Separation of variables and explicit theta-function solution of the classical Steklov-Lyapunov systems: a geometric and algebraic geometric background (Q691185): Difference between revisions

In a brief paper [Berl. Ber. 1900, 79--87 (1900; JFM 31.0726.02)], \textit{F. Kötter} treated the problem of the motion of a rigid body in the ideal incompressible fluid, described by the classical Kirchhoff equations\ \({\dot K} =K\times \frac{\partial H}{\partial K}+p\times \frac{\partial H}{\partial p},\;{\dot p}=p\times \frac{\partial H}{\partial K}\),\ where\ \(K,p\in R^3\)\ are the total angular momentum and linear momentum vectors, respectively, and \(H=H(K,p)\) is the Hamiltonian, quadratic in \(K,p\), chosen in the Steklov form. In general form, the Hamiltonian admits 2 obvious quadratic integrals of motion plus the Hamiltonian, but in the Steklov form there is a 4th quadratic integral. Exploiting the existence of 4 integrals of motion, Kötter was able to find a new set of coordinates in which the Kirchhoff equations separated, and to express the solutions in terms of theta functions. However, he didn't explain how his method worked and the change of variables and the theta function solution seemed mysterious. Later it was recognized that these Kirchhoff equations could be turned into a canonical Hamiltonian system on the 3-sphere; see e.g.[\textit{A. V. Tsiganov}, Regul. Chaotic Dyn. 9, No. 2, 77--89 (2004; Zbl 1079.70004)]. This suggested that more standard geometrical methods of Hamiltonian mechanics could be used to find an appropriate separation of variables solution for this problem, in terms of spheroconical coordinates. However, the more modern tools yielded theta function solutions that were not clearly the same as Kötter's. In the paper under review, the authors use ideas from projective geometry to introduce a symmetrical set of 6 coordinates for the Kirchhoff equations in which a transition to a particular set of spheroconical coordinates becomes more transparent and show how Kötter's original solution can be derived. They also discuss the problem of integration of Rubanovsky gyroscopic generalizations of these systems.