Take the 2-minute tour ×
Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. It's 100% free, no registration required.

I'm studying the asymptotic behavior $(n \rightarrow \infty)$ of the following formula, where $k$ is a given constant. $$ \frac{1}{n^{k(k+1)/(2n)}(2kn−k(1+k) \ln n)^2}$$

I'm trying to do a series expansion on this equation to give the denominator a simpler form so that it is easier to make an asymptotic analysis.

I used mathematica/wolframalpha to expand the formula (the documents say Taylor expansion is used). http://www.wolframalpha.com/input/?i=1%2F%28n%5E%28k+%28k+%2B+1%29%2F%282+n%29%29+%282+k+n+-+k+%281+%2B+k%29+Log%5Bn%5D%29%5E2%29

However in series expansion at $n \rightarrow \infty$, the result still has $log n$. This is actually a form I prefer, compared to the form $$a_0 + a_1x + a_2 x^2+...$$ Does anyone see how the result may be produced? Any help is much appreciated. Thanks.

share|improve this question
    
Do you need something more precise than the equivalent $1/(4k^2n^2)$? If yes, what form of expansion? –  Did May 17 '13 at 9:04
    
Yes I want more than $1/(4k^2 n^2)$. However I think you nailed it by pointing out it is the $n^{k(k +1)/(2n)}$! –  ELW May 17 '13 at 9:17
    
@Dominic: I was under the impression that doing a Taylor expansion at $n \rightarrow \infty$ is essentially replacing $n$ with $1/n^\prime$ and do $n^\rightarrow 0$. And I need more precision than $1/(4 k^2 n^2)$. Sorry if I didn't make that clear. Thanks for the comments. –  ELW May 17 '13 at 9:25

2 Answers 2

up vote 0 down vote accepted

Hint: $n^{k(k+1)/(2n)}=\mathrm e^{k(k+1)\log(n)/(2n)}=1+u+u^2/2+\ldots$ with $u=k(k+1)\log(n)/(2n)$.

This yields the expansion $$ \frac1{4k^2n^2}\left(1-\frac{(k+1)(k-2)}2\frac{\log n}n+O\left(\left(\frac{\log n}n\right)^2\right)\right). $$

share|improve this answer

Please be careful, because a Taylor expansion at $\infty$ is not a normal series expansion. For $k>0$ it will surely converge to zero, as $\ln(n)$ grows slower than $n$, and hence the denominator is not bounded, as the logarithm is much slower we can forget him for the asympotics

When we look at $$n^\frac{k(k+1)}{2n}$$ this converges to $1$ because the $n$-th root of $n$ converges to 1. So the asymptotic behaviour should be like \[ \frac{1}{(2k n)^2}=\frac{1}{4k^2n^2}\] which is also the coefficient in the taylor expansion at infinity.

share|improve this answer

Your Answer

 
discard

By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.