Good upper bound for $\sum_{n=1}^N a^{-n} n^{-b}$, for $a, b \in (0, 1]$

176 Views Asked by At

Let $a, b \in (0, 1]$ and define $S_N:=\sum_{n=1}^N a^{-n} n^{-b}$.

Question

What is a good upper bound for $S_N$ ?

Observations

By a simple (and probably careless) application of Cauchy-Schwarz, one can already get

$$ S_N \le (\max_{n=1}^N a^{-n})\sum_{n=1}^N n^{-b} \le \frac{1}{a}\begin{cases}\frac{1}{1-b}(N^{1-b}-b),&\mbox{ if }0 \le b < 1\\ \log(N) + 1,&\mbox{ if }b=1. \end{cases} $$ See this post for details.

I was wondering whether one can do much better than this, and using what kind of tools.


Edit

One of the users have suggested trying to look around incomplete gamma functions. Indeed, according to wolfram alpha, $$\int_{1}^{N} a^{-t}t^{-b}dt = \log^{b-1}(a)\left(\Gamma(1-b,\log(a))-\Gamma(1-b,N\log(a))\right), $$ where $\Gamma(1-a,x)$ defines an incomplete gamma function.

I'm wondering whether there are any upper bounds for such an expression.