This portfolio of work contributes to the remit of the National Physical Laboratory (NPL) to develop the underpinning expertise and tools for validating nascent and future optical quantum technologies based on the discrete and quantum properties of photons. This requirement overlaps with the requirement to provide validation for devices operating in the photon-counting regime. A common theme running through the portfolio is photon pairs generated through spontaneous parametric downconversion (SPDC). A Hong-Ou-Mandel (HOM) interferometer sourced with visible wavelength photon pairs from an SPDC process in beta-barium borate (BBO) was designed, built and characterised. The visibility of the HOM interference is dependent on the indistinguishability of the interfering photons, but is also influenced by imperfections of the interferometer; therefore an investigation was carried out to quantify the effects of the interferometer imperfections on the measured visibility so that the true photon indistinguishability could be measured with a quantified uncertainty. A bright source of correlated pair photons in the telecoms band based upon a pump enhanced SPDC process in periodically-poled potassium titanyl phosphate (PPKTP) was designed, built and characterised. From the characterisation measurements the source brightness was estimated to be 6.2×10⁴ pairs/ s/ mw pump. The photon pairs were further characterised through their incorporation as a source in a HOM interference experiment. The developed correlated photon pair source was at the heart of a novel scheme for the generation of polarisation entangled photon pairs, for which the design, build and characterisation work is presented. The source was demonstrated to produce two of the four maximally entangled Bell states with quantum interference visibilities of around 0.95. The generated states were also shown to break a form of Bell's inequality by around six standard deviations. The polarisation entangled photon pair source was originally built at the University of St Andrews and was later transferred to the NPL where it will extend NPL's capabilities to this key spectral region. Finally a study was carried out to investigate the possibility of a wavelength tuneable device for the absolute measurement of single photon detector quantum efficiencies based upon an established SPDC technique.

<div><div><div><p>When confined to a finite, two-dimensional area and exposed to a strong magnetic field, electrons exhibit a complicated, highly correlated quantum behavior known as the quantum Hall effect. This dissertation consists of finite size numerical investigations of this effect. One line of study develops treatment of the fractional quantum Hall effect using the hyperspherical method, in conjunction with applications to the few-body quantum Hall systems, e.g., highly-controlled atomic systems. Another line of research fully utilizes the developed numerical techniques to study on the platform of finite size fractional quantum Hall states the bulk-edge correspondence principle, which is universal for phases in topological orders. It has been demonstrated that the eigenstates associated with the entanglement spectrum reveal more information about the ground state than the spectrum alone.</p></div></div></div>

We consider the application of Few-Shot Learning (FSL) and dimensionality reduction to the problem of human motion recognition (HMR). The structure of human motion has unique characteristics such as its dynamic and high-dimensional nature. Recent research on human motion recognition uses deep neural networks with multiple layers. Most importantly, large datasets will need to be collected to use such networks to analyze human motion. This process is both time-consuming and expensive since a large motion capture database must be collected and labeled. Despite significant progress having been made in human motion recognition, state-of-the-art algorithms still misclassify actions because of characteristics such as the difficulty in obtaining large-scale leveled human motion datasets. To address these limitations, we use metric-based FSL methods that use small-size data in conjunction with dimensionality reduction. We also propose a modified dimensionality reduction scheme based on the preservation of secants tailored to arbitrary useful distances, such as the geodesic distance learned by ISOMAP. We provide multiple experimental results that demonstrate improvements in human motion classification.

