Entropy, invertibility and variational calculus of adapted shifts on Wiener space (Q1039419): Difference between revisions

Suppose \(\left( W, H, \mu \right)\) is an abstract Wiener space, and \(L\) be a positive random variable whose expectation with respect to \(\mu\) is \(1\). The article is devoted to finding necessary and sufficient conditions for \(L\) to be represented as the Radon--Nikodym derivative of the image of the Wiener measure under an adapted perturbation of identity. One of the tools the author is using is the notion of the innovation process for a stochastic process from the filtering theory. Then he proves that the innovation conjecture holds if and only if the original process is almost surely invertible. The article also gives a variational characterization of the invertibility of perturbations of the identity and the representability of a positive random variable. The latter means that there is a map \(U :W \to W\) such that \(\frac{dU \mu}{d\mu} = L\). In particular, he shows that an adapted perturbation of identity \(U=IW+u\) satisfying the Girsanov theorem, is invertible if and only if the kinetic energy of \(u\) is equal to the entropy of the measure induced with the action of \(U\) on the Wiener measure \(\mu\). In addition, the relation with the Monge--Kantorovitch measure transportation is studied. An application of these results to a variational problem related to large deviations is also given.