Regularly varying correlation functions and KMO-Langevin equations (Q1365130)

The author considers the linear stochastic delay equation \[ \dot X=-\int_{-\infty}^t \gamma(t-s)\dot X(s) ds+\alpha\dot B(t), \] where \(\alpha>0\), \(\dot B\) is a Gaussian white noise, and \(\gamma\colon(0,\infty)\to(0,\infty)\) is a function of the form \(\gamma(t)=\int_0^\infty \exp\{-t\lambda\} d\rho(\lambda)\), \(t>0\), where \(\rho\) is a Borel measure on \((0,\infty)\) such that \(\int_0^\infty(\lambda+1)^{-1} d\rho(\lambda)<\infty\). This equation is a variant of the so-called first KMO--Langevin equation introduced by Y. Okabe. In particular, the existence and uniqueness of a stationary solution is established for all \(\alpha>0\) and all measures \(\rho\) from a certain subclass \(C\). The main results of the paper, however, consist in a precise description of the asymptotic behavior of the correlation function \(R(t)=E[X(t)X(0)]\) in terms of \(\alpha\) and the asymptotic behavior of the kernel function \(\gamma\).