
Results concerning the construction of quantum Bayesian error regions as a
means to certify the quality of parameter point estimators have been reported
in recent years. This task remains numerically formidable in practice for large
dimensions and so far, no analytical expressions of the region size and
credibility (probability of any given true parameter residing in the region)
are known, which form the two principal region properties to be reported
alongside a point estimator obtained from collected data. We first establish
analytical formulas for the size and credibility that are valid for a uniform
prior distribution over parameters, sufficiently large data samples and general
constrained convex parameterestimation settings. These formulas provide a
means to an efficient asymptotic error certification for parameters of
arbitrary dimensions. Next, we demonstrate the accuracies of these analytical
formulas as compared to numerically computed region quantities with simulated
examples in qubit and qutrit quantumstate tomography where computations of the
latter are feasible.

Bayesian error analysis paves the way to the construction of credible and
plausible error regions for a point estimator obtained from a given dataset. We
introduce the concept of region accuracy for error regions (a generalization of
the pointestimator mean squarederror) to quantify the average statistical
accuracy of all region points with respect to the unknown true parameter. We
show that the increase in region accuracy is closely related to the
Bayesianregion dual operations in [1]. Next with only the given dataset as
viable evidence, we establish various adaptive methods to maximize the region
accuracy relative to the true parameter subject to the type of reported
Bayesian region for a given point estimator. We highlight the performance of
these adaptive methods by comparing them with nonadaptive procedures in three
quantumparameter estimation examples. The results of and mechanisms behind the
adaptive schemes can be understood as the region analog of adaptive approaches
to achieving the quantum CramerRao bound for point estimators.

Balanced homodyning, heterodyning and unbalanced homodyning are the three
wellknown sampling techniques used in quantum optics to characterize all
possible photonic sources in continuousvariable quantum information theory. We
show that for all quantum states and all observableparameter tomography
schemes, which includes the reconstructions of arbitrary operator moments and
phasespace quasidistributions, localized sampling with unbalanced homodyning
is always tomographically more powerful (gives more accurate estimators) than
delocalized sampling with heterodyning. The latter is recently known to often
give more accurate parameter reconstructions than conventional marginalized
sampling with balanced homodyning. This result also holds for realistic
photodetectors with subunit efficiency. With examples from first through
fourthmoment tomography, we demonstrate that unbalanced homodyning can
outperform balanced homodyning when heterodyning fails to do so. This new
benchmark takes us one step towards optimal continuousvariable tomography with
conventional photodetectors and minimal experimental components.

Wigner and Husimi quasidistributions, owing to their functional regularity,
give the two archetypal and equivalent representations of all
observableparameters in continuousvariable quantum information. Balanced
homodyning and heterodyning that correspond to their associated sampling
procedures, on the other hand, fare very differently concerning their state or
parameter reconstruction accuracies. We present a general theory of a nowknown
fact that heterodyning can be tomographically more powerful than balanced
homodyning to many interesting classes of singlemode quantum states, and
discuss the treatment for twomode sources.

We examine the momentreconstruction performance of both the homodyne and
heterodyne (doublehomodyne) measurement schemes for arbitrary quantum states
and introduce moment estimators that optimize the respective schemes for any
given data. In the largedata limit, these estimators are as efficient as the
maximumlikelihood estimators. We then illustrate the superiority of the
heterodyne measurement for the reconstruction of the first and second moments
by analyzing Gaussian states and many other significant nonclassical states.

In continuousvariable tomography, with finite data and limited computation
resources, reconstruction of a quantum state of light is performed on a
finitedimensional subspace. No systematic method was ever developed to assign
such a reconstruction subspaceonly ad hoc methods that rely on
hardtocertify assumptions about the source and strategies. We provide a
straightforward and numerically feasible procedure to uniquely determine the
appropriate reconstruction subspace for any given unknown quantum state of
light and measurement scheme. This procedure makes use of the celebrated
statistical principle of maximum likelihood, along with other validation tools,
to grow an appropriate seed subspace into the optimal reconstruction subspace,
much like the nucleation of a seed into a crystal. Apart from using the
available measurement data, no other spurious assumptions about the source or
ad hoc strategies are invoked. As a result, there will no longer be
reconstruction artifacts present in state reconstruction, which is a usual
consequence of a bad choice of reconstruction subspace. The procedure can be
understood as the maximumlikelihood reconstruction for quantum subspaces,
which is an analog to, and fully compatible with that for quantum states.

