Dissertations / Theses on the topic 'Weight matrices'

In this study, for the well-known Analytic Hierarchy Process (AHP) method a new approach to interval priority weight generation from interval comparison matrix is proposed. This method can be used for both inconsistent and consistent matrices. Also for the problems having more than two hierarchical levels a synthesizing heuristic is presented. The performances of the methods, interval generation and synthesizing, are compared with the methods that are already available in the literature on randomly generated matrices.

Eukaryotní DNA se váže kolem nukleozomů, čím ovplyvnuje vyšši strukturu DNA a přístup k vazebním mistům pro všeobecní transkripční faktory a oblasti genů. Je proto důležité vědet, kde se nukleozomy vážou na DNA, a jak silná tato vazba je, abychom mohli porozumět mechanizmům regulace genů. V rámci projektu byla implementována nová metoda pro predikci nukleozomů založená na rozšíření Skrytých Markovových modelů, kde jako trénovací a testovací sada posloužila publikována data z Brogaard et al. (Brogaard K, Wang J-P, Widom, J. Nature 486(7404), 496-501 (2012). doi:10.1038/nature11142). Správne predikováno bylo zhruba 50% nukleozomů, co je porovnatenlný výsledek s existujícimi metodami. Okrem toho byla provedena řada experimentů popisující vlastnosti sekvencí nukleozomů a ich organizace.

This dissertation originates from a classical result that the lp-stability of the convolution operator associated with a summable sequence are equivalent to each other for different p . This dissertation is motivated by the recent result by C. E. Shin and Q. Sun (Journal ofFunctional Analysis, 256(2009), 2417-2439), where the lp-stability of infinite matrices in the Gohberg-Baskakov-Sjostrand class are proved to be equivalent to each other for different p. In the dissertation, for an infinite matrix having certain off-diagonal decay, its weighted lp-stability for different p are proved to be equivalent to each other and hence a result by Shin and Sun is generalized.<br>Ph.D.<br>Department of Mathematics<br>Sciences<br>Mathematics PhD

Computing the weighted geometric mean of large sparse matrices is an operation that tends to become rapidly intractable, when the size of the matrices involved grows. However, if we are not interested in the computation of the matrix function itself, but just in that of its product times a vector, the problem turns simpler and there is a chance to solve it even when the matrix mean would actually be impossible to compute. Our interest is motivated by the fact that this calculation has some practical applications, related to the preconditioning of some operators arising in domain decomposition of elliptic problems. In this thesis, we explore how such a computation can be efficiently performed. First, we exploit the properties of the weighted geometric mean and find several equivalent ways to express it through real powers of a matrix. Hence, we focus our attention on matrix powers and examine how well-known techniques can be adapted to the solution of the problem at hand. In particular, we consider two broad families of approaches for the computation of f(A) v, namely quadrature formulae and Krylov subspace methods, and generalize them to the pencil case f(A\B) v. Finally, we provide an extensive experimental evaluation of the proposed algorithms and also try to assess how convergence speed and execution time are influenced by some characteristics of the input matrices. Our results suggest that a few elements have some bearing on the performance and that, although there is no best choice in general, knowing the conditioning and the sparsity of the arguments beforehand can considerably help in choosing the best strategy to tackle the problem.

O objetivo deste trabalho é comparar as estruturas de vizinhanças espaciais ou matrizes de pesos espaciais da classe de modelos autorregressivos e de médias móveis espaço-temporais (STARMA). O modelo STARMA é empregado para descrever dados de séries temporais espacialmente localizados, ele é caracterizado pela dependência linear defasada tanto no espaço quanto no tempo. Foram realizadas simulações utilizando vários modelos de covariância espaço-temporal para comparar diferentes estruturas de construção da matriz de pesos espaciais com a finalidade de identificar a melhor matriz. As matrizes espaciais com pesos exponenciais apresentaram os melhores desempenhos de ajuste dos modelos STAR; e mostram uma estabilidade em relação à medida de ajuste. Por fim para ilustração, será ajustado um modelo STARMA para um conjunto de dados mensais do índice FIPEZAP de preço imobiliário de venda para apartamentos de dois dormitórios de seis cidades metropolitanas de São Paulo.<br>The objective of this work is to compare spatial neighborhoods structures, or the same as spatial weights matrices of the class of space-time autoregressive and moving average models STARMA. The STARMA model is used to describe spatially localized time series datas, it is characterized by the linear dependence lagged both in space and time. Simulations were performed using several space-time covariance models to compare different structures of construction of the weight matrix with the purpose of identifying the best matrix. The spatial matrices with exponential weights presented the best adjustment performances of the STAR models ans showed a stability in relation to the adjustment measure. Finally, for illustration, a STARMA model will be adjusted for a set of monthly data of the FIPEZAP real estate price index for two bedroom apartments in six metropolitan cities of São Paulo.

This thesis consists of three studies that centre around forecasting realised volatility based on high-frequency financial data. Accurate volatility forecasts are used extensively in many financial applications. The methods used here draw on econometric models and machine learning techniques. The empirical studies are based on fifty stocks and two stock indices. The thesis has established new perspectives on forecasting the realised volatility asset returns to improve financial decision-making.

A lot of interesting research is currently being done in the field of neuroscience, a recent subject being the effort to analyse the the human brain connectome and its functional connectivity. One way this is done is by applying graph-theory based network analysis, such as centrality, on data from fMRI measurements. This involves creating a graph representation from a correlation matrix containing the correlations over time between all measured voxels. Since the input data can be very big, this results in computations that are too memory and time consuming for an ordinary computer. Researchers have used different techniques to work around this problem, examples include thresholding correlations when creating the adjacency matrix and using a smaller input data with lower resolution.This thesis proposes three ways to compute two different centrality measures, degree centrality and eigenvector centrality, on fully weighted adjacency matrices that are built from complete correlation matrices computed from high resolution input data. The first is reducing the problem by doing the calculations in optimal order and avoiding the construction of the large correlation matrix. The second solution is to distribute and do the computations in parallel on a large computer cluster using MPI. The third solution is to calculate as large sets as possible on an ordinary laptop using shared-memory parallelism with OpenMP. Algorithms are presented for the different solutions, and the effectiveness of the implementations of them is tested.

