A general stochastic maximum principle for singular control problems.
A stochastic integral equation corresponding to a probability space is considered. This equation plays the role of a dynamical system in many problems of stochastic control with the control variable . One constructs stochastic processes , connected with a Markov chain and with the space . The expected values of (i = 1,2) are respectively the expected value of an integral representation of a solution x(t) of the equation and that of its derivative .
We prove, by means of Malliavin calculus, the convergence in of some properly renormalized weighted quadratic variations of bi-fractional Brownian motion (biFBM) with parameters and , when and .