The main topic of this thesis is linear optics and the implementation of quantum operations (measurements, quantum channels, and unitary rotations) on optical systems. \nIn the opening chapter the basic notions needed to understand the rest of the thesis will be explained. \nThese notions include defining a quantum state, measurement, quantum channel and the linear optics tool set. \n \nThe work in this thesis takes both fundamental and practical approaches to studying linear optical networks. \nFor instance in the first chapter a proof is provided that shows that any unitary on a single mode Fock state can be realised with linear optics. \nThe proof is constructive, however the approach to realising the unitary is not suitable for experimental implementation \nbecause it requires complicated ancilla states. \nAs in the KLM proposal the procedure works only stochastically however by allowing the size of the ancilla to grow \nthe probability of failure can be made arbitrarly small. \n \nFurthermore we investigate the realisation of arbitrary channels in a specific encoding that we call a $d$-rail encoding. \nThe only ancilla state that we allow is a vacuum ancilliary state and further restrictions were considered (e.g. photon counting). \nA proof is provided that using these resources only random unitaries can be applied deterministically using linear optics. An expression for the optimal probability of success for realising \nmore general channels with these resources is also discussed. \n \nAs a final topic we also investigate the realisation of a quantum non-demolition measurement onto the dual rail qubit space. \nThe investigation is a blend of both fundamental and practical approaches. \nTo begin we employ a modified KLM-like procedure and show that the scheme can be realised perfectly but stochastically. \nThe probability that the proper measurement is made can be made arbitrarly close to one using a suitably large ancilla state. \nIn addition we consider an existing scheme \\cite{gisin10a} which uses practical sources (two single photon sources) to perform the measurement. \nThe scheme does not realise the true measurement but instead has a free parameter in it which is the transmittiviy of a beamsplitter. The measurement will project onto a space that has a vacuum component. By adjusting the transmittivity of this beamsplitter the vacuum component can be made arbitrarly small but only at the expense of the probability of success of the procedure. \nIn this thesis a modification that can be made to eliminate the vacuum component without changing the sources is introduced. The modification is surprisingly simple and only involves the addition of a single beamsplitter. In the proposal for the original amplifier it was used in simulations for DIQKD that included device imperfections. \nTo show the improvement of our modification these DIQKD simulations are reproduced using the modified amplifier and its results are compared to the results of the original amplifier.
The main topic of this thesis is linear optics and the implementation of quantum operations (measurements, quantum channels, and unitary rotations) on optical systems. \nIn the opening chapter the basic notions needed to understand the rest of the thesis will be explained. \nThese notions include defining a quantum state, measurement, quantum channel and the linear optics tool set. \n \nThe work in this thesis takes both fundamental and practical approaches to studying linear optical networks. \nFor instance in the first chapter a proof is provided that shows that any unitary on a single mode Fock state can be realised with linear optics. \nThe proof is constructive, however the approach to realising the unitary is not suitable for experimental implementation \nbecause it requires complicated ancilla states. \nAs in the KLM proposal the procedure works only stochastically however by allowing the size of the ancilla to grow \nthe probability of failure can be made arbitrarly small. \n \nFurthermore we investigate the realisation of arbitrary channels in a specific encoding that we call a d-rail encoding. \nThe only ancilla state that we allow is a vacuum ancilliary state and further restrictions were considered (e.g. photon counting). \nA proof is provided that using these resources only random unitaries can be applied deterministically using linear optics. An expression for the optimal probability of success for realising \nmore general channels with these resources is also discussed. \n \nAs a final topic we also investigate the realisation of a quantum non-demolition measurement onto the dual rail qubit space. \nThe investigation is a blend of both fundamental and practical approaches. \nTo begin we employ a modified KLM-like procedure and show that the scheme can be realised perfectly but stochastically. \nThe probability that the proper measurement is made can be made arbitrarly close to one using a suitably large ancilla state. \nIn addition we consider an existing scheme \cite{gisin10a} which uses practical sources (two single photon sources) to perform the measurement. \nThe scheme does not realise the true measurement but instead has a free parameter in it which is the transmittiviy of a beamsplitter. The measurement will project onto a space that has a vacuum component. By adjusting the transmittivity of this beamsplitter the vacuum component can be made arbitrarly small but only at the expense of the probability of success of the procedure. \nIn this thesis a modification that can be made to eliminate the vacuum component without changing the sources is introduced. The modification is surprisingly simple and only involves the addition of a single beamsplitter. In the proposal for the original amplifier it was used in simulations for DIQKD that included device imperfections. \nTo show the improvement of our modification these DIQKD simulations are reproduced using the modified amplifier and its results are compared to the results of the original amplifier.