<pre><pre><p> </p><p>Recent progress in machine learning has been mainly due to the availability of large amounts of annotated data used for training complex models with deep architectures. Annotating this training data becomes burdensome and creates a major bottleneck in maintaining machine-learning databases. Moreover, these trained models fail to generalize to new categories or new varieties of the same categories. This is because new categories or new varieties have data distribution different from the training data distribution. To tackle these problems, this thesis proposes to develop a family of transfer-learning techniques that can deal with different training (source) and testing (target) distributions with the assumption that the availability of annotated data is limited in the testing domain. This is done by using the auxiliary data-abundant source domain from which useful knowledge is transferred that can be applied to data-scarce target domain. This transferable knowledge serves as a prior that biases target-domain predictions and prevents the target-domain model from overfitting. Specifically, we explore structural priors that encode relational knowledge between different data entities, which provides more informative bias than traditional priors. The choice of the structural prior depends on the information availability and the similarity between the two domains. Depending on the domain similarity and the information availability, we divide the transfer learning problem into four major categories and propose different structural priors to solve each of these sub-problems.</p><p> </p><p>This thesis first focuses on the unsupervised-domain-adaptation problem, where we propose to minimize domain discrepancy by transforming labeled source-domain data to be close to unlabeled target-domain data. For this problem, the categories remain the same across the two domains and hence we assume that the structural relationship between the source-domain samples is carried over to the target domain. Thus, graph or hyper-graph is constructed as the structural prior from both domains and a graph/hyper-graph matching formulation is used to transform samples in the source domain to be closer to samples in the target domain. An efficient optimization scheme is then proposed to tackle the time and memory inefficiencies associated with the matching problem. The few-shot learning problem is studied next, where we propose to transfer knowledge from source-domain categories containing abundantly labeled data to novel categories in the target domain that contains only few labeled data. The knowledge transfer biases the novel category predictions and prevents the model from overfitting. The knowledge is encoded using a neural-network-based prior that transforms a data sample to its corresponding class prototype. This neural network is trained from the source-domain data and applied to the target-domain data, where it transforms the few-shot samples to the novel-class prototypes for better recognition performance. The few-shot learning problem is then extended to the situation, where we do not have access to the source-domain data but only have access to the source-domain class prototypes. In this limited information setting, parametric neural-network-based priors would overfit to the source-class prototypes and hence we seek a non-parametric-based prior using manifolds. A piecewise linear manifold is used as a structural prior to fit the source-domain-class prototypes. This structure is extended to the target domain, where the novel-class prototypes are found by projecting the few-shot samples onto the manifold. Finally, the zero-shot learning problem is addressed, which is an extreme case of the few-shot learning problem where we do not have any labeled data in the target domain. However, we have high-level information for both the source and target domain categories in the form of semantic descriptors. We learn the relation between the sample space and the semantic space, using a regularized neural network so that classification of the novel categories can be carried out in a common representation space. This same neural network is then used in the target domain to relate the two spaces. In case we want to generate data for the novel categories in the target domain, we can use a constrained generative adversarial network instead of a traditional neural network. Thus, we use structural priors like graphs, neural networks and manifolds to relate various data entities like samples, prototypes and semantics for these different transfer learning sub-problems. We explore additional post-processing steps like pseudo-labeling, domain adaptation and calibration and enforce algorithmic and architectural constraints to further improve recognition performance. Experimental results on standard transfer learning image recognition datasets produced competitive results with respect to previous work. Further experimentation and analyses of these methods provided better understanding of machine learning as well.</p><p> </p></pre></pre>

In this thesis we present the realization of a source of 1.5-cycle carrier-envelope phase stable laser pulses in the mid-infrared spectral region. We used ytterbium laser system generating 1 µm pulses as a pump of setup, where the beam is split into several parts and interact in nonlinear optical media. 2 µJ pulses with duration 18 fs at 50 kHz repe- tition rate are produced. By spectral broadening in crystal GGG, 9,9 fs pulses can be achieved. The mid-IR pulses was characterized by third harmonics generation frequency resolved optical gating in the interferometric configuration. Fourier filtering of the mea- sured interferogram allows for the complete reconstruction of amplitude and phase of the ultrashort pulses generated by our setup. The pulses will in future serve for experimental investigation of ultrafast strong-field phenomena in solids. 1

