Recall the aggregate model with transition matrix:
![]()
where

and C changes depending on the state. That is,
from j=1 to m.
Here, each column sum in the transition matrix must be one.

where we can see a little difference in each point. It depends on the parameters and
transition matrix but gaps tend to be wider as the number of rows in transition matrix
becomes larger. The movement of K depends on the random seed in common.