Pseudo-random Graphs

Abstract

truly random graphs. He also supplied several examples of pseudo-random graphs and discussed many of their properties. Thomason's papers undoubtedly defined directions of future research for many years.Another cornerstone contribution belongs to Chung, Graham and Wilson [26] who in 1989 showed that many properties of different nature are in certain sense equivalent to the notion of pseudo-randomness, defined using the edge distribution. This fundamental result opened many new horizons by showing additional facets of pseudo-randomness.Last years brought many new and striking results on pseudo-randomness by various researchers. There are two clear trends in recent research on pseudo-random graphs. The first is to apply very diverse methods from different fields (algebraic, linear algebraic, combinatorial, probabilistic etc.) to construct and study pseudo-random graphs. The second and equally encouraging is to find applications, in many cases quite surprising, of pseudo-random graphs to problems in Graph Theory, Computer Science and other disciplines. This mutually enriching interplay has greatly contributed to significant progress in research on pseudo-randomness achieved lately.The aim of this survey is to provide a systematic treatment of the concept of pseudo-random graphs, probably the first since the two seminal contributions of Thomason [79], [80]. Research in pseudo-random graphs has developed tremendously since then, making it impossible to provide full coverage of this subject in a single paper. We are thus forced to omit quite a few directions, approaches, theorem proofs from our discussion. Nevertheless we will attempt to provide the reader with a rather detailed and illustrative account of the current state of research in pseudo-random graphs.Although, as we will discuss later, there are several possible formal approaches to pseudorandomness, we will mostly emphasize the approach based on graph eigenvalues. We find this approach, combining linear algebraic and combinatorial tools in a very elegant way, probably the most appealing, convenient and yet quite powerful.This survey is structured as follows. In the next section we will discuss various formal definitions of the notion of pseudo-randomness, from the so called jumbled graphs of Thomason to the (n, d, λ)-graphs defined by Alon, where pseudo-randomness is connected to the eigenvalue gap. We then describe several known constructions of pseudo-random graphs, serving both as illustrative examples for the notion of pseudo-randomness, and also as test cases for many of the theorems to be presented afterwards. The strength of every abstract concept is best tested by properties it enables to derive. Pseudo-random graphs are certainly not an exception here, so in Section 4 we discuss various properties of pseudo-random graphs. Section 5, the final section of the paper, is devoted to concluding remarks. Definitions of pseudo-random graphsPseudo-random graphs are much more of a general concept describing some graph theoretic phenomenon than of a rigid wel...

Non-Backtracking Random Walks Mix Faster

“…This notion was introduced by the first author in the 80's, motivated by the fact that if λ is much smaller than d, then the graph has strong pseudo-random properties. We mention a few of the properties of these graphs, and refer the readers to [11] for an extensive survey of the subject. Let G = (V, E) denote an (n, d, λ)-graph.…”

A Study of Tabu Search for Coloring Random 3-Colorable Graphs Around the Phase Transition

“…While many classes of random graphs exist (Bollobás, 2001;Krivelevich & Sudakov, 2006), we focus our study on three well-known classes of 3-colorable graphs: Uniform, equipartite, and flat.…”

k-Wise Independent Random Graphs

“…For our purposes, α-jumbledness means that (as expected in G(N, p) graphs) for all vertex-sets U, V , the number of edges that pass from U to V should be p|U ||V | ± α |U ||V |. Jumbledness and quasirandomness had been studied extensively (see [31] and its many references), and serve in Graph Theory as the common notion of resemblance to random graphs. In particular, G(N, p) graphs are known to exhibit (the best possible) jumbledness parameter, α = Θ( √ pN ).…”