The graph spectrum of barycentric refinements (1508.02027v1)
Abstract: Given a finite simple graph G, let G' be its barycentric refinement: it is the graph in which the vertices are the complete subgraphs of G and in which two such subgraphs are connected, if one is contained into the other. If L(0)=0<L(1) <= L(2) ... <= L(n) are the eigenvalues of the Laplacian of G, define the spectral function F(x) as the function F(x) = L([n x]) on the interval [0,1], where [r] is the floor function giving the largest integer smaller or equal than r. The graph G' is known to be homotopic to G with Euler characteristic chi(G')=chi(G) and dim(G') >= dim(G). Let G(m) be the sequence of barycentric refinements of G=G(0). We prove that for any finite simple graph G, the spectral functions F(G(m)) of successive refinements converge for m to infinity uniformly on compact subsets of (0,1) and exponentially fast to a universal limiting eigenvalue distribution function F which only depends on the clique number respectively the dimension d of the largest complete subgraph of G and not on the starting graph G. In the case d=1, where we deal with graphs without triangles, the limiting distribution is the smooth function F(x) = 4 sin2(pi x/2). This is related to the Julia set of the quadratic map T(z) = 4z-z2 which has the one dimensional Julia set [0,4] and F satisfies T(F(k/n))=F(2k/n) as the Laplacians satisfy such a renormalization recursion. The spectral density in the d=1 case is then the arc-sin distribution which is the equilibrium measure on the Julia set. In higher dimensions, where the limiting function F still remains unidentified, F' appears to have a discrete or singular component.
Collections
Sign up for free to add this paper to one or more collections.
Paper Prompts
Sign up for free to create and run prompts on this paper using GPT-5.