|
||
Title: Markov Chain Visitations Post by william wu on Feb 4th, 2003, 3:21am A problem I was discussing with a friend a few hours ago; thought you guys might be interested. I think it's originally from topcoder.com. The solution algorithm may seem surprisingly simple. Markov Chain Visitations You are given a Markov Chain M, defined by a set of states S and a transition probability matrix Pij. Write an algorithm that computes the average number of times you will visit a certain subset of states S', if you move k steps starting from an initial state v. (In other words, if you start at v and start walking k steps according to the probabilities given by Pij, on average how many times will you hit a vertex in the set S'?) |
||
Title: Re: Markov Chain Visitations Post by towr on Feb 4th, 2003, 7:32am ...[hide] From what I remember from the last time I used them, it might be the sum of all incoming probabilities (given that all transitions together sum to one) for every state in the subset. Assuming v is random.. [/hide]... |
||
Powered by YaBB 1 Gold - SP 1.4! Forum software copyright © 2000-2004 Yet another Bulletin Board |