[ES] En esta Tesis Doctoral, se propone diferentes técnicas de acoplo y conversión modal destinadas a aumentar la capacidad de transporte en sistemas de telecomunicaciones sobre fibra óptica. En particular, el objetivo principal es el desarrollo de la tecnología necesaria para conseguir una multiplexación modal utilizando un número limitado de modos, de manera controlada. Para ello, se estudian dos escenarios MDM con dos longitudes de onda distinta. Por un lado, usando la longitud de onda de 850 nm sobre SSMF favoreciendo la utilización de componentes ópticos y electro-ópticos de coste mucho menor que sus equivalentes en la banda C+L. Esta novedosa tecnología de transmisión permitirá una nueva generación de interconexiones ópticas de muy alta capacidad aplicable a enlaces chip-a-chip, a backplanes ópticos y también a clústeres de computación de altas prestaciones y centros de conmutación de red. Por otro lado, usando la longitud de onda de 1550 nm sobre guías ópticas basadas en SOI, es decir, Si (silicio) sobre sustrato de SiO2 (óxido de silicio) favoreciendo la utilización de dispositivos basados en tecnología integrada que ofrecen un menor tamaño, mejor repetibilidad y robustez que los dispositivos basados en fibra óptica. Para ello, se propone el uso de acopladores ópticos fusionados siendo un elemento indispensable a la hora de multiplexar y demultiplexar los distintos modos ópticos en un enlace MDM a 850 nm. Esta técnica permite multiplexar/demultiplexar los modos ópticos cuando el tipo de acoplador óptico utilizado es simétrico (DC, del inglés directional coupler), siendo necesario la utilización de un conversor de modos. También se estudia la posibilidad de convertir el modo óptico mediante la utilización de un acoplador óptico asimétrico (ADC, del inglés asymmetrical directional coupler), no siendo necesario utilizar un conversor de modos y simplificando el esquema MDM. Además, en esta tesis doctoral también se propone y evalúa el diseño de un conversor de modos mecánico basado en SSMF. Esta técnica permite obtener el primer modo de orden superior con una alta calidad y sin la necesidad de utilizar un ADC. Después de esto, se propone y evalúa la posibilidad de utilizar acopladores comerciales (diseñados a 1550 nm) a la longitud de onda de 850 nm permitiendo de esta forma reducir la necesidad de utilizar acopladores ópticos y conversores modales específicamente diseñados en dicha longitud de onda. Esta técnica reduciría los costes del sistema al necesitar un menor número de dispositivos y aprovechar los dispositivos diseñados a 1550 nm, siendo más económicos que los diseñados a 850 nm. En esta Tesis también se propone el uso de ADCs en guías strip basadas en SOI para la conversión y multiplexación de los modos ópticos desde la guia fundamental a la guia de dos modos, a la longitud de onda de 1550 nm. Para ello se estudia y demuestra experimentalmente diferentes diseños con el fin de obtener el diseño más robusto frente a las tolerancias de fabricación consiguiendo un rendimiento óptimo. Además, el uso de DCs sobre guías ridge es comúnmente utilizado y ofrece mejores prestaciones que el basado en guías strip, por ese motivo esta Tesis estudia y evalúa el uso de ADCs sobre guías ridge mediante el método de análisis de los índices efectivos de los supermodos par e impar. De esta forma se realiza una comparación entre los diseños óptimos de ambas estructuras (strip y ridge) con el objetivo de averiguar qué diseño ofrece mejores prestaciones. Por último, se propone y estudia el diseño de un acoplador grating capaz de multiplexar y demultiplexar los modos ópticos del modo fundamental y del primer orden superior desde la guia óptica a la fibra óptica y viceversa. Para ello se proponen diferentes diseños con el objetivo de conseguir un diseño más tolerante y eficiente frente a los errores por desalineamiento obteniendo un acoplo óptimo. / [CAT] En aquesta Tesi Doctoral, es proposen diferents tècniques d'acoblament i conversió modal destinades a augmentar la capacitat de transport en sistemes de telecomunicacions sobre fibra òptica. En particular, l'objectiu principal és el desenrotllament de la tecnologia necessària per a aconseguir una multiplexació modal utilitzant un número limitat de modes, de manera controlada. Per a això, s'estudien dos escenaris MDM amb dos longituds d'onda distinta. D'una banda, usant la longitud d'ona de 850 nm sobre SSMF afavorint la utilització de components òptics i electro-òptics de cost molt menor que els seus equivalents en la banda C+L. Aquesta nova tecnologia de transmissió permetrà una nova generació d'interconnexions òptiques de molt alta capacitat