On the spectral radius of the (L,\kappa)-lazy Markov chain

From MaRDI portal
Publication:6428200




Abstract: We consider an (L,kappa)-lazy operation on an irreducible Markov transition probability P with state space S where LsubsetS and kappain[0,1). For each xinL and yinS, this (L,kappa)-operation replaces P(x,y), the transition probability from x to y, by kappa1x=y+(1−kappa)P(x,y). We are interested in how L and kappa influence the spectral radius hokLappa of this new transition probability. We first show that hokLappa is non-decreasing and continuous in kappa. We then show that: (1) If L is nonempty and finite, then P being rho-transient is equivalent to that the growth of displaystyle(hokLappa)kappain[0,1) exhibits a phase transition: There exists a critical value kappac(L)in(0,1) such that kappamapstohokLappa is a constant on [0,kappac(L)] and increases strictly on [kappac(L),1); (2) For every kappain(0,1), if SsetminusL is nonempty and finite, then hokLappa=hokSappa if and only if P is not strictly rho-recurrent.














This page was built for publication: On the spectral radius of the $(L,\kappa)$-lazy Markov chain

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q6428200)