Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. Join them; it only takes a minute:

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

Let $f \in C^1(\mathbb R, \mathbb R)$ and suppose $$ \tag{H} \vert f(x) \vert\le \frac{1}{2}\vert x \vert + 3, \quad \forall x \in \mathbb R. $$

Then, every solution of $$ x'(t)+x(t)+f(x(t))=0, \quad t \in \mathbb R $$ is bounded on $[0,+\infty]$.

First of all, I would like to say that the text has been copied correctly: I mean, it's really $[0,+\infty]$ so I suppose the author wants me to prove $\displaystyle \lim_{t \to +\infty} x(t) <\infty$. Indeed, boundness on $[0,+\infty)$ is quite obvious, because we have global existence (its enough to write the equation as $x'=-x-f(x)$ and to observe that the RHS is sublinear thanks to $(H)$).

So, we have to prove $\displaystyle \lim_{t \to +\infty} x(t) <\infty$. How can we do? I've got some ideas but I can't conclude. First, I observe that the problem is autonomous: this implies that solutions are either constant either monotonic.

First idea: I've fixed $x_0 \in \mathbb R$ and I've written the equivalent integral equation: $$ x(t) = x_0 - \int_0^t [x(s)+f(x(s))]ds $$

Taking the absolute value and making some rough estimates, we get $$ \vert x(t) \vert \le \vert x_0 \vert + \left\vert \int_0^t [x(s)+f(x(s))]ds \right\vert \le \vert x_0 \vert + \int_0^t \frac{3}{2}\left\vert x(s) \right\vert +3 ds $$ but now I don't know how to conclude. Gronwall's lemma? But how can I use it?

Second idea: if $x_0 \in \mathbb R$ is s.t. $x_0 +f(x_0) \neq 0$, the solution of the Cauchy problem is not constant. I can divide both members of equation and I obtain (integrating on $[0,t]$) $$ -t=\int_0^t \frac{dx}{x+f(x)} $$ Now I let $t \to +\infty$ but... what can I conclude?

Thanks in advance for your help.

share|cite|improve this question
In regards to your first paragraph - does global existence imply boundedness by definition? Otherwise, how? What does sublinearity mean (this?) and how does a derivative's sublinearity imply boundedness? – anon Jul 17 '12 at 0:42
@anon I use a global existence theorem: you can find the statement here. As you can see, if the RHS has a growth which is at most linear we have global existence. In my opinion, this implies that the solutions are bounded on $[0,+\infty)$ (no vertical asymptotes). The only point where they could become unbounded is therefore $+\infty$: we have to prove that $\lim_{t \to + \infty} x(t) < \infty$. Do you agree? – Romeo Jul 17 '12 at 8:32
"Bounded" does not mean "no vertical asymptotes." Bounded means the range is contained in an interval, i.e. the function has constant upper and lower bounds. In the DE $x'=x$ the RHS is at most linear, but $x=e^t$ is certainly not bounded. – anon Jul 17 '12 at 12:29
I do agree with you about the definition of "bounded". Anyway, since we have global existence, the function $x(\cdot)$ exists and is of course continuous on $[0,+\infty)$: continuity implies that $x(\cdot)$ is bounded on every compact subset $[a,b] \subset [0,+\infty)$: therefore, the only problems of boundness can occur when $t \to +\infty$. Hope now it's clear what I meant. – Romeo Jul 17 '12 at 15:03
A note, then: bounded functions need not have limits at infinity. (Consider $\sin$.) – anon Jul 17 '12 at 15:12
up vote 4 down vote accepted

By assumptions $$ x'(t) = -x(t) -f(x(t)) \leq -x(t) + \frac 1 2 \lvert x(t) \rvert + 3 $$ Let's suppose there is a point $t_0$ such that $$ x(t_0) > \max\{x(0), 6\} $$

Since $x(t_0)$ is positive the following relation is satisfied $$ x'(t_0) \leq -x(t_0) + \frac 1 2 \lvert x(t_0) \rvert + 3 = -\frac 1 2 x(t_0) + 3 < 0 $$

So the $\max$ of $x(t)$ on the interval $[0, t_0]$ must occur at $t_1\in (0, t_0)$, but that leads to contradiction because: $$ 0 = x'(t_1) \leq - \frac 1 2 x(t_1) + 3 < 0 $$ where the equal sign is there because $t_1$ is an extremum point of a differentiable function on an open interval.

