Bayesian Inversion and Sampling Methods

Statistic makes thinking a way through probabilistic formulation. As an alternative solution for many inverse problems, statistical inversion seeks for an ensemble of solutions instead of a unique one via a sampling process. The probabilistic equation is governed by the rule of Bayesian inference. In this chapter, we will introduce those fundamental concepts including Bayesian inference, Markov chain Monte Carlo method, as well as Metropolis-Hastings random-walk algorithm. Following these, we will show the magic that connects an inverse problem with Bayesian formula. One simple example will be presented to demonstrate the power of statistical inversion and the use on directional EM LWD data.