Recent refinements in Industry 4.0 and Machine Learning demonstrate the positive effects of using deep learning models for intelligent maintenance. The primary benefit of Deep Learning (DL) is its capability to extract attributes and make fast, accurate, and automated predictions without supervision. However, DL requires high computational power, significant data preprocessing, and vast amounts of data to make accurate predictions for intelligent maintenance. Given the considerable obstacles, meta-learning has been developed as a novel way to overcome these challenges. As a learning technique, meta-learning aims to quickly acquire knowledge of new tasks using theminimal available data by learning through meta-knowledge. There has been less research in the area of using meta-learning for Predictive Maintenance (PdM) and we considered it necessary to conduct this review to understand the applicability of meta-learning’s capabilities and functions to PdM since the outcomes of this technique seem to be rather promising. The review started with the development of a methodology and four research questions: (1) What is the taxonomy of meta-learning for PdM?, (2) What are the current state-of-the-art methodologies? (3) Which datasets are available for meta-learning in PdM?, and (4) What are the open issues, challenges, and opportunities of meta-learning in PdM?. To answer the first and second questions, a new taxonomy was proposed and meta-learnings role in predictive maintenance was identified from selected 55 papers. To answer the third question, we determined which types of datasets and their characteristics exist for this domain. Finally, the challenges, open issues, and opportunities of meta-learning in predictive maintenance were examined to answer the final question. The results of the research questions provided suggestions for future research topics.

