Solving equations with semimartingale noise (Q2121580)

Consider the equation: \[ g(t)=f^{\prime \prime}(t)+X^{\prime}(t) f^{\prime}(t), \quad t \in[0,1], \] where \(f\) is unknown function, \(g\) is given and \(X\) represents some potential, which may not be differentiable. In this paper, the authors deal with this equation when \(X\) is a continuous semimartingale, that is a martingale plus a process with finite variation trajectories. This generalizes the case of coefficients being the white noise. To solve the equation the authors use the idea of finding explicitly the inverse of the ill-posed differential operator: \[ L f(t)=f^{\prime \prime}(t)+X^{\prime}(t) f^{\prime}(t), \] which boils down to finding the associated Green kernel. To find the kernel the authors give explicitly two homogeneous solutions in terms of the Dolean-Dade exponential. Moreover, several examples with explicit calculations are given.