Superstability and rigorous asymptotics in singularly perturbed state-dependent delay-differential equations (Q2430950)

From MaRDI portal
scientific article
Language Label Description Also known as
English
Superstability and rigorous asymptotics in singularly perturbed state-dependent delay-differential equations
scientific article

    Statements

    Superstability and rigorous asymptotics in singularly perturbed state-dependent delay-differential equations (English)
    0 references
    0 references
    0 references
    8 April 2011
    0 references
    The equation \[ \varepsilon \dot x(t) = -x(t) - kx(t - 1 -x(t)) \] with \(k >1\), is studied in the limit \(\varepsilon \to 0\). Earlier work of the authors proved that in this limit there exist slowly oscillating periodic solutions, approaching a sawtooth-like limit profile. The profile consists of portions increasing from \(-1\) to \(k\) with slope 1 and a discontinuous jump back to \(-1\). The present paper gives a more detailed description of the solution behavior, in particular, near the extremal points. For example, the maximum is shown to be located at \(-\varepsilon [ |\log(\varepsilon)| + 2 \log(k-1)]/(k-1) + o(\varepsilon)\). These results are obtained by a very precise and careful analysis of the solutions on four different consecutive intervals, which together cover one period. A suitable ansatz, for example \[ x(t) = t-1 - \varepsilon \sigma(t) +\varepsilon \exp[-\eta( (t - \varepsilon T_3)/\varepsilon^2)] \] (with unknown functions \( \sigma\) and \(\eta\)), is used in each interval and then substituted for the `historical time' \( t -1 -x(t)\) in the analysis on the subsequent interval. An important technique is to {\parindent=6,5mm \begin{itemize}\item[(1)] derive ODEs for functions such as \( \sigma\) and \( \eta\) which are satisfied if and only if \(x\) solves the state-dependent delay equation; \item[(2)] use arguments of the type `A solution starting in a region \(R\) can leave \(R\) only at points where the vector field points to the exterior of \(R\)' for these ODEs, and thus to restrict the behavior of the ODE's solutions like \( \sigma, \eta\). \end{itemize}} The additional information allows also to study the variational equation along the periodic solutions, and to obtain that 1 is a simple Floquet multiplier, while all other multipliers are \( O(\varepsilon)\). (With the companion paper that directly follows this work in the same journal, one obtains asymptotic stability from these properties of the monodromy operator.) A fixed point index argument applied to a return map on a suitable cone then shows that the periodic solutions are unique, since each stable solution contributes +1 to the total index, which in turn was shown to be +1 in earlier work of the authors with P. Paraskevopoulos. The paper shows clearly how intricate even the analysis of this `linear-looking' type of state-dependent delay equation is, and the techniques developed here may well have model character for (seemingly) more complicated equations.
    0 references
    delay-differential equations
    0 references
    state-dependent delay
    0 references
    singular perturbation
    0 references
    periodic solution
    0 references
    limit profile
    0 references
    stability
    0 references
    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