Pathwise uniqueness for a class of SDE in Hilbert spaces and applications (Q982497)

The moral of the paper is to prove that partial differential equations, which may have non-unique solutions, have unique solutions provided they are perturbed by a suitable noise. A stochastic partial differential equation \(dX_t=AX_t\,dt+F(t,X_t)\,dt+B(t,X_t)\,dt+\sqrt Q\,dW_t\) in a Hilbert space is considered with a Lipschitz non-linearity \(F\), a Hölder non-linearity \(B\), a selfadjoint bounded operator \(Q\) and a cylindrical Wiener process \(W\). The main result of the paper states that pathwise uniqueness holds for the equation above provided several technical hypotheses upon the entries are fulfilled and this (positive) result is further applied on a parabolic equation for which pathwise uniqueness is eventually shown to fail if there is no noise. A rather analytic proof is based on a regularity result for a non-homogeneous infinite dimensional Kolmogorov equation.