A graph G is said to be determined by its generalized spectra (DGS for short) if for any graph H, H and G are cospectral with cospectral complements implies H is isomorphic to G. In Wang (2017) [14], the author gave a simple condition for a graph to be DGS. More precisely, let G be a n-vertex graph with adjacency matrix AG, and let W=[e,AGe,…,AGn−1e] be the walk-matrix of G, where e is the all-one vector. A theorem of Wang states that if 2−⌊n/2⌋detW is odd and square-free, then G is DGS.However, the above theorem fails for regular graphs. In this paper, we first generalize the notion of DGS by introducing a new matrix associated with graph G, which is essentially a rank-one perturbation of the adjacency matrix of G and is obtained by introducing a graph-vectorξG associated with G. For a fixed graph-vector ξG, define Φ(G)=det(λI−(AG+tξGξGT)). A graph G is determined byΦ(G) (Φ-DS for short) if for any graph H, Φ(G)=Φ(H) implies that H is isomorphic to G. It is not difficult to show that if ξG is the all-one vector, then G is Φ-DS if and only if it is DGS. Thus, the notion of Φ-DS generalizes that of DGS in a natural way. Moreover, we give a simple arithmetic condition for a large family of regular graphs to be Φ-DS, which is analogous to the aforementioned theorem. Numerical experiments are also presented to illustrate the effectiveness of our results.
Read full abstract