Stability threshold of Couette flow for Boussinesq equations in ℝ 2 \mathbb{R}^{2}
Yubo Chen, Wendong Wang, Guoxu Yang · 2025
Read Stability threshold of Couette flow for Boussinesq equations in ℝ 2 \mathbb{R}^{2} by Yubo Chen, Wendong Wang, Guoxu Yang (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2508-11908.txt.
Introduction
Yubo Chen and co-authors settle a stability question for a classic flow in a 2025 mathematics paper. The setup pairs Couette flow, fluid sliding in straight layers, with the Boussinesq equations, which couple motion to temperature.
Earlier proofs covered a strip that repeats sideways; this paper handles the whole flat plane, where sideways waves can stretch without limit. For starts that are small in a precise sense and calm at low sideways frequencies, the flow settles back down rather than churning into turbulence. The team pins down how small is small enough and builds a new multiplier tool to tame the temperature equation's tricky echoes. The result extends a hard stability theory to open space.
Catalog record
| CREATOR | Yubo Chen, Wendong Wang, Guoxu Yang |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 229 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2508-11908
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Stability threshold of Couette flow for Boussinesq equations in ℝ 2 \mathbb{R}^{2} Yubo Chen, Wendong Wang, Guoxu Yang Abstract . This paper establishes the asymptotic stability threshold for the Couette flow ( y , 0 ) (y,0) under the 2D Boussinesq system in ℝ 2 \mathbb{R}^{2} . It was proved that for initial perturbations in Sobolev spaces with controlled low horizontal frequencies, the stability threshold is at most { 1 3 + , 2 3 + } \left\{\frac{1}{3}+,\frac{2}{3}+\right\} , extending the known threshold results from the periodic case 𝕋 x × ℝ y \mathbb{T}_{x}\times\mathbb{R}_{y} to the whole space. The core innovations are twofold: First, the ⟨ D x − 1 ⟩ \langle D_{x}^{-1}\rangle control on the initial data simultaneously resolves horizontal frequency singularities and optimizes integral indices when applying Young’s convolution inequality. Second, we develop a modified multiplier ℳ 3 \mathcal{M}_{3} that effectively absorbs the | D x | 1 / 3 |D_{x}|^{1/3} derivative structure induced by the temperature equation while handling nonlinear echo cascades. 1. Introduction In this paper, we investigate the nonlinear stability of the Couette flow (y,0) for the two-dimensional Boussinesq equations in \mathbb{R}^{2} : \displaystyle\left\{\begin{array}[]{l}\partial_{t}v-\nu\Delta v+v\cdot\nabla v+\nabla p=\vartheta e_{2},\\ \partial_{t}\vartheta-\mu\Delta\vartheta+v\cdot\nabla\vartheta=0,\\ \operatorname{div}v=0,\\ \left.v\right|_{t=0}=v_{\text{in}}(x,y),\quad\left.\vartheta\right|_{t=0}=\vartheta_{\text{in}}(x,y),\end{array}\right. where v(t,x,y)=\left(v^{1},v^{2}\right) denotes the velocity field, \vartheta the temperature and p the pressure. Moreover, e_{2}=(0,1)^{T} , \nu is the viscosity coefficient and \mu is the thermal diffusivity. The Boussinesq approximation simplifies fluid dynamics by assuming that density variations are negligible except in the buoyancy term (typically due to temperature or concentration gradients). The Boussinesq approximation and Boussinesq equations provide a balanced approach to modeling buoyancy-driven