Parameter dependent rough SDEs with applications to rough PDEs

Abstract

In this paper we generalize Krylov's theory on parameter-dependent stochastic differential equations to the framework of rough stochastic differential equations (rough SDEs), as initially introduced by Friz, Hocquet and L\^e. We consider a stochastic equation of the form $$ dX_t^\zeta = b_t(\zeta,X_t^\zeta) \ dt + \sigma_t(\zeta,X_t^\zeta) \ dB_t + \beta_t (\zeta,X_t^\zeta) d\mathbf{W}_t,$$ where $\zeta$ is a parameter, $B$ denotes a Brownian motion and $\mathbf{W}$ is a deterministic H\"older rough path. We investigate the conditions under which the solution $X$ exhibits continuity and/or differentiability with respect to the parameter $\zeta$ in the $\mathscr{L}$-sense, as defined by Krylov. As an application, we present an existence-and-uniqueness result for a class of rough partial differential equations (rough PDEs) of the form $$-du_t = L_t u_t dt + \Gamma_t u_t d\mathbf{W}_t, \quad u_T =g.$$ We show that the solution admits a Feynman--Kac type representation in terms of the solution of an appropriate rough SDE, where the initial time and the initial state play the role of parameters.