The total external length of the evolving Kingman coalescent

TL;DR

This paper studies the asymptotic behavior of the total external length in the evolving Kingman coalescent, showing convergence to a stationary Gaussian process as the number of leaves grows large.

Contribution

It establishes the Gaussian limit process for the external length of the evolving Kingman coalescent under a specific time scaling, linking it to internal lengths and a branching process coupling.

Findings

External length process converges to a Gaussian process

Covariance function of the limit is explicitly derived

Internal lengths behave asymptotically as multivariate Gaussian

Abstract

The evolving Kingman coalescent is the tree-valued process which records the time evolution undergone by the genealogies of Moran populations. We consider the associated process of total external tree length of the evolving Kingman coalescent and its asymptotic behaviour when the number of leaves of the tree tends to infinity. We show that on the time-scale of the Moran model slowed down by a factor equal to the population size, the (centred and rescaled) external length process converges to a stationary Gaussian process with almost surely continuous paths and covariance function $c(s,t)=\Big( \frac 2 {2+|s-t|} \Big)^2$. A key role in the evolution of the external length is played by the internal lengths of finite orders in the coalescent at a fixed time which behave asymptotically in a multivariate Gaussian manner (see Dahmer and Kersting (2015)). A coupling of the Moran model with a critical branching process is used. We also derive a central limit result for normally distributed sums endowed with independent random coefficients.