WorksMathematics
Large deviation principles for component counts in the dense regime
Abstract
This paper develops large deviation principles for component counts in dense random geometric graphs. We consider a homogeneous Poisson point process on the flat -dimensional torus and study the number of isolated components of fixed order at connection radii above the connectivity scale. Two types of large deviation behaviour are investigated. First, we prove a Donsker–Varadhan large deviation principle in the regime where the expected number of such components tends to infinity. Second, we study the opposite extreme regime, where this expectation tends to zero, and establish fixed order tail measure limits for the associated -component isolation-volume point process. As a consequence, we obtain functional large deviations for the moving threshold component count process.