O estudo das características produtivas visa a obtenção de animais possuidores de genótipos com potencial para altas produções mas que sejam também adaptados ao meio ambiente de criação extensiva no Brasil. Esses animais produtivos e adaptados poderiam otimizar os custos, tornando assim, sua criação sustentável em escala comercial. Apesar das características reprodutivas serem de extrema importância como objetivo de seleção, sua utilização não tem sido ampla em programas de melhoramento genético por serem consideradas de baixa herdabilidade. Entre as características produtivas existentes em bovinos de corte pode ser citada a caracterítica de peso adulto, que deve ser avaliado e monitorado o aumento do tamanho das fêmeas pela resposta indireta a seleção para peso, afim de confrontar os resultados economicamente. Dentro deste contexto foram avaliados dados de 74.547 animais com o objetivo de analisar o peso adulto e o peso adulto com repetibilidade (PAVr) em animais Nelore estimando a herdabilidade de características de desenvolvimento ponderal, utilizando modelos lineares; Estimar a correlação entre peso adulto (PAV) e características de peso ao sobreano (PS), conformação (CONF), precocidade (PREC), musculosidade (MUSC) e altura de garupa (AG); Determinar se há ou não vantagem em incluir o peso adulto em programas de seleção na raça Nelore. Os componentes de variância e de covariância foram estimados pelo método de máxima verossimilhança restrita (MTDFREML) utilizando-se o modelo animal. As estimativas dos coeficientes de herdabilidade foram de 0,43 para PAV e 0,41 para PAVr; 0,38 para PS; 0,23 para CONF; 0,19 para PREC; 0,22 para MUSC e 0,35 para AG. Os coeficientes de correlação genética obtidos variaram de 0,37 a 0,97. Os resultados sugerem que ganhos genéticos podem ser obtidos com a seleção para a característica de peso adulto, em monitoramento do tamanho de fêmeas na raça Nelore. A correlação entre as característica em estudo demonstra haver uma tendência de alteração das características de tamanho a partir da seleção para peso adulto.<br>The study of the productive traits aims the obtention of animals that have genotypes with potencial for high productions but that are also adaptable to the environment of the extensive creation in Brazil. These productive and adaptable animals could improve the costs, making therefore, their creation sustainable on a commercial scale. Although the reproductive characteristics have an extremely importance as an objective of the selection, its utilization have not been ample in genetic improvement programs for being considered as a low heritability trait. Between the productive traits existent in beef cattle it can be cited the mature weight trait, that must be evaluated and monitorated for the size increase control of the females by the indirect answer to weight selection, in order to confront, economically, the results. In this context it was evaluated data from 74.547 animals with the objective of analyze the mature weight and the adult weight with repeated measurements (MWr) in Nelore animals estimating the heritability of the ponderal development traits, using linear models; Estimate the correlation between mature weight (MW) and the post-weaning weight trait (PW), conformation (CONF), precocity (PREC), musculature (MUSC) and hip height (HH); Determine if there is an advantage in the inclusion of the mature weight in selection programs of the Nelore breed. The components of variance and covariance were estimated by the method of Restricted Maximum Likelihood Multiple Trait Derivative Free (MTDFREML) using the animal model. The estimation of the heritability coefficients were 0,43 for MW and 0,41 for MWr; 0,38 for PW; 0,23 for CONF; 0,19 for PREC; 0,22 for MUSC and 0,35 for HH. The genetic correlation coefficients varied from 0,37 to 0,97. The results suggest that genetic gains can be obtained with the selection for the mature weight characteristic, with the monitoration of the female sizes for Nelore breed. The correlation between the characteristics in study shows that there is a tendency for alteration in size characteristics through the selection for mature weight.

A alta variabilidade no peso ao nascer de leitões e o nascimento de leitões com pesos muito inferiores a média da leitegada evidencia a restrição de crescimento intrauterino, acentuada a partir da seleção genética de matrizes suínas mais prolíferas. Esses animais desenvolvem-se de maneira inferior aos seus contemporâneos e tem aumentado os riscos de morte ou descarte, acarretando menores índices produtivos. O objetivo deste estudo foi avaliar os efeitos do desenvolvimento intrauterino no crescimento, longevidade, produtividade e desempenho reprodutivo de futuras matrizes reprodutoras suínas. Além disso, avaliar a influência do desenvolvimento corporal e sua relação com puberdade e desempenho reprodutivo. No primeiro estudo, o peso ao nascer foi dividido em oito grupos e foram determinadas as curvas de crescimento e o desempenho reprodutivo até o terceiro parto de 1495 leitoas Landrace x Large White (DanBred®). Nas leitoas que pertenciam ao grupo mais leve, ou seja, com menos de 1 kg, o potencial de crescimento foi inferior do que as leitoas do grupo mais pesado, com menor ganho de peso diário (GPD) em todas as fases avaliadas (20, 70 e 170 dias) e com menor peso aos 170 dias, momento onde foi realizada a seleção para entrada no rebanho. A mortalidade e perdas cumulativas até os 170 dias foram maiores em leitoas pesando menos de 1 kg, reduzindo a oportunidade destas de serem selecionadas como futuras matrizes. Além disso, leitoas que nasceram com peso inferior a 1 kg tiveram menor número de dias de permanência no plantel e produziram quase 4,5 leitões a menos ao longo de três partos quando comparado com os outros grupos de peso ao nascer. O segundo estudo utilizou um total de 665 leitoas Landrace x Large White (DanBred®) para avaliar os efeitos da idade e taxa de crescimento no momento de exposição ao macho e suas consequências no desempenho reprodutivo. As leitoas foram retrospectivamente classificadas em grupos de acordo com idade a exposição ao macho (140-155 e 156–170 dias) e taxa de crescimento até a exposição ao macho (Baixa: 500–575 g/d; Intermediária: 580–625 g/d; e Alta: 630–790 g/d). Leitoas expostas ao macho com 140-155 dias tiveram menor manifestação de estro (60,8 vs. 77,0%) até os 30 dias do que àquelas expostas com 156-170 dias de idade. A manifestação de estro até 30 dias após a exposição ao macho foi maior para leitoas com alto ganho de peso (74,3%) do que as de ganho de peso baixo e intermediário (65,5 and 64,3%, respectivamente). A taxa de parto e o número de leitões nascidos no primeiro parto não foram afetados pela idade e pela taxa de crescimento. Os resultados dos nossos estudos apontam para a conclusão que o peso ao nascer e o desempenho de crescimento influenciam no desempenho reprodutivo das leitoas quando adultas.<br>The high variability on piglet birth weight and the birth of piglets weighing less than 1 kg show a restriction on intrauterine growth, increased by the large litter size of hyperprolific sows. These animals develop in lower rates than its contemporaries and have increased risk of death or culling, resulting in lower productivity. The aim of this study was to evaluate the effects of intrauterine development in growth, longevity, productivity and reproductive performance of sows destined to breeding herd. In addition, was to evaluate the influence of body development and its relation with puberty and reproductive performance. On the first study, using a retrospective classification into eight classes of birth weight were determined the growth and reproductive performance until the third parity of 1,495 crossbred Landrace x Large White gilts (DanBred®). Piglets from the lower birth weight group, i.e., less than 1 kg, had poorer growth performance when compared with the higher class, with lower body weight and average daily weight gain in all stages of development evaluated. Mortality and cumulative losses until 170 days of life were greater on piglets weighing less than 1 kg at birth, reducing the opportunity for their selection as future breeders. Furthermore, sows born weighing less than 1 kg had lower number of days in the breeding herd and produced almost 4.5 less piglets along three parities than the other gilts. A second study used a total of 665 Landrace x Large White gilts (DanBred®) to evaluate the effects of age and growth rate until the onset of boar exposure on first oestrus manifestation and reproductive performance. Gilts were retrospectively classified in groups according to their age at boar exposure (140-155 and 156–170 days) and into classes according to their growth rate from birth to boar exposure (Low: 500–575 g/d; Intermediate: 580–625 g/d; and High: 630–790 g/d). Gilts exposed to boar at 140-155 days had lower oestrus manifestation (60.8 vs. 77.0%) within 30 days than those exposed at 156-170 days of age. Lower percentages of gilts in oestrus within 30 days after boar exposure were observed in Low and Intermediate growth rate gilts (65.5 and 64.3%) than in High growth rate gilts (74.3%). Farrowing rate and number of total born litter size were affected neither by age or growth rate at onset of boar exposure. The results of our studies point to the conclusion that the birth weight and the developmental performance have influence on the reproductive performance of gilts as sows.

