Modelos de atividade baseados em COSMO se apresentam como uma interessante alternativa para a predição do comportamento de substâncias em mistura. Estes modelos dependem apenas de informações das substâncias puras, como volume e área superficial e outra informação conhecida como perfil sigma, que podem ser determinados através de cálculos de química quântica computacional. O presente trabalho tem por objetivo criar e disponibilizar gratuitamente um banco de dados de perfis sigma para um vasto conjunto de moléculas utilizando o software GAMESS. Para a criação desse banco de dados com informações confiáveis de perfis sigma, foram testados diferentes níveis de teoria para o cálculo da estrutura eletrônica das moléculas, bem como conjuntos de bases distintos, considerando funções de polarização e efeitos difusos. Para realizar a validação dos perfis sigma obtidos, foram estimados valores de coeficientes de atividade em diluição infinita, obtidos através do modelo COSMO-SAC para um conjunto de 689 misturas, as quais não apresentam ligação de hidrogênio. A correlação com os valores de coeficiente de atividade em diluição infinita (IDAC) experimentais, foram construídos gráficos do logaritmo natural de IDAC experimentais e aqueles calculados pelo modelo. O tempo computacional total para obter os perfis sigma também foi avaliado com o intuito de escolher a melhor combinação de teoria e função de base. Dentre os métodos analisados, a teoria de Hartree-Fock (HF) com a função de base 6-311G(d,p) apresentou um bom custo benefício no que se refere ao tempo computacional e a precisão dos resultados frente a dados experimentais, apresentando um coeficiente de determinação superior a 0,94. A partir desses resultados, foi possível criar outros perfis sigma para um número maior de moléculas e disponibilizar essas informações em um banco de dados criado na plataforma do GitHub. / COSMO-based activity coefficient models are widely used to predict non-ideality in mixtures. They require only pure substance properties, such as the -profile, a representation of the screening charge density along the molecular surface. The present work aims to create and freely distribute a -profile database which can be easily extended by the community using the free GAMESS quantum chemistry (QC) package. Different combinations of QC methods and basis sets were tested in the calculation of the -profiles, aiming to define an efficient approach to create the database. The accuracy of COSMO-SAC predictions using these -profiles was evaluated by comparison with experimental infinite dilution activity coefficients and phase equilibrium diagrams. Among the studied alternatives, the Hartree-Fock method with the 6-311G(d,p) basis set led to results with a good balance between accuracy and computational cost. Our database was created using this approach, and it is already available to be used and extended by the community without requiring any commercial software.

Neste trabalho, o modelo F-SAC (Functional-Segment Activity Coefficient) foi revisado e estendido de modo a representar, simultaneamente, coeficiente de atividade à diluição infinita (IDAC), equilíbrio líquido-líquido (LLE) e equilíbrio líquido-vapor (VLE), em amplos intervalos de temperatura. Originalmente, o modelo F-SAC foi calibrado essencialmente com dados experimentais de IDAC para diversas misturas. Investigouse, portanto, a possibilidade de o modelo levar também em consideração sistemas que apresentem LLE. Para a presente investigação, dois conjuntos de misturas industrialmente interessantes foram selecionados: hidrocarbonetos (alcanos, cicloalcanos, alcenos, cicloalcenos e compostos aromáticos) + n-formilmorfolina (NFM) e hidrocarbonetos + água. Foram propostas duas modificações na equação que calcula a energia de interação entre os grupos: (i) inspirado pelo comportamento típico observado na expansão isobárica de líquidos, um novo parâmetro foi proposto de modo a diminuir o (pequeno) erro sistemático do modelo com a temperatura no termo sem associação; (ii) em misturas associativas, utilizou-se uma equação similar para representar a dependência da energia de ligação de hidrogênio entre dois segmentos com a temperatura. Na estimação de parâmetros do modelo foram utilizados 1130 pontos experimentais de LLE, para 105 sistemas binários e 16 ternários, juntamente com um 815 dados experimentais de IDAC, resultando numa nova tabela de parâmetros. Ao total, foram estimados 42 parâmetros dos grupos e subgrupos. Os resultados do F-SAC foram comparados, quando possível, com diferentes versões do UNIFAC encontradas na literatura. O modelo proposto foi capaz de correlacionar a maior parte dos dados de IDAC e LLE utilizados. Para os dados de IDAC, foram observados desvios médios absolutos de 0,139 e 0,608 unidades de ln para o F-SAC e o UNIFAC(PSRK), respectivamente. Para os dados de LLE, foram observados desvios semelhantes. Finalmente, o F-SAC foi utilizado para representar dados de VLE para os sistemas de compostos aromáticos + NFM. Observou-se desvios médios relativos de 0,038 e 0,082 no cálculo da pressão de bolha utilizando-se o F-SAC e o UNIFAC(PSRK), respectivamente. Estes resultados comprovam a boa capacidade preditiva do modelo, uma vez que nenhum dado de VLE foi utilizado nas estimações. Conclui-se, desta forma, que o modelo FSAC pode ser usado para representar, simultaneamente, IDAC, LLE e VLE de misturas com e sem ligação de hidrogênio, utilizando um único conjunto de parâmetros. / In this work, the F-SAC model (Functional -Segment Activity Coefficient) has been revised and extended in order to simultaneously represent activity coefficient at infinite dilution (IDAC), liquid-liquid equilibrium (LLE) and vapor-liquid equilibrium (VLE), in a wide range of temperature. Originally, the F- SAC model was calibrated essentially with IDAC experimental data for several mixtures. In the present work, the possibility of also taking into account LLE was investigated. For this study, two sets of industrially interesting mixtures were selected : hydrocarbons (alkanes, cycloalkanes, alkenes, cycloalkenes and aromatics) + n-formylmorpholine (NFM) and hydrocarbons + water. Modifications were suggested in the equation that computes the interaction energy for each contact between two segments: (i) inspired by the typical behavior observed in the isobaric expansion of liquids, a new parameter has been proposed in order to reduce the (small) systematic error with the temperature for non-associating systems; (ii) for associating mixtures, a similar equation was proposed to represent the temperature dependence of the hydrogen bond energy between two segments. In the parameterization of the model, 1130 LLE experimental points were used, for 105 binary and 16 ternary systems, along with 815 IDAC experimental points, resulting in a new parameter table. A total of 42 parameters for groups and subgroups were calibrated. The F-SAC results were compared, whenever possible, with different versions of the UNIFAC model, freely available in the literature. The proposed model was able to correlate most of the IDAC and LLE data. For IDAC data, absolute average deviations were 0.139 and 0.608 ln units using F-SAC and UNIFAC(PSRK), respectively. Similar deviations were observed for LLE data. Finally, F-SAC was used to represent VLE data for aromatics + NFM systems. The absolute relative deviations were 0.038 and 0.082 in the bubble pressure calculations for F-SAC and UNIFAC(PSRK), respectively. The VLE results indicate a strong predictive ability of the suggested model, since no VLE data was used in the parameter estimation procedure. The results also indicate that the F-SAC model can be used to represent, simultaneously, IDAC, LLE and VLE for associating and non-associating mixtures, with a single set of parameters.

