Central limit theorems for multivariate semi-Markov sequences and processes, with applications (Q4935336)

Let \(\{(J_k, Y_k), k\geq 1\}\) be a semi-Markov sequence with state space \(\{1,2,\dots, n\}\times R\) and semi-Markov kernel \(F(y)= \{F_{i,j}(y)\}\), where \(F_{i,j}(y)= P(J_k= j, Y_k\leq y\mid J_{k- 1}= i)\), \(1\leq i, j\leq n\), and \(\{J_k\}\) is an irreducible and aperiodic Markov chain. For such a process the following CLT holds \[ {1\over \sqrt k\sigma} \Biggl(\sum^k_{i= 1} Y_i- km\Biggr)@> D>> N(0,1),\quad\text{as }k\to\infty, \] where \(\sigma\) and \(m\) are known constants depending on \(F(y)\). The author generalizes the above CLT for a multivariate semi-Markov sequence \(\{(J_k,{\mathbf Y}_k)\}\) where \({\mathbf Y}_k= (Y_{k,1}, Y_{k,2},\dots, Y_{k,p})\), \(k\geq 1\), are random vectors. Next, the author applies that theorem to multivariate stationary renewal processes and to counting processes associated with Markov renewal processes.