Stochastic differential equations

When a system is acted upon by exterior disturbances, its time-development can often be described by a system of ordinary differential equations, provided that the disturbances are smooth functions. But for sound reasons physicists and engineers usually want the theory to apply when the noises belong to a larger class, including for example “white noise.” If the integrals in the system derived for smooth noises are reinterpreted as Itô integrals, the equations make sense; but in nonlinear cases they often fail to describe the time-development of the system. In this paper (extending previous work of the author) a calculus is set up for stochastic systems that extends to a theory of differential equations. When the equations are known that describe the development of the system when noises are smooth, an extension to the larger class of noises is proposed that in many cases gives results consistent with the smooth-noise case and also has “robust” solutions, that change by small amounts when the noises undergo small changes. This is called the “canonical” extension.Nevertheless, there are certain systems in which the canonical equations are inappropriate. A criterion is suggested that may allow us to distinguish when the canonical equations are the right choice and when they are not.