Improved bounds for a condition number for Markov chains (Q1434425)

Let \(P\) and \(\widetilde P=P+E\) be transition matrices of some finite (\(n\) states) homogeneous ergodic Markov chains, \(\pi\) and \(\widetilde\pi\) be their stationary distributions. The article is devoted to the condition number \(\kappa_4\), i.e. the constant for which \[ |\pi-\widetilde\pi|_\infty\leq\kappa_4|E|_\infty. \] New upper bounds for \(\kappa_4\) are obtained in terms of \(Q=(q_{ij})_{i,j=1}^n=I-P\) and its group inverse \(Q^{\#}\). E.g. \[ \kappa_4\leq { \delta_2+\sigma_2\delta_3+\dots+\sigma_{n-2}\delta_{n-1}+\sigma_{n-1} \over \chi }, \] where \[ \sigma_k=\max_{j_1,j_2}(q_{j_2,j_1}+q_{j_2,j_2})\cdots \max_{j_1\dots j_k}(q_{j_k,j_1}+\dots+q_{j_k,j_k}), \] \[ \delta_k=\max_{j_1\dots j_k} {\prod_{i=1}^n q_{ii}\over q_{j_1j_1}\dots q_{j_kj_k}}, \chi=\prod_{\lambda_j\not=1}(1-\lambda_j), \] \(\lambda_j\) are eigenvalues of \(P\). Numerical results show that these bounds are approximately two times better than the estimate of \textit{C. D. Meyer} for \(n=10\) [SIAM J. Matrix. Anal. Appl. 15, No.~3, 715--728 (1994; Zbl 0809.65143)].