The accuracy in determining the quantum state of a system depends on the type
of measurement performed. Homodyne and heterodyne detection are the two main
schemes in continuousvariable quantum information. The former leads to a
direct reconstruction of the Wigner function of the state, whereas the latter
samples its Husimi $Q$~function. We experimentally demonstrate that heterodyne
detection outperforms homodyne detection for almost all Gaussian states, the
details of which depend on the squeezing strength and thermal noise.

We present a simple and efficient Bayesian recursive algorithm for the
datapattern scheme for quantum state reconstruction, which is applicable to
situations where measurement settings can be controllably varied efficiently.
The algorithm predicts the best measurements required to accurately reconstruct
the unknown signal state in terms of a fixed set of probe states. In each
iterative step, this algorithm seeks the measurement setting that minimizes the
variance of the datapattern estimator, which essentially measures the
reconstruction accuracy, with the help of a datapattern bank that was acquired
prior to the signal reconstruction. We show that with this algorithm, it is
possible to minimize the number of measurement settings required to obtain a
reasonably accurate state estimator by using just the optimal settings and, at
the same time, increasing the numerical efficiency of the datapattern
reconstruction.

We reveal that quadrature squeezing can result in significantly better
quantumestimation performance with quantum heterodyne detection (of H. P. Yuen
and J. H. Shapiro) as compared to quantum homodyne detection for Gaussian
states, which touches an important aspect in the foundational understanding of
these two schemes. Taking singlemode Gaussian states as examples, we show
analytically that the competition between the errors incurred during tomogram
processing in homodyne detection and the ArthursKelly uncertainties arising
from simultaneous incompatible quadrature measurements in heterodyne detection
can often lead to the latter giving more accurate estimates. This observation
is also partly a manifestation of a fundamental relationship between the
respective data uncertainties for the two schemes. In this sense, quadrature
squeezing can be used to overcome intrinsic quantummeasurement uncertainties
in heterodyne detection.

We introduce an operational and statistically meaningful measure, the quantum
tomographic transfer function, that possesses important physical invariance
properties for judging whether a given informationally complete quantum
measurement performs better tomographically in quantumstate estimation
relative to other informationally complete measurements. This function is
independent of the unknown true state of the quantum source, and is directly
related to the average optimal tomographic accuracy of an unbiased state
estimator for the measurement in the limit of many sampling events. For the
experimentallyappealing minimally complete measurements, the transfer function
is an extremely simple formula. We also give an explicit expression for this
transfer function in terms of an ordered expansion that is readily computable
and illustrate its usage with numerical simulations, and its consistency with
some known results.

We report an experiment in which one determines, with least tomographic
effort, whether an unknown twophoton polarization state is entangled or
separable. The method measures whole families of optimal entanglement
witnesses. We introduce adaptive measurement schemes that greatly speed up the
entanglement detection. The experiments are performed on states of different
ranks, and we find good agreement with results from computer simulations.

In quantumstate tomography on sources with quantum degrees of freedom of
large Hilbert spaces, inference of quantum states of light for instance, a
complete characterization of the quantum states for these sources is often not
feasible owing to limited resources. As such, the concepts of informationally
incomplete state estimation becomes important. These concepts are ideal for
applications to quantum channel/process tomography, which typically requires a
much larger number of measurement settings for a full characterization of a
quantum channel. Some key aspects of both quantumstate and quantumprocess
tomography are arranged together in the form of a tutorial review article that
is catered to students and researchers who are new to the field of quantum
tomography, with focus on maximumlikelihood related techniques as instructive
examples to illustrate these ideas.

We report a controllable method for producing mixed twophoton states via
Spontaneous Parametric DownConversion with a twotypeI crystal geometry. By
using variable polarization rotators (VPRs), one obtains mixed states of
various purities and degrees of entanglement depending on the parameters of the
VPRs. The generated states are characterized by quantum state tomography. The
experimental results are found to be in good agreement with the theory. The
method can be easily implemented for various experiments which require the
generation of states with controllable degrees of entanglement or mixedness.

