4
Abstract
In recent years, the scientifical community has succeeded in experimentally building
simple quantum systems on which series of measurements are successively acquired along
quantum trajectories, without any reinitialization of their state (density operator) by
the physicist. The subject of this thesis is to adapt the quantum tomography techniques
(state and parameters estimation) to this frame, to take into account the feedback of the
measurement on the state, the decoherence and experimental imperfections.
During the measurement process, the evolution of the quantum state is then governed
by a hidden-state Markov process (Belavkin quantum filters). Concerning continuous-time
measurements, we begin by showing how to discretize the stochastic master equation,
while preserving the positivity and the trace of the quantum state, and so reducing to
discrete-time quantum filters. Then, we develop, starting from trajectories of discrete-
time measurements, some maximum-likelihood estimation techniques for initial state and
parameters. This estimation is coupled with its confidence interval. When it concerns the
value of parameters (quantum process tomography), we provide a result of robustness
using the formalism of particular filters, and we propose maximization technique based
on the calculus of gradient by adjoint method, well adapted to the multi-parametric case.
When the estimation concerns the initial state (quantum state tomography), we give an
explicit formulation of the likelihood function thanks to the adjoint states, show that its
logarithm is a concave function of the initial state and build an intrinsic expression of
the variance, obtained from asymptotic developments of bayesian means, lying on the
geometry of the space of density operators.
These estimation techniques have been applied and experimentally validated for two
types of quantum measurements : discrete-time non-destructive measurements of photons
in the group of cavity quantum electro-dynamics of LKB at Collège de France, diffusive
measurements of the fluorescence of a supra-conducting qubit in the quantum electronics
group of LPA at ENS Paris.
Keywords
Quantum tomography ; Quantum filters ; Stochastic master equation ; Superconducting
qubit ; Cavity quantum electrodynamics ; Maximum-likelihood ; Bayesian estimation ;
Gradient calculus by adjoint method ; Asymptotic expansion of Laplace’s integrals