Weak convergence to the coalescent in neutral population models (Q4935338)

For a large class of discrete time models the ancestry of a sample of \(n\) individuals (or gens) from the current population having total size \(N\) can be traced back, under suitable time-scaling, when \(N\to\infty\); see, e.g., the author's paper, ibid. 35, No. 2, 438-447 (1998; Zbl 0913.60022), and references therein. Having established in that work the convergence of finite-dimensional distributions of the \textit{ancestral} process to those of the \textit{n-coalescent} (i.e., a certain-continuous time Markov chain), the author shows that under the same assumptions the weak convergence in \(D_E([0,\infty))\) occurs. Here \(E\) is the set of all equivalence relations on \(\{1,\ldots,n\}\) and \(D_E([0,\infty))\) is the set of all E-valued cadlag functions on \([0,\infty)\). This result, besides the classical exchangeable models, covers models with deterministic variable populations and possibly non-exchangeable offspring variables. It is used to approximate hitting probabilities of the time-scaled ancestral process. A certain type of mutation process can also be imposed upon the genealogical process.