[EN] Space-division multiplexing fibers emerged as a promising solution to overcome the imminent capacity crunch of conventional singlemode fiber networks. Despite these fibers were initially conceived as distribution media for long-haul high-capacity digital communications, they can be applied to a wide variety of scenarios including centralized radio access networks for wireless communications, data-center interconnects, Microwave Photonics signal processing and fiber sensing. Particular interest is raised by emerging communications paradigms, such as 5G and The Internet of Things, which require a full integration between the optical fiber and the wireless networks segments. Microwave Photonics, discipline that focuses on the generation, processing, control and distribution of radiofrequency signals by photonics means, is called to play a decisive role. One of the major challenges that Microwave Photonics has to overcome to satisfy next-generation communication demands relates to the reduction of size, weight and power consumption while assuring broadband seamless reconfigurability and stability. There is one revolutionary approach that has however been left untapped in finding innovative ways to address that challenge: exploiting space, the last available degree of freedom for optical multiplexing. In this Thesis, we propose to exploit the inherent parallelism of multicore and few-mode fibers to implement sampled discrete true time delay lines, providing, in a single optical fiber, a compact and efficient approach for both Microwave Photonics signal distribution and processing. For the multicore fiber approach, we study the influence of the refractive index profile of each heterogeneous core on the propagation characteristics as to feature specific group delay and chromatic dispersion values. We designed and fabricated two different heterogeneous trench-assisted 7-core fibers that behave as sampled true time delay lines. While one of them was fabricated by using 7 different preforms to feature a plenary performance, the other one employed a single preform with the aim of minimizing fabrication costs. In the case of few-mode fibers, we propose the implementation of a tunable true time delay line by means of a custom-designed fiber with a set of inscribed long period gratings that act as mode converters to properly tailor the sample group delays. We designed and fabricated a true time delay line on a 4-mode fiber by inscribing 3 long period gratings at specific positions along the fiber link. As a proof-of-concept validation, we experimentally demonstrated different Microwave Photonics signal processing functionalities implemented over both multicore and few-mode fiber approaches. This work opens the way towards the development of distributed signal processing for microwave and millimeter wave signals in a single optical fiber. These true time delay lines can be applied to a wide range of Information and Communication Technology paradigms besides fiber-wireless communications such as broadband satellite communications, distributed sensing, medical imaging, optical coherence tomography and quantum communications. / [ES] La multiplexación por división espacial en fibras ópticas surgió como una solución prometedora al inminente colapso en la capacidad de las redes de fibra monomodo convencionales. Aunque estas fibras fueron concebidas inicialmente como medio de distribución en comunicaciones digitales de larga distancia y alta capacidad, pueden emplearse en una amplia variedad de escenarios, incluyendo redes de acceso radio centralizadas para comunicaciones inalámbricas, interconexiones en centros de datos, así como procesado de señal en Fotónica de Microondas y sensado en fibra. Los paradigmas de comunicaciones emergentes despiertan un interés particular, como 5G y el Internet de las Cosas, que requieren una integración total entre el segmento de red de fibra óptica y el inalámbrico. La Fotónica de Microondas, disciplina que se focaliza en la generación, procesado, control y distribución de señales de radiofrecuencia por medio de la fotónica, está destinada a jugar un papel decisivo. Uno de los mayores desafíos que la Fotónica de Microondas debe superar para satisfacer los requisitos de las nuevas generaciones de comunicaciones se basa en la reducción de tamaño, peso y consumo de potencia, mientras se garantiza reconfiguración y estabilidad de banda ancha. Encontramos aquí un enfoque revolucionario capaz de abordar este desafío de una manera innovadora que, sin embargo, no ha sido aprovechado en este contexto: la explotación del espacio, el último grado de libertad para multiplexación óptica. En esta Tesis, proponemos explotar el paralelismo inherente de las fibras ópticas multinúcleo y de pocos modos para implementar líneas de retardo en tiempo real muestreadas que proporcionan, en una sola fibra óptica, una solución compacta y eficiente tanto para distribución como para procesado de señales de Fotónica de Microondas. En el caso de fibras multinúcleo, estudiamos la influencia del perfil de índice de refracción de cada núcleo heterogéneo en las características de propagación para que exhiba unos valores concretos de retardo de grupo y dispersión cromática. Diseñamos y fabricamos dos fibras distintas de 7 núcleos con zanjas que se comportan como líneas de retardo en tiempo real muestreadas. Mientras que una de ellas se fabricó utilizando 7 preformas diferentes para garantizar un funcionamiento completo, la segunda se fabricó utilizando una única preforma con el objetivo de minimizar costes de fabricación. En el caso de fibras de pocos modos, proponemos la implementación de líneas de retardo en tiempo real sintonizables mediante el uso de una fibra específicamente diseñada y la inscripción de un conjunto de redes de difracción de periodo largo que actúan como conversores de modos para ajustar adecuadamente el retardo de grupo de las muestras. Diseñamos y fabricamos una línea de retardo en tiempo real en una fibra de 4 modos mediante la inscripción de 3 redes de difracción de periodo largo en posiciones concretas a lo largo de enlace de fibra. Como validación de prueba de concepto, demostramos experimentalmente diferentes funcionalidades de procesado de señal de Fotónica de Microondas implementadas en fibras multinúcleo y de pocos modos. Este trabajo abre el camino hacia el desarrollo del procesado de señal distribuido para señales de microondas y ondas milimétricas en una única fibra óptica. Además, las líneas de retardo en tiempo real desarrolladas pueden aplicarse a una amplia variedad de paradigmas de Tecnologías de la Información y Comunicaciones más allá de las comunicaciones radio sobre fibra, como es el caso de las comunicaciones de banda ancha por satélite, el sensado distribuido, la imagen médica, la tomografía óptica coherente y las comunicaciones cuánticas. / [CA] La multiplexació per divisió espacial en fibres òptiques va sorgir com una solució prometedora a l'imminent col·lapse en la capacitat de les xarxes de fibra monomode convencionals. Encara que estes fibres foren concebudes inicialment com a mitjà de distribució en comunicacions digitals de llarga distància i alta capacitat, poden emprar-se en una àmplia varietat d'escenaris, incloent xarxes d'accés radio centralitzades per a comunicacions sense fils, interconnexions en centres de dades, així com processat de senyal en Fotònica de Microones i sensat en fibra. Els paradigmes de comunicacions emergents desperten un interès particular, com el 5G i la Internet de les Coses, que requereixen una integració total entre els segments de xarxa de fibra òptica i el de sense fils. La Fotònica de Microones, disciplina que es focalitza en la generació, processat, control i distribució de senyals de radiofreqüència per mitjà de la fotònica, està destinada a jugar un paper decisiu. Un dels majors desafiaments que la Fotònica de Microones ha de superar per satisfer els requisits de les noves generacions de comunicacions es basa en la reducció de grandària, pes i consum de potència, mentre es garanteix reconfiguració i estabilitat de banda ampla Trobem ací un enfocament revolucionari capaç d'abordar aquest desafiament d'una manera innovadora que, no obstant això, no ha sigut aprofitat encara en este context: la explotació de l'espai, l'últim grau de llibertat per a multiplexat òptic. En aquesta Tesi, proposem explotar el paral·lelisme inherent de les fibres òptiques multinucli i de pocs modes per a implementar línies de retard en temps real de mostres discretes que proporcionen, en una sola fibra òptica, una solució compacta i eficient tant per a distribució com per a processat de senyals de Fotònica de Microones. En el cas de fibres multinucli, estudiem la influència del perfil d'índex de refracció de cada nucli heterogeni en les característiques de propagació perquè exhibisca uns valors concrets de retard de grup i dispersió cromàtica. Dissenyem i fabriquem dues fibres distintes de 7 nuclis amb rases que es comporten com a línies de retard en temps real mostrejades. Mentre que una d'elles es va fabricar utilitzant 7 preformes diferents per a garantir un funcionament complet, la segona va fabricar-se utilitzant una única preforma amb l'objectiu de minimitzar costos de fabricació. En el cas de fibres de pocs modes, proposem la implementació de línies de retard en temps real sintonitzables mitjançant l'ús d'una fibra específicament dissenyada i la inscripció d'un conjunt de xarxes de difracció de període llarg que actuen com a convertidors de modes per tal d'ajustar adequadament el retard de grup de les mostres. Dissenyem i fabriquem una línia de retard en temps real en una fibra de 4 modes mitjançant la inscripció de 3 xarxes de difracció de període llarg en posicions concretes al llarg de l'enllaç de fibra. Com a validació de proba de concepte, demostrem experimentalment diferents funcionalitats de processat de senyal de Fotònica de Microones implementades en fibres multinucli i de pocs modes. Aquest treball obri el camí cap al desenvolupament del processat de senyal distribuït per a senyals de microones i ones mil·limètriques en una única fibra òptica. A més, aquestes línies de retard en temps real poden aplicar-se a una àmplia varietat de paradigmes de Tecnologies de la Informació i Comunicacions més enllà de les comunicacions radio sobre fibra, com es el cas de les comunicacions de banda ampla per satèl·lit, el sensat distribuït, la imatge mèdica, la tomografia òptica coherent i les comunicacions quàntiques. / Agradezco al Ministerio de Economía y Competitividad del Gobierno de España por la financiación recibida mediante la ayuda FPI. / García Cortijo, S. (2020). Distributed radiofrequency signal processing based on space-division multiplexing fibers [Tesis doctoral no publicada]. Universitat Politècnica de València. https://doi.org/10.4995/Thesis/10251/147858 / TESIS

