The influence of the distributor configuration on the mass transfer and bubble sizes in a 2-D FBR was studied for two types of distributor configurations: <ul><li> A novel multi-vortex (MV) distributor with tuyéres directed vertically and horizontally at different heights. </li><li> A standard perforated plate distributor (baseline). </li></ul> The ozone decomposition reaction over FCC catalyst was used as an indirect meas-ure of the interphase mass transfer in the bed. The ranged between 0.1 m/s and 0.35 m/s, with air as a fluidizing medium at ambient conditions. The MV distributor displayed a significant improvement in the conversion cies (χmeasured / χPFR). For all velocities an improvement of between 0% and 30% was measured (average improvement of 14.7%). At Uo ≤ 0.2 m/s the improvement distributed evenly about the mean improvement, with a minimum improvement of 10%. The Uc for the respective distributors were determined using absolute pressure fluctuations (baseline distributor) and visual observations (MV distributor) and it was found that the Uc for the MV distributor was lower than that for the baseline: <ul><li> Uc Baceline = 0.30 m/s.</li><li> Uc MV distributor = 0.25 m/s.</li></ul> The decrease in Uc indicated that the MV distributor induces faster onset of turbulent behaviour in the FBR which negates interphase mass transfer limitations in the FBR. The bubble sizes were measured visually and compared to a pressure signal decomposition technique. The bubble size growth for the MV distributor was estimated at .14 times that of the baseline. Two models were compared to the experimental results, the Kunii-Levenspiel three-phase model and the Thompson generalized bubble-turbulent model. The fitting parameters showed that the mass transfer for the MV distributor is significantly larger than that of the baseline. In addition the MV distributor decreased the axial dispersion in the FBR as Uo → Uoo, which improved the reactor performance to that of an ideal PFR. This phenomenon was observed at Uo > 0.33 m/s. / Dissertation (MEng)--University of Pretoria, 2011. / Chemical Engineering / unrestricted

Online decision making is a kind of machine learning problems where decisions are made in a sequential manner so as to accumulate as many rewards as possible. Typical examples include multi-armed bandit (MAB) problems where an agent needs to decide which arm to pull in each round, and network routing problems where each router needs to decide the next hop for each packet. Thompson sampling (TS) is an efficient and effective algorithm for online decision making problems. Although TS has been proposed for a long time, it was not until recent years that the theoretical guarantees for TS in the standard MAB were given. In this thesis, we first analyze the performance of TS both theoretically and practically in a special MAB called combinatorial MAB with sleeping arms and long-term fairness constraints (CSMAB-F). Then, we apply TS to a novel reactive network routing problem, called \emph{opportunistic routing without link metrics known a priori}, and use the proof techniques we developed for CSMAB-F to analyze the performance. / Graduate

Stochastic bandit algorithms are increasingly being used in the domain of recommender systems, when the environment is very dynamic and the items to recommend are frequently changing over time. While traditional approaches consider a single bandit instance which assumes all users to be equal, recent developments in the literature showed that the quality of recommendations can be improved when individual bandit instances for different users are considered and clustering techniques are used. In this work we develop an algorithm which clusters users based on the context at disposal using a Bayesian framework called Thompson Sampling, opposed to a similar algorithm called CAB recently presented at ICML 2017 (International Conference on Machine Learning), which uses a deterministic strategy. We show extensively through experiments on synthetic and real world data that the perfor- mance of our algorithm is in line with its deterministic version CAB when it comes to the quality of recommendations. On the other hand, our approach is relatively faster when considering running time. / Stokastiska bandit-algoritmer används allt mer för rekommenderande system där omgivningen är väldigt dynamisk och de rekommenderade föremålen ständigt byts ut. Medan traditionella tillvägagångssätt använder en enkel bandit-instans som antar alla användare att vara likadana, har litteraturen under senare tid visat att kvaliteten av rekommendationerna kan förhöjas med användarspecifika bandit-instanser och med hjälp av klustringsalgoritmer. I detta projekt har en algoritm utvecklats som klustrar användare baserat på kontexten för disposition genom att använda ett Bayesianskt ramverk som kallas Thompson-sampling, till skillnad från den liknande algoritmen CAD som nyligen presenterades vid ICML 2017 (International Conference on Machine Learning), som använder en deterministisk strategi. Projektet visar med stor omfattning, med hjälp av experiment som använt både syntetiska och reella data, att den framtagna algoritmens genererade rekommendationer håller likvärdig kvalitet som den deterministiska varianten av CAB. Däremot ger vårt presenterade tillvägagångssätt högre prestanda avseende tidsåtgång.