The different feeding schedules used in the poultry industry contribute to improve the nutrients distribution in the animal, as well as egg production. The study was carried out to analize the corporal composition of broiler breeders fed in different feeding schedules and their effect in body weight and body weight gain, as well as egg production. The broiler breeders were fed in three feeding schedules. The feeding schedules evaluated were: 1- a single feeding at 08:00 am, 2- twice daily feeding (50% at 08:00 am and 50% at 3:00 pm), 3- a single feeding at 3:00 pm. The hens were allocated in a completely randomized design of 3 treatments with 5 replications of 22 females and 3 males per repetition, to evaluate the corporal composition parameters were slaughtered two hens per repetition. Body weight, body weight gain, egg production, dry matter, mineral matter, crude protein, ether extract, calcium and phosphorus levels were evaluated. The feeding schedule late had higher body weight (P=0.0273) and body weight gain (P=0.0981) than early feeding schedule used. The first feeding schedule showed better egg production (P=0.0300) than the third feeding schedule. The percentage of ash (P=0.0741) and phosphorus (P=0.0875) were higher in broilers breeders fed at 8 am than hens fed at 3 pm, not different from the treatment 2. The calcium content were not affected by feeding scheduler used. The broiler breeder feeding at 8:00 am had more level of dry matter, crude protein and lower ether extract than feeding schedule late. Broiler breeder fed with feeding schedule late had lower egg production and higher body weight gain than others, consequently the most body fat deposition.<br>A utilização de horários de arraçoamento distintos ao empregado pela indústria visa contribuir para a melhoria na distribuição dos nutrientes requeridos para mantença, assim como para a produção das aves. O presente experimento foi conduzido com o intuito de analisar a composição corpórea de reprodutoras de corte avícolas submetidas a diferentes horários de arraçoamento e o seu impacto mediante o peso e ganho corporal, bem como, a produção de ovos. Os animais foram submetidas a 3 tratamentos: 1- horário de arraçoamento único às 08h00min, 2- horário de arraçoamento dual (50% de manhã 08h00min e 50% à tarde 15h00min), 3- horário de arraçoamento único às 15h00min. Foi utilizado um delineamento inteiramente casualizado com três tratamentos e cinco repetições de vinte e duas aves, sendo que para parâmetros referentes à composição corporal foram utilizadas duas aves abatidas para cada repetição. Foram avaliados: Peso e ganho corporal, taxa de postura, matéria seca, matéria mineral, proteína bruta, extrato etéreo, cálcio e fósforo. O horário de arraçoamento tardio apresentou aves mais pesadas (P=0,0273), com consequente ganho de peso mais elevado (P=0,0981) do que o horário matinal. O tratamento 1 apresentou a maior produção total de ovos (P=0,0300) em relação ao horário 3 A porcentagem de cinzas (P=0,0741) e fósforo (P=0,0875) foi maior em matrizes de corte alimentadas às 08h00min do que aves alimentadas às 15h00min, não diferindo do tratamento 2. Aves arraçoadas no período da manhã apresentaram maior teor de matéria seca e umidade, refletindo em um maior percentual de proteína bruta e consequentemente em um menor porcentual de extrato etéreo do que o horário de arraçoamento tardio. Reprodutoras avícolas alimentadas em horários tardios apresentaram uma menor produção de ovos com maior ganho de peso e consequentemente, maior deposição de gordura corporal.

Les packagings d’électronique embarquée sont actuellement en alliages d’aluminium. A partir d’une étude de sélection des matériaux, complétée par une simulation numérique thermique,nous avons démontré qu’un matériau composite constitué d’une matrice aluminium et de fibres de carbone à forte conductivité thermique, représente un fort potentiel de gain de masse sur ces équipements. Cependant, le couplage de ces deux matériaux génère des problèmes d’élaboration en raison d’incompatibilités fortes parmi lesquelles un mouillage très faible du carbone par l’aluminium liquide et une réactivité chimique élevée qui conduit à la formation de carbures d’aluminium préjudiciables pour le matériau final. Deux voies d’élaboration distinctes ont été envisagées : Une voie liquide où l’utilisation d’un agent de mouillage (un sel fluoré) a permis d’obtenir la montée par capillarité du métal dans des mèches de fibres. Une voie solide basée sur une technique originale d’empilements de feuillets d’aluminium et de fibres de carbone avec le procédé de Spark Plasma Sintering (SPS). .La seconde technique s’est révélée prometteuse en permettant d’obtenir des échantillons multicouches sans porosités, un endommagement très limité des fibres et une architecture contrôlée.Notre étude a montré que la formation de carbures d’aluminium est limitée. De plus, une meilleure compréhension du SPS ou l’application d’un revêtement sur les fibres devraient permettre d’éviter la formation de ces carbures. Les tentatives de caractérisations mécanique et thermique effectuées sur ces échantillons donnent un premier aperçu de l’efficacité du renforcement de l’aluminium par les fibres de carbone<br>Embedded electronic packagings are currently made of aluminum. A first study – basedupon a material selection method completed by numerical analysis – showed that a metal matrixcomposite made of aluminum and highly thermal conductive continuous carbon fibers represents ahigh potential upon weight savings for those equipments. Though, coupling these componentsrepresents numerous challenges due to their incompatibility such as a really low wetting of carbonliquidaluminum system and its unavoidable chemical reactivity that leads to the formation ofaluminum carbides that are harmful for the final material. Two manufacturing routes were considered: A liquid route using a wetting agent (fluorinated salts) led the metal to rise alongcarbon fibers by capillarity. A solid route based upon a novel technique of aluminum foils and carbon fibersstacking using the Spark Plasma Sintering (SPS) process.This second technique revealed to be very promising and allowed to obtain multilayer samples with noporosities, highly limited fiber damages and controlled composite architecture. Our study shows thataluminum carbides formation is limited. Moreover, a deeper comprehension of SPS process or thedeposit of fiber coatings would prevent this carbide formation. Attempts of mechanical and thermalcharacterization led upon such samples give a first overview of the efficiency of the aluminumreinforcement by carbon fibers

