This is a calculus problem from a high school math contest in Greece,from 2012.

I wish to know some solutions for this. I attempted to solve it.

Let $f:\Bbb{R} \to \Bbb{R}$ differentiable such that $\lim_{x \to +\infty}f(x)=+\infty$ and $\lim_{x \to +\infty}\frac{f'(x)}{f(x)}=2$.Show that $$\lim_{x \to +\infty}\frac{f(x)}{x^{2012}}=+\infty$$

Here is my attempt:


Now $\ln{\frac{f(x)}{x^{2012}}}=\ln{f(x)}-2012\ln x=\ln{x}\left( \frac{\ln{f(x)}}{\ln{x}}-2012\right)$

Now from hypothesis we see that $\lim_{x \to +\infty}\ln{f(x)}=+\infty$

By L'Hospital's rule we have that $$\lim_{x \to +\infty}\frac{\ln{f(x)}}{\ln{x}}=\lim_{x \to +\infty}x \frac{f'(x)}{f(x)}=2(+\infty)=+\infty$$

Thus $$\lim_{x \to +\infty}\ln{x}\left( \frac{\ln{f(x)}}{\ln{x}}-2012\right)=+\infty$$

Finally $\lim_{x \to +\infty}\frac{f(x)}{x^{2012}}=+\infty$

Is this solution correct?

If it is,then are there also better and quicker ways to solve this?

Thank you in advance.

  • 1
    $\begingroup$ Can't we simply say that f(x) is equivalent to exp(2x) when x goes to infinity? $\endgroup$ – Jeanba Dec 31 '19 at 13:43
  • 2
    $\begingroup$ Marios.I think it is fine.+1. One remark .While correct I personally try to avoid something like $2(+\infty)$. $\endgroup$ – Peter Szilas Dec 31 '19 at 16:09
  • $\begingroup$ @Jeanba Maybe, if you can work out how to be sufficiently exact about what "equivalent when $x$ goes to infinity" means. It's certainly a good intuitive argument for why it ought to be true. $\endgroup$ – aschepler Dec 31 '19 at 22:13

I think you can just use de l’Hopital inductively:

Given $n \in \Bbb{N}$, we have (since the numerator and denominator diverge): $\lim_{x \to \infty}\frac{f(x)}{x^n} = \lim_{x \to \infty} \frac{f’(x)}{nx^{n-1}} = \lim_{x \to \infty} \frac{f(x) \frac{f’(x)}{f(x)}}{nx^{n-1}} = \frac{2}{n} \lim_{x \to \infty} \frac{f(x)}{x^{n-1}} $

So inductively we get:

$\lim_{x \to \infty}\frac{f(x)}{x^n} = \frac{2^n}{n!} \lim_{x \to \infty}f(x) = \infty$

| cite | improve this answer | |
  • 1
    $\begingroup$ user622002.Nice answer. $\endgroup$ – Peter Szilas Dec 31 '19 at 14:35
  • 1
    $\begingroup$ This is really slick. $\endgroup$ – Randall Jan 1 at 1:35

Here is an approach that only uses the easily proved fact that

$\underset{x\to\infty }\lim\frac{e^x}{p(x)}=\infty\ \text{whenever}\ p \ \text{is a polynomial}. \tag1$

Indeed, there is an $x_0\in \mathbb R^+$ such that $\frac{f'(x)}{f(x)}>1$ for all $x>x_0.$ Then,

$\displaystyle\int^x_{x_0}\frac{f'(t)}{f(t)}dt>x-x_0\Rightarrow \ln f(x)-\ln f(x_0)>x-x_0\Rightarrow f(x)>(f(x_0)e^{-x_0})\cdot e^x \tag2.$

Divide $(2)$ by $x^{2012}$ and invoke $(1)$ to conclude.

| cite | improve this answer | |
  • $\begingroup$ +1.... before the constest,no theory of integration was covered but nice solution though,i will keep it in mind..is my approach correct? $\endgroup$ – Marios Gretsas Dec 31 '19 at 15:00
  • 1
    $\begingroup$ Matematleta.Nice. $\endgroup$ – Peter Szilas Dec 31 '19 at 17:03
  • $\begingroup$ @Matematleta Not sure if the assumption $f \to \infty$ really is superfluous. Couldn't $f$ and $f'$ be both negative so that you can not even write down the expression $\operatorname{ln} f (x)$? $\endgroup$ – Bruno Krams Jan 2 at 18:31
  • $\begingroup$ By the way and in response to @MariosGretsas' objection: There is no need to use an integral in this proof. One could argue that - under appropriate conditions - the mean value theorem yields $$\operatorname{ln} f(x) - \operatorname{ln} f(x_0) = \frac{f'(\xi)}{f(\xi)} \cdot ( x - x_0) \qquad \text{for some } \xi \in (x_0, x)$$ and from this equation one can deduce your estimate. $\endgroup$ – Bruno Krams Jan 2 at 18:36
  • 1
    $\begingroup$ @Matematleta +1 for your l’Hospital-free proof. I appreciate that $\endgroup$ – Bruno Krams Jan 3 at 9:52

Credit to user622002.

Show $\lim_{x \rightarrow \infty} \dfrac{f(x)}{x^n}=\infty$, $n \in \mathbb{N}$.


Base case: $n=0$ √.


$\lim_{x \rightarrow \infty} \dfrac{f(x)}{x^n}=\infty$;


$\lim_{x \rightarrow \infty}\dfrac{f(x)}{x^{n+1}}=$

$\lim_{x \rightarrow \infty}\dfrac{f(x)(f'(x)/f(x))}{(n+1)x^n}$;

For large enough $x$: $f'(x)/f(x)>1$;


$ \dfrac{f(x)(f'(x)/f(x)}{(n+1)x^n}$.

Taking limits, invoking the hypothesis for the left hand side, we get

$\lim_{x \rightarrow \infty} \dfrac{f(x)}{x^{n+1}}=\infty$.

| cite | improve this answer | |
  • $\begingroup$ Is my solution ok though? $\endgroup$ – Marios Gretsas Dec 31 '19 at 14:36
  • $\begingroup$ @PeterSzilas, you wrote hypothesis: $\displaystyle\lim_{x\to\infty}\frac{f(x)}{x^n}=0$ and it should be $n\in\mathbb N_0$ because $0\in\mathbb N_0$ $\endgroup$ – Invisible Dec 31 '19 at 15:50
  • $\begingroup$ VerkovtsevaKatya. Thanks for your comment. Actually as far as I know $0 \in \mathbb{N}$, if you want the positive integers one could write for example $\mathbb{Z^+}$. $\endgroup$ – Peter Szilas Dec 31 '19 at 16:04
  • $\begingroup$ According to the original Peano axiomatisation that we use at our uni, $1$ isn't a successor, therefore we define $\mathbb Z^+:=\mathbb N=\{1,2,3,\ldots\}$. However, I searched the internet to check it and saw the modern version includes $0$. My mistake for correcting you, your statement was legitimate as well. $\endgroup$ – Invisible Dec 31 '19 at 16:46
  • $\begingroup$ VerkovtsevaKatya. No mistake correcting me:) Happy New Year. $\endgroup$ – Peter Szilas Dec 31 '19 at 16:59

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.