flows by simplifying density variations while retaining essential physics. They are fundamental in studying thermal convection, environmental flows, and heat transfer where gravitational effects dominate; see [ 63 , 28 ] . Consider the perturbation of the system ( 1 ) under the Couette flow (y,0) and the constant temperature, and letting u=v-(y,0) and \theta=\vartheta-a , where a\in\mathbb{R} , we have \left\{\begin{array}[]{l}\partial_{t}u-\nu\Delta u+y\partial_{x}u+\left(u^{2},\,0\right)+u\cdot\nabla u+\nabla p=(\theta+a)e_{2},\\ \partial_{t}\theta-\mu\Delta\theta+y\partial_{x}\theta+u\cdot\nabla\theta=0,\\ \operatorname{div}u=0,\\ \left.u\right|_{t=0}=u_{\text{in}}(x,y),\quad\left.\theta\right|_{t=0}=\theta_{\text{in}}(x,y).\end{array}\right. Define \omega=\partial_{y}u^{1}-\partial_{x}u^{2} be the vorticity, and it follows that \displaystyle\left\{\begin{array}[]{l}\partial_{t}\omega-\nu\Delta\omega+y\partial_{x}\omega+u\cdot\nabla\omega=\partial_{x}\theta,\\ \partial_{t}\theta-\mu\Delta\theta+y\partial_{x}\theta+u\cdot\nabla\theta=0,\\ u=\nabla^{\perp}(-\Delta)^{-1}\omega=\left(-\partial_{y},\partial_{x}\right)(-\Delta)^{-1}\omega,\\ \left.\omega\right|_{t=0}=\omega_{\mathrm{in}}(x,y),\quad\left.\theta\right|_{t=0}=\theta_{\mathrm{in}}(x,y).\end{array}\right. (1.5) When \vartheta=0 , ( 1 ) is reduced to the famous Navier-Stokes equations. The transition threshold problem was formulated by Bedrossian, Germain and Masmoudi [ 4 ] as follows: Given a norm \|\cdot\|_{X} , determine a \beta=\beta(X) so that \displaystyle\left\|u_{0}\right\|_{X}\ll Re^{-\gamma}\Rightarrow\text{ stability, } \displaystyle\left\|u_{0}\right\|_{X}\gg Re^{-\gamma}\Rightarrow\text{ instability. } The exponent \gamma is referred to as the transition threshold. Similarly, for the Boussinesq system, it is interesting to consider the following stability threshold problem: Given norms \|\cdot\|_{Y_{1}} and \|\cdot\|_{Y_{2}} , find \alpha=\alpha\left(Y_{1},Y_{2}\right) and \beta=\beta\left(Y_{1},Y_{2}\right) such that \displaystyle\left\|u_{\mathrm{in}}\right\|_{Y_{1}}\leq\nu^{\alpha}\text{ and }\left\|\theta_{\mathrm{in}}\right\|_{Y_{2}}\leq\nu^{\beta}\Rightarrow\text{ stability, } \displaystyle\left\|u_{\mathrm{in}}\right\|_{Y_{1}}\gg\nu^{\alpha}\text{ or }\left\|\theta_{\mathrm{in}}\right\|_{Y_{2}}\gg\nu^{\beta}\Rightarrow\text{ instability. } For \mathbb{T}\times\mathbb{R} , Deng–Wu–Zhang [ 28 ] proved the asymptotic stability of the steady state if the initial perturbations (u_{\mathrm{in}},\theta_{\mathrm{in}}) satisfy \left\|u_{\mathrm{in}}\right\|_{H^{2}}\leq\epsilon_{0}\nu^{\frac{1}{2}},\quad\left\|\theta_{\mathrm{in}}\right\|_{H^{1}}+\left\|\left|D_{x}\right|^{\frac{1}{6}}\theta_{\mathrm{in}}\right\|_{H^{1}}\leq\epsilon_{1}\nu^{\frac{11}{12}}. (1.6) The threshold obtained was improved by Zhang–Zi [ 63 ] to \left\|u_{\text{in}}\right\|_{H^{s+1}}\leq\epsilon_{0}\nu^{\frac{1}{3}},\quad\left\|\theta_{\text{in}}\right\|_{H^{s}}+\nu^{\frac{1}{6}}\left\|\left|D_{x}\right|^{\frac{1}{3}}\theta_{\text{in}}\right\|_{H^{s}}\leq\epsilon_{1}\nu^{\frac{5}{6}}, where s>7 . Later, the threshold was improved again by Niu–Zhao [ 46 ] to \left\|u_{\mathrm{in}}\right\|_{H^{s+1}}\leq\epsilon_{0}\nu^{\frac{1}{3}},\quad\left\|\left\langle\partial_{x}\right\rangle\theta_{\mathrm{in}}\right\|_{H^{s}}\leq\epsilon_{0}\nu^{\frac{2}{3}}, where