.: Quantitative asymptotic stability of the quasi-linearly stratified densities in the IPM equation on the three fundamental domains. arXiv:2210.11437
Jo, M.J., Kim, J., Lee, J.: The quasi-geostrophic approximation for the rotating stratified Boussinesq equations. arXiv:2209.02634
Ning, J.: Global regularity and long-time behavior of the solutions to the 2D Boussinesq equations without diffusivity in a bounded domain. J. Math. Fluid Mech. 19(1), 105–121 (2017)
Kim, J., Lee, J.: Stratified Boussinesq equations with a velocity damping term. Nonlinearity 35(6), 3059–3094 (2022)
Kiselev, A., Park, J., Yao, Y.: Small scale formation for the 2D Boussinesq equation. arXiv:2211.05070
Kukavica, I., Wang, W.: Long time behavior of solutions to the 2D Boussinesq equations with zero diffusivity. J. Dynam. Differ. Equ. 32(4), 2061–2077 (2020)
Lai, M.-J., Pan, R., Zhao, K.: Initial boundary value problem for two-dimensional viscous Boussinesq equations. Arch. Ration. Mech. Anal. 199(3), 739–760 (2011)
Lai, S., Jiahong, W., Zhong, Y.: Stability and large-time behavior of the 2D Boussinesq equations with partial dissipation. J. Differ. Equ. 271, 764–796 (2021)
Larios, A., Lunasin, E., Titi, E.S.: Global well-posedness for the 2D Boussinesq system with anisotropic viscosity and without heat diffusion. J. Differ. Equ. 255(9), 2636–2654 (2013)
Lee, S., Takada, R.: Dispersive estimates for the stably stratified Boussinesq equations. Indiana Univ. Math. J. 66(6), 2037–2070 (2017)
Liu, X., Wang, M., Zhang, Z.: Local well-posedness and blowup criterion of the Boussinesq equations in critical Besov spaces. J. Math. Fluid Mech. 12(2), 280–292 (2010)
Majda, A.: Introduction to PDEs and Waves for the Atmosphere and Ocean, Courant Lecture Notes in Mathematics, 9. New York; American Mathematical Society, Providence, RI, New York University, Courant Institute of Mathematical Sciences (2003)
Majda, A.J., Bertozzi, A.L.: Vorticity and Incompressible Flow. Cambridge Texts in Applied Mathematics, 27. Cambridge University Press, Cambridge (2002)
Ma, T., Wang, S.: Dynamic bifurcation and stability in the Rayleigh-Bénard convection. Commun. Math. Sci. 2(2), 159–183 (2004)
Pedlosky, J.: Geophysical Fluid Dynamics. Springer-Verlag, New York (1987)
Tao, L., Jiahong, W., Zhao, K., Zheng, X.: Stability near hydrostatic equilibrium to the 2D Boussinesq equations without thermal diffusion. Arch. Ration. Mech. Anal. 237(2), 585–630 (2020)
Takada, R.: Strongly stratified limit for the 3D inviscid Boussinesq equations. Arch. Ration. Mech. Anal. 232(3), 1475–1503 (2019)
Takada, R.: Long time solutions for the 2D inviscid Boussinesq equations with strong stratification. Manuscripta Math. 164(1–2), 223–250 (2021)
Wan, R.: Global well-posedness for the 2D Boussinesq equations with a velocity damping term. Discrete Contin. Dyn. Syst. 39(5), 2709–2730 (2019)
Zhai, Xiaoping: On some large solutions to the damped Boussinesq system. Appl. Math. Lett. 111, Paper No. 106621, MR4126358 (2021)
Widmayer, K.: Convergence to stratified flow for an inviscid 3D Boussinesq system. Commun. Math. Sci. 16(6), 1713–1728 (2018)
Acknowledgements
J. Jang’s research is supported in part by the NSF DMS-grant 2009458. J. Kim is supported by a KIAS Individual Grant (MG086501) at Korea Institute for Advanced Study.
Author information
Authors and Affiliations
Corresponding author
Ethics declarations
Competing Interests
The authors have no competing interests to declare that are relevant to the content of this article.
Additional information
Communicated by F.-H. Lin.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Rights and permissions
Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law.
About this article
Cite this article
Jang, J., Kim, J. Asymptotic stability and sharp decay rates to the linearly stratified Boussinesq equations in horizontally periodic strip domain. Calc. Var. 62, 141 (2023). https://doi.org/10.1007/s00526-023-02474-x
Received:
Accepted:
Published:
DOI: https://doi.org/10.1007/s00526-023-02474-x