O objetivo central dos dois estudos que compõem essa tese foi estudar a manipulação de aminoácidos em linhagens hiperprolíficas, nas quais o desafio do crescimento e produção de leite é ainda maior. No primeiro trabalho, foram utilizadas leitoas Landrace x Large White gestantes distribuídas em quatro tratamentos: Argiblend – oferta de 1% de arginina do D25-D80 e de 20g de blend do D81-D112 de gestação; Arginina – oferta de 1% de arginina do D25 ao D80; Blend – oferta de 20g de blend do D81 ao D112, e Controle - dieta controle durante toda a gestação. Não houve efeito (P>0,05) dos tratamentos sobre o peso e ganho de peso das matrizes, bem como para número de nascidos totais, nascidos vivos, natimortos, mumificados, peso ao nascer, coeficiente de variação do peso ao nascer, percentual de leitões ≤850g e ≤1000g, peso de placenta, eficiência placentária e expressão do fator de crescimento do endotélio vascular. Houve efeito da interação do tratamento e classe de prolificidade (P<0,05) sobre os percentuais de leitões leves. Na classe de baixa prolificidade (até 14 leitões nascidos), menores percentuais de leitões com peso ≤850g foram observados nos tratamentos Arginina e Blend em comparação ao Controle (P<0,05). Ainda dentro das fêmeas de baixa prolificidade, o tratamento Blend teve maior peso de leitegada e peso médio de nascidos, além de menor percentual de leitões com peso ≤1000g do que as fêmeas Controle (P<0,05). O uso de suplementação de arginina 1% do D25-D112, blend de aminoácidos do D81-D112 ou a combinação de ambos, não afeta o peso médio ao nascer e nem o coeficiente de variação do peso, em leitegadas de matrizes hiperprolíficas. Porém, arginina e blend diminuem o percentual de leitões leves e blend aumenta o peso médio ao nascer nas leitegadas com até 14 leitões. O objetivo do segundo estudo foi investigar o efeito da suplementação de arginina na dieta de lactação sobre a composição do leite, desempenho e sobrevivência de leitões lactentes. Sessenta e quatro matrizes lactantes de linhagem Landrace x Large White de ciclo 1 a 7 foram distribuídas em dois tratamentos: 1) suplementação diária com 1% de Arginina via top dressing durante toda a lactação e 2) Controle. Não houve efeito do tratamento sobre o peso médio individual, peso da leitegada e ganho de peso diário dos leitões, tanto no D10 como no D20 da lactação (P>0,05). A interação entre o dia da pesagem e o tratamento não afetou nenhuma dessas variáveis (P>0,05). No geral, as taxas de sobrevivência dos leitões foram 90,3% e 88,3% no D10 e D21, respectivamente, sem diferença entre os tratamentos (P>0,05). Não houve efeito do dia da lactação (D10 ou D20), do tratamento ou da interação entre eles, sobre a proteína bruta e conteúdo de aminoácidos do leite (P>0,05). A suplementação de 1% de arginina utilizando top dressing na dieta de lactação de fêmeas suínas, de ciclo 1 a 7, não influencia o desempenho e sobrevivência dos leitões na maternidade e não altera a relação lisina:arginina bem como o teor de proteina e de aminoácidos no leite.<br>The central objective of both studies was to test the effect of amino acid supplementation in prolific sows, in which fetal growth and milk production is a big challenge.The first study was conducted to evaluate the effect of arginine and/or amino acids blend (lysine, methionine, threonine and tryptophan) supplementation in gestation diet of prolific primiparous sows on the placental efficiency and piglet’s birth weight. Pregnant gilts were distributed into four treatments: Argiblend – supplemented with 1% of arginine from D25 to D80 and 20g of blend from D81 to D112 of gestation; Arginine - supplemented with 1% of arginine from D25 to D80; Blend - 20g of blend from D81 to D112; and Control – basal diet from D25 to D112. There was no treatment effect (P>0.05) on the gestation body weight gain of sows, total number of piglets born, born alive piglets, stillborn piglets, mummified fetuses, average birth weight, coefficient of whithin-litter weight variation, proportion of low birth weight piglets (≤850g or ≤1000g), placental efficiency and vascular endothelial growth factor expression. The effect of amino acid supplementation was also evaluated taking into account the prolificity of females, which were separated in two categories: high (>14 piglets born) and low (≤14 piglets born) prolificity. In low prolificity females, the proportion of piglets with ≤850g was lower in arginine and blend treatments compared with the Control group. In addition, females receiving blend had total litter birth weight and average birth weight greater and the proportion of piglets with ≤1000g lower than in Control group (P<0.05). In the present study, supplementing gestation diets with arginine and/or amino acid blend does not influence the average birth weight and birth weight variation in primiparous prolific sows. However, arginine and blend decrease the proportion of light piglets at birth and increase the average birth weight when the litter size is ≤ 14 piglets. The objective of the second study was to investigate the effect of arginine supplementation in the lactation diet on milk composition and the performance and survival of suckling piglets. Sixty four lactating Landrace x Large White sows, parity 1 to 7, were assigned in two treatments: Arginine – control diet supplemented daily with 1% arginine by top dressing during all lactation, and Control – control cornsoybean based diet. The sows remained with 12.8 piglets on average after cross fostering. The litters were weighed on D1, D10 and D21 of lactation and piglet mortality was recorded. Samples of milk (60 ml) were collected at D10 and D20 of lactation, from all functional teats. There was no effect of arginine supplementation on the average individual weight, litter weight and average daily gain of piglets at both D10 and D20 of lactation (P>0.05). The interaction between the day of weighing and treatment did not affect any of these parameters (P>0.05). Overall, survival rates were 90.3% and 88.3% until D10 and D21, respectively, with no difference between treatmens (P>0.05). There was no effect of the lactation day (D10 or D20), treatment or the interaction between them on the crude protein and amino acid content in milk (P>0.05). Supplementation with 1% of arginine by top dressing, in lactation diet of parity 1 to 7 sows, does not affect the performance and survival of piglets, as well as does not alter the amino acid content or lysine:arginine ratio of milk.

