Nonlinear stability of 2-D Couette flow for the compressible Navier-Stokes equations at high Reynolds number

Minling Li, Chao Wang, Zhifei Zhang · 2025

Read Nonlinear stability of 2-D Couette flow for the compressible Navier-Stokes equations at high Reynolds number by Minling Li, Chao Wang, Zhifei Zhang (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-2508-07291.txt.

Introduction

Minling Li, Chao Wang, and Zhifei Zhang prove that a simple shear flow stays stable even when a fluid barely resists motion, in a paper published in two thousand twenty-five. The flow is Couette flow, where fluid layers slide past each other at a steady rate.

They work with the compressible Navier-Stokes equations in two dimensions at high Reynolds number. If the starting state sits close enough to the pure shear, with the gap shrinking as the Reynolds number grows, the flow exists for all time and stays close. Formal estimates suggest this closeness demand cannot be eased. The proof uses a Fourier multiplier method and new variables that untangle the system. It captures faster decay from mixing while taming the lift-up effect that can stretch disturbances.

Catalog record

CREATORMinling Li, Chao Wang, Zhifei Zhang
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE241 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2508-07291

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Nonlinear stability of 2-D Couette flow for the compressible Navier-Stokes equations at high Reynolds number Minling Li, Chao Wang, Zhifei Zhang Abstract In this paper, we investigate the nonlinear stability of the Couette flow for the two-dimensional compressible Navier–Stokes equations at high Reynolds numbers( R ​ e Re ) regime. It was proved that if the initial data ( ρ i ​ n , 𝒖 i ​ n ) (\rho_{in},\bm{u}_{in}) satisfies ‖ ( ρ i ​ n , 𝒖 i ​ n ) − ( 1 , y , 0 ) ‖ H 4 ​ ( 𝕋 × ℝ ) ≤ ϵ ​ R ​ e − 1 \|(\rho_{in},\bm{u}_{in})-(1,y,0)\|_{H^{4}({\mathbb{T}}\times{\mathbb{R}})}\leq\epsilon Re^{-1} for some small ϵ \epsilon independent of R ​ e Re , then the corresponding solution exists globally and remains close to the Couette flow for all time. Formal asymptotics indicate that this stability threshold is sharp within the class of Sobolev perturbations. The proof relies on the Fourier-multiplier method and exploits three essential ingredients: (i) the introduction of “good unknowns” that decouple the perturbation system; (ii) the construction of a carefully designed Fourier multiplier that simultaneously captures the enhanced dissipation and inviscid-damping effects while taming the lift-up mechanism; and (iii) the design of distinct energy functionals for the incompressible and compressible modes. 1 Introduction In this paper, we consider the two-dimensional isentropic compressible Navier-Stokes equations in a domain {\mathbb{T}}\times{\mathbb{R}} : \displaystyle\left\{\begin{array}[]{l}\partial_{t}\rho+\mathop{\rm div}\nolimits_{\bm{x}}(\rho\bm{u})=0,\\ \partial_{t}(\rho\bm{u})+\mathop{\rm div}\nolimits_{\bm{x}}(\rho\bm{u}\otimes\bm{u})+\frac{1}{M}\nabla_{\bm{x}}\mathcal{P}(\rho)=\mu\Delta_{\bm{x}}\bm{u}+(\mu+\mu^{\prime})\nabla_{\bm{x}}\mathop{\rm div}\nolimits_{\bm{x}}\bm{u},\end{array}\right. where \rho(t,\bm{x}) , \bm{u}(t,\bm{x})=(u^{1},u^{2})^{T}(t,\bm{x}) and \mathcal{P}(\rho) represent the density, the velocity and the pressure, respectively. The pressure \mathcal{P}(\rho) here is assumed to be a smooth function in a neighborhood of 1 with \mathcal{P}^{\prime}(1)>0 . Here, M>0 is the Mach number, \mu and \mu^{\prime} are the shear viscosity and the bulk viscosity coefficients, respectively, satisfying the following physical condition \displaystyle\mu>0,\quad\mu+\mu^{\prime}\geq 0. Without loss of generality, we assume that \mathcal{P}^{\prime}(1)=1 and M=1 in this paper. The initial data of ( 1 ) is given by \displaystyle(\rho,\bm{u})|_{t=0}=(\rho_{in},\bm{u}_{in})(\bm{x}). In this paper, we study the nonlinear stability of the Couette flow given by \displaystyle(\bar{\rho},\bar{\bm{u}})(t,\bm{x})=(1,(y,0)^{T}). Thus, we introduce (n,\bm{v}) to be the perturbation of (\rho,\bm{u}) around (\bar{\rho},\bar{\bm{u}}) , i.e., \displaystyle n\triangleq\rho-\bar{\rho},\quad\bm{v}\triangleq\bm{u}-\bar{\bm{u}}, (1.3) which satisfies the following perturbation system \displaystyle\left\{\begin{array}[]{l}\partial_{t}n+y\partial_{x}n+\mathop{\rm div}\nolimits_{\bm{x}}\bm{v}=F_{1},\\[4.30554pt] \partial_{t}\bm{v}+y\partial_{x}\bm{v}+(v^{2},0)^{T}+\nabla_{\bm{x}}n-\mu\Delta_{\bm{x}}\bm{v}-(\mu+\mu^{\prime})\nabla_{\bm{x}}\mathop{\rm div}\nolimits_{\bm{x}}\bm{v}=\bm{F}_{2},\\[4.30554pt] (n,\bm{v})|_{t=0}=(n_{in},\bm{v}_{in})(\bm{x})=(\rho_{in}-\bar{\rho},\bm{u}_{in}-\bar{\bm{u}}),\end{array}\right. where \displaystyle F_{1}\triangleq \displaystyle-\bm{v}\cdot\nabla_{\bm{x}}n-n\mathop{\rm div}\nolimits_{\bm{x}}\bm{v}, \displaystyle\bm{F}_{2}\triangleq \displaystyle-\bm{v}\cdot\nabla_{\bm{x}}\bm{v}-\left(\frac{\mathcal{P}^{\prime}(1+n)}{1+n}-1\right)\nabla_{\bm{x}}n-\frac{n}{1+n}\left(\mu\Delta_{\bm{x}}\bm{v}+(\mu+\mu^{\prime})\nabla_{\bm{x}}\mathop{\rm div}\nolimits_{\bm{x}}\bm{v}\right). (1.7) 1.1 Background and known results The stability of plane Couette flow has been investigated since the seminal works of Rayleigh [ 37 ] and Kelvin [ 27 ] . Romanov [ 15 , 39 ] proved that Couette flow is spectrally stable for all Reynolds numbers, a result that appears to contradict both numerical simulations and physical experiments [ 10 , 20 , 40 , 41 , 42 , 46 ] , which indicate that shear flows become unstable and undergo transition to turbulence at sufficiently high Reynolds numbers. This discrepancy is known as the Sommerfeld paradox. To elucidate the mechanisms underlying transition to turbulence, Trefethen et al. [ 42 ] first formulated the transition threshold problem: namely, to quantify the magnitude of perturbations required to trigger instability and to determine their scaling with the Reynolds number. More recently, Bedrossian, Germain, and Masmoudi [ 5 , 6 ] provided a rigorous mathematical framework for this problem, stated as follows. Given a norm \|\cdot\|_{X} , find a \beta=\beta(X) so that \displaystyle\|u_{0}\|_{X}\leq Re^{-\beta}\to\textrm{stability}, \displaystyle\|u_{0}\|_{X}\geq Re^{-\beta}\to\textrm{instability}. The exponent \beta is referred to as the transition threshold. A substantial body of work in applied mathematics and physics has been devoted to estimating the exponent \beta (see, e.g., [ 2 , 10 , 16 , 28 , 32 , 38 , 43 , 46 ] ). Over the past decade, rigorous mathematical results have appeared in rapid succession. For the two-dimensional incompressible case without physical boundaries (i.e., {\mathbb{T}}\times{\mathbb{R}} ), Bedrossian, Masmoudi, and Vicol [ 8 ] proved that \beta=0 when the perturbation space X is of Gevrey class 2+ . If X is taken to be a Sobolev space, subsequent works [ 36 , 44 ] have established the upper \beta\leq\frac{1}{3} . In the finite channel {\mathbb{T}}\times[-1,1] , Chen, Li, Wei, and the third author [ 13 ] derived \beta\leq\frac{1}{2} for the two-dimensional problem under the no-slip boundary condition. Further two-dimensional stability results for Couette flow can be found in [ 9 , 35 , 30 , 7 ] . The picture changes significantly in three dimensions because of a new linear mechanism—the lift-up effect—that induces transient growth of disturbances. Consequently, the transition threshold problem becomes substantially more difficult. For the domain {\mathbb{T}}\times{\mathbb{R}}\times{\mathbb{T}} , Bedrossian, Germain, and Masmoudi [ 3 , 4 , 5 ] obtained \beta\leq 1 when X is a Gevrey space and \beta\leq\frac{3}{2} when X is a Sobolev space. Wei and the third author [ 45 ] later improved the Sobolev result to \beta\leq 1 . Most recently, Chen, Wei, and the third author [ 14 ] treated the channel geometry {\mathbb{T}}\times[-1,1]\times{\mathbb{T}} under the no-slip boundary condition and again derived \beta\leq 1 for Sobolev perturbations. Although the stability of compressible flows has a long history, the stability of compressible Couette flow remains far less understood than its