Stochastic Simulation - with an Application of the Metropolis-Hastings Algorithm in the Strauss Mode

Specialeforsvar ved Sebastian Nielsen

Titel: Stochastic Simulation - with an Application of the Metropolis-Hastings Algorithm in the Strauss Mode

Abstract: The topic and title of this Master's thesis is Stochastic Simulation. We will introduce and justify a variety of simulations methods including inversion, acceptance-rejection sampling, Monte Carlo integration and importance sampling. On the foundation of Markov chain theory we will introduce Markov chain Monte Carlo methods. The Metropolis-Hastings algorithm and the Gibbs sampler will provide several examples of Markov chain Monte Carlo methods. The existence of a stationary distributions and ergodicity of the corre-sponding Markov chains will be proven for a selection of these methods. As an application we will implement a version of the Metropolis-Hastings algorithm in the Strauss model. Simulations will be performed to investigate properties of the parameters in the Strauss model

Vejleder: Anders Rønn-Nielsen
Censor:    Lars Nørvang Andersen, Aarhus Universitet