Stochastic reaction-diffusion equations driven by jump processes (Q1650762): Difference between revisions

An equation \[ du(t)+Au(t)\,dt=F(t,u(t))\,dt+\int_ZG(t,u(t);z)\tilde\eta(dz,dt),\quad u(0)=u_0 \] in a separable UMD type \(p\) Banach space is considered where \(p\in(1,2]\), \(A\) is a linear, positive, closed and densely defined operator with the bounded imaginary power property, \(-A\) is assumed to generate a compact analytic contraction semigroup, the non-linearities \(F\) and \(G\) are separately continuous, \(G\) is bounded, \(F\) is dissipative with a polynomial growth and can be approximated by a sequence of bounded, separately continuous functions that are uniformly dissipative with the same polynomial growth and \(\tilde\eta\) is a compensated Poisson random measure on a measure space \((Z,\mathcal Z)\) with the intensity measure \(\nu\otimes\lambda\) (where \(\lambda\) denotes the Lebesgue measure on \(\mathbb R\)). The authors provide sufficient assumptions of technical nature that guarantee existence of a martingale mild solution to the equation above, which has cadlag paths in a fractional domain of \(A\). The problem and the existence result is then reformulated for equations driven by Lévy processes in a separable Banach space, i.e. the driving compensated Poisson random measure \(\tilde\eta\) is replaced by a Lévy process \(L\) in the equation. The existence results are consequently applied to particular reaction-diffusion equations with polynomial drifts driven by \(\alpha\)-stable Lévy processes, and to equations with properly elliptic operators of an arbitrary order (or their fractional powers), non-linear but bounded continuous drift and diffusion coefficients and driven by space-time Poissonian white noises.