chapter  2
18 Pages


The acceptance/rejection (AR) method was the first major protocol for perfect simulation. The basic method gives a way of sampling from a random variable given some condition on that variable. When used with an auxiliary variable, this method can be extended to sampling from any bounded density. Other bounds on probabilities such as Markov’s inequality and Chernoff bounds can actually be viewed as AR algorithms, and these bounds assist in simulating from rare events.