s>5 . For \mathbb{T}\times\mathbb{I} with non-slip boundary data, Masmoudi–Zhai–Zhao [ 45 ] obtained the same threshold as ( 1.6 ). For \mathbb{T}\times\mathbb{I} with Navier-slip boundary data, the threshold was obtained by Wang–Yang [ 54 ] as follows: \left\|u_{\text{in}}\right\|_{H^{4}}+\nu^{-\frac{7}{12}}\left\|\theta_{\text{in}}\right\|_{H^{3}}+\nu^{-\frac{1}{3}}\left\|\partial_{x}\theta_{\text{in}}\right\|_{H^{3}}\leq\epsilon_{0}\nu^{\frac{1}{3}}. In the whole space \mathbb{R}\times\mathbb{R} , Wang–Wang [ 52 ] proved the asymptotic stability of the steady state if the initial perturbations (u_{\mathrm{in}},\theta_{\mathrm{in}}) satisfy \left\|\omega_{\mathrm{in}}\right\|_{H^{1}\cap L^{1}}\leq\epsilon_{0}\nu^{\frac{3}{4}},\quad\left\|\theta_{\mathrm{in}}\right\|_{H^{1}\cap L^{1}}\leq\epsilon_{1}\nu^{\frac{5}{4}}. Recently, for the Navier-Stokes equations on \mathbb{R}\times\mathbb{R} , Arbon–Bedrossian [ 1 ] proved \displaystyle\sum_{0\leq j\leq 1}\|\langle\partial_{x}\rangle^{m}\langle\frac{\partial_{x}}{\nu}\rangle^{-\frac{j}{3}}\partial_{y}^{j}w_{in}\|_{L^{2}_{x,y}}+\|w_{in,k}\|_{L^{\infty}_{k}L_{y}^{2}}\leq\delta_{1}\frac{\nu^{\frac{1}{2}}}{1+\ln(\frac{1}{\nu})^{1/2}}, where m\in(1/2,1) , \langle\cdot\rangle:=\sqrt{1+(\cdot)^{2}} and \delta_{1}>0 is a constant. Then, the result was improved in [ 38 ] by Li–Liu–Zhao \left\|\left\langle D_{x},D_{y}\right\rangle^{6}\left\langle\frac{1}{D_{x}}\right\rangle^{4}\omega_{\mathrm{in}}\right\|_{L^{2}}+\left\|\left\langle D_{x},D_{y}\right\rangle^{5}\left\langle\frac{1}{D_{x}}\right\rangle^{4}\omega_{\mathrm{in}}\right\|_{L^{1}}\leq\nu^{\frac{1}{3}+\delta}, where \delta>0 is small, D=\frac{1}{i}\partial and \langle D_{x},D_{y}\rangle:=\sqrt{1+D_{x}^{2}+D_{y}^{2}} . For the 3D Boussinesq equations via the Couette flow, we refer to [ 17 , 18 , 23 ] and the reference therein. Motivated by the progress in [ 38 ] , it is interesting whether the threshold holds for the Boussinesq system ( 1.5 ). Our main result is follows: Theorem 1.1 . Suppose that \nu=\mu . For arbitrarily small 0<\delta<1 , \frac{1-\delta}{2}<\epsilon<\frac{1}{2} , there exists 0<c,\nu_{0}<1 such that for all 0<\nu\leq\nu_{0} , if the initial data \omega_{\mathrm{in}} and \theta_{\mathrm{in}} satisfy \displaystyle \displaystyle\|\langle D_{x},D_{y}\rangle^{6}\langle\frac{1}{D_{x}}\rangle^{4}\omega_{\mathrm{in}}\|_{L^{2}\cap L^{1}}\leq\nu^{\frac{1}{3}+\delta}, (1.7) \displaystyle\|\langle D_{x},D_{y}\rangle^{6}\langle\frac{1}{D_{x}}\rangle^{4}\langle D_{x}\rangle\theta_{\mathrm{in}}\|_{L^{2}\cap L^{1}}\leq\nu^{\frac{2}{3}+2\delta}, then ( 1.5 ) has a pair of unique global solution (\omega(t),\,\theta(t)) with the following stability estimate: \displaystyle\|e^{c\nu^{\frac{1}{3}}\lambda(D_{x})t}\langle D_{x}\rangle \displaystyle\langle\tfrac{1}{D_{x}}\rangle^{\epsilon}\omega\|_{L^{\infty}_{[0,+\infty)}L^{2}} \displaystyle+\nu^{-\frac{1}{3}-\delta}\|e^{c\nu^{\frac{1}{3}}\lambda(D_{x})t}\langle D_{x}\rangle\langle\tfrac{1}{D_{x}}\rangle^{\epsilon}\langle D_{x}\rangle^{\frac{1}{3}}\theta\|_{L^{\infty}_{[0,+\infty)}L^{2}}\leq C\nu^{\frac{1}{3}+\delta}, where \lambda(k)=\min\{1,|k|^{\frac{2}{3}}\} . Remark 1.2 . When the temperature \theta vanishes, the system reduces to the Navier-Stokes equations, as studied in Arbon–Bedrossian [ 1 ] and Li–Liu–Zhao [ 38 ] . The above