This DPhil thesis presents two key works towards practical applications of quantum optics. Both works are novel and achieve competitive state-of-the-art results. Today's most widely used method of encoding quantum information in optical qubits is the dual-rail basis, often carried out through the polarisation of a single photon. On the other hand, many stationary carriers of quantum information | such as atoms | couple to light via the single-rail encoding in which the qubit is encoded in the number of photons. As such, interconversion between the two encodings is paramount in order to achieve cohesive quantum networks. In the first part of this thesis, we demonstrate this by generating a hybrid entangled resource between the two encodings and using it to teleport a dual-rail qubit onto its singlerail counterpart. Our key results yield an average fidelity of F = (92:8±2:2)% for the teleportation and F = (89:7 ± 2:1)% for entanglement swapping, thus confirming the applicability of this scheme towards a real-world implementation. This work completes the set of tools necessary for the interconversion between the three primary encodings of a qubit in the optical field: single-rail, dual-rail and continuous-variable. A remarkable aspect of quantum theory is that certain measurement outcomes are entirely unpredictable to all possible observers. Such quantum events can be harnessed to generate numbers whose randomness is asserted based upon the underlying physical processes. In the second part of this thesis, we formally introduce and experimentally demonstrate an ultrafast optical quantum randomness generator that uses a totally untrusted photonic source and whose idea we have patented. While considering completely general quantum attacks, we certify randomness at a rate of 1:1 Gbps with a rigorous security parameter of 10-20. Our security proof is entirely composable, thereby allowing the generated randomness to be utilised for arbitrary applications in cryptography and beyond.
Quantum optics has successfully been used to test fundamental principles in quantum physics and to demonstrate the potential of quantum-enhanced technologies. Linear quantum optics, in which large quantum states of light are produced by optical interference of smaller quantum states, has proved to be particularly fruitful. Further progress will rely both on developing improved experimental tools to manipulate and measure quantum light, and on expanding and refining the range of applications of these technologies. This thesis presents our contributions towards both of these endeavours. We first focus on some of the components necessary for linear quantum optics, starting with the requirement for reconfigurable interference between several optical modes. We propose a novel design for interferometers that satisfy this requirement, which is based on a new mathematical decomposition of unitary matrices used to describe optical interference. We show that our design is more efficient than previously known designs. We also experimentally demonstrate a modular approach to building these devices, which is based on the assembly of multiple UV-written integrated photonic chips. These chips are characterised, and three of them are assembled into a structure shown to enable a wide range of optical transformations. We then study methods of photon detection, showing how photon detectors can be calibrated and discussing the operation of superconducting photon number resolving transition edge sensors. Next, we study two applications of linear optics. We examine the applicability of a proposal for simulating molecular spectroscopy using quantum optics in the presence of experimental imperfections. Our findings are illustrated with a proof of principle experiment in which we simulate part of the vibronic spectrum of the tropolone molecule. Finally, we study a class of optical devices, known as optical Ising machines, that has been shown to find solutions to difficult combinatorial problems. Describing the optical pulses in these devices in phase space as Gaussian quasi-probability distributions that evolve stochastically, we analyse the computational mechanism of these machines and show in theory that they can be simplified without affecting their performance.
