On the finite cyclicity of open period annuli (Q963734): Difference between revisions

Consider a real analytic vector field \(X_0\) defined on a real analytic surface \(S\) without boundary. An open period annulus \(\Pi\) of \(X_0\) is a union of periodic orbits which is bianalytic with the standard annuli \(\mathcal{S}^1 \times (0,1)\). In this paper, relatively compact period annuli are considered, that is, the closure \(\bar{\Pi} \subset S\) is assumed to be compact. Given an analytic family of analytic vector fields \(X_\lambda\), with \(\lambda \in (\mathbb{R}^n,0)\), defined on \(S\), the cyclicity of \(\Pi\) with respect to the deformation \(X_\lambda\) is the maximal number of limit cycles of \(X_\lambda\) which tend to \(\Pi\) as \(\lambda\) tends to zero. This paper deals with Hamiltonian vector fields, that is \(X_0\) has a first integral with isolated critical points in a complex neighborhood of \(\bar{\Pi}\), and generic Darbouxian vector fields, that is all the singular points of \(X_0\) in a neighborhood of \(\bar{\Pi}\) are orbitally analytically equivalent to linear saddles. The main result of the paper establishes that the cyclicity of the open period annulus \(\Pi\) of a Hamiltonian or a generic Darbouxian vector field with respect to any of the described families \(X_\lambda\) is finite. To prove the finite cyclicity of \(\Pi\), it suffices to show its finite cyclicity with respect to a given one-parameter deformation \(X_\varepsilon\) (with \(\varepsilon \in (\mathbb{R},0)\)) [\textit{L. Gavrilov}, Ergodic Theory Dynam. Systems, 28, No.~5, 1497--1507 (2008; Zbl 1172.37020)]. Then, consider the first return map associated to \(\Pi\) for \(X_\varepsilon\): \[ t \to t + \varepsilon^k M_k(t) + \cdots, \] where \(t \in (0,1)\) and \(k \geq 1\). The function \(M_k(t)\) is called the Poincaré-Pontryagin-Melnikov function and it is analytic for \(t \in (0,1)\). The finite cyclicity of \(\Pi\) is proved if the function \(M_k(t)\) has a finite number of zeros in \((0,1)\), that is, the finite cyclicity follows from the nonaccumulation of zeros of \(M_k(t)\) at \(0\) or \(1\). In the Hamiltonian case, \(M_k(t)\) satisfies a Fuchsian equation whose properties ensure that \(M_k(t)\) has a finite number of zeros in \((0,1)\). In the generic Darbouxian case, the authors follow the work [\textit{D. Novikov}, Geom. Funct. Anal., 18, No.~5, 1750--1773 (2008; Zbl 1167.32005)] and make use of the properties of the Mellin transformation on iterated integrals.