I was trying to solve the limit:

$$\lim_{n \to \infty} \sqrt[n]{\frac{(3n)!}{(5n)^{3n}}}$$

By using the root's criterion for limits (which is valid in this case, since $b_n$ increases monotonically):

$$L= \lim_{n \to \infty} \frac{a_n}{b_n} = \lim_{n \to \infty} \frac{a_{n+1}-a_n}{b_{n+1}-b_n}$$

Now I realise using Sterling's formula would make everything easier, but my first approach was simplifying the factorial after applying the criterion I mentioned before. So, after a few failed attempts I looked it up on Mathematica and it said that $\frac{(3(n+1))!}{(3n)!}$ (which is one of the fractions you have to simplify) equals $3(n+1)(3n+1)(3n+2)$. Since I can't get there myself I want to know how you would do it.

Just so you can correct me, my reasoning was:

$$\frac{(3(n+1))!}{(3n)!} = \frac{3\cdot 2 \cdot 3 \cdot 3 \cdot 3 \cdot 4 \cdot (...) \cdot 3 \cdot (n+1)}{3 \cdot 1 \cdot 3 \cdot 2 \cdot 3 \cdot 3 \cdot (...) \cdot 3 \cdot n } = $$ $$= \frac{3^n(n+1)!}{3^{n}n!} = \frac{(n+1)!}{n!} = n+1$$

Which apparently isn't correct. I must have failed at something very silly. Thanks in advance!

  • 1
    $\begingroup$ $(3(n+1)!)$ is the product of all the natural numbers from $1$ to $3(n+1)$, and $3\cdot1\cdot 3\cdot 2\ldots\cdot 3(n+1)$ is the product of the numbers that are multiples of $3$. $\endgroup$ – ajotatxe Jan 22 '17 at 19:08
  • $\begingroup$ @Manuel Also you can apply Stirling's approximation. $\endgroup$ – Behrouz Maleki Jan 22 '17 at 19:11
  • $\begingroup$ @ajotatxe I don't have (3(n+1)!), but (3(n+1))! (the factorial includes everything between the parentheses). $\endgroup$ – mar Jan 22 '17 at 19:17
  • $\begingroup$ This question is somewhat similar: Need to simplify ratio test: $\frac{(2(k+1))!}{(2k)!} $ - it might also be useful for you. I found it using Approach0. $\endgroup$ – Martin Sleziak Jan 23 '17 at 6:49

By Stirling's inequality the answer is clearly $\left(\frac{3}{5e}\right)^3$. To prove it, you may notice that by setting $$ a_n = \frac{(3n)!}{(5n)^{3n}} $$ you have: $$ \frac{a_{n+1}}{a_n} = \frac{(3n+3)(3n+2)(3n+1)(5n)^{3n}}{(5n+5)^{3n+3}} = \frac{\frac{3n+3}{5n+5}\cdot\frac{3n+2}{5n+5}\cdot\frac{3n+1}{5n+5}}{\left(1+\frac{1}{n}\right)^{3n}}\to\frac{\left(\frac{3}{5}\right)^3}{e^3}$$ as $n\to +\infty$.

  • 2
    $\begingroup$ I don't want to know the answer to the limit, but how I can go from a reason of factorials to the polynomial Mathematica gave me. But thanks, I'll use your answer to check if I got it right, haha. $\endgroup$ – mar Jan 22 '17 at 19:15
  • $\begingroup$ @Manuel: the reason is simple: if you have a positive sequence such that $\frac{a_{n+1}}{a_n}\to c$, then $\sqrt[n]{a_n}\to c$, too. $\endgroup$ – Jack D'Aurizio Jan 22 '17 at 19:16
  • 1
    $\begingroup$ And $$\frac{(3(n+1))!}{(3n)!}=\frac{(3n+3)!}{(3n)!}=(3n+3)(3n+2)(3n+1).$$ $\endgroup$ – Jack D'Aurizio Jan 22 '17 at 19:18
  • 1
    $\begingroup$ Oh, that makes sense. Why did I not see it before? Thanks a lot, Jack. $\endgroup$ – mar Jan 22 '17 at 19:19

I thought it might be useful to present an approach that relies on elementary tools only. To that end, we proceed.

First, we write

$$\begin{align} \frac{1}{n}\log((3n!))&=\frac1n\sum_{k=1}^{3n}\log(k)\\\\ &=\left(\frac1n\sum_{k=1}^{3n}\log(k/n)\right)+3\log(n) \tag 1 \end{align}$$

Now, note that the parenthetical term on the right-hand side of $(1)$ is the Riemann sum for $\int_0^3 \log(x)\,dx=3\log(3)-3$.

