The moderate deviations principle for the trajectories of compound renewal processes on the half-line (Q2058319): Difference between revisions

In this paper the authors consider i.i.d. random vectors \((\tau _{i},\zeta _{i}),i\geq 2\) and independent of it a random vector \((\tau _{1},\zeta _{1})\) . Now consider the sums \(T_{n}=\sum_{i=1}^{n}\tau _{i}\) and \( Z_{n}=\sum_{i=1}^{n}\zeta _{i}\) and the random process \(\nu (t)=\max \left\{ k\geq 0:T_{k}<t\right\} \) with \(\nu (0)=0\). Among others, a compound renewal process is defined as \(Z(t)=Z_{\nu (t)}\) with \(Z(0)=0\). The authors suppose that Cramer's condition holds for both \((\tau _{1},\zeta _{1})\) and \((\tau _{2},\zeta _{2})\) and then consider \(Z_{1}(t)=Z(t)-at\) and \( Z_{2}(t)=Z(t)-a_{\zeta }\nu (t)\), where \(a_{\zeta }=E\zeta _{2}\), \(a_{\tau }=E\tau _{2}\) and \(a=a_{\zeta }/a_{\tau }\). Now the authors fix a function \( x(T)\) so that \(\lim_{T\rightarrow \infty }x(T)/\sqrt{T}=\infty \) and \( \lim_{T\rightarrow 0}x(T)/\sqrt{T}=0\). The main object of the paper is to study the processes \(z_{1,T}(t)=Z_{1}(tT)/X(T)\) and \( z_{2,T}(t)=Z_{2}(tT)/x(T)\). There are some technical lemmas involved in proving the main results of the paper.