A representation free quantum stochastic calculus (Q1188108)

The purpose of quantum stochastic calculus is the behaviour of quantum systems driven by the usual interactions of quantum physics, and by a ``quantum noise''. In classical probability, one first studied the Wiener (Brownian) noise and the corresponding stochastic differential equations, then the Poisson noises, then martingale and semimartingale noises. In quantum probability the role of Brownian and Poisson processes together is taken up by the creation, annihilation and number operator families on the Boson Fock space over \(L^ 2(\mathbb{R}_ +)\), and the creation and annihilation operators over Fermion Fock space have provided another much studied example. The problem in this paper is to describe generally what a ``noise'' should be, to build a theory of stochastic integration and stochastic differential equations for such noises, and to give a unitarity criterion for solutions of s.d.e.'s. The set-up consists of a (complex, separable) Hilbert space \({\mathcal H}\) and an increasing family of von Neumann algebras \({\mathcal A}_ t\) of operators on \({\mathcal H}\). The ``noises'' will be adapted operator processes, i.e. families of operators \(M_ t\), possibly unbounded, with a common dense domain \({\mathcal D}\) on which their adjoints also act. These processes must be measurable and adapted to the family \({\mathcal A}_ t\) in a suitable sense (essentially, they commute with the commutant \({\mathcal A}_ t'\) of \({\mathcal A}_ t)\). Among these processes are both the integrators (noises) and the integrands of the theory. One of the difficulties is the necessity of defining both right and left stochastic integrals. The integrators are ``additive processes'' satisfying suitable inequalities, of the kind which allow to decompose the process into a ``martingale'' plus a ``drift term''. The theory of stochastic integrals and s.d.e.'s is good enough to include the known results on the existence and unitarity of s.d.e.'s with bounded coefficients with respect to Boson noise, due to Hudson-Parthasarathy. A new noise described by A. Boukas is also shown to be included.