Parametric randomization, complex symplectic factorizations, and quadratic-exponential functionals for Gaussian quantum states
For researchers in quantum control and filtering, this provides a new computational tool for performance criteria in quantum risk-sensitive problems.
The paper develops a method for computing quadratic-exponential functionals (QEFs) for Gaussian quantum states by combining complex symplectic factorizations with parametric randomization, reducing QEFs to exponential moments of classical Gaussian random variables. This enables recursive computation of such moments.
This paper combines probabilistic and algebraic techniques for computing quantum expectations of operator exponentials (and their products) of quadratic forms of quantum variables in Gaussian states. Such quadratic-exponential functionals (QEFs) resemble quantum statistical mechanical partition functions with quadratic Hamiltonians and are also used as performance criteria in quantum risk-sensitive filtering and control problems for linear quantum stochastic systems. We employ a Lie-algebraic correspondence between complex symplectic matrices and quadratic-exponential functions of system variables of a quantum harmonic oscillator. The complex symplectic factorizations are used together with a parametric randomization of the quasi-characteristic or moment-generating functions according to an auxiliary classical Gaussian distribution. This reduces the QEF to an exponential moment of a quadratic form of classical Gaussian random variables with a complex symmetric matrix and is applicable to recursive computation of such moments.