The impact of geometry on monochrome regions in the flip Schelling process (Q2088882)

Using probability- and graph-theoretical approaches, the authors consider the model of self-organized segregation of agents of two types placed on a graph. In a nutshell, the model assumes that an agent tends to change his type depending on the predominant type of his neighborhood. For random geometric and Erdos-Rényi graphs, general results are shown for the expected fraction of monochrome edges (i.e. with similar-type vertices) of a graph. Ability of joint neighborhood of two vertices to push them towards a similar type in the process of segregation is linked to a concept of decisiveness that is formalized as the excess of the number of majority-type vertices over the minority type. It is shown that greater decisiveness of joint neighborhood of two vertices as compared to decisiveness of their exclusive neighborhoods leads to elevated probability of the two vertices falling into the same type. Further, assuming a simple probabilistic model when initial types of vertices are randomly set with probabilities \(1/2\) for each of the two types, it is shown that decisiveness is related to the sheer size of the neighborhood. Formal results are supplemented by simulations. For random geometric graphs, where the vertices are connected depending on their spatial proximity and, therefore, common neighborhoods are easily formed, it is shown that segregation is likely to occur. For the Erdos-Rényi graphs lacking spatial structure, on the other hand, the model eventually leads to a monochrome-edges state.