Com o melhoramento genético das últimas décadas, houve decréscimo no peso do leitão ao nascer e, com isso, estratégias nutricionais que envolvem o terço final de gestação para melhorar o peso ao nascer são estudadas. O objetivo desse estudo foi avaliar os efeitos da ingestão de duas quantidades de ração no terço final de gestação no peso do leitão ao nascimento, desempenho produtivo e reprodutivo subsequente. Nesse estudo, 407 fêmeas foram alimentadas com 1,8 ou 2,2 kg/d com uma dieta a base de milho e farelo de soja (3,25 Mcal EM kg-1, 13% PB e 0,78% de SID Lis) a partir do 90° dia de gestação até o parto. As fêmeas foram pesadas aos 90, aos 112 dias de gestação e no desmame. Os leitões nascidos vivos e natimortos foram pesados em até 12 horas após o nascimento. Para análise dos efeitos sobre o consumo voluntário de ração e desempenho dos leitões na fase lactacional, 53 fêmeas de cada grupo foram aleatoriamente selecionadas. Neste subgrupo, as fêmeas foram pesadas em até 24 horas após o parto e o consumo voluntário de ração foi registrado em intervalos de 4 dias. Além disso, o peso dos leitões ao desmame foi mensurado nesse subgrupo. Houve aumento no ganho de peso entre 90 e 112 dias de gestação nas fêmeas alimentadas com 2,2 kg/d comparativamente às alimentadas com 1,8 kg/d (P < 0,001). Não foram encontrados efeitos sobre o peso ao nascimento e coeficiente de variação dos leitões tanto de nulíparas, quanto de pluríparas (P > 0,05). Entretanto, em pluríparas alimentadas com 2,2 kg/d houve um aumento no tamanho da leitegada e peso total da leitegada ao nascimento comparativamente às fêmeas alimentadas com 1,8 kg/d (P < 0,05). Não foram encontradas diferenças nas porcentagens de leitões natimortos, fetos mumificados e leitões com peso <1000 g ao nascimento (P < 0,05). Fêmeas alimentadas com 1,8 kg/d durante a gestação tiveram maior consumo voluntário de ração durante a lactação quando comparadas às fêmeas alimentadas com 2,2 kg/d. Não foram encontradas evidências de diferença entre os tratamentos no desempenho dos leitões até o desmame, perda corporal das fêmeas durante a lactação, intervalo desmame estro, tamanho da leitegada subsequente e taxa de remoção (P > 0,05) Em conclusão, o aumento no consumo de ração no terço final de gestação aumentou o ganho de peso das fêmeas, não implicou em efeitos sobre o peso individual dos leitões ao nascer, mas influenciou negativamente o consumo de ração na lactação. Além disso, não foram encontradas evidências de efeitos do tratamento no desempenho da leitegada ou desempenho reprodutivo subsequente.<br>The genetic improvement in the swine industry in the last decades induced a reduction on the individual piglet birth weight, thus nutritional strategies in the last third of gestation to improve birth weight have been studied. The objective of this study was to evaluate the effect of two levels of feed intake in the last third of gestation on piglet birth weight, subsequent productive, and reproductive female performance. In this study, 407 females were fed either 1.8 or 2.2 kg/day of a corn-soybean meal based diet (3.25 Mcal ME kg-1, 13% CP and 0.78% SID Lys) from 90th day of gestation up to farrowing. Females were weighed at 90 and 112 days of gestation and at weaning. Piglets born alive and stillborn were weighed up to 12 h after birth. To analyze the effects on voluntary feed consumption and litter growth during lactation, 53 sows of each group were randomly selected. In this subsample, sows were weighed up to 24 h after farrowing and the voluntary feed consumption was recorded every 4 days. Furthermore, piglet weaning weight was collected from this sub-sample. There was an increase in sow weight gain from d 90 to 112 in the females fed 2.2 kg/d compared to 1.8 kg/d (P < 0.001). There was no effect of feed level on the individual piglet birth weight and within-litter weight variation for both gilts and sows (P > 0.05); however, in sows fed 2.2 kg/d there was an increase in litter size and total litter weight at farrowing compared to 1.8 kg/d (P < 0.05). There were no differences in percentages of stillborn piglets, mummified fetuses and piglets with <1000 g at birth (P > 0.05). Females fed 1.8 kg/d during gestation had higher feed intake during lactation compared to females fed 2.2 kg/d (P < 0.05). There was no evidences for differences between the treatments in piglet growth until weaning, weight loss of sows during lactation, weaning-to-estrus interval, subsequent litter size and culling rate (P > 0.05). In conclusion, increased feed intake in the last third of gestation increased sow body weight gain, with no evidence for effects on individual piglet birth weight, but had a negative influence on feed intake during lactation. Furthermore, there was no evidence of treatment effects on litter growth or subsequent reproductive performance.