aplicable a enllaços chip-a-chip, a backplanes òptics i també a clústers de computació d'altes prestacions i centres de commutació de xarxa. D'altra banda, usant la longitud d'ona de 1550 nm sobre guies òptiques basades en SOI, és a dir, Si (silici) sobre substrat de SiO2 (òxid de silici) afavorint la utilització de dispositius basats en tecnologia integrada que ofereixen una menor grandària, millor repetibilitat i robustesa que els dispositius basats en fibra òptica. Per a això, es proposa l'ús d'acobladors òptics fusionats sent un element indispensable a l'hora de multiplexar i demultiplexar els distints modes òptics en un enllaç MDM. Aquesta tècnica permet multiplexar/demultiplexar els modes òptics quan el tipus d'acoblador òptic utilitzat és simètric (DC, de l'anglès directional coupler), sent necessari la utilització d'un convertidor de modes. També s'estudia la possibilitat de convertir el mode òptic per mitjà de la utilització d'un acoblador òptic asimètric (ADC, de l'anglès asymmetrical directional coupler), no sent necessari utilitzar un convertidor de modes i simplificant l'esquema MDM. Es mes, en aquesta tesi doctoral també es proposa i avalua el disseny d'un convertidor de modes mecànic basat en SSMF. Aquesta tècnica permet obtindre el primer mode d'orde superior amb una alta qualitat sense la necessitat d'utilitzar un ADC. Després d'açò, es proposa i avalua la possibilitat d'utilitzar acobladors comercials (dissenyats a 1550 nm) a la longitud d'ona de 850 nm permetent d'esta manera reduir la necessitat d'utilitzar acobladors òptics i convertidors modals específicament dissenyats en la dita longitud d'ona. Aquesta tècnica reduiria els costos del sistema al necessitar un menor nombre de dispositius i aprofitant els dispositius dissenyats a 1550 nm, sent més econòmics que els dissenyats a 850 nm. En aquesta Tesi també es proposa l'ús de ADCs en guies strip basades en SOI per a la conversió i multiplexació dels modes òptics des de la guia fonamental a la guia de dos modes, a la longitud d'ona de 1550 nm. Per a això s'estudia i demostra experimentalment diferents dissenys a fi de obtindré el disseny més robust enfront les toleràncies de fabricació aconseguint un rendiment òptim. A més, l'ús de DCs sobre guies ridge és comunament utilitzat i ofereix millors prestacions que el basat en guies strip, per eixe motiu aquesta Tesi estudia i avalua l'ús de ADCs sobre guies ridge per mitjà del mètode d'anàlisi dels índexs efectius dels supermodes parell i imparell. D'aquesta manera es realitza una comparació entre els dissenys òptims de les dos estructures (strip i ridge) amb l'objectiu d'esbrinar quin disseny ofereix millors prestacions. Finalment, es proposa i estudia el disseny d'un acoblador grating capaç de multiplexar i demultiplexar els modes òptics del mode fonamental i del primer orde superior des de la guia òptica a la fibra òptica i viceversa. Per a això es proposen diferents dissenys amb l'objectiu d'aconseguir un disseny més tolerant i eficient enfront dels errors per desalineament obtenint un acoblament òptim. / [EN] In this Ph.D. thesis, different mode coupling and mode conversion techniques with the aim to increase the transport capacity in telecommunications systems over optical fiber are proposed. Concretely, the main aim is the development of the technology to achieve MDM using a limited controlled number of modes. Two different MDM scenarios based on two distinct wavelengths have been considered. On one hand, using the 850 nm wavelength over SSMF favors the use of optical and electro-optical devices with costs much lower than their equivalent in the C+L band. This novel transmission technology enables a new generation of very high capacity optical interconnections applicable to chip-to-chip links, to optical backplanes, and also to high-performance computing clusters and network switching centre interconnections. On the other hand, using the 1550 nm wavelength over optical waveguides based on SOI, i.e., Si (Silicon) above SiO2 substrate (silicon oxide), allows the use of integrated devices offering a less size, better repeatability and robustness in comparison with the optical fiber devices. Fused fiber couplers are proposed as key elements to (de)multiplex different fiber modes in a MDM link at 850 nm. The use of a symmetric directional coupler (DC) as a (de)multiplexer requires the use of an additional mode converter. The use of an asymmetrical directional coupler (ADC) as optical (de)multiplexer and mode converter is proposed, avoiding the necessity of an additional mode converter and simplifying the MDM scheme. Furthermore, in this Ph.D. thesis it is also proposed