The level matrix of a tree and its spectrum

Abstract

Given a rooted tree $T$ with vertices $u_1,u_2,\ldots,u_n$, the level matrix $L(T)$ of $T$ is the $n \times n$ matrix for which the $(i,j)$-th entry is the absolute difference of the distances from the root to $v_i$ and $v_j$. This matrix was implicitly introduced by Balaji and Mahmoud~[{\em J. Appl. Prob.} 54 (2017) 701--709] as a way to capture the overall balance of a random class of rooted trees. In this paper, we present various bounds on the eigenvalues of $L(T)$ in terms of other tree parameters, and also determine the extremal structures among trees with a given order. Moreover, we establish bounds on the mutliplicity of any eigenvalue in the level spectrum and show that the bounds are best possible. Furthermore, we provide evidence that the level spectrum can characterise some trees. In particular, we provide an affirmative answer to a very recent conjecture on the level energy (sum of absolute values of eigenvalues).