Asymptotic behaviour of sum of decreasing definite integrals

109 Views Asked by At

I would like to calculate:

\begin{equation*}g(K, T) = \displaystyle \sum_{k=1}^{K} \sum_{t = 1}^{T} \int_{0}^{1} \left(1 - z^k\right)^t \, dz. \end{equation*}

If no closed form solution exists, I would like to find a tight upper bound. Trivially we have

\begin{equation*} g(K,T) = O(KT), \end{equation*}

but we can do better than this. One can show that

\begin{equation*}\displaystyle \int_{0}^{1} \left(1 - z^k\right)^t \, dz < \left(1 - \left(\frac{1}{2}\right)^\frac{1}{t}\right)^\frac{1}{k}, \end{equation*}

and further that

\begin{equation*} 1 - \left(\frac{1}{2}\right)^{\frac{1}{t}} \le \frac{1}{t}, \end{equation*}

which can be combined and fiddled with to provide the improved bound,

\begin{equation*} g(K,T) = O(KT^{1-\frac{1}{K}}). \end{equation*}

Can we do better than this?

1

There are 1 best solutions below

0
On

The integral over $z$ can be reduced to a Beta function or Euler Integral of the First Kind ${\rm B}\left(x,y\right)$: $$ \int_{0}^{1}\left(1 - z^{k}\right)^{t}\,{\rm d}z = {1 \over k}{\rm B}\left({1 \over k}, t + 1\right) $$

For the time being, we don't see a further 'nice' reduction.