A stochastic Tikhonov theorem in infinite dimensions (Q2502189)

The authors consider a two-scale system of stochastic differential equations in Hilbert space with one fast and one slow component. Under certain natural assumptions the authors prove that, as the time scales separate, the slow variable converges uniformly in probability on the given time interval \([0,T]\), while the fast variable converges uniformly in probability on each compact subinterval of \((0,T]\), i.e. away from the origin. This is an infinite-dimensional stochastic version of Tikhonov's theorem, which can be applied to stochastic partial differential equations of the diffusion-reaction type. A main difference with the proofs in the finite-dimensional case is that the authors only deal with mild solutions and rely on a factorization method for Ornstein-Uhlenbeck-type processes.