Minimal quasi-stationary distributions under null \(R\)-recurrence (Q5936982): Difference between revisions

Let us consider a Markov chain \(\{X_n\}_{n\in\mathbb{N}}\) or simply \(\{X_n\}\) on the state space \(S= \mathbb{N}\) with transition matrix \(P= (P_{ij})_{i,j\in S}\). It is considered that there exists a proper subset \(B\subset S\) of absorbing states, and let \(B^c\) be its complementary set of transient states. Now \(P^n_{ij}\) is the notation for transient probability \(P(X_n= j\mid X_0= i)\), \(i,j\in S\). The behavior of the chain before absorption is given by the matrix \(P_{B^c}= (P_{ij})_{i,j\in B^c}\). The purpose of the authors is to obtain conditions under which an \(R\)-null chain with finite mass solution \(x\) to the systems \(xP_{B^c}={1\over R} x\), \(P_{B^c} y={1\over R} y\) has \(i\)-LCD (LCD is referred to as limiting conditional distribution) for every \(i\in B^c\). Some auxiliary results are proved in Section 2, where they obtain sufficient conditions for the probability distribution \(\mu\) being the minimal quasi-stationary distribution (QSD). The main result is proved in Section 3 (Theorem 1.1 which is explained previously). The paper is a good and very interesting study.