1: \begin{proof}

2: First notice that for any $0<t<t_{max}$, $f(t)\le \st + a t^{1/3}$.

3: A consequence of this inequality is that if $t_{max}<+\infty$, then the limit of $f$ when $t$ goes to $t_{max}$ can only be $0$.

4: 

5: Now, suppose that $t_{max}=+\infty$ but that $f$ does not go to infinity. Then there are $M>0$ and a sequence $(t_n)_n\ge1$ with $\lim_n t_n=+\infty$ such that for any $n\ge 1$, $f(t_n)\le M$.

6: We can choose $n$ such that $\frac{a}{3} t_n^{-2/3} - \frac{\pi^2\sigma^2}{2M^2}<0$.

7: 

8: Then it is easy to see that $f$ decreases after $t_n$.

9: Indeed, for $t\ge t_n$,

10: \begin{equation*}

11: f'(t)=\frac{a}{3} t^{-2/3} - \frac{\pi^2\sigma^2}{2f(t)^2}\le \frac{a}{3} t_n^{-2/3} - \frac{\pi^2\sigma^2}{2M^2}<0.

12: \end{equation*}

13: To be rigorous, we must consider $t_*:=\inf\left\{t\ge t_n, f'(t)>\frac{a}{3} t_n^{-2/3} - \frac{\pi^2\sigma^2}{2M^2}\right\}$ and notice that if we assume $t_*<+\infty$, then $f$ decreases in a neighborhood of $t_*$ and the inequality still holds on this neighborhood, which contradicts the definition of $t_*$.

14: 

15: We have proved that $f'(t)$ is less than a negative constant for $t\ge t_n$, which implies that $f$ reaches zero in finite time.

16: \end{proof}

17: