1
$\begingroup$

Quick question about Picard–Lindelöf theorem.

We know the existence and uniqueness of a local solution around $[t_0-\epsilon,t_0+\epsilon ]$. But what if we have Lipschitz continuity everywhere, does this mean that a unique solution exists say on the whole of $\mathbb{R}$.

For example $\frac{d}{dx}(x(t))=x(t)^3$ and $x(0)=1$ can we conclude without solving this that the solution will exist everywhere and be unique? I don't know how to apply the Picard–Lindelöf theorem here.

$\endgroup$

2 Answers 2

1
$\begingroup$

You need a global Lipschitz condition with the same constant to apply a global version of the theorem.

Your example $\dot x = x^3$ does not possess a global Lipschitz constant since its derivative $3x^2$ is unbounded. And indeed, the solution $x^{-2}=C-2t$ has a couple limitations on its domain.

However, there is a theorem of Cauchy that tells you that if the ODE function is continuously differentiable (and thus locally Lipschitz), then any solution can be uniquely continued to the boundary of its domain, which may be the $x$ infinity at finite time.

$\endgroup$
4
  • $\begingroup$ So in out case the theorem of Cauchy would imply global existence of the solution but not necessarily the uniqueness? $\endgroup$
    – Caps
    May 18, 2016 at 9:44
  • $\begingroup$ No, it says that the solution is unique as long as it exists, and in this example there is a blow-up in finite time, $x(t)=\frac{x_0}{\sqrt{1-2x_0^2·t}}$. But it does not guarantee a global solution. That you only get with a global Lipschitz constant. $\endgroup$ May 18, 2016 at 9:54
  • $\begingroup$ Do you have any reference to the theorem of Cauchy? That happens to be the version I need. (Of course I can work out the details but I want to credit) $\endgroup$
    – Symbol 1
    Sep 21, 2022 at 18:43
  • $\begingroup$ No, sorry. I only know this as a corollary in the theory of the maximal solution. I believe that the singling out this result as separate theorem can be most likely found in the french literature. $\endgroup$ Sep 21, 2022 at 19:12
1
$\begingroup$

Consider standard example $$ \dot{y}=y^2+1, y(0)=0 $$ which has a solution $y(t)=\tan t$. Function $f(t,y)=y^2+1$ satisfies Lipshitz continuity condition over, say, $[0,2]$ but solution blows up at $t=\frac{\pi}{2}$. What is wrong in above reasoning?

$\endgroup$

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .