Lyapunov exponents of linear stochastic functional differential equations. II: Examples and case studies (Q1370227)

Being a sequel to its first part [Ann. Inst. Henri Poincaré, Probab. Stat. 32, No. 1, 69-105 (1996; Zbl 0862.60044)], the present paper is aimed to examine examples of linear stochastic functional differential equations (e.g., \(dx(t)= \int_{[-r,0]} x(t+s) d \nu(s) dW(t)\)) in two distinct classes: regular [see the first author, in: Diffusion processes and related problems in analysis, Vol. II: Stochastic flows. Prog. Probab. 27, 141-169 (1992; Zbl 0763.60030)] and singular, depending on whether stochastic flows exist or not. In the regular cases, i.e., the one-dimensional linear delay equation with Poisson noise \[ dx(t)= \biggl\{ \nu_1 x(t)+ \mu_1 x(t-r)+ \int_{-r}^0 x(t+s) \sigma_1 (s)ds \biggr\} dt + \biggl\{ \nu_2 x(t)+ \int_{-r}^0 x(t+s) \sigma_2 (s)ds \biggr\} dM(t), \] questions of existence of the stochastic semiflow, its Lyapunov spectrum and upper bounds on the top Lyapunov exponent are studied in a general scheme. For singular equations \[ dx(t)= \sigma x(t-r) dw(t), \qquad t >0, \;(x(0),x_0)=(v, \eta) \in R \times L^2([-r,0],R), \] \[ dx(t)= \sigma\Bigl( \int_{[-1,0]} x(t+s)d \nu(s)\Bigr) dW(t), \qquad t \geq 0, \;\sigma \in R \setminus\{0\},\;\eta\in C([-1,0],R), \] estimates on the maximal exponential growth rate are obtained.