Neste trabalho, o modelo F-SAC (Functional-Segment Activity Coefficient) foi revisado e estendido de modo a representar, simultaneamente, coeficiente de atividade à diluição infinita (IDAC), equilíbrio líquido-líquido (LLE) e equilíbrio líquido-vapor (VLE), em amplos intervalos de temperatura. Originalmente, o modelo F-SAC foi calibrado essencialmente com dados experimentais de IDAC para diversas misturas. Investigouse, portanto, a possibilidade de o modelo levar também em consideração sistemas que apresentem LLE. Para a presente investigação, dois conjuntos de misturas industrialmente interessantes foram selecionados: hidrocarbonetos (alcanos, cicloalcanos, alcenos, cicloalcenos e compostos aromáticos) + n-formilmorfolina (NFM) e hidrocarbonetos + água. Foram propostas duas modificações na equação que calcula a energia de interação entre os grupos: (i) inspirado pelo comportamento típico observado na expansão isobárica de líquidos, um novo parâmetro foi proposto de modo a diminuir o (pequeno) erro sistemático do modelo com a temperatura no termo sem associação; (ii) em misturas associativas, utilizou-se uma equação similar para representar a dependência da energia de ligação de hidrogênio entre dois segmentos com a temperatura. Na estimação de parâmetros do modelo foram utilizados 1130 pontos experimentais de LLE, para 105 sistemas binários e 16 ternários, juntamente com um 815 dados experimentais de IDAC, resultando numa nova tabela de parâmetros. Ao total, foram estimados 42 parâmetros dos grupos e subgrupos. Os resultados do F-SAC foram comparados, quando possível, com diferentes versões do UNIFAC encontradas na literatura. O modelo proposto foi capaz de correlacionar a maior parte dos dados de IDAC e LLE utilizados. Para os dados de IDAC, foram observados desvios médios absolutos de 0,139 e 0,608 unidades de ln para o F-SAC e o UNIFAC(PSRK), respectivamente. Para os dados de LLE, foram observados desvios semelhantes. Finalmente, o F-SAC foi utilizado para representar dados de VLE para os sistemas de compostos aromáticos + NFM. Observou-se desvios médios relativos de 0,038 e 0,082 no cálculo da pressão de bolha utilizando-se o F-SAC e o UNIFAC(PSRK), respectivamente. Estes resultados comprovam a boa capacidade preditiva do modelo, uma vez que nenhum dado de VLE foi utilizado nas estimações. Conclui-se, desta forma, que o modelo FSAC pode ser usado para representar, simultaneamente, IDAC, LLE e VLE de misturas com e sem ligação de hidrogênio, utilizando um único conjunto de parâmetros. / In this work, the F-SAC model (Functional -Segment Activity Coefficient) has been revised and extended in order to simultaneously represent activity coefficient at infinite dilution (IDAC), liquid-liquid equilibrium (LLE) and vapor-liquid equilibrium (VLE), in a wide range of temperature. Originally, the F- SAC model was calibrated essentially with IDAC experimental data for several mixtures. In the present work, the possibility of also taking into account LLE was investigated. For this study, two sets of industrially interesting mixtures were selected : hydrocarbons (alkanes, cycloalkanes, alkenes, cycloalkenes and aromatics) + n-formylmorpholine (NFM) and hydrocarbons + water. Modifications were suggested in the equation that computes the interaction energy for each contact between two segments: (i) inspired by the typical behavior observed in the isobaric expansion of liquids, a new parameter has been proposed in order to reduce the (small) systematic error with the temperature for non-associating systems; (ii) for associating mixtures, a similar equation was proposed to represent the temperature dependence of the hydrogen bond energy between two segments. In the parameterization of the model, 1130 LLE experimental points were used, for 105 binary and 16 ternary systems, along with 815 IDAC experimental points, resulting in a new parameter table. A total of 42 parameters for groups and subgroups were calibrated. The F-SAC results were compared, whenever possible, with different versions of the UNIFAC model, freely available in the literature. The proposed model was able to correlate most of the IDAC and LLE data. For IDAC data, absolute average deviations were 0.139 and 0.608 ln units using F-SAC and UNIFAC(PSRK), respectively. Similar deviations were observed for LLE data. Finally, F-SAC was used to represent VLE data for aromatics + NFM systems. The absolute relative deviations were 0.038 and 0.082 in the bubble pressure calculations for F-SAC and UNIFAC(PSRK), respectively. The VLE results indicate a strong predictive ability of the suggested model, since no VLE data was used in the parameter estimation procedure. The results also indicate that the F-SAC model can be used to represent, simultaneously, IDAC, LLE and VLE for associating and non-associating mixtures, with a single set of parameters.