incompressible counterpart. The compressible system is inherently more complicated, and the associated lift-up effect is markedly stronger. To date, most investigations have been numerical or physical (see, e.g., [ 18 , 19 , 23 , 24 , 33 , 22 , 11 , 12 , 17 ] ). On the mathematical side, Kagei [ 26 ] established asymptotic stability of Couette flow for small Reynolds numbers, while Li and Zhang [ 29 ] later treated the case with slip boundary conditions. Regarding the transition threshold problem, Antonelli, Dolce, and Marcati [ 1 ] proved linear stability and enhanced dissipation for two-dimensional Couette flow on {\mathbb{T}}\times{\mathbb{R}} at high Reynolds number; they recovered the t^{1/2} growth of the L^{2} norm previously observed in the physical literature [ 21 ] . Zeng, Zi, and the third author [ 47 ] obtained analogous results on {\mathbb{T}}\times{\mathbb{R}}\times{\mathbb{T}} . Most recently, Huang, Li, and Xu [ 25 ] addressed the nonlinear problem and showed that \beta\leq 11/3 when the perturbation space X is taken to be a Sobolev space. The goal of this paper is to investigate the nonlinear stability of the Couette flow for the two-dimensional compressible Navier–Stokes equations and seek the optimal threshold exponent \beta . 1.2 Main result Due to the different physical properties of the velocity field \bm{v} , we decompose it into two distinct components: the compressible part and the incompressible part. Specifically, we define \displaystyle d\triangleq\mathop{\rm div}\nolimits_{\bm{x}}\bm{v},\quad\omega\triangleq\mathop{\rm curl}\nolimits_{\bm{x}}\bm{v}=\nabla^{\perp}_{\bm{x}}\cdot\bm{v}=-\partial_{y}v^{1}+\partial_{x}v^{2}. (1.8) Thus, we have \displaystyle\bm{v}= \displaystyle\nabla_{\bm{x}}\Delta^{-1}_{\bm{x}}\mathop{\rm div}\nolimits_{\bm{x}}\bm{v}+\nabla^{\perp}_{\bm{x}}\Delta^{-1}_{\bm{x}}\mathop{\rm curl}\nolimits_{\bm{x}}\bm{v}=\nabla_{\bm{x}}\Delta^{-1}_{\bm{x}}d+\nabla^{\perp}_{\bm{x}}\Delta^{-1}_{\bm{x}}\omega. (1.9) In view of the system ( 1 ), we find that (n,d,\omega) satisfies the following system \displaystyle\left\{\begin{array}[]{l}\partial_{t}n+y\partial_{x}n+d=F_{1},\\[4.30554pt] \partial_{t}d+y\partial_{x}d+2\partial_{x}\partial_{y}\Delta^{-1}_{\bm{x}}d+2\partial_{x}^{2}\Delta^{-1}_{\bm{x}}\omega+\Delta_{\bm{x}}n-\nu\Delta_{\bm{x}}d=\mathop{\rm div}\nolimits_{\bm{x}}{\bm{F}}_{2},\\[4.30554pt] \partial_{t}\omega+y\partial_{x}\omega-d-\mu\Delta_{\bm{x}}\omega=\mathop{\rm curl}\nolimits_{\bm{x}}{\bm{F}}_{2},\\[4.30554pt] (n,d,\omega)|_{t=0}=(n_{in},d_{in},\omega_{in})(\bm{x})=(n_{in},\mathop{\rm div}\nolimits_{\bm{x}}\bm{v}_{in},\mathop{\rm curl}\nolimits_{\bm{x}}\bm{v}_{in})(\bm{x}),\end{array}\right. where \nu\triangleq 2\mu+\mu^{\prime} . For any s\in{\mathbb{N}}^{+} , we say that q(x,y)\in{\mathcal{H}}^{s}({\mathbb{T}}\times{\mathbb{R}}) whenever \displaystyle\|q\|_{{\mathcal{H}}^{s}}^{2}\triangleq\sum_{|\alpha|\leq s}\int_{{\mathbb{T}}\times{\mathbb{R}}}|(\partial_{x},\partial_{y}+t\partial_{x})^{\alpha}q|^{2}\,dxdy<+\infty. For any function q , we also define \displaystyle{\mathbb{P}}_{0}q\triangleq\frac{1}{2\pi}\int_{{\mathbb{T}}}q(x,y)\,dx,\quad{\mathbb{P}}_{\not=}q\triangleq q-{\mathbb{P}}_{0}q. Now we are in a position to state our main result. Theorem 1.1 . There exists a small \epsilon independent of \mu,\nu , such that if 0<\nu,\mu\leq 1 , \nu has the same order of \mu , and the initial data (n_{in},d_{in},w_{in}) satisfies \displaystyle\|(n_{in},\bm{v}_{in})\|_{H^{4}({\mathbb{T}}\times{\mathbb{R}})}\leq\epsilon\mu, (1.14) then the system ( 2.1 ) admits a unique global solution (n,d,\omega) , such that for any t\geq 0 , \displaystyle\|({\mathbb{P}}_{0}n,{\mathbb{P}}_{0}\Lambda^{-1}_{\bm{x}}d,{\mathbb{P}}_{0}\omega)\|_{L_{t}^{\infty}({\mathcal{H}}^{3}({\mathbb{T}}\times{\mathbb{R}}))}+\mu^{\frac{1}{6}}\|({\mathbb{P}}_{\not=}n,{\mathbb{P}}_{\not=}\Lambda^{-1}_{\bm{x}}d,{\mathbb{P}}_{\not=}\omega)\|_{L_{t}^{\infty}({\mathcal{H}}^{3}({\mathbb{T}}\times{\mathbb{R}}))}\leq C\|(n_{in},\bm{v}_{in})\|_{H^{4}({\mathbb{T}}\times{\mathbb{R}})},

Read the complete text