Global divergence of spatial coalescents (Q2428502)

Spatial \(\Lambda\)-coalescent processes are investigated on a locally finite graph \(G\). These processes can be considered as a model of ancestral lineage merging for a spatially segregated population in which the spatial motion of individuals must be taken into account. In this model stochastically independent \(\Lambda\)-coalescence takes place at each site in \(G\) and particles move as independent continuous random walks on \(G\). For a process starting from \(n\) particles the authors derive lower and upper asymptotic bounds (in probability) for the number \(N(t)\) of the particles in the moment \(t\) as \(n\to\infty\). It is shown that \(N(t)\to\infty\) as \(n\to\infty\) but in a very slow rate, e.g., as \(\sim(\log^* n)^d\) for \(G=R^d\), where \(\log^*\) means the inverse tower function. The behavior of spatial Beta-coalescent processes is also considered.