2022/06/15 by Michael Färber, Farber, Michael, Lewin Strauss +1
Computer Science · Mathematics · #05C48 #05C50 #05C63 #05C81 #Combinatorics (math.CO) #FOS: Mathematics #Graph theory and applications #Markov Chains and Monte Carlo Methods #Topological and Geometric Data Analysis
paper · pdf · doi:10.48550/arxiv.2206.07688
openalex publication_date 2022/06/15 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28
In this paper we study spectra of Laplacians of infinite weighted graphs. Instead of the assumption of local finiteness we impose the condition of summability of the weight function. Such graphs correspond to reversible Markov chains with countable state spaces. We adopt the concept of the Cheeger constant to this setting and prove an analogue of the Cheeger inequality characterising the spectral gap. We also analyse the concept of the dual Cheeger constant originally introduced in \citeB14, which allows estimating the top of the spectrum. In this paper we also introduce a new combinatorial invariant, k(G,m), which allows a complete characterisation of bipartite graphs and measures the asymmetry of the spectrum (the Hausdorff distance between the spectrum and its reflection at point 1∈ \Bbb R). We compare k(G, m) to the Cheeger and the dual Cheeger constants. Finally, we analyse in full detail a class of infinite complete graphs and their spectra.