When is a stochastic integral a time change of a diffusion? (Q912482)

Let \(Y_ t\) denote an Ito diffusion. The author obtains necessary and sufficient conditions for a time change of an n-dimensional stochastic Ito integral \(X_ t\) to have the same law as \(Y_ t\). This is a generalization of a similar problem arising in filtering theory, in that the innovation process is a Brownian motion w.r.t. the observation \(\sigma\)-field. Conditions for a function \(\phi\) to map an Ito diffusion into a time change of Ito diffusion is explicitly given in terms of the generators of the diffusions. Applications to conformal martingales in \({\mathbb{C}}^ n\) are also discussed. Results from semi-martingale theory are utilized. The method of this paper allows the time changes to have jumps.