3

I was trying to evaluate $$ \lim_{n\to \infty} \left(\dfrac{n!}{n^n}\right)^{\dfrac{1}{n}} $$ without resorting to a certain theorem that states : $$ \limsup_{n\to \infty}\; a_n\,^{\frac{\Large 1}{\Large n}} = \lim_{n\to \infty} \left(\dfrac{a_{n+1}}{a_n}\right). $$

With my own idea:

  • By using the discreet definition of factorial using the product notation (also converting $n^n$ into discreet product) :

$$ n! = \prod_{0\leq k \leq n-1} (n-k)= n\cdot\left(\prod_{1\leq k \leq n-1} (n-k)\right) $$

$$ n^n = \prod_{1\leq k \leq n} n = n\cdot\left(\prod_{1\leq k \leq n-1} n \right) $$

And by dividing $n-k$ by $n$ and duplicating the limit (one for evaluating the limit inside product notation and another for reminding myself that $x$ goes to $\infty$ ) into two I get this limit :

$\displaystyle \lim_{n\to\infty} \prod_{1\leq k \leq n-1} \lim_{n\to\infty}\left(1-\dfrac{k}{n}\right)^{\dfrac{1}{n}}$ (1)

Evaluating the inner limit by logarithm and L'Hopital get me into this point :

$a = \displaystyle \lim_{n\to\infty} \dfrac{1}{n} - \dfrac{1}{n-k}$, $1 < k < n-1$

where $y = \displaystyle \lim_{n\to\infty} \left(1-\dfrac{k}{n}\right)^{\dfrac{1}{n}}$, and $a = \ln(y)$

My Question :

$1$) Does $a$ exist as $k$ varies and goes to $n-1$ ? (Continue to Q$2$)

$2$) How to evaluate it ? (Continuation of Q$1$)

$3$) Is my way of doing the limit at equation (1) valid (Duplicating the limit) ??

$4$) Any other way to do $\displaystyle \lim_{n\to \infty} \left(\dfrac{n!}{n^n}\right)^{\dfrac{1}{n}}$ without that sequence theorem I mentioned earlier

My opinion about Q$1$ : because the main limit (at the very beginning) exist and has a value, I think somehow $a$ exist, but what confuse me is that $a$ varies between $0$ and $-1$ ($a = -1$ when $k$ goes to $n-1$).

Q$2$ : I don't have any idea as $k$ value is varying (at least I know that when $k$ is small then $a$ goes to $0$, and when $k$ value is approaching $n-1$ then $a$ goes to $-1$)

Alessio K
  • 10,599
  • You can't have two nested $\lim_{n\to\infty}$ since the first one already binds the variable $n$. Why do you think you need two limits? – Christoph Jun 21 '21 at 16:20
  • @Christoph The 1 is a typo, the inequality is just my way of writing that $k$ goes from 1 to $n-1$ – JangoHypno Jun 21 '21 at 16:22
  • I think i need two limits because the upper bound is a variable – JangoHypno Jun 21 '21 at 16:24
  • Well, you don't. – Christoph Jun 21 '21 at 16:24
  • Your duplication of the limit is not correct — you get that the limit is 1, but it is $e^{-1}$. – Timur Bakiev Jun 21 '21 at 16:24
  • Any other solution beside mine and Without the sequence theorem with the supremum is appreciated (for Q4) – JangoHypno Jun 21 '21 at 16:25
  • Strangely by some subtitution to change the inner limit ( by replacing $-\dfrac{k}{n}$ with subtitution), i was able to change the inner limit to equal 1, and then because $x$ goes to $\infty$ i can just do $1 = 1+\dfrac{1}{x}$ inside the product notation to get into the correct answer : $\dfrac{1}{e}$) – JangoHypno Jun 21 '21 at 16:29
  • 1
    Does this help? https://math.stackexchange.com/questions/28476/finding-the-limit-of-frac-n-sqrtnn – Gary Jun 21 '21 at 16:52
  • @Gary ah i overlooked that second answer, thanks for that. – JangoHypno Jun 21 '21 at 16:56
  • @Gary but i think that's kind of circular logic, because what if i dont know the value of the limit yet ?? – JangoHypno Jun 21 '21 at 17:01
  • @JamboRambo No, you seem to misunderstood something. – Gary Jun 21 '21 at 17:39
  • @Gary of what ? – JangoHypno Jun 21 '21 at 17:52
  • @JamboRambo You only need to know that $\left(1+\frac{1}{n}\right)^n \to e$ which is a basic fact (defintion). – Gary Jun 21 '21 at 18:15
  • I know that, what i meant by circular logic is that by using the limit definition of $e$ then that mean i would have known before that the limit value involve some multiple/power of e . There was an answer that was wrong earlier but he didn't use circular logic because he didn't know that the value of the limit is $\dfrac{1}{e}$ – JangoHypno Jun 21 '21 at 18:17