and evaluated the design of a mechanical mode converter at 850 nm using a SSMF. This technique permits to obtain the first high order mode with high quality and without the necessity of using an ADC. After that, it is analyzed and investigated the employment of commercial optical couplers (designed at 1550 nm) at 850 nm wavelength operation, thus avoiding the use of optical couplers and mode converters specifically designed at 850 nm wavelength. The MDM system costs are reduced as fewer devices are required and commercial components designed at 1550 nm are cheaper than the counterparts at 850 nm. In this Ph.D. thesis it is also considered the employment of ADCs over strip waveguides based on SOI technology for the conversion and multiplexing of the optical modes, from single-mode waveguide to high order mode waveguide at the 1550 nm wavelength. Thus, it has been studied and experimentally investigated different designs aimed to achieve the most robust configuration, in which the yield is less affected by the fabrication tolerances. Furthermore, the use of DCs over ridge waveguides is commonly employed and it offers better performance than strip waveguides. For this reason, the Ph.D. thesis studies and evaluates the use of ADCs with ridge waveguides by considering the effective refractive indexes of the even and odd supermodes analysis. In this way, a comparison between strip and ridge structures is done in order to find the optimum design that offer the best features. Finally, it is analyzed the design of a grating coupler capable of multiplexing and demultiplexing the fundamental and the high order mode from the waveguide to the optical fiber and vice versa. Thus, different designs are evaluated in order to achieve a design more robust and efficient to the coupling misalignments. / García Rodríguez, D. (2018). Few-Mode Transmission Technology for Ultra-High Capacity Optical Networks [Tesis doctoral no publicada]. Universitat Politècnica de València. https://doi.org/10.4995/Thesis/10251/115938 / TESIS

The continued growth of the optoelectronics industry and the emergence of wearable and flexible electronics will continue to place an ever increasing pressure on replacing ITO, the most widely used transparent conducting electrode (TCE). Among the various candidates, graphene shows the highest optical transmittance in addition to promising electrical transport properties. The currently available large-scale synthesis routes of graphene result in polycrystalline samples rife with grain boundaries and other defects which limit its transport properties. Chemical doping of graphene is a viable route towards increasing its conductivity and tuning its work function. However, dopants are typically present at the surface of the graphene sheet, making them highly susceptible to degradation in environmental conditions. Few-layers graphene (FLG) is a more resilient form of graphene exhibiting higher conductivity and performance stability under stretching and bending as contrasted to single-layer graphene. In addition FLG presents the advantage of being amenable bulk doping by intercalation. Herein, we explore non-covalent doping routes of CVD FLG, such as surface doping, intercalation and combination thereof, through in-depth and systematic characterization of the electrical transport properties and energy levels shifts. The intercalation of FLG with Br2 and FeCl3 is demonstrated, showing the highest improvements of the figure of merit of TCEs of any doping scheme, which results from up to a five-fold increase in conductivity while maintaining the transmittance within 3% of that for the pristine value. Importantly the intercalation yields TCEs that are air-stable, due to encapsulation of the intercalant in the bulk of FLG. Surface doping with novel solution-processed metal-organic molecular species (n- and p-type) is demonstrated with an unprecedented range of work function modulation, resulting from electron transfer and the formation of molecular surface dipoles. However, the conductivity increases compared modestly to intercalation as the electron transfer is limited to the uppermost graphene layers. Finally, a novel and universal multi-modal doping strategy is developed, thanks to the unique platform offered by FLG, where surface and intercalation doping are combined to mutually achieve high conductivity with an extended tunability of the work function. This work presents doped-FLG as a prospective and versatile candidate among emerging TCEs, given the need for efficient and stable doping routes capable of controllably tuning its properties to meet the criteria of a broad range of applications.

