Perron-Frobenius Theorem: Markov Chain -> Matrices

244 Views Asked by At

I am interested in finding out a way how to transform the stochastic results of perron-frobenius for markov chains to any matrix. I am aware that perron-frobenius was originally proofed with linear algebra for non-negative matrices and the results are easily applicable to markov chains. However, I am interested in the other direction. How can you generalise the result that a finite positive recurrent markov chain has a unique left eigenvector - to any matrix? My only constraint is that a markov chain has a row sum = 1 and matrices obviously have no such contraints. Best regards, Carl