I have been using results from this paper in calculations. In sections 2.4 and 3.4 they perform a canonical transformation into new coordinates consisting of constants of motion.
My question is specifically about eq. 2.65. There they claim that the flow generated by $H_0$ on some observable $f$ is given by the Poisson bracket $\{H_0, f\}$. To me this seems entirely backwards as for example Hamiltonian flows are given by the bracket $\{\cdot, H\}$.
However it's entirely possible that my Hamiltonian mechanics is simply too rusty.