REG3A est un effecteur de l'immunité innée qui joue un rôle important dans l'homéostasie du glucose et qui favorise le maintien d'une flore intestinale commensale aux propriétés anti-inflammatoires. Au cours de notre projet de recherche, nous avons étudié le rôle de REG3A dans le contexte de la maladie métabolique d'origine nutritionnelle et développé pour cela un modèle de maladie métabolique combinant un régime hyperlipidique avec des variations de poids. Les souris C57Bl/6 sauvages et REG3A-TG (exprimant REG3A dans le foie) ont été nourries pendant 1 à 3 cycles avec un régime riche en gras administré en continu ou entrecoupé de périodes d'alimentation standard. Nos résultats montrent que la succession de périodes de gain et de perte de poids favorise le développement d'une MASH chez les souris sauvages, caractérisée par une réduction de la sensibilité à l'insuline, la présence d'inflammation, de ballonnement hépatocytaire et d'une fibrose. L'analyse protéomique des foies révèle que l'alimentation riche en gras augmente le métabolisme lipidique tout en réduisant la voie de biosynthèse du cholestérol. L'expression de REG3A réduit l'hépatopathie inflammatoire et le développement de fibrose chez la souris, réduction qui est corrélée à l'activation de la voie de signalisation de l'IFN-γ/STAT1. L'alimentation hyperlipidique induit chez la souris sauvage le remodelage de la matrice extracellulaire du foie et cela, bien en amont de l'apparition de la fibrose, matrice inflammatoire remodelée qui altère la capacité d'hépatocytes en culture primaire de répondre à l'insuline. L'expression de REG3A impacte également sur la composition de cette matrice inflammatoire qui préserve alors sa capacité à sensibiliser les hépatocytes à répondre à l'insuline. En conclusion, nos travaux révèlent les effets pervers de l'alternance de périodes de gain et perte de poids chez la souris (régime yo-yo) qui favorise la sévérité de la maladie métabolique du foie d'origine nutritionnelle et le rôle protecteur que joue l'effecteur de l'immunité innée REG3A dans cette maladie<br>REG3A is an effector of innate immunity which plays an important role in glucose homeostasis and promotes the maintenance of a commensal gut flora with anti-inflammatory properties. In our research project, we investigated the role of REG3A in the context of nutritionally-induced metabolic disease, using a metabolic disease model combining a hyperlipidic diet with weight cycling. Wild-type C57Bl/6 and REG3A-TG mice (expressing REG3A in the liver) were fed with a high-fat diet for 1-3 cycles, either continuously or interspersed with standard feeding periods. Our results show that successive periods of weight gain and loss promote the development of MASH in wild-type mice, characterized by reduced insulin sensitivity, inflammation, hepatocyte ballooning and fibrosis. Proteomic analysis of livers reveals that high-fat diet increases lipid metabolism in mice while reducing the cholesterol biosynthesis pathway. REG3A expression reduces inflammatory hepatopathy and fibrosis development in mice, a reduction that correlates with activation of the IFN-γ/STAT1 signaling pathway. In wild-type mice, hyperlipidic feeding induces remodeling of the extracellular matrix in the liver, well before the onset of fibrosis. This remodelled inflammatory matrix impairs the ability of primary cultured hepatocytes to respond to insulin. REG3A expression also has an impact on the composition of this inflammatory matrix, which then preserves its ability to sensitize hepatocytes to respond to insulin. In conclusion, our work reveals the perverse effects of alternating periods of weight gain and loss in mice (yo-yo diet), which promotes the severity of nutritionally induced metabolic liver disease, and the protective role played by the innate immune effector REG3A in this disease

La réparation de lésions cérébrales reste de nos jours un grand challenge de l'ingénierie tissulaire. De même, l'établissement de modèles in vitro représentatifs du tissu cérébral est un sujet qu'il reste à explorer, surtout du point de vue des biomatériaux à utiliser pour soutenir la croissance cellulaire. Ce travail consiste donc au développement d'un biomatériau de type hydrogel adapté à la survie et la croissance de cellules souches neurales humaines. Les gels de type (supra)moléculaires, qui ont la particularité de se former par auto-assemblage, pourraient présenter de nombreux avantages pour ce type d'application. En effet, ils possèdent entre autres des propriétés mécaniques, une biodisponibilité et une structuration aux échelles micro et nanométriques originale qui font de ces gels des supports intéressants pour la culture de neurones. Une famille de gélifiants supramoléculaires a donc été synthétisée, caractérisée et étudiée en tant que support de culture cellulaire. Ces gélifiants sont de type alkylgalactonamide, c'est-à-dire dérivés d'un sucre - le galactose - et d'une chaîne grasse de type alkyle. Ils forment des hydrogels par refroidissement jusqu'à température ambiante d'une solution de gélifiant préalablement portée à haute température. Le refroidissement entraîne la formation de fibres micrométriques par auto-assemblage des molécules. Il a été établi qu'un contrôle précis du refroidissement lors de leur préparation permettait l'obtention d'hydrogels plus homogènes, plus résistants, qui sont compatibles avec les conditions de culture cellulaire et avec des fibres plus longues. Ces hydrogels se sont révélés être biocompatibles, avec une survie et une croissance en trois dimensions des cellules souches neurales humaines en leur sein. Ces dernières développent de longs neurites et expriment des marqueurs de la différentiation neuronale (ß3-tubuline) ou gliale (GFAP), notamment sur l'un des hydrogels. Le dernier aspect de ce travail a consisté à l'utilisation de nouvelles techniques de structuration en 3D des matériaux afin de construire avec ces hydrogels des supports centimétriques avec une forme bien définie. Une technique de filage en voie humide basée sur l'échange de solvants (wet spinning) a alors été mise au point permettant d'extruder l'hydrogel de façon contrôlée et directement à température ambiante. Elle permet d'obtenir de fins filaments de gel très réguliers soutenus par des fibres nanométriques de largeur monodispersée. Des essais ont également été effectués pour appliquer cette technique à l'impression 3D. Au final, ce projet démontre que certains gels moléculaires peuvent présenter des propriétés particulièrement adaptées pour l'ingénierie tissulaire notamment avec des cellules souches neurales et ouvrent également des perspectives dans le domaine de la mise en forme de ces matériaux délicats<br>Nowadays, repairing brain lesions is still one of the main challenges of tissue engineering. Meanwhile, work still has to be done for the creation of representative in vitro brain tissue models, especially when it comes to the biomaterials used to support cell growth. This present work consists in the development of a hydrogel as a biomaterial for the survival and growth of human neural stem cells. These (supra)molecular gels, the property of which is to form by self-assembly, may present many advantages for this kind of application. Indeed, their mechanical properties, their bioavailability and their microstructure - among others - make them interesting candidates for neuron culture. One family of supramolecular gelators have thus been synthesized, characterized and tested as cell culture scaffolds. Those gelators are alkylgalactonamides, which means they are derived from a sugar - the galactose - and an alkyl fatty chain. They form hydrogels by cooling down to room temperature after a first dissolution at high temperature. The cooling provokes the self-assembly of the molecules resulting in the formation of the fibers. During their preparation, it has been found that a controlled cooling rate enabled the formation of more homogeneous and more stable hydrogels that are compatible with cell culture conditions and with longer fibers. These hydrogels have shown a good biocompatibility as well as a good cell survival and a three-dimensional growth of human neural stem cells. The latter grew long neurites and expressed markers of neuronal (ß3-tubulin) and glial differentiation (GFAP), especially on one of the hydrogels. The last part of this work was to use new 3D material structuring techniques in order to further construct well-defined centimetric scaffolds with these hydrogels. A technique of wet spinning based on solvent exchange was developed and enabled the direct and controlled extrusion of the hydrogel at room temperature. Thin and regular hydrogel filaments composed of monodisperse nanometric fibers can thus be obtained. Trials have also been done to apply this method to 3D printing. In the end, this project shows that some molecular gels can display properties particularly adapted for tissue engineering, especially with neural stem cells, and it also opens perspectives for the shaping of these delicate materials

