1: \begin{proof}

2: We just show one of the implications, since the reverse one is analogous. Assume point~\eqref{it:fp}, and let us show point~\eqref{it:befp}. Since $g$ is a classical solution to \eqref{fp}, it lies in $\mathcal{C}_t^1\mathcal{C}_v^2$. Hence the calculations \eqref{eq:radialbefp}--\eqref{eq:varu} are rigorous and we see that $f$ satisfies \eqref{befp} in $(0,\infty) \times (\R^2 \setminus \{0\})$ (the point $v=0$ being a possible problem due to the radial change of variables.) But due to Lemma~\ref{lem:C2-cont},

3:   $f$ belongs to $\mathcal{C}_t^1\mathcal{C}_v^2$ and a continuity argument entails that $f$ satisfies \eqref{befp-pde} in $(0,\infty) \times \R^2$. Finally, $f$ satisfies the initial condition in the weak-$*$ sense since the change $\Lambda$ is continuous for the weak-$*$ convergence of measures.

4: \end{proof}

5: