January 12th, 2007, 6:10 am
to add to what wheeb said, differentiability is generally proven as finding a number say M, so that as dt->0 abs(df/dt-M)<e where e is a some arbitrarily chosen small real number.but for f=w, it turns out there is no M, because dw=N(0,1)*sqrt(dt) and so dw/dt=N(0,1)/dt