Jämförelse av olika metoder att generera Bernoullifördelade slumptal givet deras summa
In this master?s thesis the problem of simulating conditional Bernoulli distributed stochastic variables, given the sum, is considered. Three simulation methods are considered, namely the acceptance/rejection technique, Bondesson?s method and the Markov chain Monte Carlo method.To compare the three methods the bias and the standard deviations of the simulated variables are evaluated. The results of the simulation study shows that the Markov chain Monte Carlo method is not the best method for this type of simulation. Both the other methods were quite suitable for the task. The acceptance/rejection technique is a little bit more time consuming than Bondesson?s method, but on the other hand the acceptance/rejection technique is easier to implement.