Averaging principle for dissipative dynamical systems with rapidly oscillating right-hand sides (Q4334989)

The averaging techniques originated by \textit{N. N. Bogolyubov} [On some statistical methods in mathematical physics, Akad. Nauk Ukr. SSR (1945)] have been used in studies of differential equations, generalized and extended to equations with unbounded coefficients, to abstract evolution systems. The present author, who is a well known researcher in several areas of applied mathematics considers here the abstract system \[ \partial_tu= N(u)+ f(\omega t), \] where \(f\) is a known function and \(\omega\) is a ``large'' dimensionless parameter. (Some dimension theorists may argue with this statement, but in the present work this detail is of no importance.) It is assumed that \(f\) can be averaged, that is \[ \lim_{t\to\infty}(1/t) \int^t_0 f(s)ds= f_0 \] exists. The averaged equation is \(\partial_t\check u=N(\check u)+ f_0\). The famous first theorem of Bogolyubov asserts that if \(t\) is contained in a finite interval \([0,T]\) then for any \(\check u(0)= \check u_0\), \(|u(t)-\check u(t)|\leq\eta_T(\omega)\to 0\) as \(\omega\to\infty\). The corresponding statement for the unbounded interval is known as Bogolyubov's second theorem. Here he asserts that if the stationary solution of the averaged equation exists and the point \(u_0\) of the averaged solution is hyperbolic, then for \(\omega>\omega_0\) a bounded solution of the original equation exists and it is exponentially stable, and if \(f\) is almost periodic then this solution is also almost periodic. The theory of attractors for such almost periodic solutions has been developed by \textit{G. R. Sell} [Trans. Am. Math. Soc. 127, 241-262 and 263-283 (1967; Zbl 0189.39602)]. The present work re-examines the averaging theory in application to the Navier-Stokes equations. This is a review of some existing results, which are always complemented by the author's insight and new theorems, plus entirely new insight into the theory of attractors for the Navier-Stokes systems, culminating in the theorem on Hausdorff dimension for the kernel section of the non-autonomous system. (A kernel is the collection of all bounded full trajectories of the system.) This work contains several sections. Aside from introductory comments and definitions (1), they are: 2. Navier-Stokes equations, their various properties, estimates, and averaging schemes. 3. A hyperbolic equation with dissipation: \[ \partial^2_t u+\gamma\partial_tu= \Delta u-f(u)+\varphi(\omega t), \] with \(u|_{\partial\Omega}=0\), \(\gamma>0\), while \(\Omega\subseteq\mathbb{R}^3\). 4. Attractors derived from averaged equations, plus bibliography. According to the author's summary the most important and basic result contained in this impressive work is the theorem on convergence of the solutions of the non-autonomous equation to the attractor of the autonomous averaged equation as \(\omega\to\infty\). This work certainly deserves a careful study. A discussion in a seminar on differential equations would be a perfect place to go into many details that are too difficult to be fully absorbed in the first reading.