Heteroclinic orbits in systems with slowly varying coefficients (Q687485): Difference between revisions

The authors consider the system (*) \(\dot\xi=f(\xi,\eta,\varepsilon)\), \(\dot \eta=\varepsilon g(\xi,\eta,\varepsilon)\) with \(\xi\in\mathbb{R}^ n\), \(\eta\in\mathbb{R}^ m\), \(f\) and \(g\) smooth and bounded. They assume that for each \(\eta\), \(\dot\xi=f(\xi,\eta,0)\) has two hyperbolic equilibria \(u^ -(\eta)\) and \(u^ +(\eta)\) of the same type, and a heteroclinic solution \(u(t,\eta)\) that joins them. Each \(u(t,\eta)\) is supposed to be a nondegenerate heteroclinic solution, in the sense that \(\dot u(t,\eta)\) spans the space of bounded solutions of \(\dot \xi=D_ 1 f(u(t,\eta),\eta,0)\xi\). The authors define a Melnikov-type integral \(\Delta(\eta)\), and assume that it has a simple zero at \(\eta=\alpha\). They then prove that for each \(\varepsilon\neq 0\) there is an \((m-1)\)- parameter family of solutions \((\xi(t,\mu,\varepsilon),\eta(t,\mu,\varepsilon))\) of (*), \(\mu\in\mathbb{R}^{m-1}\), such that each solution satisfies a phase condition, and in addition \(\sup_{t\in\mathbb{R}} \|\xi(t,\mu,\varepsilon)- u(t,\eta(t,\mu,\varepsilon),\varepsilon)\| \to 0\) as \(\varepsilon\to 0\). The authors' work is inspired by \textit{K. J. Palmer's} paper [J. Differ. Equations 65, 321-360 (1986; Zbl 0611.34021)] and is related to work of \textit{S. Wiggins} [Global bifurcations and chaos, Berlin (1988; Zbl 0661.58001)].