Let $f$ be lipschitz and consider the IVP $\dot x=f(x),\ x(t_0)=x_0$ with a solution $\varphi$. If $f(x_0)>0$, then $\varphi$ is monotonically increasing.

I have a question about the following reasoning for this fact:

Let's assume $\dot\varphi (t_1)=0$ for some $t_1\in\mathbb R$. Set $y:=\varphi(t_1)$, then we have $f(y)=0$ and thus $x_1(t)\equiv y$ solves the DE $\dot x=f(x)$. Because $x_1$ and $\varphi$ intersect in $y$, they must be equal, $\varphi =x_1$, which is a contradiction to $\dot\varphi(x_0)=f(x_0)\neq 0$.

Why must $\varphi$ and $x_1$ be equal? Picard-Lindelöf guarantees a unique solution since $f$ is lipschitz, but this is regarding the IVP. And, as I see it, $x_1 (t)\equiv y$ is a solution for the IVP if and only if $y=x_0$, but that would be a contradiction since $f(x_0)>0$, so we don't have the trivial solution $x(t)\equiv x_0$. Or, in other words, the constructed $x_1$ in the reasoning doesn't solve the IVP. Why should it be equal then with a solution that does(!) solve it?


I'm very glad I ran across this statement, since I've been unfamiliar with it!

Keep in mind though that here, Picard‒Lindelöf is not applied to the original initial value problem, but to the IVP

$$ \begin{cases} x_1'(t) = f(x_1(t)) \\ x_1(t_1) = y; \end{cases} $$

the endpoint in the argument above for equality is $t_1$, not $t_0$.

  • 1
    $\begingroup$ Do you maybe mean the IVP with $x_1(t_1)=y$? For yours, $\varphi$ isn't a solution. $\endgroup$ – Buh Jul 16 '18 at 19:54
  • $\begingroup$ I do indeed @Buh $\endgroup$ – AlgebraicsAnonymous Jul 17 '18 at 6:08

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.