5
$\begingroup$

I asked a very similar question here. But now this is different. Suppose $f(t)$ is differentiable and $c$ is a finite constant, then the following statement looks correct, but in fact it is not: \begin{equation} \lim\limits_{t \to \infty} f'(t) = 0 \implies \lim\limits_{t \to \infty} f(t)=c \end{equation} A counter-example is $f(t)=\ln(t)$. Now the question is, what condition should be used for the above statement to be true?

$\endgroup$
  • $\begingroup$ The inverse implication is always true: $$ \begin{equation} \lim\limits_{t \to \infty} f(t)=c\implies \lim\limits_{t \to \infty} f'(t) = 0 \end{equation}$$ Thus, we find that $\lim\limits_{t \to \infty} f'(t) = 0 $ is a necessary condition for the function being convergent at $x\to\infty$. I think that what you're actually looking for is a statement $A$ so that $$\left( \lim\limits_{t \to \infty} f'(t) = 0\right) \land A \implies \lim\limits_{t \to \infty} f'(t) = 0 $$ is true, but so that the formula $$ A \implies \lim\limits_{t \to \infty} f'(t) = 0 $$ doesn't hold in general. $\endgroup$ – Sudix Jul 24 '19 at 15:32
  • 2
    $\begingroup$ The inverse implication is NOT true. See the link in my question. $\endgroup$ – winston Jul 27 '19 at 21:01
3
$\begingroup$

An equivalent condition is that $$ \int_a^\infty f'(t)\,dt $$ converges as an improper Riemann integral for some $a\in\Bbb R$. This happens for instance if $|f'(x)|\le C\,x^{-p}$ for some $C\ge0$ and $p>1$.

| cite | improve this answer | |
$\endgroup$
  • $\begingroup$ Is there any proof of your result? $\endgroup$ – winston Dec 18 '18 at 16:39
  • 1
    $\begingroup$ $$f(x)=f(a)+\int_a^xf'(t)\,dt,\quad x>a.$$ $\endgroup$ – Julián Aguirre Dec 18 '18 at 16:40
  • $\begingroup$ True if $f'$ is Riemann integrable on bounded intervals. $\endgroup$ – zhw. Dec 18 '18 at 17:14
  • $\begingroup$ Does it mean that $\lim\limits_{t \to \infty} f'(t)=0$ does not make any difference to the condition? Doesn't it somehow relax your condition in some way? $\endgroup$ – winston Dec 18 '18 at 20:21
  • $\begingroup$ If $\lim_{x\to\infty}f'(x)$ exists, it must be $0$. But it may happen that the limit does not exist. $\endgroup$ – Julián Aguirre Dec 19 '18 at 8:28
0
$\begingroup$

One way you could potentially change your statement so that it is true is to assert that the function is bounded asymptotically (as $t \rightarrow \infty$).

If $\lim\limits_{t \rightarrow \infty} f'(t) = 0$ and $\exists n,M$ such that $\forall t > n, |f(t)| \le M$, then $\exists c$ such that $\lim\limits_{t \rightarrow \infty} f(t) = c$

I don't have a proof of this currently, but I believe it should be true, and in fact I believe that this also works in reverse. I will provide a proof if I can come up with one.

| cite | improve this answer | |
$\endgroup$
0
$\begingroup$

Two ways for a differentiable (and hence, continuous) function $f(t)$ to satisfy $\lim_{t \to \infty} f(t) = c$:

(1) $f(t) = c$; i.e., is a constant function

(2) $f(t)$ has a horizontal asymptote $y=c$ in the positive x-direction.

Hence, restrict $f(t)$ to be either of these two function types and your original assertion $\lim_{t \to \infty} f'(t) \Rightarrow \lim_{t \to \infty} f(t)$ will be true.

| cite | improve this answer | |
$\endgroup$

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.