The number of tetrahedra sharing the same metric invariants via symbolic and numerical computations (Q2048154)

It was asked by Marcin Mazur as a problem in the American Mathematical Monthly whether there can be several non-congruent tetrahedra with the same volume \(V\), circumradius \(R\), and facet areas \(A_1\), \(A_2\), \(A_3\), and \(A_4\). This question was answered positively in [\textit{P. Lisoněk} and \textit{R. B. Israel}, in: Proceedings of the 2000 international symposium on symbolic and algebraic computation, ISSAC '00, St. Andrews, Scotland, GB, 2000. New York, NY: ACM Press. 217--219 (2000; Zbl 1326.68361)] and an infinite family of tetrahedra with the same values for \(V\), \(R\), and \(A_1\) to \(A_4\) was given in [\textit{L. Yang} and \textit{Z. Zeng}, in: Proceedings of the 2005 international symposium on symbolic and algebraic computation, ISSAC'05, Beijing, China, July 24--27, 2005. New York, NY: ACM Press. 362--364 (2005; Zbl 1352.51018)]. However, this family of tetrahedra is such that \(A_2\), \(A_3\), and \(A_4\) coincide and a proof was announced in 2013 by Lu Yang and Zhenbing Zeng that, when \(A_1\), \(A_2\), \(A_3\), and \(A_4\) are pairwise distinct, there cannot be more than eight non-congruent such tetrahedra. This proof was published by \textit{Z. Zeng} et al. [in: Proceedings of the 44th international symposium on symbolic and algebraic computation, ISSAC '19, Beijing, China, July 15--18, 2019. New York, NY: Association for Computing Machinery (ACM). 363--370 (2019; Zbl 1467.52019)] and an alternative proof of the same result by \textit{Y.-L. Tsai} [J. Symb. Comput. 77, 162--174 (2016; Zbl 1405.51013)]. This paper gives more details on the proof by Zhenbing Zeng, Lu Yang, and Lydia Dehbi [loc. cit]. A degree nine polynomial \(R_1\) whose roots contained in \(]-1,1[\) each correspond to at most one tetrahedron with volume \(V\), circumradius \(R\), and facet areas \(A_1\) to \(A_4\) is derived explicitly. It is shown that this polynomial is non-negative in \(-1\) and \(1\), and as a consequence that it has at most eight roots in \(]-1,1[\). The authors further report numerical computations and discuss whether \(R_1\) might have less than eight roots in this interval.