| statdistr {DTMCPack} | R Documentation |
Computing Stationary Distribution
Description
This function computes the stationary distribution of a markov chain (assuming one exists) using the formula from proposition 2.14.1 of Resnick: pi=(1,...1)(I-P+ONE)^(-1), where I is an mxm identity matrix, P is an mxm transition matrix, and ONE is an mxm matrix whose entries are all 1. This formula works well if the number of states is small, but since it directly computes the inverse of the matrix, it is not tractable for larger matrices. For larger matrices 1/E(FPTime(n)) is a rough approximation for the long run proportion of time spent in a state n.
Usage
statdistr(tmat)
Arguments
tmat |
Markov chain transition matrix, must be a square matrix and rows must sum to 1. |
Value
Returns a stationary distribution: mxm matrix which represents the long run percentage of time spent in each state.
Author(s)
Will Nicholson
References
Resnick, "Adventures in Stochastic Processes"
Examples
data(hh)
statdistr(hh)