I wonder whether there is a closed form for this sum $$ S_n:=\displaystyle\sum_{k=0}^n \dfrac{4^k}{4^k+5^k}$$ The question asks to express the sum in terms of $n$ then to deduce the limit of $\dfrac{S_n}{n+1}$. I tried to use the following sum as an auxiliary sum $$T_n:=\displaystyle\sum_{k=0}^n\dfrac{5^k}{4^k+5^k}$$ noticing that $S_n + T_n = n+1$. Any thoughts about this ? thanks.

  • 4
    $\begingroup$ I don't know how to express the sum in terms of $n$, but I am a little confused. Don't we have $\lim_{n\to\infty}S_n=\sum_{k=0}^\infty\frac{4^k}{4^k+5^k}\leq\sum_{k=0}^\infty(\frac45)^k=\frac1{1-\frac45}=5$ and hence $\lim_{n\to\infty}\frac{S_n}{n+1}=0$? $\endgroup$ – SmileyCraft Dec 16 '18 at 12:49
  • 1
    $\begingroup$ Also, Wolfram Alpha gives a closed form involving the $q$-digamma function, which I've never heared of, and I doubt you have. What makes you think there is a closed form of the sum? $\endgroup$ – SmileyCraft Dec 16 '18 at 12:54
  • $\begingroup$ Thanks, the first part of the question in a paper (which I doubt is wrong) asks to express $S_n$ in terms of $n$ before computing the limit of the quotient $\dfrac{S_n}{n+1}$ $\endgroup$ – Oussama Sarih Dec 16 '18 at 13:00
  • 2
    $\begingroup$ @OussamaSarih, can you link to the paper, or reproduce verbatim what it says about expressing $S_n$? Given that the first three $S_n$'s are $1/2$, $17/18$, and $985/738$, I'd be a little surprised to see any kind of simple closed formula. (Even well written papers make occasional mistakes.) $\endgroup$ – Barry Cipra Dec 16 '18 at 13:04
  • $\begingroup$ It's a homework sheet in french for highschool, I'll post a screenshot as soon as I get from one from the students who asked me to solve it. $\endgroup$ – Oussama Sarih Dec 16 '18 at 15:39

As hinted by SmileyCraft, I cannot think of any simple closed form to express $S_n$.

My best guess is to use Big $\mathcal{O}$ notation. As you thought, $S_n = n+1 - T_n$ then write $T_n = \sum_{k=0}^n u_k$ with $u_k = \frac{1}{1+\left(\frac{4}{5}\right)^k}\underset{k\to\infty}{=}1 + \mathcal{O}\left(\frac{4}{5}\right)^k$. Now $\left(\frac{4}{5}\right)^k$ is a positive real sequence thus sommable in Big $\mathcal{O}$ notation.

This yields: $$ S_n \underset{n\to \infty}{=} n+1 - \left[(n+1) + \mathcal{O}(1) \right] = \mathcal{O}(1) $$

since $\sum_{k=0}^n \left(\frac{4}{5}\right)^k \underset{n\to \infty}{=} \mathcal{O}(1)$.

Then $\frac{S_n}{n+1} = \mathcal{O}(\frac{1}{n+1})$.

Note that it is very similar and perfectly equivalent to what SmileyCraft does but it does give you an expression of $S_n$ (though trivial) depending on $n$.


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.