Sign up ×
Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. It's 100% free, no registration required.

My question is about differential equations. I want to prove this theorem

Theorem: Let $u(t)$ be a bounded solution in $[0,\infty[$ of the equation $u''+cu'+f(u)=0$, where c is a positive constant and $f()$ is a continuous function. Then $\lim u(t) = l$ exists and $f(l) = 0$. Plus, $\lim u'(t) = 0$.

I'll sketch the proof I have here and i'll point the five questions I have.

Lemma 1: $u'$ is bounded (This one I can prove)

Lemma 2: $\int_{0}^{\infty}{u'(s)^2 ds} < \infty$ (This one I can prove too). Observation: by multiplying u' in the equation we have $\frac{d}{dt}(\frac{u'^2}{2} + F(u)) + cu'^2 = 0$, so $\frac{u'^2}{2} + F(u)$ is decreasing in [0,$\infty$[. Note: $F' = f$.

Lemma 3: $\lim u'(t) = 0$. In the demonstration we are informed that $u''$ is bounded due to the 1st lemma and I can prove it. But I don't understand where I am able to use this information. I think it's easy to see due to lemma 2 that if $u'$ is convergent them the limit is zero, but why this limit exist? That's my first question...

Lemma 4: The limit $\lim F(u(t))$ exists. We know that $\frac{u'^2}{2} + F(u)$ is decreasing in $[0,\infty[$. But why does the limit exists?... 2nd question.

Proof (using the lemmas): Suppose that $\lim \inf u(t)$ is different from $\lim \sup u(t)$. So there exists $a < b$ such that given any number $r$ in $[a,b]$ there exists $t$ arbitrarily large such that $u(t) = r$. Due to the lemma 4, $F$ is constant in $[a,b]$. (Why??? 3rd question) So $f = 0$ in $[a,b]$. Let $r=\frac{a+b}{2}$, $\delta$>0 such that $\frac{2\delta}{c} < \frac{b-a}{4}$ and $T>0$ such that $u(T)=r$ and $|u'(t)|<\delta$ for all $t \geq T$. For $t \geq T$ and at least in the interval $[T,S[$ such that $t\in[T,S[ \Rightarrow u(t)\in]a,b[$, we have: $u'' + cu'$ = 0.

So, $u=A+Be^{-ct}, A+Be^{-cT}=r$ and $|B|ce^{-ct}<\delta$. Finally we have $$ |A+Be^{-ct}-r|\leq \frac{1}{c}|cBe^{-cT} - cBe^{-ct}| \leq \frac{2\delta}{c} < \frac{b-a}{4}, $$ what means that for all $t \ge T$ the solution is $A+Be^{-ct}$ and only takes values in [$\frac{3a+b}{4}$ , $\frac{a+3b}{4}$] (4th question: why this interval??), a contradiction.

So, lim $u(t) = l$ exists. And now the equation $u'' + cu' + f(u) = 0$ itself shows that lim $u''(t$) exists and is equal to $0$ (5th and last question: why lim $u''(t)$ exists and it's $0$???).

So, $f(l) = 0$. The proof is completed

share|cite|improve this question
(1) It's not at all easy to see your actual questions (and the more difficult it is for your readers, the less likely it is that you will get a useful answer. (2) You should learn to use latex (for the same reason as in (1). –  Chris Godsil Jan 22 '13 at 22:09
I'll try to help with your "first question" as to why $\lim u'(t)$ exists and equals zero (in a rough and sketchy way). You know already that $u'$ is $C^1$ and that its derivative $u''$ is bounded, and now you have that $u' \in L^2[0,\infty)$. Suppose now that $\lim u'(t)$ doesn't exist; as $u'$ is square integrable this must mean that there are spikes in $u'$ as you go further out; moreover the "width" of these spikes must be diminishing due to square integrability. But then that means that $u''$ must be unbounded, or else that the maximum height of the spikes is diminishing to zero. –  A Blumenthal Jan 22 '13 at 22:56
For your second question, equipped with the answer from the first, you know that $(u')^2/2 + F \circ u(t)$ is decreasing, and hence has a limit (perhaps equal to $-\infty$; see if you can't rule that out), so that because $u' \rightarrow 0$, you must have that $F \circ u(t) \rightarrow 0$ as a consequence. –  A Blumenthal Jan 22 '13 at 22:59
For the third question, you know that $F \circ u$ has a limit for long times; the idea here is that $u$ is crossing over the interval $[a,b]$ infinitely often for long times, and so if $F$ attained two distinct values on $[a,b]$, then $\lim F \circ u$ wouldn't exist, as the $\limsup$ and the $\liminf$ wouldn't coincide. –  A Blumenthal Jan 22 '13 at 23:02
Thanks for all the answers! I know now how to solve my 'question 4' - it's really easy =D –  Raphaël Jan 23 '13 at 0:32

Your Answer


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

Browse other questions tagged or ask your own question.