Periodic solutions of second-order differential equations with two-dimensional Lie point symmetry algebra (Q708577): Difference between revisions

The authors consider second order, autonomous, \(C^\infty\) differential equations \[ \ddot{x} = f (x,\dot{x}), \] which they study also with the help of the equivalent first order system \(\dot{x} = y\), \(\dot{y} = f (x,y)\) defined in \(U \subseteq \mathbb R^2\). Let \(S\) be the Lie-point symmetry algebra of the equation in its original (second order) form; according to a well known classification going back to Lie and Tresse, the dimension \(s\) of \(S\) can be equal to 1,2,3 or 8. Based on the dimension and structure of \(S\), which is reflected in the properties of certain determinants \(V_{ij}\) built with components of (prolongations of) the symmetry vector fields, the authors obtain several results on the existence and localization of periodic orbits of the equation. Thus, e.g., they are able to state (Corollary 3) that if \(s \geq 2\) and there is such a determinant \(V_{ij}\) with a given structure, then the system does not have a periodic annulus in \(U\). Moreover (Theorem 6) if \(s = 2\) and \(V_{12} = 0\), then the system has no periodic solution in \(U\), and its limit sets are given by stationary points (\(y=0\)). Special attention is devoted to Lienard systems \[ \ddot{x} = - p(x) - q(x) \dot{x} \] with \(p\) and \(q\) polynomial functions of degree \(m\) and \(n\) respectively. Within this class, it is stated (Theorem 7) that the case \(s=3\) is impossible, and cases with \(s=2,8\) are possible only for certain values -- explicitly identified -- of \(m,n\). In Theorem 9 it is shown that for \(s \geq 2\) there can be no limit cycles.