The Smith normal form of distance matrices of high dimensional trees
arXiv:2510.04471
Abstract
Graham-Lovász-Pollak \cite{GL,GP} obtained the celebrated formula for the determinant of the distance matrix for any tree with vertices. Later, Hou and Woo \cite{HW} extended this formula to the Smith normal form (SNF) obtaining that $\SNF({\sf D}(T_{n+1}))={\sf I}_2\oplus 2{\sf I}_{n-2}\oplus [2n]$, for any tree with vertices. A -{\it tree} is either a complete graph on vertices or a graph obtained from a smaller -tree by adjoining a new vertex together with edges connecting it to a -clique. If and are -cliques in a -tree , a -{\it walk} between and is a finite sequence , where , , and the -cliques and are incident to the same -clique . For , the -{\it distance} from the -cliques and is the number of -cliques in a minimum -walk from and , and is denoted by $\dist^d(τ,τ')$. Let denote the number of -cliques in the -tree . Then the -distance matrix of the -tree is the matrix, indexed by the -cliques of , such that the -entry is if , and $\dist^d(τ_i,τ_j)$ otherwise. Here, we show that, for and fixed, the SNF of the -distance matrix is the same for any -tree with vertices. Specifically, for any -tree with vertices such that , the Smith normal form of is which extends Graham-Lovász-Pollak and Hou-Woo results.