Explicit metrics for a class of two-dimensional cubically superintegrable systems (Q478644)

A superintegrable system is, roughly speaking, a system that allows more integrals of motion than degrees of freedom. The study of superintegrable dynamical systems has received many important developments reviewed recently in [\textit{W. Miller jun.} et al., J. Phys. A, Math. Theor. 46, No. 42, Article ID 423001, 97 p. (2013; Zbl 1276.81070)]. While integrable systems on the cotangent bundle \(T^*M\) of a \(n\)-dimensional manifold \(M\) require a set of functionally independent observables \((H, Q_1,\dots, Q_{n-1})\) which are all in involution for the Poisson bracket, a superintegrable system is made out of \(\nu \geq n\) functionally independent observables \((H, Q_1,\dots, Q_{\nu-1})\), with the constraints \(\{H,Q_i\}=0\). Let us observe that for two-dimensional manifolds, a superintegrable system is necessarily maximal since \(\nu=3\). As is apparent from [loc. cit.], the large amount of results for superintegrable models is restricted to quadratically superintegrable ones, which means that the integrals \(Q_i\) are either linear or quadratic in the momenta, and the metrics on which these systems are defined are either flat or of constant curvature. \textit{V. S. Matveev} and \textit{V. V. Shevchishin} [J. Geom. Phys. 61, No. 8, 1353--1377 (2011; Zbl 1218.53087)] have given a complete classification of all (local) Riemannian metrics on surfaces of revolution, namely \(G =(dx^2+ dy^2)/h^2_x\), \(h_x=dh/dx\), which have a superintegrable geodesic flow, with integrals linear and cubic in momenta, opening the way to the new field of cubically superintegrable models. Superintegrability is then achieved provided the function \(h\) be a solution of following non-linear first-order differential equations, namely (i) \(h_x(A_0 h^2_x +\mu^2 A_0 h^2 - A_1 h + A_2) = A_3\frac{\sin(\mu x)}{\mu}+ A_4 \cos(\mu x)\), (ii) \(h_x(A_0 h^2_x -\mu^2 A_0 h^2 - A_1 h + A_2) = A_3\frac{\sinh(\mu x)}{\mu}+ A_4 \cosh(\mu x)\), (iii) \(h_x(A_0 h^2_x - A_1 h + A_2) =A_3 x + A_4\) for some real constants \(A_i\) and \(\mu \geq 0\). The aim of this article is on the one hand to integrate explicitly the three differential equations and, on the other hand, to determine, by a systematic case study, all special cases which lead to superintegrable models globally defined on simply-connected, closed, Riemann surfaces. The authors obtain, in local coordinates, the explicit form of the two-dimensional, superintegrable systems of Matveev and Shevchishin involving linear and cubic integrals. This enables them to determine for which values of the parameters these systems are indeed globally defined on \(S^2\).