Pullback and forward attractors for a 3D LANS-\(\alpha\) model with delay (Q874708): Difference between revisions

The authors consider a rather involved system of PDEs representing Navier-Stokes with delay. The system in question lives on a smooth, bounded open set \(\Omega\subseteq\mathbb R^3\) and has the following form: \[ \begin{aligned} &\partial_t(u-\alpha\Delta u)+\nu(Au-\alpha\Delta u)+(u\nabla)(u-\alpha\Delta u)\\&-\alpha\nabla u^*\cdot\Delta u+\nabla p=f(t)+F(t,u_t)\text{ in }\Omega\times (\tau,\infty)\\ & \text{div}\,u =0 \text{ in }\Omega\times (\tau,\infty) \text{ and }u=Au=0\text{ on }\partial\Omega\times (\tau,\infty)\\ & u(\tau)=u_0\text{ in }\Omega\text{ and }u(t)=\phi(t-\tau)\text{ for }t\in(\tau-h,\tau).\end{aligned}\tag{1} \] Here \(\alpha,\tau,\nu\) are constants, while \(u_t(s)=u(t+s)\) for \(s\in(-h,0)\), that is, (1) is a system which incorporates delay via \(F(\,,\,)\). Finally, \(A\) is the Stokes operator \(P\Delta\). Systems of this type have been studied by many authors, among others the present ones. Prior to study (1), the authors introduce an abstract model which contains (1) as special case. This model is based on a number of objects, subject to various conditions, which will only be indicated. One is given two real Hilbert spaces \(U,H\) with \(U\) compactly and densely embedded in \(H\); \(\|\;\|_U\) and \(|\;|_H\) are the respective norms. Spaces \(C_H=C^0([-h,0];H)\), \(C_U=C^0([-h,0];U)\) and \({\mathcal L}^s_H=L^2((-h,0);H)\) are then introduced. Next one is given an operator \(\widetilde A\in\mathbb L({\mathcal U},{\mathcal U}^*)\), whose restriction to \(\{u\in {\mathcal U}/\widetilde Au\in H\}\) is selfadjoint. Furthermore one has a bilinear mapping \(\widetilde B:{\mathcal U}\times {\mathcal U}\to{\mathcal U}^*\), subject to some estimates. One is also given a mapping \(F:\mathbb R\times C_H\to H\), subject to four conditions, two of them of Lipschitz type. Finally one has an element \(\widetilde f\in{\mathcal L}^2_{\text{loc}}(\mathbb R,{\mathcal U}^*)\) and two initial conditions \(u_0 \in H\), \(\phi\in{\mathcal L}^2_H\). Based on this objects, one can formulate the intended abstract model equations, which assume the form: \[ \begin{aligned} &u(t)+\int^t_\tau\widetilde Au(s)\, ds+\int^t_\tau\widetilde B(u(s),u(s))\,ds=u_0+\int^t_\tau \widetilde f(s)\,ds+\int^t_\tau\widetilde F(s,u_s)\,ds,\quad t\geq t,\\ & u(t)=\phi(t-\tau),\quad t\in(\tau-h,h).\end{aligned}\tag{2} \] A solution to (2) is an element \(u\) belonging to \[ {\mathcal L}^2(\tau,T;{\mathcal U})\cap {\mathcal L}^2(\tau-h,T;H)\cap {\mathcal L}^\infty(\tau,T;H) \] for all \(T>\tau\), which satisfies (2) on \({\mathcal U}^*\) for \(t\in [\tau,\infty)\) and moreover \(u(t)=\phi(t-\tau)\) for \(t\in(\tau-h,\tau)\). Starting point for the main part of the paper is Theorem 1 which asserts that under the various assumptions on \(\widetilde A,\widetilde B,\dots,\) (2) admits a unique global solution \(u\) which also satisfies \(u\in C^0([\tau,\infty),H)\). Theorem 1 is not proved; it is only stated that the proof is by a Galerkin argument. This is somewhat surprising: as shown at the end of the paper, (2) includes (1) as special case, which in turn seems to be more complex than Navier-Stokes in \(3d\); one would expect difficulties with global existence and uniqueness. Next the authors introduce those concepts which are needed in order to handle attractors and related notions. Based on (2) and Theorem 1 the notion of process is introduced: \[ {\mathcal U}(t,\tau),\quad t\geq \tau,\quad t,\tau\in\mathbb R\text{ such that }U(t,s)U(s,\tau)= U(t,\tau)\text{ for }t\geq s\geq \tau,\tag{3} \] which reduces to the notion of solution semigroup for (2) in the autonomous case. By considering (2) as an initial value problem in different ways, one is led to consider two other processes \(\widetilde U(t,\tau)\), \(S(t,\tau)\), \(t\geq \tau\), related to \(U(t,\tau)\), \(t\geq \tau\) in some specific way. In terms of \(U,\widetilde U, S\) forward attractors and pullback attractors are then introduced, the latter being counterparts into the past of the familiar forward attractors. Based on this frame, the authors now prove various theorems asserting the existence of forward and pullback attractors for the processes \(U, \widetilde U, S\). In the last section it is shown that problem (1) can indeed be considered as subcase of (2) when properly interpreted; thus all theorems proved for (2) specialise to (1).