9
$\begingroup$

A set of positive integers $d_1, \dots, d_n$ describe two n-dimensional closed lattice tetrahedron: $$ T = \left\{ (x_1, \dots, x_n) \in \mathbb{R}^n: \sum_{i=1}^n \frac{x_i}{d_i} \leq 1 \textrm{ and all } x_i \geq 0 \right\}.$$ Let $L_T(t)$ be the Ehrhart polynomial of $T$, i.e. the number of integer lattice points in $tT$.

When are all coefficients of $L_{T}(t)$ positive?

Improve this question
$\endgroup$
9
  • 1
    $\begingroup$ Let me take a look at the case $n=2$ to get an idea of what one might expect. In this case the triangle $tT$ in question has vertices $$(td_1,0),(0,td_2),(0,0).$$ Pick's formula gives the number of points as $${\rm Area}+\frac 12\sharp({\rm pts ~on~the~boundary})+1 =\frac 12t^2 d_1d_2 + \frac 12t(\gcd(d_1,d_2)+d_1+d_2)+1 $$ so the coefficients are all positive. $\endgroup$
    – Lev Borisov
    Feb 20, 2016 at 20:18
  • $\begingroup$ To add some information: 1. Pommersheim has given an explicit formula for $n=3$ based on Dedekind sums. I can post the polynomial on request. 2. In the case that $d_1 = \dots = d_n$, then the coefficients are the absolute value of the Stirling numbers of the first kind and therefore always positive. $\endgroup$
    – Jiro
    Feb 21, 2016 at 8:16
  • $\begingroup$ @LevBorisov: Thanks for the start! Any idea how to proceed from there? This paper of Beck gives a description of all coefficients via the Residue Theorem, however I cannot derive anything about the coefficient signs. $\endgroup$
    – Jiro
    Feb 25, 2016 at 8:39
  • 3
    $\begingroup$ $T_2$ does not have integer vertices, so it does not have an Ehrhart polynomial. Let $M$ be the least common multiple of $d_1,\dots,d_n$. You could ask for each $0\leq j<M$ whether the polynomial $p_j(t)=L_{T_2}(tM+j)$ has positive coefficients. $\endgroup$
    – Richard Stanley
    Feb 25, 2016 at 13:38
  • 1
    $\begingroup$ I agree with Lev that this is a difficult problem, already for $n=3$. As Sebastian mentioned, one can compute these polynomials through Dedekind sums (and there generalized versions, for $n \ge 5$), but those are delicate... $\endgroup$
    – matthias beck
    Feb 25, 2016 at 18:21

0

Reset to default

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.