Optimal validity region for the Lorentzian polynomial criterion

Determine the exact set of positive integer pairs (a,b) for which the Hessian of \(\partial_x f\), where \(f(x_1,x_2,x,y)=(ax+y)[(bx+x_1+y)(bx+x_2+y)+xy]\), is Lorentzian, and identify conditions under which \(f\) itself is Lorentzian beyond the sufficient cases \(a=1\) and \(a=b\).

Background

Lemma 3.2 proves that the polynomial f(x1,x2,x,y)=(ax+y)[(bx+x1+y)(bx+x2+y)+xy]f(x_1,x_2,x,y)=(ax+y)[(bx+x_1+y)(bx+x_2+y)+xy] is Lorentzian when a=1a=1 or a=ba=b.

The authors explicitly state that these hypotheses are sufficient but not necessary. They report numerical evidence that the Hessian is Lorentzian for (a,b)=(3,2)(a,b)=(3,2) and (4,2)(4,2), but not for (2,1)(2,1) or (5,2)(5,2), and leave the exact region of validity unresolved.

References

The hypotheses of Lemma~\ref{p3.2} are sufficient but not necessary, and the exact region of validity appears to be complicated. For example, numerical experiments suggest that the Hessian of $\partial_xf$ is Lorentzian for $(a,b)=(3,2)$ and $(4,2)$, but not for $(2,1)$ or $(5,2)$. Thus, the conditions $a=1$ and $a=b$ are not expected to be optimal.

— Lorentzian polynomials and log-concavity of the independence polynomials of graphs  (2609.37553 - Liu et al., 29 Sep 2026) in Remark following Lemma 3.2 in Section 3