The Rayleigh-Bénard problem for compressible fluid flows (Q2689655)

The authors study the Rayleigh-Bénard problem, describing a fully compressible fluid confined between two parallel plates, as described by the system of PDEs \begin{align*} \partial_t \varrho + \operatorname{div}_x (\varrho\, \mathbf{u}) & =0,\\ \partial_t (\varrho\, \mathbf{u}) + \operatorname{div}_x (\varrho\, \mathbf{u}\otimes \textbf{u}) + \nabla_x p(\varrho, \vartheta) & = \operatorname{div}_x\mathbb{S} + \varrho\nabla_x G,\\ \partial_t (\varrho \, e(\varrho,\vartheta)) + \operatorname{div}_x(\varrho\, e(\varrho,\vartheta)\textbf{u}) + \nabla_x\mathbf{q} & = \mathbb{S}:\mathbb{D}_x\mathbf{u} - p(\varrho, \vartheta) \operatorname{div}_x \textbf{u}. \end{align*} The equations are set on \(\Omega=\mathbb{T}^2\times (0,1)\), \(\mathbb{T}^2\) denoting the two-dimensional torus with periodic b.c.; \(\mathbf{u}\) is the fluid velocity, \(\varrho\) the fluid mass density, \(\vartheta\) the temperature, \(\mathbb{S}\) the viscous stress tensor, \(G\) the gravitation potential, \(\mathbf{q}\) the heat flux and \(\mathbb{D}_x \mathbf{u}= \mathrm{Sym} (D_x\mathbf{u})\). Under the additional constitutive assumptions of the fluid being Newtonian (yielding an explicit formula for \(\mathbb{S}=\mathbb{S}(\vartheta,\mathbb{D}_x\mathbf{u})\)) and the heat flux being given by Fourier's law, this is usually referred as the \textit{Navier-Stokes-Fourier system}. Mathematically, in the physically relevant framework of compressible, heat conducting fluids with Dirichlet boundary conditions on \(\mathbf{u}\), global existence of smooth solutions and uniqueness of weak ones are both open problems, which makes the analysis of their long time behaviour challenging. Nontheless, leveraging on the theory of weak solutions proposed by \textit{N. Chaudhuri} and \textit{E. Feireisl} [Appl. Anal. 101, No. 12, 4076--4094 (2022; Zbl 1496.35277)], the following results are established: \begin{itemize} \item[a)] The system is dissipative in the sense of Levinson, namely any global-in-time weak solution eventually enters a bounded absorbing set. \item[b)] Any bounded family of global solutions is precompact in suitable topologies and any of its accumulation points represents a weak solution to the same problem. \end{itemize} Using these results, the authors are then able to show the existence of a global compact attractor \(\mathcal{A}\), which is non-empty and time-shift invariant. The Krylov-Bogolyubov theorem then yields the existence of an invariant measure supported on \(\mathcal{A}\) (in other terms, a stationary statistical solution); convergence of the ergodic averages by means of the Birkhoff-Kinchin theorem is also discussed. Uniqueness of the invariant measure however remains an outstanding open problem.