Soit sigma g,n une surface orientable de genre g avec n trous. Le groupe modulaire de sigma g,n agit sur divers complexes, comme le complexe de courbes et le complexe de décomposition en pantalons. Il a été prouvé, selon une approche initialement établie par Ivanov, que le groupe d'automorphismes de chacun de ces complexes est isomorphe au groupe modulaire. Cela implique notamment que le groupe des automorphismes extérieurs d'un sous-groupe d'indice fini du groupe modulaire est fini. Le but de cette thèse est de démontrer un résultat similaire s'appliquant à des surfaces de type infini de genre zéro. Pour cela, on définit un groupe modulaire asymptotique de ces surfaces, puis un complexe cellulaire localement infini sur lequel le groupe modulaire agit naturellement. On fait apparaitre des propriétés du groupe des automorphismes de chaque complexe en faisant agir les automorphismes sur des graphes auxiliaires. Le premier groupe modulaire étudiée est isomorphe au groupe de Thompson T. Le second est une extension du groupe modulaire universel de genre zéro. / Let sigma g,n be an orientable surface of genus g with n punctures. The mapping class group of sigma g,n acts on several complexes, for instance the curve complex or the pants complex of the surface. It is proved that the automorphism group of each of these complexes are isomorphic to the mapping class group. This implies in particular that the group of outer automorphisms of a finite index subgroup is finite. The purpose of this thesis is to prove a similar result on some surfaces of infinite type and genus zero. For this, we define an asymptotic mapping class group of these surfaces, and then a locally infinite cellular complex where the mapping class group acts naturally. It brings up some properties of the automorphism group of each cellular complex by making automorphisms act on auxiliary graphs. The first studied asymptotic mapping class group is isomorphic to the Thompson group T. The second one is an extension of the universal mapping class group of genus zero.

The aim of this student thesis is to achieve a translation of an excerpt of a previously untranslated book from English to Swedish. The translated material has been collected from the autobiographical book Kingdom of Fear by the author Hunter S. Thompson and consists partly of paratexts. The translation has been performed in accordance with a principle that was formed within the theoretical frames of Gideon Toury’s Descriptive Translation Studies. Accordingly, the discussion about the translation is based on the terms adequacy and acceptancy. The paper includes analyses of relevant aspects of the translated material, for example genre, paratext, and persona related problems. To illustrate the translation process, comments on the translation are also enclosed, which deal with solutions regarding grammatical and syntactical changes, imagery and cultural-specific expressions. / Syftet med detta examensarbete är att åstadkomma en översättning av ett utdrag från ett tidigare oöversatt verk från engelska till svenska. Det översatta materialet är hämtat från det autobiografiska verket Kingdom of Fear av författaren Hunter S. Thompson och består bland annat av paratexter. Översättningen har utförts efter en princip som formulerats inom de teoretiska ramarna för Gideon Tourys Descriptive Translation Studies. Därmed utgår diskussionen kring översättningen från termerna adekvans och acceptans. Arbetet innefattar analyser på relevanta aspekter av det översatta materialet, exempelvis genre, paratext och personaproblematik. För att åskådliggöra översättningsprocessen medföljer även översättningskommentarer som behandlar lösningar angående grammatiska och syntaktiska ändringar, bildspråk och kulturspecifika uttryck.

