1: \begin{proof}[Proof of Lemma~\ref{l:ennio}]

2: %We argue by contradiction: we assume that there is a distributional solution of the Cauchy problem~\eqref{e:cpburgers}.  By contradiction, let $u \in L^2_{\mathrm{loc}} ([0, 1] \times \R)$ be a nonnegative distributional solution of the Cauchy problem~\eqref{e:cpburgers} satisfying \eqref{e:430}.

3: The conservation law \eqref{e:cpburgers} implies

4: \begin{equation}

5: \label{e:conserviamo}

6:     \int_{a}^{b} u(t, x) dx =

7:     \int_\R u(t, x) dx 

8:     =

9:     \int_{a}^{b} \bar u(x) dx 

10:  \quad \text{for a.e. $t>0$}. 

11: \end{equation}

12: We perform some \emph{formal} computations, which can be made rigorous by arguing as in the proof of Counterexample~\ref{l:ce1}: by \eqref{e:cpburgers}, the Jensen inequality, and the previous equality, we have

13: \begin{equation}

14: \label{e:baricentro}

15: \begin{split}

16: \frac{d}{dt}

17:  \int_{a}^{b}& 

18:   x \, u(t, x) dx  =

19:  \int_{a}^{b} x \, \partial_t u (t, x) dx 

20: =

21:  - \int_{a}^{b} 

22: x \, \partial_x \big[ u^2 \big](t, x) dx

23:  \\ & 

24:  = \Big[ - x \, u^2 (t, x)\Big]^{x = a}_{x = b}

25:  + 

26:  \int_{a}^{b} u^2 (t, x)dx 

27: \ge 

28:   \frac{1

29:  }{b-a}

30:   \left( \int_{a}^{b}  u (t, x)dx \right)^2.

31: \end{split}

32: \end{equation} 

33: Integrating in time, we get \eqref{e:nono}.

34: %On the other hand, since $u \ge 0$, 

35: %In particular, the left hand side in \eqref{e:nono} attains positive values when

36: %$$

37: %   \int_{-1}^{0} \! 

38: %   x u(t, x) dx  \leq 0,

39: %$$

40: %which contradicts~\eqref{e:nono} 

41: %$t > 1/2$, which contradicts the fact that $u$ is nonnegative. 

42: \end{proof}

43: