Use this URL to cite or link to this record in EThOS:
Title: Self-organised criticality in random graph processes
Author: Yeo, Dominic
ISNI:       0000 0004 6500 5988
Awarding Body: University of Oxford
Current Institution: University of Oxford
Date of Award: 2016
Availability of Full Text:
Access from EThOS:
Full text unavailable from EThOS. Please try the link below.
Access from Institution:
In the first half of this thesis, we study the random forest obtained by conditioning the Erdös-Rényi random graph G(N,p) to include no cycles. We focus on the critical window, in which
p(N) = 1+λN-1/3
, as studied by Aldous for G(N,p). We describe a scaling limit for the sizes of the largest trees in this critical random forest, in terms of the excursions above zero of a particular reflected diffusion. We proceed by showing convergence of the reflected exploration process associated to the critical random forests, using careful enumeration of classes of forests, and the asymptotic properties of uniform trees. In the second half of this thesis, we study a random graph process where vertices have one of k types. An inhomogeneous random graph represents the initial connections between vertices, and over time new edges are added homogeneously, as in the classical random graph process. Each vertex is frozen at some rate, resulting in the removal of its entire component. This is a version of the frozen percolation model introduced by R\'ath, which (under mild conditions) exhibits self-organised criticality: the dynamics first drive the system to a critical state, and from then on maintain it in criticality. We prove a convergence result for the proportion of vertices of each type which survive until time t, and describe the local limit in terms of a multitype branching process whose parameters are critical and given by the solution to an unusual differential equation driven by Perron--Frobenius eigenvectors. The argument relies on a novel multitype exploration process, leading to a concentration result for the proportion of types in all large components of a near-critical inhomogeneous random graph; and on a stronger convergence result for mean-field frozen percolation, when the initial graphs may be random.
Supervisor: Goldschmidt, Christina ; Martin, James Sponsor: Engineering and Physical Sciences Research Council
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral
EThOS ID:  DOI: Not available