Globally analytical solutions of the compressible Oldroyd-B model without retardation (Q6571037)

The author considers the compressible Oldroyd-B model that describes the motion of a viscoelastic fluid with memory: \(\rho _{t}+\operatorname{div}(\rho u)=0\) , \((\rho u)_{t}+\nabla \cdot (\rho u\otimes u)-\mu \omega (\Delta u+\nabla \operatorname{div}u)+\nabla \pi =\operatorname{div}\tau \), \(\nu (\tau _{t}+u\cdot \nabla \tau +Q(\tau ,\nabla u))+\tau =2\mu (1-\omega )\mathbb{D}(u)\), where \(\rho \in \mathbb{R}\) is the density, \(u\in \mathbb{R}^{n}\) (\(n=2,3\)) the velocity, \(\tau \in \mathbb{R}^{n\times n}\) the non-Newtonian part (a symmetric matrix) of the stress tensor, and \(\pi \in \mathbb{R}\) the scalar pressure. The constants \(\mu >0\), \(\nu >0\), and \(\omega \in \lbrack 0,1]\) are the viscosity, relaxation time, and coupling constant. The bilinear term \(Q\) has the form: \(Q(\tau ,\nabla u)=\tau W(u)-W(u)\tau -b(\mathbb{D}(u)\tau +\tau \mathbb{D}(u))\), where \(b\in \lbrack -1,1]\) is a parameter, \(\mathbb{D}(u)=\frac{1}{2} (\nabla u+(\nabla u)^{T})\) is the deformation tensor, and \(W(u)=\frac{1}{2} (\nabla u-(\nabla u)^{T})\) is the vorticity tensor. In the present paper, the author considers the case where \(\omega =0\) and assumes that \(\mu =1=\nu \). The initial conditions \((\rho ,u,\tau )\mid _{t=0}=(\rho _{0},u_{0},\tau _{0})\), \(x\in \mathbb{R}^{n}\), are added to the system posed in \(\mathbb{R}^{n}\times (0,+\infty )\). \N\NThe main result proves that if the initial data \((\rho _{0},u_{0},\tau _{0})\) are such that \(e^{2\lambda _{0}\Lambda }(\rho _{0}-1,u_{0},\tau _{0})\in \overset{.}{B}^{n/2-1}\cap \overset{.}{B} ^{n/2}\), for some constant \(\lambda _{0}>0\), there exist two positive constants \(\epsilon _{0}\) and \(C_{0}\), independent of \(\lambda _{0}\), such that for any \(0<\varepsilon \leq \epsilon _{0}\), if \(\left\Vert e^{2\lambda _{0}\Lambda }(\rho _{0}-1,u_{0},\tau _{0})\right\Vert _{\overset{.}{B} ^{n/2-1}\cap \overset{.}{B}^{n/2}}\leq \epsilon \lambda _{0}\), the system admits a unique solution \((\rho (t),u(t),\tau (t))\) satisfying, for any \( t\in (0,+\infty )\), \(\left\Vert e^{\lambda _{0}\Lambda }(\rho -1,u,\tau )(t)\right\Vert _{\overset{.}{B}^{n/2-1}\cap \overset{.}{B} ^{n/2}}+\left\Vert e^{\lambda _{0}\Lambda }(\rho -1,u,\tau )\right\Vert _{L_{t}^{1}(\overset{.}{B}^{n/2})}\leq C_{0}\left\Vert e^{2\lambda _{0}\Lambda }(\rho _{0}-1,u_{0},\tau _{0})\right\Vert _{\overset{.}{B} ^{n/2-1}\cap \overset{.}{B}^{n/2}}\). Here \(\overset{.}{B}^{n/2-1}\) and \( \overset{.}{B}^{n/2}\) are Besov spaces. \N\NFor the proof, the author introduces the linearized system \(a_{t}+\operatorname{div}u=0\), \(u_{t}+\nabla a-\operatorname{div}\tau =0\), \( \tau _{t}+\tau -2\mathbb{D}(u)=0\), with \(a=\rho -1\). Defining further \( \mathfrak{g}=\nabla a-\operatorname{div}\tau \), he writes the linear problem as: \(\mathfrak{ g}_{t}-\operatorname{div}\tau +2\nabla \operatorname{div}u+\Delta u=0\), \(u_{t}+\mathfrak{g}=0\), \(\operatorname{div}\tau _{t}+\operatorname{div}\tau -\nabla \operatorname{div}u-\Delta u=0\). \N\NThe key step of the proof is the derivation of an a priori estimate on the solution to the linearized problem, by using a Littlewood-Paley decomposition, Chemin-Lerner spaces and a result by \textit{M. Paicu} and \textit{P. Zhang} in [Commun. Math. Phys. 307, No. 3, 713--759 (2011; Zbl 1237.35129)], among other tools. The author distinguishes between high and low frequencies.