Markov Chain simulation

难道算不出来？

come on. not everything is as simple as what you learn from the book, where
an MC only have a couple of states and usually the stationary state can be
analytically derived. In practice, Monte-Carlo simulation might be the only
way to obtain the stationary state of an MC.

记得有个德国人编写的软件（网上发布）可以计算出来。只要给出转移条件和状态空间，稳态分布就可以计算出来。

When the state space is so large that it is exponential in some dimension
parameters, counting the whole space is a very hard problem. In mathematics,
it is categorized into so called #P problem (the counting problem corresponding to NP). Therefore, if you could give the state space to the German guy,
you would have already solved NP, which would make you deserve the highest
math award in the world :)