3
$\begingroup$

I want to compute/prove that the following sum is positive: $$ \sum_{i = 0}^n \left[\frac{D(n - i, i)}{d} \sum_{j = m}^d s(d, j) \binom{j}{m} (d - i)^{j - m}\right] > 0 $$ where $s(d, j)$ is the stirling number of the first kind, $D(n - i, i)$ is the Delannoy number, $d = 2 n + 2\geq 4$ is an even number, and $0 \leq m \leq d$. The first step could be computing either $$ a_i = \sum_{j = m}^d s(d, j) \binom{j}{m} (d - i)^{j - m} $$ which seems to be a big positive number when $i = 0$ and then alternates signs as i increases. I suspect that $a_i$ could be obtained by exponential generating functions, or that it could be the coefficient of a Fourier transform as in this question, but I haven't found it.

Or compute: $$ d_j = \sum_{i = 0}^n \frac{D(n - i, i)}{d} (d - i)^{j - m} $$ Intuitively, a closed formula is unlikely so some sophisticated estimation would do the job, but I'm quite inexperienced in that. Thanks!

$\endgroup$
3
  • $\begingroup$ Why divide by $d$ if $d$ is constant? $\endgroup$ Commented May 10 at 6:15
  • $\begingroup$ @FedorPetrov, it's unnecessary, I just wrote it as how it came up in the first place $\endgroup$
    – Zhi Wang
    Commented May 10 at 11:32
  • $\begingroup$ That's a typo, thanks for pointing it out! $\endgroup$
    – Zhi Wang
    Commented May 11 at 2:59

0

You must log in to answer this question.

Browse other questions tagged .