| Popis: |
The spread of influence on social networks can be modeled by the activation of vertices in a random graph model. Each edge is associated with a probability of an activated vertex activating its neighbor in each update. When the probability associated with each edge does not depend on direction, a natural question is, does this local symmetry indicate a global symmetry, i.e. the influence of any pair of vertices on each other is symmetric after arbitrary time. We define a simple criterion to capture this global symmetry, and prove that this criterion is met with a method based on random matrices. |
