On Newton equations which are totally integrable at infinity (Q2628973)
From MaRDI portal
scientific article
Language | Label | Description | Also known as |
---|---|---|---|
English | On Newton equations which are totally integrable at infinity |
scientific article |
Statements
On Newton equations which are totally integrable at infinity (English)
0 references
19 July 2016
0 references
The author considers a system of Newtonian-type equations with a potential enery \(u(g,t)\) that is time-depenent and periodic. Specifically, \(u:\mathbb{T}^n\times S^1\to\mathbb{R}\), where \(\mathbb{T}^n\) is the \(n\)-torus, \(S^1\) the circle, and where \(q= (q_1,q_z,\dots,q_n)\) and \(t\) are standard coordinates on \(\mathbb{R}^n\) and \(\mathbb{R}\). A Riemannian metric \(g\) is fixed on the torus and the system of Newtonian equations determined by \(g\) is given by \[ \nabla_q\dot q=-\nabla_q u.\tag{\(*\)} \] The extremals of the action functional \(\int({1\over 2}\dot q^2- u(q,t))\,dt\) coincide with the solutions of \((*)\) and the associated Hamiltonian flow can be described in the standard way. The author's focus is the set \({\mathcal M}\subset T^*\mathbb{T}^n\times S^1\) swept out by the orbits of the Hamiltonian flow \((p(t),q(t),t)\) such that the corresponding extremals have no conjugate points. \({\mathcal M}\) is a closed invariant subset of \(T^*\mathbb{T}^n\times S^1\). The author's main theorem is the following. If \(n\geq 3\) and the set \({\mathcal M}\) contains a neighborhood of infinity (i.e., \(\{|p|>R\}\subset{\mathcal M}\) for some positive constant \(R\)), then the metric \(g\) must be Euclidean, and the potential \(u\) does not depend on \(q\). As a corollary, the same conclusions hold under the assumption that system \((*)\) is ``totally integrable at infinity'' (which means that there is a neighborhood of infinite in \(T^*\mathbb{T}^n\) filled by Lagrangian tori homologous to the zero section that stagy invariant under the time-map of the Hamiltonian flow.) The author provides a counterexample showing that the theorem is false for \(n=1\). He also notes that the corresponding results for \(n=2\) are unknown; in any case, his methods do not work in this case.
0 references
Hamiltonian system
0 references
total integrability at infinity
0 references
time-dependent periodic potential
0 references