1: \begin{proof}

2: Computing formally the evolution of the log-moment in \eqref{KS.e} in a similar fashion to \cite{CC2}, we find for the test function $\phi(x)=\ln

3: (1+|x|^2)$

4: after integrating by parts

5: \begin{eqnarray*}

6: \frac{d}{dt}N_{\e}&=&\int_{\mathbb{R}^2} \pa_t\r_\e\, \phi dx =-\int_{\R^{2}}\r_\e \nabla h_\e[\r_\e]\cdot \nabla \phi dx  + \e \int_{\mathbb{R}^2}\r_\e \Delta \phi  dx

7: \nonumber\\

8: &\leq& \frac{1}{2}\int_{\R^{2}}\r_\e | \nabla \phi|^2 dx + \frac12\int_{\R^{2}}\r_\e |\nabla h_\e[\r_\e]|^2dx + \e \int_{\mathbb{R}^2}\r_\e \Delta \phi

9: dx\,.

10: \end{eqnarray*}

11: Computing the derivatives of $\phi$ we see

12: \[|\nabla \phi|=\left|\frac{2 x}{1+|x|^2}\right|\leq 1\,,\qquad |\Delta \phi| = \frac{4}{(1+|x|^2)^2}\leq 4\,.\]

13: We thus obtain

14: \begin{eqnarray*}

15: \frac{d}{dt}N_{\e}\leq \frac{1}{2}((1+8\e)M+  {\cal D}_\e[\r_\e])\,.

16: \end{eqnarray*}

17: Integration in time and making use of the energy - energy dissipation inequality \eqref{EED.e} and the uniform bound on ${\cal E}_\e$ from below in \eqref{bound.below} gives

18: \begin{eqnarray*}

19: N_\e(t)\leq N_\e (0)+\frac{1}{2}(1+8\e)M t + {\cal E}_\epsilon(\rho_0) - \cal E_*+\e C(1+t)t  \leq C(1+ t+\epsilon t^2)

20: \end{eqnarray*}

21: The argument can easily be made rigorous by using compactly supported approximations of $\phi$ on $\mathbb{R}^2$ as test functions, see e.g. also

22: \cite{BDP}. The

23: proof is concluded by referring to Lemma 3.2 in \cite{CC} for the proof of uniform boundedness of $\r_\e$.

24: \end{proof}

25: