The Hamiltonian formalism of the Inverse Problem

Abstract

A big-isotropic structure is a generalization of the notion of Dirac structure, due to Vaisman. We discuss the inverse problem of deciding if a vector field is Hamiltonian having a big-isotropic structure as underlying geometry. In [1] we have considered this question for the special case of replicator equations. Here we generalize that approach to any vector field that can be written in the form $X(u)=(\B\eta)(u),$ where $\B=\B(u)$ is a matrix and $\eta=\eta(u)$ is a vector. For a linear system we show that, if the representing matrix of the system has at least one pair of positive-negative non-zero eigenvalues, or a zero eigenvalue with at least one 3-dimensional Jordan block associated to it, then the linear system has a Hamiltonian description with respect to a big-isotropic structure. As a byproduct, we find a class of linear systems with zero eigenvalue that are Hamiltonian with respect to a big-isotropic structure but not a Dirac structure. Moreover, we prove that every linear Hamiltonian system, having a big-isotropic structure as underlying geometry, is completely integrable in the sense of Zung [12]. For linear systems, in the both Hamiltonian formulation and complete integrability cases, explicit descriptions of the geometric structures, the Hamiltonian functions, the first integrals and the commuting flows are provided.