摘要
It is known that nearly uncoupled irreducible stochastic matrices must possess sub-dominant eigenvalues near λ=1. It is nature to ask whether the converse is true. Hortfieland Meyer [2] gave a positive answer. They introduced the notion of uncoupling measureof Stochastic matrices. For an n×n stochastic matrix P the uncoupling measure of P is de-fined as σ（p）=min((sum from i∈M<sub>1</sub>,j∈M<sub>1</sub>(P<sub>ij</sub>))+(sum from i∈M<sub>1</sub>,j∈M<sub>1</sub>(P<sub>ij</sub>)), where the minimum is taken over
基金
Supported by the National Natural Science Foundation of China
