Nonlinear stability of a pendulum with variable length in elliptic orbit

From MaRDI portal





The authors of this article consider a pendulum attached to a satellite describing an elliptical orbit around a Newtonian center of attraction, where the rod of the pendulum varies proportionally to the radius vector of the orbit. A similar mathematical model was considered by \textit{J. L. Synge} [Proc. R. Ir. Acad., Sect. A 60, 6 p. (1959; Zbl 0084.20004)]. It turns out that the model under consideration is actually a Hamilton system with a Hamiltonian \N\[\N H=y^2/2+\frac{1}{2(1+e\cos{\nu})} \biggl[\frac{\beta^2+3\beta +3} {(1+\beta )^3}\biggr]x^2\N-\frac{1}{24(1+e\cos{\nu})} \biggl[\frac{12+9\beta + 10\beta^2+5\beta^3+\beta^4} {(1+\beta )^5}\biggr]x^4, \N\tag{i}\N\]\Nwhere the parameter \(\beta\) is related to the length of the pendulum rod, while \(e\) is the eccentricity of the orbit. Here \(\nu\) is so-called the true anomaly. The rod of the pendulum is varying with the radius vector of the ellipse according to the law \(b = r\beta\), where \(\beta > 0 \). Further the authors obtain the Euler-Lagrange equation and the corresponding equation of motion. After transforming the participating quantities and rescaling, they obtain the Hamilton \(2\pi \)-periodic in \(\nu\) system with a Hamiltonian \(H=H(\theta ,p_{\theta},\nu , \beta ,e)\) having the form: \N\[\NH=p_{\theta}^{2}/2- (1+e\cos{\nu})^{-1}\biggl( \beta^{-1}\cos{\theta}+\beta^{-2} (1+2\beta\cos{\theta}+ \beta^{2})^{-1/2} \biggr). \N\tag{ii}\N\] \NThis Hamiltonian corresponds to a Hamilton system with two equilibrium points for the pendulum, that is, \((p_{\theta}=0,\theta =0)\) and \((p_{\theta}=0, \theta =\pi )\). In both equilibrium positions, the pendulum is parallel to the \(Oz\) axis. After introducing the variables \(x\) and \(y\), the expansion of the Hamiltonian in \textbf{(ii)} around each one of the equilibrium positions \(p_{\theta}=y\), \(\theta = x\) and \(p_{\theta}=y\), \(\theta = x+\pi \) takes the form \N\[\N H=y^2/2+\frac{1}{2(1+e\cos{\nu})} \biggl[\frac{\beta^2\pm 3\beta +3} {(1\pm \beta )^3}\biggr]x^2\N-\frac{1}{24(1+e\cos{\nu})} \biggl[\frac{12\pm 9\beta + 10\beta^2\pm 5\beta^3+\beta^4} {(1\pm \beta )^5}\biggr]x^4+\cdots , \N\tag{iii}\N\]\Nwhere the ``\(+\) sign corresponds to the solution \( p_{\theta} = y\), \(\theta = x\) and \(\beta > 0\) and ``\(-\) sign to \( p_{\theta} = y\), \(\theta = x+\pi \) and \(0<\beta <1\). Thus the expansion of the Hamiltonian function at both equilibrium positions can be written as \N\[\NH=y^2/2+\frac{1}{2(1+e\cos{\nu})} \biggl[\frac{\beta^2 + 3\beta +3} {(1 + \beta )^3}\biggr]x^2\N-\frac{1}{24(1+e\cos{\nu})} \biggl[\frac{12 + 9\beta + 10\beta^2 + 5\beta^3+\beta^4} {(1\pm \beta )^5}\biggr]x^4+\cdots , \N\tag{iv}\N\]\Nwhere \(-1<\beta <0\) for \( p_{\theta} = y\), \(\theta = x+\pi \), and \(\beta >0\) for \( p_{\theta} = y\), \(\theta =x\). Actually the Hamiltonian \(H\) can be expressed as \(H=H_2+H_4+\cdots \) with \(H_2=y^2/2+(\alpha x^2/2) (1+e\cos{\nu})^{-1}\) and \(H_4=-(\gamma x^4/24) (1+e\cos{\nu})^{-1}\). The authors use the quadratic part \(H_2\) to analyse the linear stability of the considered problem. For sufficiently small values of \(e > 0 \) \textit{J. L. de Menezes Neto} and \textit{H. E. Cabral} [Regul. Chaotic Dyn. 25, No. 4, 323--329 (2020; Zbl 1465.70049)] have shown that there exist expressions for the curves that separate the regions of linear stability and instability. Further the authors consider the nonlinear stability. To this end, they use the Deprit-Hori method and Kolmogorov-Arnold-Moser theory (KAM), following known techniques developed by Markeev.











This page was built for publication: Nonlinear stability of a pendulum with variable length in elliptic orbit

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q7012413)