Chutes and ladders in Markov chains (Q5952048): Difference between revisions

Let \(\{X_n, n\geq 0\}\) be an irreducible Markov chain with finite state space \({\mathcal X}\) and transition probability \(p(x,y)\), \(x,y\in{\mathcal X}\). Fix a state \(w\) and define a new Markov chain \(\{\overline X_n, n\geq 0\}\) with transition probability \(\overline p(x,y)= p(x,y)\), \(x\neq w\). It is considered the question how much can be stationary distribution be perturbed by altering one row of the transition probability. Examples of random walks on a finite discrete torus, the Bessel random walks and random walks on the Sierpiński gasket are discussed in detail. Limiting behaviour of suitably rescaled stationary distributions is also obtained.