Another proof of the averaging principle for fully coupled dynamical systems with hyperbolic fast motions (Q2574723)

From MaRDI portal
scientific article
Language Label Description Also known as
English
Another proof of the averaging principle for fully coupled dynamical systems with hyperbolic fast motions
scientific article

    Statements

    Another proof of the averaging principle for fully coupled dynamical systems with hyperbolic fast motions (English)
    0 references
    0 references
    30 November 2005
    0 references
    The author considers a system of differential equations (*) for \(X^\varepsilon= X^\varepsilon_{x,y}\) and \(Y^\varepsilon= Y^\varepsilon_{x,y}\), \[ dX^\varepsilon /dt=\varepsilon\cdot B(X^\varepsilon(t),Y^\varepsilon(t)), \quad dY^\varepsilon/dt= b (X^\varepsilon(t),Y^\varepsilon(t)),\tag \(*\) \] with the initial conditions \(X^\varepsilon (0)= x\), \(Y^\varepsilon(0)=y\) on the product \(\overline{\mathcal X}\times M\), where \({\mathcal X}\subset\mathbb{R}^\alpha\) is a bounded connected open set, \(\overline{\mathcal X}\) is its closure and \(M\) is a compact \(C^2\) Riemannian manifold, and where \(B(x,y)\), \(b(x,y)\) are vector fields on \({\mathcal X}\) and \(M\), respectively, which are assumed to be bounded and Lipschitz continuous. To generalize the results in the author's previous paper [Ergodic Theory Dyn. Syst. 24, 847--871 (2004; Zbl 1055.37025)], the author assumes that each flow \(F^t_x(x\in{\mathcal X})\) on \(M\) given by \(dF^t_x (y)/dt=b(x,F^t_xy)\), \(F^0_xy=y\) possesses a basic hyperbolic attractor \(\Lambda_x\). Moreover, the author assumes the boundedness of the \(C^1\)-norm of \(B(x,y)\) and \(C^2\)-norm of \(b(x,y)\) for \((x,y)\in \overline{\mathcal X}\times M\). Denoting by \(\overline B(x)\) the average of \(B(x,y)\) with respect to some measure \(\mu^0_x\) on \(\Lambda_x\), we see that \(\overline B(x)\) is \(C^1\) in \(x\) and so \(\overline B\) is bounded and Lipschitz continuous. Thus, we know the existence of uniqueness of the solutions of the equations (*) and the equations \(d\overline X_x^\varepsilon(t)/dt=\varepsilon\cdot\overline B(\overline X_x^\varepsilon(t))\), \(\overline X^\varepsilon_x(0)=x\). Under these assumptions, the author proves the convergence in average in the averaging principle, i.e., the average of the suprema of \(|X^\varepsilon_{x,y}(t)-\overline X_x^\varepsilon(t)|\) for \(0\leq t<T/ \varepsilon\), with respect to a probability measure on \({\mathcal X}_T\times M\) converges to 0 for \(\varepsilon\to 0\), where \({\mathcal X}_T\) is some open set in \({\mathcal X}\). The author asserts that we can formulate the corresponding results in the discrete time setup with difference equations, instead of differential equations (*).
    0 references
    averaging principle
    0 references
    slow and fast motions
    0 references
    coupled systems
    0 references
    convergence in average
    0 references
    difference equations
    0 references
    differential equations
    0 references

    Identifiers