Small noise expansion of moment lyapunov exponents for two-dimensional systems (Q4366138)

The moment Lyapunov exponent \(g(p)\) of a random dynamical system is given by \(g(p)= \lim_{t\to 0} t^{-1}\log \mathbb{E}(|x(t; x_0)|^p)\). Small intensity \((\varepsilon\ll 1)\) perturbations of real-noise or white noise dynamical systems are studied, e.g., in the white noise case \[ dx= Ax dt+ \varepsilon \sum^m_{i=1} B_ix\circ dW_i(t), \] where \(x\in \mathbb{R}^d\), \(A,B_i\in \mathbb{R}^{d\times d}\), and \(W_i\) are independent Brownian motions. The aim of the paper is to approximate the moment Lyapunov exponent \(g_\varepsilon(p)\) for small \(\varepsilon\). In the unperturbed case \((\sim\varepsilon=1)\), the approximation \(g(p)= p\lambda+ p^2/2g''(0)+ o(p^2)\), \(\lambda\) being the top Lyapunov exponent, is known, cf. the first author, \textit{W. Kliemann} and \textit{E. Oeljeklaus} [in: Lyapunov exponents. Lect. Notes Math. 1186, 85-125 (1986; Zbl 0588.60047)] and the first author, \textit{E. Oeljeklaus} and \textit{E. Pardoux} [in: ibid., 129-159 (1986; Zbl 0588.60049)]. The strategy of the present approach is to use this approximation also for \(g_\varepsilon(p)\) and to obtain expansions for \(\lambda^\varepsilon\sim \sum \varepsilon^j\lambda_j\) and \(g''_\varepsilon(0)\equiv V^\varepsilon\sim \sum \varepsilon^jV_j\). The values of \(\lambda_j\), \(V_j\) are now determined in a recursive scheme involving only the infinitesimal generator of the system and (the polar coordinate versions of) \(A\), \(B_i\). The expansion is done up to order \(n\). The starting point for this recursion is the adjoint Fokker-Planck equation (i.e. the Kolmogorov backward equation) rather than the previously used forward equation. The real and white noise situations in dimension \(d=2\) are thoroughly discussed and concrete formulae (up to order \(o(\varepsilon^2)\) and/or \(o(p^2)\)) for \(\lambda^\varepsilon\) and \(g_\varepsilon(p)\) derived. Finally, two applications to stochastic bifurcation theory and structural and mechanical systems are considered.