Rough path stability of (semi-)linear SPDEs (Q2447287)

The authors consider the stochastic partial differential equation \[ du+L(t, x, u, Du, D^2u)dt=\sum_{k=1}^d\Lambda_k(t, x, u, Du)\circ dZ_t^k \] with scalar initial data \(u(0, .)=u_0(.)\) on \(\mathbb R^n\) in which \(L\) is a semi-linear second-order operator of the form \[ L(t, x, r, p, X)= -\text{Tr}[A(t,x). X] + b(t,x) . p + c(t,x, r) \] and \(\Lambda_k\) is a collection of first-order differential operators which are affine linear in \(r\) and \(p\). The aim of the paper is to give meaning to this equation when a semi-martingale \(Z\) is replaced by a rough path. The main results are the existence of a unique solution in the space of bounded uniformly continuous functions on \([0,T]\times \mathbb R^n\) which is of course obtained by the locally uniform limit of a sequence of viscosity solutions, stability and continuity with respect to initial data. The resulting theory of rough PDEs can then be used to give meaning and then existence, uniqueness and stability to a large class of stochastic partial differential equations. By combining well-known Wong-Zakai-type results of the \(L^2\)-theory of SPDEs with convergence of piecewise linear approximations to enhanced Brownian motion, the authors show that the solutions provided by the above theory when applied to this type of Brownian motion, are in fact usual \(L^2\)-solutions of variational approach. The usual way to deal with the above equation is to find solutions in a suitable functional analytic setting such that solutions involve in suitable Sobolev spaces. The authors also discuss the equivalence of this solution concept with the rough PDE approach.