I spend too much time on this one,and I just can't find a way to prove anything. Please, will someone help me?

Does $$ \sum _{n=2}^{\infty}\frac{1}{\ln\left(n!\right)}$$ converge?

P.S. Does anybody knows where can I find (online) these kind of tasks (I mean, with this kind of difficulty). Thank you (in advance). :)

  • 1
    $\begingroup$ Well, you don't want to start at n=1 I would assume $\endgroup$ – imranfat Aug 23 '16 at 21:07

No, it does not.

It is standard to see that $\ln n! \sim_{n\to\infty} n\ln n$, so that the series with positive terms $\sum_n \frac{1}{\ln n!}$ and $\sum_n \frac{1}{n \ln n}$ have same nature by theorems of comparison.

But the latter diverges, as a Bertrand series.$^{(\dagger)}$

Theorem. (Bertrand series) The series $\sum_{n=2}^\infty \frac{1}{n^a(\ln n)^b}$ converges if, and only if, (i) $a>1$ or (ii) $a=1$ and $b>1$.

  • $\begingroup$ Standard with or without Stirling ? $\endgroup$ – Gabriel Romon Aug 23 '16 at 21:10
  • 1
    $\begingroup$ Might want to reference Stirling's approximation . $\endgroup$ – Alexis Olson Aug 23 '16 at 21:10
  • 1
    $\begingroup$ @AlexisOlson See the link provided. No need for Stirling, actually. $\endgroup$ – Clement C. Aug 23 '16 at 21:12
  • $\begingroup$ @LeGrandDODOM without: Look at $(\sum_{k=1}^n \ln k)/(n\ln n)$ with Stolz-Cesaro. $\endgroup$ – zhw. Aug 23 '16 at 21:14
  • 2
    $\begingroup$ @zhw. The comments are not really the place for a debate on "what is simpler, and what does 'simple' mean in terms of a tradeoff between conciseness of the proof and level of the tools used in it." As for the question itself, all that is needed is that $\ln n! \leq n\ln n$, which is actually immediate since $\ln n! = \sum_{k=1}^n \ln k \leq \sum_{k=1}^n \ln n$. $\endgroup$ – Clement C. Aug 23 '16 at 21:28

For all $n >1$, we have $n^n > n!$. Therefore, $\displaystyle \frac{1}{n^n} < \frac{1}{n!}$, and further $\displaystyle \frac{1}{\ln(n^n)} < \frac{1}{\ln(n!)}$ since $\ln(x)$ is an increasing function. By a property of logarithms, we have $\displaystyle \frac{1}{\ln(n^n)} = \frac{1}{n\ln(n)}$.

Since $\displaystyle \sum_{n=2}^\infty \frac{1}{n \ln(n)}$ can be shown to diverge with, e.g., the integral test, it follows that $\displaystyle \sum_{n=2}^\infty \frac{1}{\ln(n!)}$ diverges per the comparison test.


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.