A counterexample and a threshold theorem for graphs with two distinct eigenvalues
Rattachement africain : ir, us. Niveau de preuve : code pays fourni par la source.
Le résumé fourni par la source
For a graph G on n vertices, q(G) denotes the minimum number of distinct eigenvalues among the real symmetric matrices whose off-diagonal zero pattern is prescribed by G. Barrett, Fallat, Furst, Nasserasr, Rooney and Tait (arXiv:2411.12917; ELA 42 (2026)) conjectured (their Conjecture 5.5) that if e(complement(G)) <= n-2 then q(G) = 3 precisely when complement(G) is a double star together with an isolated vertex, and q(G) = 2 otherwise. We show that G = K_{1,1,3} is a counterexample: its complement K_3 u 2K_1 has 3 = n-2 edges and is not a forest, yet q(K_{1,1,3}) = 3. We then prove that within the range e(complement(G)) <= n-2 the known obstruction produces exactly two families of graphs with q > 2, namely the double stars S_{a,b} u K_1 and the single graph K_3 u 2K_1; this yields the amended conjecture and explains why no obstruction of this kind exists below the threshold n-3 of the authors' Conjecture 1.1. On the constructive side we prove that every graph G with e(complement(G)) = n-2 whose complement is non-bipartite with a connected non-trivial component, other than K_{1,1,3}, satisfies q(G) = 2, exhibiting a bipartite/non-bipartite asymmetry at the threshold: the exceptions are an infinite bipartite family against a single non-bipartite graph. The counterexample and an infinite non-bipartite family attaining q = 2 have been formally verified in Lean 4 / Mathlib.
Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.
Le contrôle bibliographique ouvert
Les institutions déclarées
Une affiliation ne permet pas de déduire la nationalité d’un auteur.