中文

The Narayana transformation

组合数学 2026-07-02 v1

摘要

For mZ0m\in\mathbb{Z}_{\geq 0}, let Nn,m(x)=2F1(n,nm;m+1;x), N_{n,m}(x)={}_2F_1(-n,-n-m;m+1;x), which specializes to the Narayana polynomials of types BB and AA for m=0m=0 and m=1m=1, respectively. We prove that the associated basis transformation TNm(k=0nakxk)=k=0nakNk,m(x) T_{N_m}\left(\sum_{k=0}^n a_kx^k\right)=\sum_{k=0}^n a_kN_{k,m}(x) maps every real-rooted polynomial with nonnegative coefficients to a real-rooted polynomial. The proof is based on the rectangular additive convolution of polynomials. We then apply this result to products of lower triangular matrices and obtain a general criterion ensuring that their row generating functions remain real-rooted. As consequences, we recover this property for powers and products of several classical triangular matrices, including Pascal's triangle, the Stirling triangles, and the Narayana triangles of types AA and BB. We conclude with conjectures concerning the squares of the Eulerian and Delannoy triangles.

引用

@article{arxiv.2607.01572,
  title  = {The Narayana transformation},
  author = {Jianxi Mao and Lijie Wang},
  journal= {arXiv preprint arXiv:2607.01572},
  year   = {2026}
}

备注

12 pages