No label defined (Q4936662)

A description is provided for two inverse problems connected with nonlinear dynamics: 1. Consider the dynamic system \[ \frac{d x_i}{dt} = P_i (x_1 , \ldots , x_k) , \quad i = 1, \ldots , k, \tag{1} \] in \({\mathbb{R}}^k\) such that the movements of the particles along its trajectories are subjected to the stochastic process action described by the Fokker-Plank-Kolmogorov equation. The author formulates the inverse problem of determining the functions \( P_i (x) \), \( x \in{\mathbb{R}}^k \), from the probability distributions and the scattering coefficients of this stochastic process and the surfaces in \({\mathbb{R}}^k \) which contain the most probable movements of system (1). This inverse problem is reduced to an underdetermined system of differential equations. In the case \( k = 2 \), under some additional conditions, explicit formulas for \( P_1 \) and \( P_2 \) are given. 2. For the iterative process in \({\mathbb{R}}^k \) \[ x_{n+1} = f( x_n) , \quad n = 0,1,2, \ldots, \] the author considers the problem of determining the mapping \( f\: {\mathbb{R}}^k \to {\mathbb{R}}^k \) from the known probability distribution \( \rho (x) \) of the random vector-valued functions \( x_n \) for large values of \( n \). Some particular examples of this problem are described in the case \( k = 1 \). For higher-dimensional spaces, under the assumptions that the inverse mapping \( f^{-1} \) exists and can be represented as \( f^{-1} (x) = \text{grad } h(x) \) for some smooth function \( h \), it is shown that the distribution \( \rho (x) \) satisfies the Monge-Ampère equation. Under some additional restrictions, the existence of this inverse mapping \( f^{-1} \) is shown.