Numerical determination of the basin of attraction for exponentially asymptotically autonomous dynamical systems (Q534459)

The authors consider numerical methods for determining the basin of attraction for a non-autonomous dynamical system. Well-known methods for autonomous equations focus on a bounded set of phase space. But, for non-autonomous systems, any relevant subset of phase space -- because it has time as one of its coordinates -- is unbounded, and previous numerical methods can not be applied. To overcome this problem, the authors introduce a transformation of phase space that transforms the time variable \(t\) to a new variable \(\tau\), mapping the infinite time interval to a finite compact one. To make this work, they need to make the following assumption: in the non-autonomous differential equation \(\dot x= f(x,t)\), the function \(f(x,t)\) tends to a limit function \(g(x)\) exponentially fast as \(t\to\infty\). They also assume similar behavior for all relevant derivatives of \(f\) and \(g\). Having made the transformation just described under the assumptions they impose, the authors then use Lyapunov functions to determine basins of attraction. The Lyapunov function is characterized as the solution of a suitable linear first-order partial differential equation. It is approximated by radial basins functions in a numerical implementation. As an example, the authors apply their method to the exponentially asymptotically autonomous one-dimensional equation \[ \dot x=-x+ x^3+ e^{-2t}x^2 \] and show numerical approximations of the basins of attraction in \((x,\tau)\) coordinates.