The winter ecology of mountain caribou (Rangifer tarandus caribou) in the North Thompson watershed of British Columbia was investigated over winters 1978-79 and 1979-80. The main objective of the study was to evaluate caribou movements and habitat use in relation to indices of energy cost of locomotion in snow and to forage availability, especially arboreal lichens. These data were collected in habitats from valley bottom to alpine throughout winter. Largely because of their high arboreal lichen productivity, mature forests are regarded by wildlife managers as essential winter habitat of caribou, leading to conflicts with forest harvesting. Data were compared between mature forests and other habitat types, to evaluate their importance to caribou. An index of caribou locomotion cost in snow was caribou track depth in snow. A significant regression between caribou track depth and human sinking depth in snow permitted an estimate of caribou locomotion costs in all habitats. Locomotion costs often were greater in cutovers than in mature forests, and broadly increased with elevation; while temporal trends were cyclical, due to alternating accumulations of fresh, soft snow followed by settlement and maturation of the surface snow. Analysis of fecal and rumen samples, and feeding-site inspections were used to determine caribou winter food habits. Arboreal lichens (Alectoria sp. and Bryoria spp.) dominated the diet by mid winter because terrestrial forage availability declined due to deep and crusted snowpacks. For each habitat, the absolute abundance of arboreal lichen was inventoried, and then this data together with snowpack measurements were used to estimate the relative availability of arboreal lichen over winter. Arboreal lichen availability was greatest in mature forests, and generally increased with elevation; it also increased within habitats as snow deepened and elevated caribou to higher forest canopy levels where greater quantities of lichen occurred. For the first time, radio telemetry was used to determine mountain caribou movements and habitat use. Observations of non-radiocollared caribou were also used in some analyses. In both used and unused habitats, estimated energy costs of locomotion and the availability of arboreal lichen were treated as indices of energy expenditure and of energy intake of foraging. These indices were qualitatively integrated in a net energy balance relationship to evaluate caribou movements and habitat use. Caribou appeared to follow a general optimizing strategy, balancing their energy expenditure for locomotion in snow against the energy available from forage, when both terrestrial and arboreal forages are considered. Throughout winter, caribou preferably used mature forests, which offered much greater energetic benefits than cutovers and immature forests. As snow in subalpine (Engelmann Spruce - Subalpine Fir Zone) and alpine summer habitats deepened over early winter, caribou migrated to lower subalpine and lower slope and valley (Interior Cedar - Hemlock Zone), mature forest habitats. Caribou locomotion conditions and forage availability, primarily of terrestrial forages, were most favourable at these lower elevations, despite lower arboreal lichen availability, because of snowfall interception by the forest canopy and lower snowfall. Firmer, mid-winter snowpack conditions allowed caribou to reascend to late winter range in higher elevation subalpine forests, which, because of greater arboreal lichen availabilities combined with moderated locomotion conditions, became the most favourable habitats. Minor elevational shifts during this period occurred in response to fluctuations in locomotion conditions caused by cycles of snow accumulation and snow settlement. This study confirmed that mature forests are required habitat for caribou throughout winter, by providing critical arboreal lichen forage, and compared to cutovers, having lower locomotion costs and greater availability of terrestrial forage. Proposed forest reserves above 1680 m elevation in the upper subalpine are insufficient therefore to ensure essential caribou winter habitat. Mature forests from valley bottoms to the lower subalpine must also be reserved. / Land and Food Systems, Faculty of / Graduate

(in English) The Master's Thesis surveys the origins of academic folkloristics in Finland characterised by the birth of so called Finnish school of folkloristics and historic- geographic method. It breaks downs the method's specifics, theoretical starting points and practical applications. The thesis is also concerned with the work of Finnish folklorist Satu Apo who uses emic units and her own structural classification for studying folk narrative.

