Persistence and smooth dependence on parameters of periodic orbits in functional differential equations close to an ODE or an evolutionary PDE (Q2676529)

The paper establishes persistence of hyperbolic periodic solutions for an ODE \[ \dot x(t) = f(x(t)) \] in \(\mathbb{R}^n\) under perturbation to an equation of the type \[ \dot x(t) = f(x(t)) + \varepsilon P(x_t, \gamma), \] where in this case \(x_t\) stands for \(x\) restricted to \([t-h, t + h]\) (i.e., delayed \textit{and} advanced arguments are possible), and \( \gamma\) is a parameter. At this level of generality the equation does in general not define a dynamical system. Therefore the approach to search for a solution of the equation in a space of periodic functions (functions from \(S^1\) to \( \mathbb{R}^n\)), with the frequency \( \omega\) as flexible parameter, is natural -- the obtained periodic solutions may be not `surrounded' by other solutions. Theorems 4.6 and 4.7 establish a main result in a context of \(C^{\ell + \mathrm{Lip}} \) functions, expressing persistence of the periodic orbit for small \( \varepsilon\) and smooth dependence of the orbit and its frequency on the parameter. Based on Lipschitz-type assumptions and on hyperbolicity of the unperturbed orbit, the method of the proofs presented in Section 5 is the contracting mapping theorem, in combination with interpolation inequalities for \(C^{\ell}\) norms. Sections 6 and 7 show how the general results cover different applications, such as perturbation to equations with state-dependent or distributed delay (but so far not to small delay such as \(\dot x(t) = f(x(t- \varepsilon))\), and to non-autonomous time-periodic equations. Section 8 then addresses the small delay case. It is a bit surprising that the results of Jaroslav Kurzweil are not mentioned here (as is the misspelling of the name Feynman in references [88] and [89]). The authors indicate how the previous methods have to be modified for the proof. One needs higher smoothness assumptions on the nonlinearities, or, with an alternative approach, can establish the contraction property also for a \(C^1\)-type metric. In both cases, an analogue of Theorem 4.6 (Theorem 8.2) is obtained for the small delay case. Section 9 considers a model for the \(n\)-body problem of electrodynamics, with retarded forces `arriving' with implicit delays, and with periodic forcing, since the Hamiltonian character of the system without delay would violate the hyperbolicity assumptions. It is shown that the implicit equations for the delays can be suitably solved, and then Theorem 8.2 applies. A different definition of hyperbolicity (essentially a backward contracting, a forward contracting, and a one-dimensional neutral bundle along the unperturbed trajectory) leads to an approach in the spirit of shadowing in Section 10, which is in principle also viable for non-periodic solutions, but used only for periodic solutions on this paper. Section 11 considers evolution equations in a Banach space, with parabolic type PDE, also of higher order (e.g., the Kuramoto-Sivashinsky eqation) in mind. Again, an analogue of Theorem 4.6 (Theorem 11.2) is proved by a contraction method. It has an assumption of the type that a linearized equation along the unperturbed periodic orbit admits a suitable solution. This assumption, and the smoothness requirements, are the major difficulties for concrete applications. In a degree increasing towards the end of the work (understandable in view of the length), arguments are sketched only, or the reader is referred to related publications. The paper concludes with an appendix giving estimates for derivatives of compositions, interpolation inequalities for \(C^{\ell}\)-norms, and compactness type results in \(C^{\ell + \mathrm{Lip}}\)-spaces on so-called compensated domains, and an extensive list of references.