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.

How can I find $ a_{n}$ such that $$a_{n} \sim_{n \rightarrow \infty} \sum_{k=1}^n (\ln k)^{1/3} $$ ?

I tried to use integrals:

$$ \int_{k-1}^{k} \ln(t)^{1/3} \mathrm dt\leq \ln(k)^{1/3}\leq \int_{k}^{k+1} \ln(t)^{1/3} \mathrm dt$$ but I cannot compute $$\int_{k-1}^{k} \ln(t)^{1/3} \mathrm dt, \int_{k}^{k+1}\ln(t)^{1/3} \mathrm dt$$

Any idea?

share|improve this question
    
Are you looking for the closed form of $a_n$, i.e. it should have a nice expression through the elementary functions? –  Ilya Feb 22 '12 at 11:17
1  
This should help wolframalpha.com/input/… –  userNaN Feb 22 '12 at 11:34
1  
$\root3\of{\log k}$ is a very slowly growing function, so the sum should be asymptotic to $n\root3\of{\log n}$. –  Gerry Myerson Feb 22 '12 at 12:12
    
How about $1-\frac{1}{x} \leq ln(x) \leq x-1$ –  Joe Feb 22 '12 at 12:24
    
@Gingerjin, that's not a very sharp estimate when $x$ is large. Have you tried to see whether it's good enough to get an answer to the question? –  Gerry Myerson Feb 22 '12 at 23:23

2 Answers 2

From your work, it follows that $$ \sum_{k=1}^n (\ln k)^{1/3}\sim\int_{1}^{n} \ln(t)^{1/3} \, dt. $$ Now, for any $p>0$ $$ \lim_{r\to\infty}\frac{\int_{1}^{r}(\ln t)^p\,dt}{r\,(\ln r)^p}=\lim_{r\to\infty}\frac{(\ln r)^p}{(\ln r)^p+p\,r\,(\ln r)^{p-1}\dfrac{~1}{r}}=\lim_{r\to\infty}\frac{1}{1+\dfrac{p}{\ln r}}=1, $$ so that $$ \sum_{k=1}^n (\ln k)^{1/3}\sim n\,(\ln n)^{1/3}. $$ This is precisely the asymptotic behaviour given in Gerry's comment.

share|improve this answer
    
I don't understand the equality: $$\lim_{r\to\infty}\frac{\int_{1}^{r}(\ln t)^p\,dt}{r\,(\ln r)^p}=\lim_{r\to\infty}\frac{(\ln r)^p}{(\ln r)^p+p\,r\,(\ln r)^{p-1}\dfrac{~1}{r}}$$ Could you be more precise? –  Chon Feb 22 '12 at 22:22
    
@PlaneChon-Ju I think it's an application of L'Hôpital and the fundamental theorem of calculus. –  Dylan Moreland Feb 22 '12 at 22:31
    
Thank you very much! –  Chon Feb 22 '12 at 22:53

You don't need to guess $a_n$.

Using integration by parts we get $$\int_{2}^{n} \sqrt[3]{\log x} dx = n \sqrt[3]{\log n} - C - \frac{1}{3} \int_{2}^{n} (\log x)^{-2/3} dx = n \sqrt[3]{\log n}+ \mathcal{O}(n)$$

When powers of the $\log$ function are involved (say $(\log x)^\alpha$), it is usually a good idea to try integration by parts, taking $u = 1$, $v = (\log x)^\alpha$.

See Also: Euler McLaurin Summation formula.

share|improve this answer
    
$$ \int_{1}^n \ln(t)^{1/3} dt= n\ln(n)^{1/3}-\frac{1}{3}\int_{1}^{n} \frac{1}{\ln(t)^{2/3}}dt $$ But how can I show that $$\int_{1}^{n} \frac{1}{\ln(t)^{2/3}}dt=o(n\ln(n)^{1/3}) $$ –  Chon Feb 22 '12 at 22:18
    
@PlaneChon-Ju: Right, but that does not matter for the final thing. I will correct. Thanks. –  Aryabhata Feb 22 '12 at 22:19
    
@PlaneChon-Ju: $(\ln(t))^{2/3} \ge 1$ for $t \ge e$. –  Aryabhata Feb 22 '12 at 22:22

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.