We propose a new method for determining the stochastic or ordered nature of
trajectories in non-integrable Hamiltonian dynamical systems. The method c
onsists in constructing a time-series from the divergence of nearby traject
ories and then performing a power spectrum analysis of the series. Ordered
trajectories present a spectrum that consists of a few spikes while the spe
ctrum of stochastic trajectories is continuous. A test of the method with t
hree different systems, a 2-D mapping as well as a 2-D and a 3-D Hamiltonia
n, shows that the method is fast and efficient, even in the case of sticky
trajectories. The method is also applied to the motion of asteroids in the
Solar System.