I would like to know the limit of $\lim_{n\to\infty}\frac{2^{\frac{\ln n}{2}}}{n}$. I notice repeatedly taking L'opital Rule does not help. Is there any other approach where I can use to derive the limit?
-
You might simplify the numerator. – John Wayland Bales Aug 26 '18 at 02:52
-
Can I change the variable by letting $ln n = k$? – user122049 Aug 26 '18 at 02:56
-
Yes, that approach would work. – John Wayland Bales Aug 26 '18 at 02:58
-
yes, $\ln n=2k$ works. – Nosrati Aug 26 '18 at 02:58
-
1Yes. It doesn't always work for every function and every limit, but it works here. – Hans Musgrave Aug 26 '18 at 02:59
4 Answers
Hint: $$2^{(\ln n)/2}=2^{(\log_2 n)\frac{\ln 2}{2}}=\left(2^{(\log_2 n)}\right)^{(\ln 2)/2}=n^{(\ln 2)/2}$$
Now simplify further, using the laws of exponents on numerator and denominator of the original expression.

- 82,796
Let $$y=\dfrac1n2^{\frac12\ln n}\Rightarrow \ln y=\ln n\left(\dfrac{\ln 2}{2}-1\right)\to-\infty$$ as $n\to\infty$, then $y\to0$.

- 29,995
I like @vadim123's answer, but personally I find coming up with the algebra a little more intuitive if I explicitly include the base of the logarithm I'm trying to undo. We have a $\ln n$ thing going on here, so we want to incorporate $e$ somehow.
$$2^{(\ln n)/2}=\left(e^{\ln 2}\right)^{(\ln n)/2}=e^{(\ln n)(\ln 2)/2}=n^{(\ln 2)/2}$$
Just for fun there's a completely different trick you can do still using L'Hopital's rule. Note that this only computes the limit if it exists; proof of existence has to be taken care of separately.
When you make no simplifications to the top or the bottom and apply L'Hopital's rule, you get $$\lim_{n\to\infty}\frac{2^{\frac{\ln n}{2}}}{n}=\lim_{n\to\infty}\frac{\frac{\ln 2}{2}2^{\frac{\ln n}{2}}}{n}.$$
Pulling the constant $\frac{\ln 2}{2}$ out and calling the limit you want computed $x$, we find $$x=\frac{\ln 2}{2}x.$$ Solving for $x$, the limit is $0$.

- 1,979
-
2Regarding your different trick, this actually does not prove that the limit exists. Instead, you've shown if it does exist, it must be zero. – JavaMan Aug 26 '18 at 03:18
-
Agreed. If we're being pedantic, OP asked for "the limit", implicitly assuming its existence. – Hans Musgrave Aug 26 '18 at 03:22
-
Not trying to be pedantic; just correct :) I like this method though! (+1) – JavaMan Aug 26 '18 at 03:24
-
That's fair :) Do you think it's worth putting that in the answer? (and thank you for pointing it out too -- I miss details from time to time). – Hans Musgrave Aug 26 '18 at 03:26
-
-
The "just for fun" trick proves that if the limit exists, it is either $0$ or $\infty$ (or, I suppose, $-\infty$). – vadim123 Aug 26 '18 at 05:20
-
1
Set $e^y =2$, then $y \in (0,1)$ (why?).
Rewriting:
$2^{((\log n)/2)}=e^{(y/2)(\log n)}=$
$e^{\log n^{y/2}}= n^{y/2}$, $y/2 \in (0,1).$
Now consider the limit $n \rightarrow \infty$ of $\dfrac{n^{y/2}}{n}$.

- 20,344
- 2
- 17
- 28