Fisher informations and local asymptotic normality for continuous-time quantum Markov processes

TL;DR

This paper establishes that the output states of irreducible quantum Markov processes become approximately Gaussian for large times, enabling explicit calculation of quantum Fisher information and classical Fisher information for parameter estimation.

Contribution

It introduces a local asymptotic normality framework for quantum Markov processes and derives explicit formulas for quantum and classical Fisher informations in this setting.

Findings

Output states approximate quantum Gaussian states asymptotically

Explicit expression for quantum Fisher information in terms of the Markov generator

Additive statistics of measurements satisfy local asymptotic normality

Abstract

We consider the problem of estimating an arbitrary dynamical parameter of an quantum open system in the input-output formalism. For irreducible Markov processes, we show that in the limit of large times the system-output state can be approximated by a quantum Gaussian state whose mean is proportional to the unknown parameter. This approximation holds locally in a neighbourhood of size $t^{-1/2}$ in the parameter space, and provides an explicit expression of the asymptotic quantum Fisher information in terms of the Markov generator. Furthermore we show that additive statistics of the counting and homodyne measurements also satisfy local asymptotic normality and we compute the corresponding classical Fisher informations. The mathematical theorems are illustrated with the examples of a two-level system and the atom maser. Our results contribute towards a better understanding of the statistical and probabilistic properties of the output process, with relevance for quantum control engineering, and the theory of non-equilibrium quantum open systems.