Ciphers are encrypted documents created to hide their content from those who were not the receivers of the message. Different types of symbols, such as zodiac signs, alchemical symbols, alphabet letters or digits are exploited to compose the encrypted text which needs to be decrypted to gain access to the content of the documents. The first step before decryption is the transcription of the cipher. The purpose of this thesis is to evaluate an automatic transcription tool from image to a text format to provide a transcription of the cipher images. We implement a supervised few-shot deep-learning model which is tested on different types of encrypted documents and use various evaluation metrics to assess the results. We show that the few-shot model presents promising results on seen data with Symbol Error Rates (SER) ranging from 8.21% to 47.55% and accuracy scores from 80.13% to 90.27%, whereas SER in out-of-domain datasets reaches 79.91%. While a wide range of symbols are correctly transcribed, the erroneous symbols mainly contain diacritics or are punctuation marks.

Ce manuscrit décrit l’étude et la mise en œuvre d’une source laser ultrarapide à taux de répétition élevé dans l’infra-rouge moyen, pour des applications à la physique des champs forts et à la spectroscopie moléculaire multidimensionnelle. Cette source est basée sur le phénomène d’amplification paramétrique optique à dérive de fréquence, qui permet la génération d’impulsions de quelques cycles optiques.Tout d’abord, nous présentons les applications de ces sources, ainsi que leurs paramètres importants, débouchant sur un cahier des charges pour la source à l’étude. Un état de l’art des sources paramétriques présentées dans la littérature scientifique nous permet ensuite d’appréhender la diversité des architectures et des performances atteintes. En particulier, nous soulignons les points cruciaux que sont la nature et les performances du laser de pompe, le mécanisme de génération du signal à amplifier, ainsi que la robustesse de la synchronisation temporelle entre le signal et la pompe.Nous étudions ensuite la possibilité d’émettre un signal autour de la longueur d’onde de 1,55 µm à partir d’impulsions femtoseconde de pompe à 1,03 µm par génération de supercontinuum dans un cristal massif de YAG. Nous menons ainsi une étude détaillée des propriétés de la partie infra-rouge du supercontinuum obtenu, en termes de contenu spectral, cohérence, propriétés statistiques tir à tir et long terme, et propriétés spatiales. Cette étude nous permet de conclure sur la validité de cette approche pour générer le signal à amplifier.Nous arrivons donc à définir une architecture inédite basée sur l’utilisation d’un laser de pompe basé sur un amplificateur à fibre dopée ytterbium de forte énergie délivrant des impulsions de 300 fs 400 µJ à la cadence de 125 kHz. La durée courte rendue possible par le choix de cette technologie de pompe nous permet de bénéficier d’un certain nombre d’avantages importants : la génération efficace de supercontinuum autour de 1,55 µm, ce qui entraine une synchronisation temporelle très robuste entre pompe et signal. D’autre part le couple étireur – compresseur est constitué de simples lames de matériaux massifs, ce qui permet une grande efficacité et une gestion simplifiée de la phase spectrale. Enfin, la courte durée de pompe augmente le seuil de dommage en intensité crête, ce qui permet l’utilisation de cristaux non linéaires courts et augmente la bande spectrale d’amplification. Des expériences supplémentaires sont menées pour étudier les phénomènes limitant la puissance au sein des cristaux de MgO:PPLN. Les étages d’amplification sont tous réalisés en géométrie colinéaire, ce qui permet d’utiliser le signal et l’idler sans introduction de chirp angulaire. Toutes ces caractéristiques permettent la génération de deux faisceaux en sortie portant des impulsions de 50 fs 20 µJ à 1550 nm et 70 fs 10 µJ à 3,1 µm. / This thesis describes the design and construction of an ultrafast high repetition rate laser source in the mid-IR, for applications in strong-field physics and multidimensional molecular spectroscopy. This source is based on optical parametric chirped-pulse amplification, allowing the generation of few-cycle pulses.We first present some applications of these lasers, along with important parameters, to define specifications for the considered source. We then briefly outline the state of the art of similar ultrafast sources described in the literature, to highlight the variety of architectures and performances. In particular, several key points are identified, namely the nature and performances of the pump laser source, the method to generate a seeding signal, and the robustness of temporal synchronization between pump and signal pulses.We proceed to study the possibility of emitting a seed signal around 1.55 µm wavelength by supercontinuum generation in a bulk YAG crystal from femtosecond pump pulses at 1.03 µm. A detailed analysis of the properties of the infrared spectral content of the supercontinuum is carried out, focusing on spectral bandwidth, coherence, shot-to-shot and long term stability, and spatial properties. This work allows us