Solutions of affine stochastic functional differential equations in the state space (Q2425810)

This article is concerned with the solutions of affine stochastic functional differential equations on \(\mathbb{R}^d\). In this case the drift of these equations is specified by a functional defined on a general semi-normed function space \(B\), which can be very specific in examples and thus is only described axiomatically. As usual the author reformulates the solutions are as stochastic processes in the space \(B\), but now his aim is to avoid any lifting tequiques onto topologically more favorable ones because the behaviour on the original space gets irrevocably lost. By representing such a process in the bidual space of \(B\) as Pettis integral, the author establishes that the associated transition functions of this process form a generalized Gaussian Mehler semigroup on \(B\). Hence the process is completely characterized there due to its Markovianity. In the sequel a sufficient and necessary condition for the underlying space B such that the transition functions are even an Ornstein-Uhlenbeck semigroup. This result is finally exploited to associate an abstract Cauchy problem in \(B\) to the stochastic functional differential equation.