A similar reasoning shows that $x(t)$ is bounded from below.

share|cite|improve this answer
Thank you very much, great proof. – Romeo Jul 17 '12 at 17:42

The goal is to show that the solution is bounded. (I have no idea what the $+\infty$ bit is about, it adds nothing as far as I can tell.)

We first need to show that the solution from any initial condition $x_0$ (at $t_0$) is defined on $[t_0,\infty)$. It is not immediately clear to me that 'we have global existence'. After that, it is fairly straightforward to show boundedness using a 'Lyapunov' function of sorts.

Let $\phi(x) =-x -f(x)$. Since $f$ is $C^1$, we have that $\phi$ is $C^1$, hence Lipschitz on any bounded set. From the usual considerations of ODEs, it is straightforward to establish that, for any $B>0$, if $|x_0| < B$, then there exists a $\delta>0$ (which depends only on $B$ and $f$) such that the solution to $\dot{x} = \phi(x)$ with initial condition $x(t_0) = x_0$ is defined on $[t_0,t_0+\delta]$. If we can establish that $|x(t_0+\delta)| < B$, then it is clear that the solution is defined on $[t_0,\infty)$. Furthermore, the solution is $C^1$.

Using the bound provided for $f$, we have:

If $x\geq 8$, $\phi(x) \leq -x -f(x) \leq -x +\frac{1}{2} |x| +3 = -\frac{1}{2} x+3 \leq -1$.

If $x\leq -8$, $\phi(x) \geq -x -f(x) \geq-x -\frac{1}{2} |x| -3 = -\frac{1}{2} x-3 \geq +1$.

Now let $x$ be as a solution to $\dot{x} = \phi(x)$, and define the 'Lyapunov' function $V(t) = \frac{1}{2} x(t)^2$. Then it is easy to see that $V$ is $C^1$ and $\dot{V}(t) = x(t) \phi(t)$. Consequently, if $|x(t)| \geq 8$, then $\dot{V}(t) \leq -8$.

Thus, if $|x_0| \geq 8$, then the solution will satisfy $|x(t)| < 8$ in finite time.

Now suppose $|x_0| < 8$, and that at some $t \in [t_0,t_0+\delta]$, we have $|x(t)| \geq 8$. Let $t_1$ be the first time at which $|x(t_1)| = 8$ ($x$ is continuous, so this is well defined). Since $V$ is $C^1$, we have, for some suitably small $\epsilon>0$, $\dot{V}(t) \leq -7$, for $t \in [t_1-\epsilon, t_1]$. Since $|x(t_1-\epsilon)| < 8$, this contradicts $|x(t_1)| = 8$, since $V(t_1) = V(t_1-\epsilon)+\int_{t_1-\epsilon}^{t_1} \dot{V}(\tau) d \tau$.

Hence $|x(t)| < 8$ for all $t \in [t_0,t_0+\delta]$, and from the considerations above, we see that $x$ is defined on $[t_0, \infty)$, and $|x(t)| < 8$, for all $t\geq t_0$.

If $\phi(x(t')) = 0$ for some $t'$, then it is clear that $x(t) = x(t')$, for all $t \geq t'$ (by uniqueness of solution). If $\phi(x(t')) \neq 0$ for all $t$, then it is clear that $\phi(x(t))$ has the same sign as $\phi(x(t'))$ for all $t\geq t'$. Hence either $x(t)$ is increasing and bounded above, or decreasing and bounded below. Hence $\lim_{t \to \infty} x(t)$ exists (and is in $[-8,8]$, of course).

share|cite|improve this answer

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

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