to conclude that supercontinuum generation is a valid approach to generate the seed signal.This leads us to define a novel architecture built around an ytterbium-doped fiber femtosecond pump source delivering 300 fs 400 µJ pulses at a repetition rate 125 kHz. The short pump pulse duration compared to bulk Yb:YAG or Nd:YVO4 based systems results in a number of important advantages. First, it allows efficient seeding at 1550 nm using supercontinuum generation directly from the pump pulses in a bulk YAG crystal, resulting in extremely robust passive pump – signal synchronization. The short pump pulse duration also allows the use of millimeter to centimeter lengths of bulk materials to provide stretching and compression for the signal and idler, which minimizes the accumulation of higher-order spectral phase. Finally, the shorter pump pulse duration increases the damage peak intensity, permitting the use of shorter nonlinear crystals to perform the amplification, which increases the spectral bandwidth of the parametric process. Additional experiments are performed to sort out the phenomena that limit power scaling in MgO:PPLN crystals. The OPCPA stages are all operated in collinear geometry, allowing the use of both signal and idler without the introduction of angular chirp on the latter. These points result in the dual generation of 70 fs 23 µJ signal pulses at 1550 nm and 60 fs 10 µJ idler pulses at 3070 nm from a simple setup.

One common shortcoming of modern computer vision is the inability of most models to generalize to new classes—one/few shot image recognition. We propose a new problem formulation for this task and present a network architecture and training methodology to solve this task. Further, we provide insights into how careful focus on how not just the data, but the way data presented to the model can have significant impact on performance. Using these method, we achieve high accuracy in few-shot image recognition tasks.

Thousands of historical ciphers, encrypted manuscripts, are stored in archives across Europe. Historical cryptology is the research field concerned with studying these manuscripts - combining the interest of humanistic fields with methods of cryptography and computational linguistics. Before a cipher can be decrypted by automatic means, it must first be transcribed into machine-readable digital text. Image processing techniques and Deep Learning have enabled transcription of handwritten text to be performed automatically, but the task faces challenges when ciphers constitute the target data. The main reason is a lack of labeled data, caused by the heterogeneity of handwriting and the tendency of ciphers to employ unique symbol sets. Few-Shot Learning is a machine learning framework which reduces the need for labeled data, using pretrained models in combination with support sets containing a few labeled examples from the target data set. This project is concerned with evaluating a Few-Shot model on the task of transcription of historical ciphers. The model is tested on pages from three in-domain ciphers which vary in handwriting style and symbol sets. The project also investigates the use of further fine-tuning the model by training it on a limited amount of labeled symbol examples from the respective target ciphers. We find that the performance of the model is dependant on the handwriting style of the target document, and that certain model parameters should be explored individually for each data set. We further show that fine-tuning the model is indeed efficient, lowering the Symbol Error Rate (SER) at best 27.6 percentage points.

The Pareto Principle, also known as the 80–20 rule, is the observation that 20% of input (e.g., workers) produce 80% of the results. Consistent with this principle, previous research has identified a group (20%) of families, the Vital Few, who provide a disproportionate amount of foster care and are more willing to foster children with special needs. The ability to predict the emergence of these families has important implications for recruitment, support and placement stability, as well as longevity of foster families. This study replicated and extended previous research by conducting a follow-up study of 107 families (90% response rate) 17 years after pre-service training. Consistent with previous research we found a small proportion (10%) of families who provide a disproportionate amount of care in terms of length of service and number of children fostered, approved to foster, adopted, and removed at families’ request. At the completion of pre-service training Vital Few families were more likely to have had previous foster parent experience and one or more children in their homes; mothers and fathers in the Vital Few were older, and fathers reported less education. Also, at pre-service training more Vital Few families said they would foster sibling groups (100 vs. 64%), but there were no other differences in terms of willingness to foster children with special needs. This study further validates the utility of the Pareto Principle for understanding foster families and, by extension, has important implications for the well-being and stability of foster children.

