2
$\begingroup$

This question is based on Zeidler II/B, Problem 30.2.

Consider the ODE: find $u:[0,T] \to \mathbb{R}^n$ s.t. $$u'(t) = F(t,u(t))$$ $$u(0) = u_0$$ given $F:[0,T]\times \mathbb{R}^n \to \mathbb{R}^n$ Caratheodory. We know that if it has a solution, it satisfies $|u(t)| \leq C$ for all $t$. Let us suppose that there is $f \in L^1(0,T)$ such that $|F(t,a)| \leq f(t)$ for all $t$ and all $a$ such that $|a| \leq 2C$.

Consider the ODE $$v'(t) = \hat F(t,v(t))$$ $$v(0) = u_0$$ where (*) $$\hat F(t,a) = \begin{cases} F(t,a) &: |a| \leq 2C\\ F(t, \frac{2Ca}{|a|}) &: |a| \geq 2C \end{cases}.$$ This ODE has a solution, since $\hat F(t,a)$ is dominated by $f$ on the whole of $\mathbb{R}^n$ by definition of $\hat F$.

How do I show that the solution $v$ of the second ODE is also a solution of the first ODE (Zeidler claims this)? Of course it's obvious if the solution $|v(t)| \leq 2C$, but how about if it's greater than $2C$?

(*) - in Zeidler, the second case is given as $F(t,2a/C|a|)$, which I think is wrong since it would not make $\hat F(t,\cdot)$ continuous.


I posted this on MSE some time ago and started a bounty a few days ago but no luck.

https://math.stackexchange.com/questions/1347443/two-odes-why-is-one-solution-the-solution-of-the-other

$\endgroup$
1
$\begingroup$

First, note that if $0<b<T$ and $u:[0,b] \to \mathbb{R}^n$ satisfies
$$u'(t) = F(t,u(t))$$ $$u(0) = u_0$$ then we also have that it satisfies $|u(t)| \leq C$ for all $t\in [0,b]$.

Now, we know that $|v(0)|=|u_0|=|u(0)| \leq C$. Suppose there is $t_1\in [0,T]$ such that $|v(t_1)| > 2C$. Then there is $t_2\in [0,t_1)$ such that, $|v(t_2)| = 2C$ and, for all $t\in [0,t_2]$, $|v(t)| \leq 2C$.

Then $v$ restricted to $[0,t_2]$ satisfies $$v'(t) = F(t,v(t))$$ $$v(0) = u_0$$

So, by our initial remark, we have that $|v(t)| \leq C$ for all $t\in [0,t_2]$. But $|v(t_2)| = 2C>C$. Contradiction. So there is no such $t_1\in [0,T]$ such that $|v(t_1)| > 2C$.

$\endgroup$

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.