It\^{o}-Stratonovich Conversion in Infinite Dimensions for Unbounded, Time-Dependent, Nonlinear Operators

We prove that a solution, in a variational framework, to the Stratonovich stochastic partial differential equation with noise $G\left(t, \Psi_t\right) \circ dW_t$ is given by a solution to the It\^{o} equation with It\^{o}-Stratonovich corrector $\frac{1}{2}\sum_{i=1}^\infty D_uG_i\left(t, \Psi_t\right)\left[G_i(t,\Psi_t)\right]dt$. Here $G_i$ denotes the action of $G$ on the $i^{th}$ component of the cylindrical noise, and $D_uG_i$ its Fr\'{e}chet partial derivative in the Hilbert space for which the It\^{o} form is satisfied. The noise operator $G$ may be time-dependent, nonlinear, and unbounded in the sense of differential operators; in the latter case, one must pass to a larger space in order to solve the Stratonovich equation. Our proof relies on martingale techniques, and the results apply to fluid equations with time-dependent and nonlinear transport noise.