Topology and mechanics. I (Q2540998)

The purpose of this paper is to study the topology of mechanical systems with symmetry. A mechanical system with symmetry consists of a 4-tuple \((M,K,V,G)\) where \(M\) is a manifold, \(K\) a Riemannian metric on \(M\), \(V\) a function on \(M\) and \(G\) a Lie group of transformations on \(M\) preserving \(K\) and \(V\), with all data smooth \((M=\) configuration space, \(T(M) =\) space of states, \(K\) defines the kinetic energy, \(V=\) potential energy). The equation of motion is a second order differential equation on \(M\) defined by the equations of Newton, Lagrange or Hamilton (see \textit{R. Abraham} [Foundations of mechanics. New York etc.: W. A. Benjamin (1967; Zbl 0158.42901)]). The author gives a description of the angular momentum of the system as a map \(J: T(M)\to\mathfrak G^*\) where \(\mathfrak G^*\) is the dual Lie algebra of G. He proves that \(J\) is an integral of the mechanical system, i.e., \(J\) is constant on orbits of the dynamical system on \(T(M)\). In the first three sections of this paper there are given several simple examples which illustrate the general situation. The main problem of the general study is to find the topological type of the integral manifolds and the bifurcation set of a mechanical system with symmetry. The integral manifolds are the sets \(I_{c,p}= (E\times J)^{-1}(c,p)\) and \(\tilde I_{c,p}=I_{c,p}/G_p\) where \((c,p)\in \mathbb R\times \mathfrak G^*\), \(E=K+V\circ\pi\) \((E\) total energy, \(\pi: T(M)\to M\) projection), and \(G_p\) is the isotropy group at \(p\) of the adjoint action of \(G\) on \(\mathfrak G^*\). The bifurcation set \(\Sigma\) of \((M,K,V,G)\) is defined to be the set of points of \(\mathbb R\times \mathfrak G^*\) over which \(E\times J\) fails to be trivial in the differentiable sense, i.e., where the topological type of \(I_{c,p}\) changes. \(\Sigma\) contains the critical values of \(E\times J\). Proposition (6.5) gives a way of getting information about \(\Sigma\) from the amended potential \(V_p(x)\), defined by \(V_p(x)=\min\{E(v)\mid v\in J_x^{-1}(p)\}\) for \(p\in\mathfrak G^*\) fixed. The main theorems consider the following example: Take \(m_1\ldots, m_n\) to be positive numbers with \(n >2\), \(M= \{(x_1, \ldots, x_n)\in (\mathbb R^2)^n\mid m_1x_1+\ldots+m_nx_n=0\}\), \(G=S^1\) (the action of \(S^1\) is the diagonal action of \(S^1\) on \((\mathbb R^2)^n\) restricted to \(M)\), kinetic energy \(K(v)= \frac12 (m_1\| v_1\|^2 +\ldots+ m_n\| v_n\|^2)\), \(V\) smooth, invariant under \(S^1\) acting on \(M\), and \(V(x)\to\infty\) as \(\| x\|\to\infty\) on \(M\). The statements are: There are at least \(n-1\) closed orbits of this problem, invariant under \(S^1\), for each \(p\neq 0\). The line \(p =\text{const.}\) in the energy-angular momentum plane \(\mathbb R^2(\mathfrak G^*=\mathbb R)\) intersects the bifurcation set \(\Sigma\) at least \(n-1\) times, if \(p\neq 0\) and \(V\) is contained in a certain Baire subset of a \(C^r\) function space of the \(V\) defined above. Suppose that in addition \(V\) is given by an inner product. Then \(\Sigma=\{(| p| a_i,p)\in\mathbb R^2\mid i=1,\ldots,n-1,\;p\in\mathbb R,\;p\neq 0\}\) where \(a_i>0\). Generically, the \(a_i\) can be expected to be distinct. Then numbering the regions of \(\mathbb R^2-\Sigma\) from the left to the right by \(R_1,\ldots,R_n\), the author shows: \((m=n-2)\) \((c,p)\in R_1\), \(I_{c,p}=\emptyset\), \(\tilde I_{c,p}=\emptyset\); \((c,p)\in R_2\), \(I_{c,p}=S^1\times S^{4n+1}\), \(\tilde I_{c,p}= S^{4n+1}\); \(3\leq k\leq n\), \((c,p)\in R_k\), \(I_{c,p}=S^{2k-3}\times S^{4n-2k+5}\), \(\tilde I_{c,p}\) is an \(S^{4n-2k+5}\) bundle over \(\mathbb P^{k-2}(C)\) (associated to a certain vector space bundle).