Classical and generalized solutions of fractional stochastic differential equations (Q2219503)

Let \(\beta\in(0,1)\), let \(f\) be a real-valued function on \([0,+\infty)\), define \[ \partial^\beta f(t)=\frac{1}{\Gamma(1-\beta)}\frac{d}{dt}\int_0^t(t-s)^{-\beta}(f(s)-f(0+))\,ds \] and consider an ordinary stochastic equation \[ \partial^\beta X(t)=aX(t)+f(t)+\partial^\gamma\int_0^t(\sigma X(s)+g(s))\,dw(s),\qquad X(0)=x_0 \] where \(\beta,\gamma\in(0,1]\), \(a,\sigma\in\mathbb R\), \(f,g\) are real-valued functions and \(w\) is a standard Brownian motion. The authors study the well-possedness and long time behaviour for classical and generalized solutions of the above equation under several sets of assumptions on \(\beta\), \(\gamma\), \(f\), \(g\), \(a\) and \(\sigma\), they provide explicit solutions (where existence holds) and classify the solutions in the weighted chaos spaces and Gaussian Volterra processes. In particular, the cases of the time fractional Ornstein-Uhlenbeck processes and the time fractional geometric Brownian motions are covered. In the last section, the well-posedness of a stochastic partial differential equation \[ \partial^\beta u(t,x)+b\Lambda^\alpha u(t,x)=\sigma\partial^\gamma\int_0^t\Lambda^\nu u(s,x)\,dw(s), \] where \(b>0\), \(\sigma\in\mathbb R\), \(\beta,\gamma\in(0,1]\), \(\alpha,\nu\in(0,2]\) and \(\Lambda=(-\Delta)^{1/2}\) is studied in \(L_2(\mathbb R^d)\).