The structural behaviour of pavements in cold regions can considerably be affected by seasonal variation in environmental factors such as temperature and moisture content. Along with the destructive effect of heavy traffic loads, climatic and environmental factors can considerably contribute to pavement deterioration. These factors can influence the structural and functional capacity of the pavement structures which, as a result, can trigger and accelerate pavement deterioration mechanisms. Studies on the influence of variation of the environmental factors on the response and behaviour of pavement materials have shown that proper consideration to these factors must be given in realistic pavement design and analysis. In flexible pavement structures, particularly with a thin hot mix asphalt (HMA) layer, unbound materials and subgrade soil largely contribute to the overall structural behaviour of the pavement system. In unbound materials, moisture content and its variation can significantly affect pavement layer stiffness and permanent deformation characteristics. Therefore, the moisture condition of pavements and its influence on the mechanical behaviour of pavement materials has been of interest among the pavement research community. A proper understanding of moisture transformation in pavement systems and its effects on pavement performance are important for mechanistic pavement design. The present summary of this doctoral thesis is based on four main parts. The first part of the thesis covers field measurements and findings from a test section along county road 126 in southern Sweden and consists of two journal papers (paper I and II) tackling different aspects of the research topic. This test section is located in a relatively wet ground condition and consists of a thin flexible pavement structure with a deep drainage system. It is instrumented with subsurface temperature, volumetric moisture content and groundwater probes. The mechanical response of the pavement structure was investigated using Falling Weight Deflectometer (FWD) measurements. The second part of the thesis (paper III and IV) are based on laboratory experiments and investigates different recent approaches that have been proposed to apply principles of unsaturated soil mechanics for incorporating seasonal variation of moisture content into the resilient modulus models using matric suction. The third part of the thesis (paper V) builds a bridge that spans between the laboratory and field investigations with an attempt to evaluate one of the predictive models presented in Paper III. The fourth part of the thesis (paper VI) mainly focuses on the laboratory-based investigation of the permanent deformation characteristic of subgrade soils. In this part, the permanent deformation characteristics of two different silty sand subgrade soils were investigated and modelled using the data obtained from repeated load traxial tests. Paper I mainly focuses on the spring-thaw weakening of the pavement structure. The environmental data collected using different sensors and the FWD tests were used to investigate variations in moisture content with thaw penetration and its influence on the stiffness of unbound layers and the pavement’s overall bearing capacity. Using the backcalculated layer stiffness and corresponding in situ moisture measurements in the unbound layers, a degree of saturation-based moisture-stiffness model was developed for the granular material and the subgrade. In Paper II, the drainage system of the structure was manually clogged during a three month period in summer to raise the groundwater level and increase the moisture content of the layers. Along with the subsurface groundwater level and moisture content monitoring, the structural response of the pavement was studied. In this research work, the FWD tests were conducted at three different load levels. The stress dependent behaviour of the unbound granular layer and the subgrade soil were further studied using the multilevel loads FWD test data. Additionally, parameters of a nonlinear stress-dependent stiffness model were backcalculated and their sensitivity to in situ moisture content was studied. In Paper III and IV, series of suction-controlled repeated load triaxial (RLT) tests were conducted on two silty sand (SM) subgrade materials. Several resilient modulus prediction models that account for seasonal moisture content variation through matric suction were summarized and after optimizing the model parameters, the capability of the prediction models in capturing the material response were evaluated. In Paper V, an attempt was made to evaluate the proficiency of one of the suction-resilient modulus models using the field moisture content and FWD measurements from the Torpsbruk test site. The backcalculated subgrade stiffness dataset at different moisture contents were compared with resilient modulus models obtained from the suction-resilient modulus predictive model. Paper VI presents an evaluation of several permanent deformation models for unbound pavement materials that incorporate the time-hardening concept using a series of multistage repeated load triaxial (RLT) tests conducted on silty sand subgrade materials. The permanent deformation tests were conducted at four different moisture contents with pore suctions measurement throughout the test. The effect of moisture content (matric suction) on the permanent deformation characteristics of the materials and the predictive model parameters were further investigated.<br><p>QC 20150324</p>

Hyaluronan (HA) and its principal receptor CD44 are known to be involved in regulating tumor cell dissemination and metastasis. It is hypothesized that the CD44-HA interaction regulates proliferation and invasion of tumor cells in dependence on the molecular weight and the presentation form of HA. To address this hypothesis, we reconstituted 3D collagen (Coll I) matrices and functionalized them with HA of molecular weight of 30-50 kDa (low molecular weight; LMW-HA) and 500-750 kDa (high molecular weight; HMW-HA). A post-modification strategy was applied to covalently immobilize HA to reconstituted fibrillar Coll I matrices, resulting in a non-altered Coll I network microstructure and stable immobilization over days. Functionalized Coll I matrices were characterized regarding topological and mechanical characteristics as well as HA amount using confocal laser scanning microscopy, colloidal probe force spectroscopy and quantitative Alcian blue assay, respectively. To elucidate tumor cell behavior, BRO melanoma cell lines with and without CD44 receptor expression were used for in vitro cell experiments. We demonstrated that only soluble LMW-HA promoted cell proliferation in a CD44 dependent manner, while HMW-HA and immobilized LMW-HA did not. Furthermore, an enhanced cell invasion was found only for immobilized LMW-HA. Both findings correlated with a very strong and specific adhesive interaction of LMW-HA and CD44+ cells quantified in single cell adhesion measurements using soft colloidal force spectroscopy. Overall, our results emphasize the importance of presentation mode and molecular weight specificity in biomaterial studies on the impact of HA on cell behavior.

This thesis is devoted to the models that are suitable for modelling spatial data. For this purpose, random fields with finite index set are used. Based on the neighbourhood relationship a spatial weight matrix is introduced which describes spatial dependencies. A recognition and testing of spatial dependence is mentioned and it is applied for macroeconomic indicators in the Czech Republic. Spatial models originated from generalization of usual time series models are subsequently combined with linear regression models. The parameter estimators are derived for selected models by three different methods. These methods are ordinary least squares, maximum likelihood and method of moments. Theoretical asymptotic results are supplemented by a simulation study that examines the performance of estimators for finite sample size. Finally, a short illustration on real data is demonstrated. Powered by TCPDF (www.tcpdf.org)

碩士<br>國立中央大學<br>數學研究所<br>94<br>In this thesis, we will study about numerical ranges of weighted permutation matrices and weighted shift matrices. Firstly, we know that if $A$ is a $3 imes3$ weighted permutation matrix, $W(A)$ has symmetry of $frac{2pi}{3}$. Thus, if there is a line segment on $partial W(A)$ then $W(A)$ is a triangle. Moreover, $A$ is normal. If $A$ is a $3 imes3$ weighted permutation matrix, $W(A)$ has symmetry of $frac{2pi}{4}$. If there is a line segment on $partial W(A)$ then $W(A)$ is a quadrangle. Moreover, $Acong A_{1}oplus A_{2}$, where $A_{1}$ and $A_{2}$ are $2 imes 2$ weighted permutation matrices. Let $A$ be a $4 imes4$ companion matrix. We will see that $W(A)$ has four line segments if and only if $A$ can be reducible. Another subject is that we are interested in finding the order of the weights of a weighted shift matrix so that the numerical radius will be the largest.

