Discretized fast-slow systems with canards in two dimensions (Q2073969)

From MaRDI portal
scientific article
Language Label Description Also known as
English
Discretized fast-slow systems with canards in two dimensions
scientific article

    Statements

    Discretized fast-slow systems with canards in two dimensions (English)
    0 references
    0 references
    0 references
    0 references
    0 references
    3 February 2022
    0 references
    Maximal canards in slow-fast systems are special trajectories that occur if the extension of an attracting slow manifold and the extension of a repelling slow manifold intersect each other. In the present work the authors are interested in the persistence of canard trajectories after discretization of a slow-fast system with a fold point. Since commonly used discretization schemes are known to destroy maximal canards the authors use the Kahan discretization, a scheme successfully used to preserve many good properties of continuous systems, such as integrability. Specifically, the authors consider the quadratic vector fields \begin{align*} \varepsilon \dot{x} & = -y+x^2+\varepsilon a_1x-a_2xy,\\ \dot{y} & = x-\lambda +a_3 y +a_4x^2, \end{align*} since in this case the Kahan discretization is an explicit method. By using a blow-up method, the existence of attracting and repelling invariant manifolds is shown while a Melnikov calculation is used to determine when the two manifolds coincide. The main result states that the one-dimensional attracting slow manifold and the one-dimensional repelling slow manifold intersect precisely for the parameter value \(\lambda = \lambda_c^h(\sqrt{\varepsilon})\) with \[ \lambda_c^h(\sqrt{\varepsilon})= -C\varepsilon +\mathcal{O}(\varepsilon^{3/2}h)\,. \] This is the discrete analog of a result by \textit{M. Krupa} and \textit{P. Szmolyan} [SIAM J. Math. Anal. 33, No. 2, 286--314 (2001; Zbl 1002.34046)] for the case of differential equations. Numerical simulations illustrate the main result. In particular, experiments with the Kahan discretization of the van der Pol equation show that the main result is not restricted to quadratic vector fields.
    0 references
    slow manifolds
    0 references
    invariant manifolds
    0 references
    blow-up method
    0 references
    loss of normal hyperbolicity
    0 references
    discretization
    0 references
    maps
    0 references
    canards
    0 references
    Kahan discretization
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references

    Identifiers

    0 references
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references