Optimal symplectic approximation of Hamiltonian flows

Long term simulations of Hamiltonian dynamical systems benefit from enforcing the symplectic symmetry. One of the several available methods to perform this symplectification is provided by the recently developed theory of extended generating functions. The theory offers an infinite supply of generator types that can be used for symplectification. Using Hofer's metric, a condition for optimal symplectification is given. In the weakly nonlinear case, the condition provides a generator type that, based on the limited information available on the system, in general gives optimal results.