As the family of two−dimensional(2D) materials has grown, two−dimensional heterostructure devices have emerged as great alternatives to replace conventional electronic materials and enable new functionality such as flexible and bendable electronics. The fabrication and performance of these devices depend critically on the understanding and ability to manipulate the mechanical interplay between the stacked materials. In this dissertation, we investigate adhesive interactions and determine the shear modulus of heterostructure devices made from Molybdenum Disulfide (MoS2). MoS2 has been attracting attention recently due to its semiconductor nature (having a direct band gap of 1.9 eV) along with its exceptional mechanical strength and flexibility. As the first step of our research, we suspended MoS2 flakes grown through chemical vapor deposition (CVD) over substrates made of metal (gold, titanium, chromium), semiconductor (germanium, silicon), insulator (silicon oxide), and semi-metal (graphite). Then, by creating pressure differences across the membrane, we forced MoS2 to bulge upward until we observe separation from the surface of the substrates. We demonstrated that MoS2 on graphite has the highest work of separation within the tested surface materials. Furthermore, we measured considerable adhesion hysteresis between the work of separation and the work of adhesion. We proposed that surface roughness and chemical interactions play a role in surface adhesion and separation of 2D materials. These experiments are critical to guiding the future design of electrical and mechanical devices based on 2D materials. Next, we measured the effective shear modulus of MoS2/few−layer graphene (FLG) heterostructures by employing a blister test. Again, by introducing a pressure differential across the suspended MoS2 membrane over the FLG substrate, the MoS2/FLG heterostructure peeled off from the silicon oxide surface once the critical pressure is exceeded. Incorporating a modified free energy model and Hencky’s axisymmetric membrane solution, we determine the average effective shear modulus of the heterostructure. This is the first experimental measurement of the shear modulus of heterostructure devices using a blister test and this platform can be extended to determine the shear modulus of other 2D heterostructures as well. / 2024-05-24T00:00:00Z

No description available.

No description available.