Relations between energy and enstrophy on the global attractor of the 2-D Navier-Stokes equations (Q818884)

The authors consider the Navier Stokes equations on \(\Omega= [0,L]^2\) with \(L\)-periodic boundary conditions. After the elimination of the pressure with the aid of the Stokes projector \(P\), the equation assumes the standard form \[ \partial_t u+\nu Au+B(u,u)=f,\quad Pf=f \tag{1} \] with \(A=-P\Delta\) the \(L\)-periodic Stokes operator, \(B(u,v)= P(u\nabla)v\) the bilinear form giving rise to the nonlinearity \(B(u,u)\) and with \(f\) the time independent exterior force. The underlying functional frame is given by the spaces \(V,H\) where \(H=\{u\in L^2 (\Omega)^2,\text{div}(u)=0\}\) while \(V=\{u\in H^1_{\text{per}}(\Omega)^2,\;\text{div}(u)=0\}\), with norms \(|u|^2=\int_\Omega |u|^2\,dx^2\) on \(H\) and \(\|u\|^2= \int_\Omega |\nabla u|^2\,dx^2\) on \(V\), where ``per'' indicates \(L\)-periodicity; in addition \(\int_\Omega u\,dx^2=0\) is required. As is known, all solutions of (1) are global, moreover (1) admits a global attractor \({\mathcal A}\). The main aim of the authors is to describe the structure of the attractor in terms of inequalities relating \(|u|\), \(\|u\|\) and \(f\). In fact, the authors prove a large number of such inequalities. For reasons of space we have to content us with three typical cases, indicating the nature of these estimates. After an extended introduction in which the authors list (mostly without proof) a number of estimates, a first result is as follows (Lemma 5.1): (I) if \(f\in D(A)\) and \(u\in{\mathcal A}\) then \[ \|u\|^2\leq\nu^{2/3}k^2_0G^{2/3}\Gamma_1^{1/3} |u|^{4/3}; \tag{2} \] here \(k_0=2\pi L^{-1}\), \(G=|f|\nu^{-2}k_0^{-2}\), and \(\Gamma_1\) expressed in terms of \(\Lambda_1=|A^{\frac{1}{2}}f|^2 k_0^{-2}|f|^{-2}\). One now considers the quotient \(\chi(u)=\|u\|\, |u|^{-1}\) for \(u\in V-\{0\}\). Theorem 5.1 asserts: (II) \(\chi(u)\) attains its absolute maximum on \({\mathcal A}-\{0\}\), and if \(0\in {\mathcal A}\) then \[ \chi(u)\leq\nu^{2/3}k^2_0G^{2/3}\Gamma_1^{1/3}|u|^{1/3}, \quad u\in{\mathcal A}-\{0\}.\tag{3} \] A rather delicate proof then leads to Theorem. 5.2 which asserts among others: (III) if \(u\in{\mathcal A}\) then \[ \|u\|^2\leq|f|\, |u| \nu^{-1}=G\nu\lambda_0|u|\tag{4} \] where \(\lambda_0>0\) is the lowest eigenvalue of \(A\). Based on these results the authors obtain a rough geometric description of \({\mathcal A}\) by looking at the set of points \(\{(|u|,\|u\|),u\in{\mathcal A}\}\). This description is then refined and improved in the subsequent sections, partly by imposing further conditions on \(f\): e.g., one assumes that \(f\) lies in an eigenspace of \(A\) or is at least close to an eigenvector \(u_0\) of \(A\). The estimates are now quite complicated and cannot be reproduced here. In sections 8,9 the behaviour of the set \(\{(|u|,\|u\|),u\in{\mathcal A}\}\) near the parabola \(E=\sqrt e\) in the energy-enstrophy plane \((e,E)\) is investigated, giving rise to very interesting, although rather involved estimates. An appendix takes care of some results used without proof in the text.