A random forest process with a variable number of giant components in the threshold window
Given a graph G, and an ordering π of its vertices, a permutation forest F(G,π) is a spanning forest of G whose components are obtained as follows. For each vertex v, connect v to its first neighbour w in G that appears after v in the ordering π. If we regard this edge (v,w) as directed forward, from v to w, then each vertex has at most one forward edge, and the components of F(G,π) are arborescences. The roots of the components formed by this process are those vertices of G with no forward edge in the ordering π. This paper shows that the permutation forests of the random graphs G<sub>n,p</sub> have a threshold for the emergence of a linear size component around p=1/n. In contrast to the w.h.p. emergence of a unique giant in G<sub>n,p</sub>, the permutation forest process has the unusual property that, with positive probability, a number of linear size components occur within the threshold window.
Authors
- Colin Cooper (ORCID: https://orcid.org/0000-0002-5264-4401)
- Tomasz Radzik (ORCID: https://orcid.org/0000-0002-7776-5461)
Publication Details
- Journal
- Discrete Applied Mathematics
- Published
- 2025-12-17
- DOI
- https://doi.org/10.1016/j.dam.2025.12.030
- Primary Topic
- Stochastic processes and statistical mechanics
- Type
- article
- Field-Weighted Citation Impact
- 0.00
Funders
- Engineering and Physical Sciences Research Council