碩士<br>東吳大學<br>數學系<br>99<br>In this paper, we study numerical ranges of weighted cyclic matrices. The relationship between numerical ranges of weighted cyclic matrices and order of their weights, and the existence of a flat portion on the boundary of the numerical ranges are examined. In section 2, we prove that if A 2 Mn is a weighted cyclic matrix,W(A) is invariant under the rotation of angles 2kpi/n, k = 1, 2,.., n −1. The numerical radius of weighted cyclic matrix is independent of the modulus of weights. Then we further discuss the numerical ranges of weights permutation. The numerical ranges of weighted cyclic matrices are not change if an arrangement of their weights in a cyclic order or a reverse order is made. Furthermore, the numerical radius of a weighted shift matrix is strictly increasing with respect to their weights. In section 3, we use the results in section 2 to discuss the flat portion on the boundary of the numerical range of a weighted cyclic matrix and give an alternative and simple proof of the numerical radius of the shift matrix.

碩士<br>東吳大學<br>數學系<br>99<br>Abstract Let A be an n-by-n complex matrix. The numerical range of A is defined as W(A) = {x Ax : x x = 1; x ?k C^n} and numerical radius w(A) = max {|z| : z 2 W(A)}. It is known that the numerical range of a weighted shift matrix is a circular disk. In this paper, we compute the numerical radius of a weighted shift matrix with weight (1, . . . , 1, r, 1, . . . , 1).

博士<br>國立中央大學<br>數學研究所<br>100<br>Assume that all aj&apos;&apos;s are nonzero and B is a n-by-n weighted shift matrix with weights bj &apos;&apos;s. We show that B is unitarily equivalent to A if and only if a1 ￠ ￠ ￠ an = b1 ￠ ￠ ￠ bn and, for some ¯xed k, 1 · k · n, jbj j = jak+j j (an+j ’ aj) for all j. Next, we show that A is reducible if and only if A has periodic weights, that is, for some ¯xed k, 1 · k · bn=2c, n is divisible by k, and jaj j = jak+j j for all 1 · j · n!k. Finally, we prove that A and B have the same numerical range if and only if a1 ￠ ￠ ￠ an = b1 ￠ ￠ ￠ bn and Sr(ja1j2; : : : ; janj2) = Sr(jb1j2; : : : ; jbnj2) for all 1 · r · bn=2c, where Sr&apos;&apos;s are the circularly symmetric functions. Let A[j] denote the (n ! 1)-by-(n ! 1) principal submatrix of A obtained by deleting its jth row and jth column. We show that the boundary of numerical range W(A) has a line segment if and only if the aj&apos;&apos;s are nonzero and W(A[k]) = W(A[l]) = W(A[m]) for some 1 · k < l < m · n. This re¯nes previous results which Tsai andWu made on numerical ranges of weighted shift matrices. In Chapter 2, we discuss re¯nements of the well-known triangle inequality and it&apos;&apos;s reverse inequality for strongly integrable functions with values in a Banach space X. We also discuss re¯nement for the Lp functions in the second kind of generalized triangle inequality . For both cases, the attainability of the equality is also investigated.

碩士<br>國立臺北科技大學<br>電資國際專班<br>107<br>Song and Chissom proposed a fuzzy time series model consisting of two main procedures which were fuzzification plus the formation of fuzzy relationships resulting to forecasting of fuzzy time series. One research has pointed out that throughout the fuzzification method, different dimensions of intervals will result in numerous predicting results, and it has been suggested that the effective dimensions of intervals were used. The projecting results that were based on operative lengths of intervals were found to outclass those based on random ones. As the lengths of partitions are two critical issues for increasing accurateness, a new hybrid model which chains the biased fuzzy relation matrices and particle swarm optimization technique is suggested. In the suggested model, some important features have been used in finding the dimensions of partitions and fuzzy relations on time series data to increase estimating correctness. The past data were used instead of its differences in the predicting model. Additionally, the FRMs were followed by the corresponding fuzzy logic relationships, the decreasing weight system allocated the biggest weights to the latest past fuzzy set of a FRM for capturing efficient ambiguous relations. Thirdly, the particle swarm optimization procedure was established and adjusted the partition lengths to find the best division for predicting by minimizing the root mean squared error function. For demonstrating the effectiveness of the proposed method, different monthly average temperature datasets for 3 cities in Taiwan namely Taipei, Taichung and Kaohsiung from the years 2010 to 2018 were used. Data from January to October for each year were used for training the model while data for November and December were used for testing. The proposed model was proved to have the advantage of obtaining good predicting results over other past representations.

Les algèbres cellulaires furent introduite par J.J. Graham et G.I. Lehrer en 1996. Elles forment une famille d’algèbres associatives de dimension finie définies en termes de « données cellulaires » satisfaisant certains axiomes. Ces données cellulaires, lorsqu’elles sont identifiées pour une certaine algèbre, permettent une construction explicite de tous ses modules simples, à isomorphisme près, et de leurs couvertures projectives. Dans ce mémoire, nous définissons ces algèbres cellulaires en introduisant progressivement chacun des éléments constitutifs d’une façon axiomatique. Deux autres familles d’algèbres associatives sont discutées, à savoir les algèbres quasihéréditaires et celles dont les modules forment une catégorie de plus haut poids. Ces familles furent introduites durant la même période de temps, au tournant des années quatre-vingtdix. La relation entre ces deux familles ainsi que celle entre elles et les algèbres cellulaires sont prouvées.<br>Cellular algebras were introduced by J.J. Graham and G.I. Lehrer in 1996. They are a class of finite-dimensional associative algebras defined in terms of a “cellular datum” satisfying some axioms. This cellular datum, when made explicit for a given associative algebra, allows for the explicit construction of all its simple modules, up to isomorphism, and of their projective covers. In this work, we define these cellular algebras by introducing each building block of the cellular datum in a fairly axiomatic fashion. Two other families of associative algebras are discussed, namely the quasi-hereditary algebras and those whose modules form a highest weight category. These families were introduced at about the same period. The relationships between these two, and between them and the cellular ones, are made explicit.