theorem extends their results to the Boussinesq setting, achieving the same stability threshold. Compared with [ 38 ] , the main difficulty arises from the nonlinear term u\cdot\nabla\theta when certain derivatives are propagated onto u . In particular, the multiplier \mathcal{M}_{3} in [ 38 ] appears insufficient for estimating |D_{x}|^{1/3}u^{\mathrm{NL}}\cdot\nabla\theta^{\mathrm{L}} (see Step VI in Proposition 5.1 ). To address this, we construct a new multiplier \mathcal{M}_{3} in ( 2.1 ), which is employed to handle the term I_{4} in Step VI: \begin{gathered}I_{4}:=\left|\int_{\mathbb{R}^{4}}\mathcal{M}(t,k,\xi)e^{2c\nu^{\frac{1}{3}}\lambda(k)t}\langle k\rangle^{2}\langle\frac{1}{k}\rangle^{2\epsilon}|k|^{\frac{1}{3}}\hat{\theta}_{k}^{\mathrm{NL}}(\xi)|l|^{\frac{1}{3}}\frac{tl(k-l)}{|l|^{2}+|\eta|^{2}}\hat{\omega}_{l}^{\mathrm{NL}}(\eta)\hat{\theta}_{k-l}^{\mathrm{L}}(\xi-\eta)dkdld\xi d\eta\right|.\end{gathered} When |k-l|<|k| , this new \mathcal{M}_{3} successfully absorbs the additional factor |l|^{1/3} due to the fact \langle 1/l\rangle^{-2/3}{|l|^{-2/3}}\lesssim 1 . Remark 1.3 . It is worth mentioning that it remains challenging to improve the transition threshold of \{\omega,\theta\} from \{\frac{1}{3}+\delta,\,\frac{2}{3}+2\delta\} to \{\frac{1}{3},\,\frac{2}{3}\} under Sobolev perturbations. The latter seems to be the best known result for the Navier-Stokes equations in \Omega=\mathbb{T}\times\mathbb{R} , as established by Masmoudi–Zhao [ 44 ] and Wei–Zhang [ 56 ] by different methods. The main difficulty lies in the singular behavior of the frequency variables (l,\eta) as |l|+|\eta|\to 0 . For instance, when dealing with the term u^{\mathrm{NL}}\cdot\nabla\theta^{\mathrm{L}} , it is difficult to use the inviscid damping to control L^{2} -norm of u^{\mathrm{NL}} . The additional \delta in the threshold, however, allows this estimate to succeed (see ( 4.12 ) in Lemma 4.2 for details). Remark 1.4 . For more precise estimates, we apply a standard quasi-linearization method to handle the difficulty arising from the buoyancy term \partial_{x}\theta . This approach is natural when making energy estimates L^{p} . The estimates of the equations for \{\omega,\theta\} are divided into linear and non-linear parts. Estimates for linear parts are determined by the size of the initial data and even the required regularity. For linear parts, there are many splendorous ideas introduced by Kelvin [ 34 ] for the incompressible Navier-Stokes equation in whole space. However, if it is coupled with the temperature equation, it seems puzzling. Here, a new observation is we calculate the Kelvin solution of the coupled system as in [ 28 ] and make some energy estimates in the sense of the Fourier transform, which will be used in the nonlinear parts. Besides, the difficulty of the nonlinear term comes from the interaction between linear and nonlinear terms. Inspired by the tricks in [ 56 ] , the calculations of the nonlinear parts will be separated into short time scale t\leq\nu^{-\frac{1}{6}} and long time scale t\geq\nu^{-\frac{1}{6}} . In the short time interval, we do not need to propagate the derivative of \partial_{x}\theta since the time integral can be handled. During the long time interval, we need to propagate some derivatives to deal with the buoyancy term. Due to the pointwise inviscid