This thesis deals with the implementation of quantum channels using linear optics. We begin with overviews of some important concepts in both quantum information and quantum optics. First, we discuss the quantum bit and describe the evolution of the states via quantum channels. We then discuss both quantum state and process tomography, methods for how to determine which states and operations we are experimentally implementing in the lab. Second, we discuss topics in quantum optics such the generation of single photons, polarization entanglement, and the construction of an entangling gate. \n \nThe first experiment is the implementation of a quantum damping channel, which intentionally can add a specific type and amount of decohering noise to a photonic qubit. Specifically, we realized a class of quantum channels which contains both the amplitude-damping channel and the bit-flip channel, and did so with a single, static, optical setup. Many quantum channels, and some gates, can only be implemented probabilistically when using linear optics and postselection. Our main result is that the optical setup achieves the optimal success probability for each channel. Using a novel ancilla-assisted tomography, we characterize each case of the channel, and find process fideilities of 0.98pm0.01 for the amplitude-damping channel and 0.976pm0.009 for the bit-flip. \n \nThe second experiment is an implementation of a protocol for quantum computing on encrypted data. The protocol provides the means for a client with very limited quantum power to use a server's quantum computer while maintaining privacy over the data. We perform a quantum process tomography for each gate in a universal set, showing that only when the proper decryption key is used on the output states, which is hidden from the server, then the action of the quantum gate is recovered. Otherwise, the gate acts as the completely depolarizing channel.
This thesis presents results under the topics of quantum correlations and quantum information processing using linear-optical networks. The first part of this thesis concerns with quantum correlations, specially in Gaussian states, which play a central role as resource in quantum protocols. The second part is about quantum information processing using linear-optical networks that are of great practical interest due to their simple physical realization. In the first part, we present a measurement-based method for verifying nonzero quantum discord, a measure of quantum correlations, in bipartite systems that can be used for both continuous and discrete-variable systems. Based on this method, we show that the only bipartite Gaussian states with zero quantum discord are product states. We also present a simple and efficient experimental method for verifying quantum correlations in Gaussian states using homodyne measurements. Moreover, we introduce an operational measure for quantifying quantum correlations in Gaussian states that is purely based on Gaussian measurements. We then illustrate the operational significance of the measure in terms of a Gaussian quantum protocol. In the second part, we present results on linear quantum-optics experiments, specifically boson sampling, an intermediate model of quantum computation that can be implemented using single-photon states, linear-optical networks, and photodetectors. We first present an efficient experimental method for characterization of linear-optical networks, which is essential for practical implementation of boson sampling. We show that using coherent states one can directly measure all the nontrivial phases and moduli of the matrix describing the network. Next, we consider the problem of boson sampling with Gaussian states. We derive a general formula for calculating the output photon-counting probabilities. We show that for input thermal states, sampling from the output probability distribution can be efficiently classically simulated. Using this, we find that permanents of positive-semidefinite Hermitian matrices can be approximated to within a multiplicative error in BPPNP. We also discuss boson sampling with squeezed vacuum states. Finally, we provide a sufficient condition for the efficient classical simulation of the general sampling problem using linear-quantum optics, which involves a multimode input quantum state to a linear-optical network and set of measurements at the output. This condition is based on the phase-space quasiprobability distributions, and suggests that negativity is a quantum resource. We apply this condition to implementations of boson sampling using single-photon or spontaneous parametric down-conversion sources. We show that above some threshold for loss and noise in the experiment, boson sampling is classically simulatable.
This thesis covers several topics in optical quantum computation. A quantum computer is a computational device which is able to manipulate information by performing unitary operations on some physical system whose state can be described as a vector (or mixture of vectors) in a Hilbert space. The basic unit of information, called the qubit, is considered to be a system with two orthogonal states, which are assigned logical values of 0 and 1. Photons make excellent candidates to serve as qubits. They have little interactions with the environment. Many operations can be performed using very simple linear optical devices such as beam splitters and phase shifters. Photons can easily be processed through circuit-like networks. Operations can be performed in very short times. Photons are ideally suited for the long-distance communication of quantum information. The great difficulty in constructing an optical quantum computer is that photons naturally interact weakly with one another. This thesis first gives a brief review of two early approaches to optical quantum computation. It will describe how any discrete unitary operation can be performed using a single photon and a network of beam splitters, and how the Kerr effect can be used to construct a two photon logic gate. Second, this work provides a thorough introduction to the linear optical quantum computer developed by Knill, Laflamme, and Milburn. It then presents this author's results on the reliability of this scheme when implemented using imperfect photon detectors. This author finds that quantum computers of this sort cannot be built using current technology. Third, this dissertation describes a method for constructing a linear optical quantum computer using nearly orthogonal coherent states of light as the qubits. It shows how a universal set of logic operations can be performed, including calculations of the fidelity with which these operations may be accomplished. It discusses methods for reducing and correcting errors and recovering from failed operations. Lastly it describes an analysis of the long distance transmission of the coherent state qubits and shows how transmission errors can be corrected.
