Find the value of $$\lim_{n\rightarrow \infty}\frac{1+\frac{1}{2}+\frac{1}{3}+\cdots +\frac{1}{n^3}}{\ln(n)}$$

My Try: Using Stolz-Cesaro,

Let $\displaystyle a_{n} = 1+\frac{1}{2}+\frac{1}{3}+\cdots \cdots +\frac{1}{n}$ and $b_{n} = \ln(n)$

So $\displaystyle \frac{a_{n+1}-a_{n}}{b_{n+1}-b_{n}} = \lim_{n\rightarrow \infty}\frac{1}{{(n+1)^3}}\cdot \frac{1}{\ln\bigg(1+\frac{1}{n}\bigg)} = 0$

Please explain if what I have done above is right.

  • $\begingroup$ I haven't heard of Cizero stolza but I would bound the numerator by $\int_1^{n} \frac 1 {x^{3}} \, dx =\frac 1 {2n^{2}}$ to show that the limit is $0$. $\endgroup$ – Kavi Rama Murthy Sep 10 '18 at 8:51
  • 1
    $\begingroup$ I wonder why so many downvotes. There was a simple mistake in asker's application of Cesaro-Stolz, but he did show effort. +1 to compensate. $\endgroup$ – Paramanand Singh Sep 10 '18 at 13:26

Let $H_n=\sum_{k=1}^n \frac 1k$. Since $H_n=\log n +O(1)$, $$H_{n^3} = 3\log n +O(1)$$ thus$$\frac{H_{n^3}}{\log n} = 3+O\left(\frac{1}{\log n} \right)=3+o(1)$$

The limit is $3$.


Use of Cesaro-Stolz is tricky here. Let's first evaluate the limit of $a_n$ given by $$a_n=\frac{1}{\log n} \sum_{k=1}^{n}\frac{1}{k}$$ Using Cesaro-Stolz we have $$\lim_{n\to \infty} a_n=\lim_{n\to\infty} \frac{1/n}{\log n-\log(n-1)}=-\lim_{n\to \infty} \frac{1}{n\log(1-(1/n))}=1$$ and then the sequence in question, say $b_n$, is given by $b_n=3a_{n^3}$ so $b_n\to 3$ as $n\to\infty$.

  • $\begingroup$ I like this creativity! (+1) $\endgroup$ – Shashi Sep 10 '18 at 11:08
  • $\begingroup$ Great simplification! $\endgroup$ – gimusi Sep 10 '18 at 12:15

Your $a_n$ should go up to $\frac1{n^3}$. Then $a_{n+1}-a_n=\frac{1}{(n+1)^3}+\frac{1}{(n+1)^3-1}\dots+\frac1{n^3+2}+\frac{1}{n^3+1}$.

I don't know whether you can arrive there by your method, but you can find the limit by using the asymptotics of harmonic numbers / definition of the euler $\gamma$ constant. Then the limit should be $3$.


Note that using Stolz-Cesaro we obtain

$$\frac{\sum_{k=1}^{(n+1)^3}\frac1k-\sum_{k=1}^{n^3}\frac1k}{\ln(n+1)-\ln n}\sim\frac{\ln(n+1)^3-\log n^3}{\ln(1+\frac1n)}=3\frac{\ln(1+\frac1n)}{\ln(1+\frac1n)}=3$$

but, as already noted by Gabriel Romon, this is not necessary and it is sufficient use that by Harmonic series

$$\frac{\sum_{k=1}^{n^3}\frac1k}{\ln n}\sim\frac{\ln n^3}{\ln n}=3$$

  • 2
    $\begingroup$ A direct use of Cesaro-Stolz leads to complications. I have tried a slight variation using subsequences. $\endgroup$ – Paramanand Singh Sep 10 '18 at 11:05

To answer your question directly. The mistake is made in $a_{n+1}-a_n$. For example, if $n = 10$, then this is given by $$a_{11}-a_{10} = 1/1331 + 1/1330 + \cdots + 1/1001$$ since $11^3 = 1331$ and $10^3=1000$.

If you insert this, you still are going to need to do something similar as Gabriel suggested by viewing the summation as upper and lower Riemann sums.

Edit2: if you now use Stolz Cesaro, which is possible, it works out as Gimusi did.

  • $\begingroup$ Why cannot we apply CS? $\endgroup$ – gimusi Sep 10 '18 at 10:04
  • $\begingroup$ I am sorry you're right. Made a mistake here. $\endgroup$ – Stan Tendijck Sep 10 '18 at 10:14

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.