We introduce a straightforward numerical coarsegraining scheme to estimate
quantum states for a set of noisy measurement outcomes, which are difficult to
calibrate, that is based solely on the measurement data collected from these
outcomes. This scheme involves the maximization of a weighted entropy function
that is simple to implement and can readily be extended to any number of
illcalibrated noisy outcomes in a measurement setup, thus offering practical
applicability for general tomography experiments without additional knowledge
or assumptions about the structures of the noisy outcomes. Simulation results
for twoqubit quantum states show that coarsegraining can improve the
tomographic efficiencies for noise levels ranging from low to moderately high
values.

This is a PhD dissertation on the latest numerical quantum estimation schemes
as of 2012, submitted to the National University of Singapore. The main content
of the thesis focuses on accessing quantum information with informationally
incomplete measurements to reconstruct quantum states of large quantum systems,
as well as to reduce the amount of resources to reconstruct quantum channels.

There exists, in general, a convex set of quantum state estimators that
maximize the likelihood for informationally incomplete data. We propose an
estimation scheme, catered to measurement data of this kind, to search for the
exact maximumlikelihoodmaximumentropy estimator using semidefinite
programming and a standard multidimensional function optimization routine.
This scheme can be used to infer the expectation values of a set of
entanglement witnesses that can be used to verify the entanglement of the
unknown quantum state for composite systems. Next, we establish an alternative
numerical scheme that is more computationally robust for the sole purpose of
maximizing the likelihood and entropy.

We present a detailed account of quantum state estimation by joint
maximization of the likelihood and the entropy. After establishing the
algorithms for both perfect and imperfect measurements, we apply the procedure
to data from simulated and actual experiments. We demonstrate that the
realistic situation of incomplete data from imperfect measurements can be
handled successfully.

We propose an iterative algorithm for incomplete quantum process tomography,
with the help of quantum state estimation, based on the combined principles of
maximumlikelihood and maximumentropy. The algorithm yields a unique estimator
for an unknown quantum process when one has less than a complete set of
linearly independent measurement data to specify the quantum process uniquely.
We apply this iterative algorithm adaptively in various situations and so
optimize the amount of resources required to estimate the quantum process with
incomplete data.

In the fourdimensional Hilbert space, there exist 16 HeisenbergWeyl (HW)
covariant symmetric informationally complete positive operator valued measures
(SIC~POVMs) consisting of 256 fiducial states on a single orbit of the Clifford
group. We explore the structure of these SIC~POVMs by studying the symmetry
transformations within a given SIC~POVM and among different SIC~POVMs.
Furthermore, we find 16 additional SIC~POVMs by a regrouping of the 256
fiducial states, and show that they are unitarily equivalent to the original 16
SIC~POVMs by establishing an explicit unitary transformation. We then reveal
the additional structure of these SIC~POVMs when the fourdimensional Hilbert
space is taken as the tensor product of two qubit Hilbert spaces. In
particular, when either the standard product basis or the Bell basis are chosen
as the defining basis of the HW group, in eight of the 16 HW covariant
SIC~POVMs, all fiducial states have the same concurrence of $\sqrt{2/5}$. These
SIC~POVMs are particularly appealing for an experimental implementation, since
all fiducial states can be connected to each other with just local unitary
transformations. In addition, we introduce a concise representation of the
fiducial states with the aid of a suitable tabular arrangement of their
parameters.

State tomography on qubit pairs is routinely carried out by measuring the two
qubits separately, while one expects a higher efficiency from tomography with
highly symmetric joint measurements of both qubits. Our numerical study of
simulated experiments does not support such expectations.

We introduce informationally complete measurements whose outcomes are
entanglement witnesses and so answer the question of how many witnesses need to
be measured to decide whether an arbitrary state is entangled or not: as many
as the dimension of the state space. The optimized witnessbased measurement
can provide exponential improvement with respect to witness efficiency in
highdimensional Hilbert spaces, at the price of a reduction in the tomographic
efficiency. We describe a systematic construction, and illustrate the matter at
the example of two qubits.