Let $l$ be a positive integer. Does the matrix $$ M_l \ := \ \left( \binom{l(2p+1)}{j} \right)_{0\leq p,j \leq[(l1)/2]} $$ have nonzero determinant?
3

$\begingroup$ Have you searched for "Vandermonde determinant"? $\endgroup$ – Jason Starr Feb 21 '15 at 16:32

$\begingroup$ Yes, but Vandermonde determinant didn't solve the problem. I found a paper that solves the problem but to a much more general case. I would like a more simpler solution, I believe it exists. $\endgroup$ – Liliam Feb 21 '15 at 16:40

1$\begingroup$ @JasonStarr How do you reduce this to Vandermonde? Just curious :) $\endgroup$ – Alex Degtyarev Feb 21 '15 at 16:50

4$\begingroup$ @AlexDegtyarev. For elements $a_0,\dots,a_{n1}$ in a commutative ring $R$, the Vandermonde determinant $\text{det}(a_i^j)_{0\leq i,j\leq n1}$ equals $0!\cdot 1!\cdots (n1)!$ times the determinant $\text{det}(\binom{a_i}{j})_{0\leq i,j\leq n1}$. Thus, the determinant above is, up to a product of factorials, a Vandermonde determinant. $\endgroup$ – Jason Starr Feb 21 '15 at 17:53

$\begingroup$ @JasonStarr: thanks, this is nice. For proof, I guess, this is the interpolation by Newton "monomials" rather than conventional ones, right? $\endgroup$ – Alex Degtyarev Feb 21 '15 at 18:52
6
$\begingroup$
$\endgroup$
There is a nice result of Gessel and Viennot that computes your determinant in terms of NE lattice paths. The original paper is available here. Aigner and Ziegler also give a nice exposition of this result in their Proofs From the Book.

$\begingroup$ Thank you Chris. The paper has a very general result, I wanted some more direct thing. I'll look in the book. $\endgroup$ – Liliam Feb 21 '15 at 17:33

$\begingroup$ @Liliam, I think if you take $k=\left\lfloor\frac{\ell1}{2}\right\rfloor$, $a_i=2i$ and $b_j=j$ in their Theorem 1, then you get something pretty close to your determinant. $\endgroup$ – Chris McDaniel Feb 21 '15 at 18:12

$\begingroup$ Chris, the book solved my problem, thank you so much. $\endgroup$ – Liliam Feb 26 '15 at 0:02