1. We want to know the posterior distribution $P(\theta)$ and where modes are, this is the goal. 2. But we cannot calculate $P(\theta)$ analytically, this is the problem. 3. However we can build a Markov Chain. 4. Sampling from the Markov Chain builds the histogram that approximates $P(\theta)$, this is the solution. [![enter image description here][1]][1] Surprisingly: * Prior can be an arbitrary normal distribution, although the choice impacts. * Proposal can be arbitrary normal distribution whose mean is current $\theta $. The questions and confusions may be: 1. How come arbitrary normal distribution can be used as the prior? 2. How come $q(\theta_i | \theta_j) * min(1, \frac { prior(\theta_j * L(E|\theta_j) * q(\theta_j | \theta_i)} { prior(\theta_j * L(E|\theta_j) * q(\theta_i | \theta_j)} ) $ can be equivalent with $P(\theta_j|\theta_i)$? [![enter image description here][2]][2] [1]: https://i.sstatic.net/pTTV1.jpg [2]: https://i.sstatic.net/RIzto.jpg