Using $(1)$, we have

$$\begin{align} \lim_{n\to \infty}\sqrt[n]{\frac{(3n)!}{(5n)^{3n}}}&=\lim_{n\to \infty}e^{\frac1n \log((3n)!)-3\log(5n)}\\\\ &=\lim_{n\to \infty}e^{\left(\frac1n\sum_{k=1}^{3n}\log(k/n)\right)+3\log(n)-3\log(5n)}\\\\ &=\lim_{n\to \infty}e^{\left(\frac1n\sum_{k=1}^{3n}\log(k/n)\right)-3\log(5)}\\\\ &= e^{3\log(3)-3-3\log(5)}\\\\ &=\left(\frac{3}{5e}\right)^3 \end{align}$$

as expected!

Tools used: Riemann sums and the continuity of the exponential function.


$$ (3(n+1))! \neq 3 \cdot 2 \cdot 3 \cdot 3 \cdot 3 \cdot 4 \cdots 3 \cdot (n+1) $$ To make it clear what the problem is, let's write the right-hand side with brackets: $$ (3 \cdot 2) \cdot (3 \cdot 3) \cdot (3 \cdot 4) \cdots (3 \cdot (n+1)) $$ That's just multiplying all the positive multiples of 3 less than $ 3(n+1) $ together; the factorial is defined as multiplying together all positive integers less than $ 3(n+1) $. So the correct expansions are \begin{align*} (3(n+1))! &= 2 \cdot 3 \cdot 4 \cdot 5 \cdot 6 \cdots (3n-2) \cdot (3n-1) \cdot 3n \cdot (3n+1) \cdot (3n+2) \cdot 3(n+1) \\ (3n)! &= 2 \cdot 3 \cdot 4 \cdot 5 \cdot 6 \cdots (3n-2) \cdot (3n-1) \cdot 3n \end{align*} which clearly have the quotient Mathematica gave you.

  • $\begingroup$ Thanks a lot, I noticed what you just said reading Jack's response. I haven't used the factorial function a lot and now I see I didn't completely understand it (or maybe I'm especially slow today). $\endgroup$ – mar Jan 22 '17 at 19:35

My favorite elementary inequality for $n!$: $ \def\lfrac#1#2{{\large\frac{#1}{#2}}} $ $(n/e)^n < n! < (n/e)^{n+1} $.

Taking n-th roots, $n/e < (n!)^{1/n} < (n/e)^{1+1/n} $.

Therefore $\sqrt[n]{\lfrac{(3n)!}{(5n)^{3n}}} =\lfrac{((3n)!)^{1/n}}{(5n)^{3}} >\lfrac{((3n/e)^{3n})^{1/n}}{(5n)^{3}} =\lfrac{(3n/e)^{3}}{(5n)^{3}} =\lfrac{27n^3}{125e^3n^{3}} =\lfrac{27}{125e^3} $.

Arguing the other way, $\sqrt[n]{\lfrac{(3n)!}{(5n)^{3n}}} =\lfrac{((3n)!)^{1/n}}{(5n)^{3}} <\lfrac{((3n/e)^{3n+1})^{1/n}}{(5n)^{3}} =\lfrac{(3n/e)^{3}(3n/e)^{1/n}}{(5n)^{3}} =\lfrac{27n^3(3n/e)^{1/n}}{125e^3n^{3}} =\lfrac{27}{125e^3}(3n/e)^{1/n} $. Since, as $n \to \infty$, $n^{1/n} \to 1 $ and $a^{1/n} \to 1 $ for any $a > 0$, $(3n/e)^{1/n} \to 1 $ so the limit is $\lfrac{27}{125e^3} $.

A more general result:

$\sqrt[n]{\lfrac{(an)!}{(bn)^{an}}} =\lfrac{((an)!)^{1/n}}{(bn)^a} \gt \lfrac{((an/e)^{an})^{1/n}}{(bn)^a} =\lfrac{(an/e)^a}{(bn)^a} =\lfrac{a^a n^a}{b^ae^a} =\lfrac{a^a}{b^ae^a} =\left(\lfrac{a}{be}\right)^a $.

The other way goes exactly as above, so that $\lim_{n \to \infty} \sqrt[n]{\lfrac{(an)!}{(bn)^{an}}} =\left(\lfrac{a}{be}\right)^a $.

  • $\begingroup$ I get the part where you say this: $\frac{((an)!)^{1/n}}{(bn)^a} \gt \frac{((an/e)^{an})^{1/n}}{(bn)^a}$ but why would that imply that $(a/be)^a$ is the result of the limit? $\endgroup$ – mar Jan 22 '17 at 20:25
  • $\begingroup$ The $n^a$ terms cancel out. $\endgroup$ – marty cohen Jan 22 '17 at 23:13
  • 1
    $\begingroup$ I made your fractions a bit bigger so that they are easier to read. Hope you don't mind, but feel free to revert if you don't like it. =) $\endgroup$ – user21820 Jan 23 '17 at 3:49
  • $\begingroup$ I don't mind editing as long as the meaning isn't changed. $\endgroup$ – marty cohen Jan 23 '17 at 4:12

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.