Submitted by Johnny Rodrigues (johnnyrodrigues@ufcg.edu.br) on 2018-08-03T17:30:28Z No. of bitstreams: 1 LUÍS ANTÔNIO DANTAS - TESE PPGEP 2012..pdf: 23430964 bytes, checksum: 08c129d8e8b4124dc38c9768179bd8a6 (MD5) / Made available in DSpace on 2018-08-03T17:30:28Z (GMT). No. of bitstreams: 1 LUÍS ANTÔNIO DANTAS - TESE PPGEP 2012..pdf: 23430964 bytes, checksum: 08c129d8e8b4124dc38c9768179bd8a6 (MD5) Previous issue date: 2012-06 / Este trabalho tem como objetivo o desenvolvimento de um programa computacional para a simulação do processo de secagem de produtos biológicos; disponibiliza, num único módulo operacional, os modelos de simulação já consolidados, associados a um banco de dados especificamente modelado, que se prestará tanto à simulação quanto ao gerenciamento de informações oriundas de experimentações. Com base nas equações para determinação de equilíbrio higroscópico, camada fina, razão de conversão, calor específico do produto e calor latente de vaporização da água do produto, foi modelado um banco de dados que permite aos usuários (técnicos, pesquisadores, estudantes, professores) o estudo, a análise ou a experimentação de qualquer produto biológico. Estão disponíveis os seguintes modelos: de Hukill, aplicado ao secador de leito fixo; de Thompson, aplicado aos secadores de leito fixo, fluxo cruzado e fluxo concorrente; e de Michigan, aplicado aos secadores de leito fixo e de fluxo cruzado, com solução numérica por volumes finitos. No aplicativo, toda a formulação matemática e os coeficientes estão dispostos em meio externo, não sendo necessárias modificações, edições ou recompilações. No modelo de Michigan, podem-se variar o número de pontos nodais, o número máximo de iterações e/ou o tempo máximo. A metodologia utilizada possibilita que se associem usuário(s), produto(s), modelo utilizado, as equações e seus coeficientes, as condições de secagem, os resultados experimentais obtidos e a análise dos resultados de secagem. Essa dinâmica operacional mostrou-se eficiente e prática, tanto na repetição de experimentos, quanto nos tempos de apresentação dos resultados. A reutilização de dados é um fator importante, uma vez que toda a gama de informações é armazenada, dispensando a utilização de ações repetitivas que maximizam erros. / The objective of this work is the development of a computational program for the simulation of the drying process of biological products. It offers, in a single operating module, the simulation models already Consolidated, associated with a database specifically modeled, which provide both simulation and managing information from drying tests. Based on the equations to determine hygroscopic equilibrium, thin layer, conversion ratio, specific heat and latent heat of vaporization of water of the product, a database was modeled that allows users (technical staff, researchers, students, teachers) to study, analyze or test any biological product. The following models are available: Hukill applied to fixed-bed dryers; Thompson applied to fixed-bed, crossflow and concurrent-flow dryers; Michigan, applied to fixed-bed and crossflow dryers with numerical solution by finite volume methods. In the application, ali mathematical formulation and coefficients are externally arranged, in such a way that modifications, edits or recompilations are not necessary. In the Michigan drying simulation model, the number of nodal points, the maximum number of iterations and/or the maximum time can be varied. The methodology used allows joining user(s), product(s), model used, the equations and their coefficients, the drying conditions, the experimental results obtained and the analysis of the drying results. Reusing data is an important factor, since the full range of information is stored, eliminating the use of repetitive actions that maximize errors.

Cette thèse concerne le groupe T de Thompson. Ce groupe simple infini et finiment présenté est généralement vu comme un sous-groupe du groupe des homéomorphismes dyadiques du cercle unité qui sont linéaires par morceaux et préservent l'orientation («T linéaire par morceaux»). Cependant, T peut aussi être vu comme: 1.- le groupe des classes d'équivalence des paires équilibrées d'arbres binaires finis («T combinatoire»), 2.- un sous-groupe du groupe des homéomorphismes de la droite projective réelle qui préservent l'orientation et sont «PSL(2,Z) par morceaux» («T projectif par morceaux»), et 3.- le groupe modulaire asymptotique de l'épaissi, dans le plan hyperbolique, de l'arbre régulier de valence 3 («T modulaire»).On montre d'abord que la copie canonique de PSL(2,Z) obtenue à partir de «T projectif par morceaux» est un sous-groupe non distordu de T. Pour cela, on transporte ce sous-groupe pour obtenir une caractérisation dans le «T combinatoire», ce qui permet d'estimer la longueur des mots de ses éléments. La non-distorsion est alors une conséquence des propriétés métriques de T établies par Burillo-Cleary-Stein-Taback. Comme corollaire, T a des sous-groupes non distordus isomorphes au groupe libre engendré par deux éléments. Qui plus est, PSL(2,Z) est aussi donné explicitement sous forme «linéaire par morceaux».Le deuxième résultat utilise «T modulaire» pour prouver qu'il y a exactement f(n) classes de conjugaison d'éléments d'ordre n dans T, où f est l'indicatrice d'Euler. Étant donné un élément de torsion t de T d'ordre n, on trouve une triangulation du disque de Poincaré qui est invariante sous l'action de T sauf dans un polygone convexe à n côtés. On construit ensuite un complexe cellulaire C contractile et simplement connexe sur lequel le groupe T agit par automorphismes, et qui est minimal pour ces propriétés. Le groupe d'automorphismes de C est essentiellement T lui même (c'est une extension de T par le groupe d'ordre 2). Ce complexe cellulaire peut être vu comme une généralisation des associaèdres deStasheff dans le cas d'un polygone convexe à une infinité de côtés. L'action de T sur C est transitive sur les arêtes et les sommets, et plus généralement, sur les cellules «de type associaèdre» de toute dimension.La partie finale décrit les premières étapes d'un programme de recherche. On utilise l'interprétation géométrique du 1-squelette de C en termes de triangulations dyadiques du disque de Poincaré pour définir un bord géométrique à l'infini. Bien qu'on ait prouvé auparavant que le 1-squelette de C n'est pas hyperbolique, la construction s'inspire de celle de Gromov et permet la description de certains points du bord. / This PhD thesis is concerned with Thompson's group T. This infinite, finitely presented, simple group is usually seen as a subgroup of the group of dyadic, piecewise linear, orientation-preserving homeomorphisms of the unit circle (piecewise linear T). However, T can also be identified to: 1.- a group of equivalence classes of balanced pairs of finite binary trees (combinatorial T), 2.- a subgroup of piecewise PSL(2,Z), orientation-preserving homeomorphisms of the projective real line (piecewise projective T), and 3.- the asymptotic mapping class group of a fattened complete trivalent tree in the hyperbolic plane (modular T). The first result shows that the canonical copy of PSL(2,Z) obtained from the piecewise projective T is a non-distorted subgroup of T. For this, one carries over this subgroup to obtain a characterization into combinatorial T, from which the word length of its elements can be estimated. Then, non-distortion follows from the metric properties of T established by Burillo-Cleary-Stein-Taback. As a corollary, T has non-distorted subgroups isomorphic to the free non-abelian group of rank 2. Furthermore, PSL(2,Z) is also explicitly given in the piecewise linear form.The second result uses modular T to state that there are exactly f(n) conjugacy classes of elements of order n, where f is the Euler function. Given a torsion element t of T of order n, a dyadic triangulation of the Poincaré disc which is invariant under the action of t modulo a convex polygon with n sides is found.The third result constructs a minimal simply-connected contractible cellular complex C on which the group T acts by automorphisms. The automorphism group of C is essentially T itself (strictly speaking it is an extension of T by the group of order 2). The cellular complex C can be seen as a generalization of Stasheff's associahedra for an infinitely sided convex polygon. The action of T on C is transitive on vertices and edges and, plus generally, on associahedral type cells in all dimensions.The final part deals with the first steps of a research project. One uses the geometric interpretation of the 1-skeleton of C in term of dyadic triangulations of the Poincaré disc to define a geometric boundary at infinity. Although the 1-skeleton of C is proved not to be hyperbolic, the construction imitates Gromov's construction of the boundary of hyperbolic spaces, and allows the description of the nature of some of the boundary points.

