Laws of large numbers for dynamical systems with randomly matched individuals (Q1201149)

``Biologists and economists have analyzed populations where each individual interacts with randomly selected individuals'' and, for very large populations, have approximated such ``complicated stochastic systems'' with deterministic systems based on averages, invoking the law of large numbers (example: random mating and Hardy-Weinberg law). This is not always correct (example is given) and ``depends on properties of the law of motion''. ``This paper gives a characterization of random mating schemes for countably infinite populations'' with several types of individuals, which is a quite hard problem in probability very skillfully handled; in particular, it shows that there is a scheme ``such that the stochastic system and the deterministic system are the same''. It also shows that, for a countably infinite population, a random matching scheme must depend on the assignment of type in the population if the probability of an individual being matched to one belonging to a subpopulation equals the proportion of the subpopulation over the whole population. This problem does not arise in finite populations. The paper shows, by examining the behaviour of the process for finitely many time periods and large enough populations, that there is a matching scheme such that the deterministic process provides good approximation for the stochastic process (upper bound on deviation provided!). However, in case of finite populations and infinite number of time periods, an example shows this is not always the case.