I was asked to evaluate the following expression:

$\lim_{n \to \infty} \left({\frac{(n+1)(n+2)(n+3)...(3n)}{n{^{2n}}}}\right)^{1/n}$

My first step was to assume that the limit existed, and set that value to $y$.

$ y = \lim_{n \to \infty} \left({\frac{(n+1)(n+2)(n+3)...(3n)}{n{^{2n}}}}\right)^{1/n}$

And then, I took the natural logarithm of both sides of the equation. I obtained the expression:

$ \ln y = \lim_{n \to \infty} \frac{1}{n} \cdot \left(\ln(1+\frac{1}{n}) + \ln(1+\frac{2}{n}) + ... + \ln(1+\frac{2n}{n})\right) $

This simplified to:

$ \ln y = \lim_{n \to \infty} \frac{1}{n} \cdot \sum_{k = 1}^{\color{Red}{2n}} \ln(1+\frac{k}{n}) $

I realize that this is similar to the form of a Riemann sum, which can then be manipulated to give the expression in the form of a definite integral. However, the part bolded in red, which is $ 2n$, throws me off. I have only seen Riemann sums be evaluated when the upper limit is $ n - k $, where $k$ is a constant.

Therefore, how would I go about evaluating this expression?

Thank you for all help in advance.

  • $\begingroup$ Hint: Use Cesaro-Stolz $\endgroup$ – Paramanand Singh Jul 25 '18 at 8:34

Consider $$\int_0^2f(x)\,dx$$ where $$f(x)=\ln(1+x).$$ Splitting $[1,2]$ into $2n$ intervals of length $1/n$ gives a Riemann sum $$\frac1n\sum_{k=1}^{2n}f(k/n)=\frac1n\sum_{k=1}^{2n}\ln\left( 1+\frac kn\right)$$ which is exactly yours.

Alternatively you could use Stirling's formula.


We use the following result: if $(a_n)$ is a sequence such that $a_n >0$ for all $n$ and $(\frac{a_{n+1}}{a_n})$ is convergent, then $(a_n^{1/n})$ is convergent and

$$\lim_{n \to \infty}\frac{a_{n+1}}{a_n}=\lim_{n \to \infty}a_n^{1/n}.$$

Now let $a_n:= \frac{(n+1)(n+2)(n+3)...(3n)}{n{^{2n}}}$.

Some easy computations give

$$\frac{a_{n+1}}{a_n}=\frac{(3n+1)(3n+2)(3n+3)}{(n+1)^3} \cdot (1- \frac{1}{n+1})^{2n} \to \frac{27}{e^2}.$$


$(2)^{1/n}\rightarrow 1$, so the answer is not altered by adding an $n$ and dropping $2n$ from the middle of the numerator. Rearrange the terms in the numerator to produce the following pairing: $$((n)(3n))((n+1)(3n-1))((n+2)(3n-2))\cdots ((2n-1)(2n+1))$$

This is $\prod\limits_{i=1}^{n} (2n-i)(2n+i) = \prod\limits_{i=1}^{n} (4n^2-i^2)$ After dividing by $n^{2n}$, you obtain $\prod\limits_{i=1}^{n} (4-(i/n)^2)$. So the limit you are looking for equals to $\exp(\lim\limits_{n\rightarrow \infty} \sum\limits_{i=1}^{n}\frac{1}{n}\log(4-\left(\frac{i}{n}\right)^2))$.

Clearly, as $n$ tends to infinity, the sum $\sum\limits_{i=1}^{n}\frac{1}{n}\log(4-\left(\frac{i}{n}\right)^2))$ tends to the integral $\int\limits_{0}^{1} \log(4-x^2) \, dx = \int\limits_{0}^{1} \log(2-x) \, dx + \int\limits_{0}^{1} \log(2+x) \, dx$.

It is easy to compute both definit integrals.

But your calculation is also good. I that case, you obtain a definite integral on $[0,2]$.


Just a comment after Lord Shark the Unknown's answer.

When I see things looking like factorials, my first reaction is to think about Stirling's approximation of them.

$$a_n=\left(\frac{\prod_{i=1}^{2n} (n+i)}{n^{2n}}\right)^{\frac 1n}=\left(\frac{(3n)!}{n^{2n}\,n!}\right)^{\frac 1n}$$

$$\log(a_n)=\frac 1n \left(\log((3n)!)-\log(n!)-2n \log(n)\right)$$ Now, Stirling's approximation $$\log(p!)=p (\log (p)-1)+\frac{1}{2} \left(\log (2 \pi )+\log \left({p}\right)\right)+\frac{1}{12 p}+O\left(\frac{1}{p^3}\right)$$ Apply it and simplify to get $$\log(a_n)=\frac 1n \left( (3\log (3)-2)n+\frac{\log (3)}{2}-\frac{1}{18 n}+O\left(\frac{1}{n^3}\right) \right)$$ $$\log(a_n)=(3\log (3)-2)+\frac{\log (3)}{2n}-\frac{1}{18 n^2}+O\left(\frac{1}{n^4}\right) $$ Now, using $$a_n=e^{\log(a_n)}=\frac{27 }{ e^2} \left(1+\frac{\log (3)}{2 n}\right)+O\left(\frac{1}{n^2}\right) $$ which shows the limit and how it is approached.


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.