Made available in DSpace on 2016-06-02T19:51:25Z (GMT). No. of bitstreams: 1 DissFA.pdf: 692673 bytes, checksum: c17603e3bc9d4ee95fd60f2c08410cec (MD5) Previous issue date: 2003-12-01 / Universidade Federal de Minas Gerais / This research have two main goals: i) the Production Planning and Control (PPC) top-authors identification according to an effective population (EF)with 592 international researchers; and ii) assessing the gap between theory and practice on the PPC area at São Carlos region. To determine the item (i) it was done a research through internet with an initial sample of 205 researchers, as not all e-mails worked, the effective sample (ES) was reduced to 168 researchers. The obtained result was estimated for the corrected effective population (CEP), that corresponds to the number of people whose e-mails worked, through the Horwitz-Thompson estimator. At the end of this stage we obtained a list containing the top-authors of the area and their main works as well as we accomplished analysis involving variables like the region, the approach and the number of years the respondents are active on PPC research. To accomplish the item (ii) we collected data from a group of São Carlos academics and from 11 important companies of the region. We lifted up several causes of the existence of this gap in the area and we also suggested several initiatives to decrease it. / Esta pesquisa possuiu dois objetivos principais: i) identificação dos principais autores do PCP na visão de uma população efetiva (PE) de 592 pesquisadores internacionais; e ii) análise da lacuna entre a teoria e a prática da área de PCP na região de São Carlos. Para determinar o item (i) foi realizada uma pesquisa via internet com uma amostra inicial de 205 pesquisadores, e que no decorrer da pesquisa foi reduzida a 168 pessoas (aqueles cujos e-mails funcionavam); chamamos este grupo de amostra efetiva (AE). O resultado obtido foi estimado para a população efetiva corrigida (PEc), que corresponde ao número de pessoas da PE cujos e-mails funcionavam, por meio do estimador de Horwitz-Thompson. Ao final desta etapa obtivemos uma lista contendo os principais autores da área e suas principais obras bem como realizamos análises envolvendo variáveis como a região, a abordagem e o tempo de pesquisa dos entrevistados. Para realizar o item (ii) levantamos dados de um grupo de acadêmicos da região de São Carlos e de onze empresas de grande porte da mesma região. Foram levantadas várias causas da existência desta lacuna na região e também foram sugeridas várias iniciativas para diminuí-la.