Tried this question here How to calculate $\lim\limits_{n \to \infty} \left( \frac{\ln(n+1)^{n+1}}{\ln n^n} \right)^n$? and was curious about the result. The answer according to Wolfram Alpha is $e$, so I wanted to try it.

$\lim\limits_{n \to \infty} \left( \frac{\ln((n+1)^{n+1})}{\ln (n^n)} \right)^n$

$\lim\limits_{n \to \infty} \left( \frac{(n+1)\ln(n+1)}{n\ln (n)} \right)^n$

$\lim\limits_{n \to \infty} \left( \frac{\ln(n+1)}{\ln(n)}\left(1 + \frac{1}{n}\right) \right)^n$

This is similar to the typical definition $\lim\limits_{n \to \infty} \left(1 + \frac{1}{n}\right)^n = e$ but it has the extra log factors.

How come these two happen to be equivalent? Is it valid to apply L'Hospital's Rule to the logs even though they're inside the $()^n$? Or can it be applied to just part of the function and not the other half? What's the correct way to handle this extra log multiplier?

For instance:

$\lim\limits_{n \to \infty}\frac{\ln(n+1)}{\ln(n)} = \lim\limits_{n \to \infty}\frac{\frac{d}{dn}\ln(n+1)}{\frac{d}{dn}\ln(n)} = \lim\limits_{n \to \infty}\frac{n}{1+n} = \lim\limits_{n \to \infty}\frac{1}{1/n+1} = 1$

but I don't think we can necessarily analyze this "separately" from the main result; I think they must be taken together somehow. I also considered squeeze theorem but couldn't think of another function approaching $e$ from the other side.

  • $\begingroup$ The log fraction goes to 1 as n goes to infinity, no? $\endgroup$ – pie314271 Jan 21 '17 at 13:58
  • $\begingroup$ Yes, but I want a more formal proof; in practice if I get in the habit of trying to eyeball things, I get it wrong $\endgroup$ – Jay Smith Jan 21 '17 at 13:58
  • $\begingroup$ The intuition says that $\ln(n+1)/\ln(n)\approx 1$ as $n$ gets large, but of course this does not constitute a rigorous proof. $\endgroup$ – Eff Jan 21 '17 at 13:59
  • 1
    $\begingroup$ @pie314271 (and Eff too): that is an invalid argument. The $(1+\frac{1}{n})$ part tends to $1$ too, but still the limit is not $1$. $\endgroup$ – TonyK Jan 21 '17 at 14:00
  • 1
    $\begingroup$ I don't think I showed that $\endgroup$ – Jay Smith Jan 21 '17 at 14:06

With a Taylor expansion-based argument:

When $n\to\infty$, we get $$ \frac{\ln(n+1)}{\ln n}= \frac{\ln n+\ln\left(1+\frac{1}{n}\right)}{\ln n} = 1+ \frac{\ln\left(1+\frac{1}{n}\right)}{\ln n} = 1 + \frac{1}{n\ln n} + o\left(\frac{1}{n\ln n}\right) \tag{1} $$ (using that $\ln(1+x)=x+o(x)$ when $x\to0$) so that $$\begin{align} \frac{\ln(n+1)}{\ln n}\left(1+\frac{1}{n}\right) &= \left(1 + \frac{1}{n\ln n} + o\left(\frac{1}{n\ln n}\right)\right)\left(1+\frac{1}{n}\right) = 1+\frac{1}{n}+\frac{1}{n\ln n} + o\left(\frac{1}{n\ln n}\right)\\ &= 1+\frac{1}{n}+o\left(\frac{1}{n}\right) \tag{2} \end{align}$$ and from (2) and the same Taylor expansion of $\ln(1+x)$ at $0$ we get $$\begin{align} \left(\frac{\ln(n+1)}{\ln n}\left(1+\frac{1}{n}\right)\right)^{n} &= e^{n\ln \left(\frac{\ln(n+1)}{\ln n}\left(1+\frac{1}{n}\right)\right)} = e^{n\ln \left(1+\frac{1}{n}+o\left(\frac{1}{n}\right)\right)} = e^{n\left(\frac{1}{n}+o\left(\frac{1}{n}\right)\right)} = e^{1+o\left(1\right)} \\&\xrightarrow[n\to\infty]{} e^1 = e \end{align}$$ as claimed.

  • $\begingroup$ Those are slick transformations / rearrangements but I don't think I would have ever seen those on my own; are they typical strategies or is there a logical process behind the rearrangement at each step? $\endgroup$ – Jay Smith Jan 21 '17 at 14:17
  • $\begingroup$ Also how do you go from $1+\frac{1}{n}+\frac{1}{n\ln n} + o\left(\frac{1}{n\ln n}\right)$ to $1+\frac{1}{n}+o\left(\frac{1}{n}\right)$? What does $o$ mean here exactly? $\endgroup$ – Jay Smith Jan 21 '17 at 14:19
  • $\begingroup$ @JaySmith I am heavily biased on using Taylor expansions whenever possible (as it's a systematic way to tackle limits, and works almost every time). With that in mind, you want to make "standard" quantities appear, with something tending to $0$: the thing that is mainly the issue here is the ratio of logarithms, so the first thing to do is massage it. $$\ln(n+1) = \ln n + \ln\left(1+\frac{1}{n}\right)$$ is a good thing to do to achieve our goals, because of that. After this step, it's pretty much on autopilot. $\endgroup$ – Clement C. Jan 21 '17 at 14:20
  • $\begingroup$ The second (and last) standard and very useful step is to write the quantity of interest $f(n)^{g(n)}$ in its exponential form $\exp(g(n)\ln f(n))$: by continuity of $\exp$, one then only has to care about the exponent $g(n)\ln f(n)$, and this does almost always make things much clearer. $\endgroup$ – Clement C. Jan 21 '17 at 14:21
  • $\begingroup$ @JaySmith $o(\cdot)$ ("little o") is the Landau notation. One writes $f(n)=o(g(n))$ (at some specified point, here when $n\to \infty$) if, basically, $\frac{f(n))}{g(n)}\xrightarrow[n\to\infty]{} 0$. In our case, we have $\frac{\frac{1}{n\ln n}+o\left(\frac{1}{n\ln n}\right)}{\frac{1}{n}} = \frac{1}{\ln n} + o\left(\frac{1}{\ln n}\right) \xrightarrow[n\to\infty]{} 0$. $\endgroup$ – Clement C. Jan 21 '17 at 14:23

Use my comment in the question mentioned to use that if $a_n\to a$ then $$\left(1+\frac{a_n}{n}\right)^n \to e^a$$ in this case $$a_n=n\frac{\ln(n+1)-\ln n}{\ln n}=\frac{1}{\ln n}\ln \left(1+\frac{1}{n}\right)^n\to 0$$ and thus $$\left(\frac{\ln (n+1)}{\ln n}\right)^n\to 1$$


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.