0
$\begingroup$

This is a cross-post from https://math.stackexchange.com/questions/1174998/estimating-an-unusual-infinite-sum, which didn't get any useful answers

I encountered the following unusual type of infinite sum: $$ S = \sum_{k=1}^{\infty} (a (1 - b^{\frac{1}{k-1}} c))^k$$ where here $a,b,c \in [0,1]$ and $a (1 - c) < 1$. (Interpret the summand when $k = 1$ as simply $a$)

This sum converges, since as $k \rightarrow \infty$ the summand converges to $(a (1-c))^k$. I want to find an upper bound for it. I'm having trouble pinning down its behavior.

An estimate up to a constant multiple would be fine with me.

Thanks for the help!

Improve this question
$\endgroup$
5
  • $\begingroup$ Did you try approximating it by an integral? $\endgroup$
    – Dima Pasechnik
    Mar 9, 2015 at 19:57
  • $\begingroup$ Apparently this does not have a closed form integral. $\endgroup$
    – David Harris
    Mar 9, 2015 at 21:03
  • $\begingroup$ Why not just bound the difficult term between 1 and (1-c)? Are you looking for a really nice function of b to use to sandwich between the two simple bounds? $\endgroup$
    – The Masked Avenger
    Mar 9, 2015 at 22:17
  • $\begingroup$ Alternarively, choose epsilon and find k_0 so that for all larger k, b^{1/k} is within epsilon of 1. Then you are a finite sum away from an upper bound involving (1 - (1- epsilon)c), which might serve your purpose. $\endgroup$
    – The Masked Avenger
    Mar 9, 2015 at 22:23
  • $\begingroup$ In the integral, I'd try to replace the part that causes the absence of a "nice" form by a simpler upper bound, e.g. taking first few terms of the Taylor expansion. $\endgroup$
    – Dima Pasechnik
    Mar 10, 2015 at 8:10

0

Reset to default

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Browse other questions tagged or ask your own question.