Single photons are the smallest excitation states of the elctromagnetic field. They can be experimentally created but there is no apparatus that performs this task with 100% efficiency. Many techniques and theorems of optical quantum information theory use single photons. In order to utilize them for information processing, it is reasonable to know if they had been successfully produced. In other words: One needs heralded single photons. How to achieve this, is the main result of this diploma thesis. The task has been to improve the efficiency of already existing single photon sources. An apparatus had to be found which postprocesses the output of such sources in order to either herald the photons or at least to improve the efficiency. This apparatus is explicitly allowed to use other optical sources, for example coherent states (laser). The situation has been formally represented by the assumption that single photon sources produce a mixture of vacuum and the single photon state. It is known that experimental set-ups also create higher photon contributions. These contributions are considered to be small in corresponding publications. For some cases they even seem to vanish at all. One can distinguish two regimes concerning the techniques which have been used in this work: linear and nonlinear optics. Linear optical tools, for example the beam splitter, do not absorb photons in the ideal case, but it is not easy to implement useful transformations by their means. Absorption and weak coupling are drawbacks which arise mainly in nonlinear optical media. This work shows that the interactions two-photon absorption and cross phase modulation can be used in schemes, which enable the heralding of single photons. Already there are some linear optical quantum gates which manipulate optical states by means of single photons. Photons do not interact with each other directly, in contrast to electrons for example. Therefore quantum gates which act on more than one mode are interesting. The effective interaction of these gates is not quitecomparable to unitary dynamics. Moreover, these gates do not function in each case. Measurements required in auxiliary modes yield different results for each single case and in order to obtain the desired effect the cases have to be selected according to the measurement results. It has been investigated, how such gates perform if they have statistical mixtures as a resource instead of perfect single-photon states. Yet, no improvement of the single-photon contribution has been accomplished by means of these apparatuses. Hereby it has not been disproved that it is possible to achieve this goal using other apparatuses and techniques of linear optics. Yet, it has been shown that an enhancement is possible but then a non-vanishing two-photon contribution is appended. One of the most basic quantum gates is the conditional sign flip. The optical 8 interaction cross phase modulation acts in quite the same way considering only single photons. This effect enables the heralding of single photons in the ideal case. Within this work it has been shown that a Mach-Zehnder interferometer, a cross phase modulating medium and a photo-detector can be employed for this purpose. One could use any source of light as input to the interferometer but a strong coherent state (laser) may be preferred for the purpose of enhancing the interaction. The absorption of two-photon states is another non-linear effect. Astonishingly this effect serves to generate a heralded single-photon state given two imperfect single-photon sources. First the photon adding tool is employed which consists of a beam splitter and a photo-detector. By means of that and the statistical mixtures of two single-photon sources one creates a mixture which contains vacuum, the single-photon state and additionally the two-photon Fock state. This mixture propagates through a Mach-Zehnder interferometer. The interferometer is adjusted such that it is transparent for incoming light. In addition to the interferometer alone a two-photon absorbing medium is placed in one of its arms. For this reason there is a chance that the two-photon contribution entering the interferometer is absorbed within this medium. We cannot exclude this unfortunate case. A detector is placed in one of the two output modes. The interferometer is adjusted such that the single photon state does not leave in the mode in which the detector is placed, but always in the other mode. The detector can only click if the two-photon state propagates through the interferometer and if it is not destroyed in the absorbing medium. If the detector then receives one photon it is obvious that another one must leave in the other mode. The whole phenomenon is very similar to the concept of null measurement. Although not a main result, it has been shown here that the so-called Hong-Ou-Mandel effect can be generalized. It has been proven that the beam splitter transforms Fock states with equal Fock number into superpositions of Fock numbers which are even in each mode. Another effect on special odd Fock states is conjectured.