There is a small, methodologically diverse body of research indicating that approximately 20% of families provide disproportionate amounts of foster caregiving, place fewer restrictions on characteristics of children they are willing to foster and actually do foster, and provide caregiving environments as good as or better than those provided by other foster families. Cherry and Orme (2013) conceptualized this phenomenon in terms of the Pareto Principle, also known as the 80-20 rule or the Vital Few, and they refer to these 20% of families as the Vital Few and the remaining 80% as the Useful Many. This review will examine and synthesize the available research on Vital Few foster families and explore next steps in the development of this body of research.

The Pareto Principle, also known as the 80-20 rule or the Vital Few, has been successfully used as a framework to identify the small proportion of highly productive foster parents who provide a disproportionate amount of care. This study replicated and extended this research using a nationally representative sample of foster families ( N=. 876) with a focus on willingness to foster, and actually fostering, children with special needs. Using latent class analysis, two classes of foster parents were identified: one accounted for 19% of respondents and the other 81%. We refer to the former as the Vital Few and the latter as the Useful Many. Vital Few respondents fostered 74.2% of foster children - 11 times more than the Useful Many, although only fostering two times longer. They also had almost 1-1/2 times as many foster children in their homes at the time of the study. Notably, the Vital Few were willing to foster more types of children with special needs and a higher percentage had actually fostered children with each of the seven types of special needs studied. The classes were similar demographically except that Vital Few respondents were less likely to work outside the home and Vital Few mothers were slightly less educated as compared to Useful Many mothers. This study further validates the utility of the Pareto Principle for understanding foster parents and, by extension, has important implications for the well-being and stability of foster children with special needs. Considerations for supporting the Vital Few, including education and training needs, are discussed.

<p>  The adiabatic hyperspherical representation has been applied to study few-body systems in both ultracold atomic physics and low energy nuclear physics, as it is a powerful tool that can be used to solve a variety of few-body Hamiltonian's across a wide range of disciplines in physics. In conjunction with the adiabatic hyperspherical representation, we utilized an explicitly correlated Gaussian basis expansion, different from the traditional hyperspherical harmonic expansion typically used with this method. In atomic physics, we applied this method to study the four-body (e^-e^-e⁺e⁺) coulombic system to study positronium-positronium collisions and to get a theoretical value of the 1<em>s</em>-2<em>s</em> scattering length. This work is published in [Phys. Rev. A 100, 012711 (2019)]. We also looked at few-body physics near the unitary limit, particularly near the <em>s</em>- and <em>p</em>-wave unitary limits where the dominant length scale is the scattering length and scattering volume. On this front, we studied universal physics in this regime for the equal-mass system. This work is published in [Phys. Rev. A 106, 023304 (2022)]. This method was further applied to few-body nuclear physics.</p> <p><br></p> <p>  We treat the three and four neutron scattering problems in the <em>N</em>-body continuum to understand and gain insight into possible few-neutron resonances, most notably whether a four-neutron resonance exists. There have been many conflicting theoretical results on whether a four-neutron resonance exists that stemmed from a recent experiment by Kisimori et al. in 2016 [Phys. Rev. Lett. 116, 052501 (2016)]. To provide further theoretical insight on this problem, we use the adiabatic hyperspherical toolkit to probe the scattering continuum and from the 4<em>n</em> density of states, conclude that there is no 4<em>n</em> resonance state. Our work on this is published in [Phys. Rev. Lett. 125, 052501 (2020)] and [Phys. Rev. C 103, 024004 (2021)]. There are other few-body systems in nuclear physics that are explored in the adiabatic hyperspherical representation, including systems like the triton, helium-3, and helium-4 nuclei to visualize and characterize the different reaction pathways the <em>N</em>-body system can fragment into at a given collision energy.</p>