High-dimensional data sets can be difficult to visualize and analyze, while data in low-dimensional space tend to be more accessible. In order to aid visualization of the underlying structure of a dataset, the dimension of the dataset is reduced. The simplest approach to accomplish this task of dimensionality reduction is by a random projection of the data. Even though this approach allows some degree of visualization of the underlying structure, it is possible to lose more interesting underlying structure within the data. In order to address this concern, various supervised and unsupervised linear dimensionality reduction algorithms have been designed, such as Principal Component Analysis and Linear Discriminant Analysis. These methods can be powerful, but often miss important non-linear structure in the data. In this thesis, manifold learning approaches to dimensionality reduction are developed. These approaches combine both linear and non-linear methods of dimension reduction.

5 Abstract Protein lipidation occupies an important place in the post-translational modification group. It plays an important role in the processes of cell differentiation or synaptic transmission. Changes in the presence of lipidated proteins may in some cases indicate the appearance of human diseases such as Huntington's disease, schizophrenia, or cancer. For this reason, their analysis has become desirable, but also problematic due to its specific characteristics. This work is focused on the development of conditions for measurement of lipopeptides produced by cleavage of proteins containing lipomodification. Two separation procedures were tested, in which the aim of work was to enrich the lipopeptides and remove as many other molecules as possible from the samples. The first test method - phase interface separation, using water and chloroform, was able to concentrate the lipopeptides on the phase interface and separate some of the added unmodified peptides from the mixture. The second method of separation was a solid phase using a C18 stationary phase. As a more suitable solid phase extraction variant, elution with solutions containing acetonitrile with trifluoroacetic acid was chosen as compared to solutions containing isopropyl alcohol. With the modified elution procedure, it was possible to separate...

Šiame darbe nagrinėjami netiesiniai daugdarų atpažinimo metodai ir daugiamačių duomenų projekcijos metodai. Siūloma jais spręsti keliais laisvės laipsniais besisukančio objekto orientacijos radimo problemą. Aprašomi MDS, Trianguliacijos, Sammon, RPM, mRPM, CCA, PCA, LLE, LE, HLLE, LTSA, SMACOF ir Isomap metodai. Su kai kuriais iš jų atliekami web-kamera gautų galvos atvaizdų tyrimai. Isomap algoritmo pagrindu sukuriama programinė įranga ir su ja taipogi atliekami galvos orientacijos tyrimai. / This paper deals with Analysis of non-linear manifold learning methods and multidimensional data projection methods. It is proposed use them in solving problem of detection of orientation of object, moving in few degrees of freedom. Methods described: MDS, triangulation, Sammon, RPM, mRPM, CCA, PCA, LLE, LE, HLLE, LTSA, SMACOF and Isomap. Some of them are used to analyze head images acquired by webcam.. Application is created which is then used to analyze head orientation by implementing Isomap method.

This work focuses on wood extractives in effluents from the CTMP plant at Skoghall Mill. Pulp and paper industry effluents contain mostly natural compounds which are part of the trees. They are toxic to aquatic life but harmless in nature, as they are present in low concentrations. Processing tons of wood, such as in a pulp mill, strongly increases the concentrations of the toxic compounds (Ali, M. and Sreekrishnan, T., 2001) which have to be treated before transferring to the aquatic environment.Extractives can be found in different forms, as micelles soluble in water, unprocessed in fibers or absorbed on the surface of fibers. It is important to know in which forms extractives are mostly present in the effluent, so that they can be treated more efficiently. It is desired to have extractives absorbed on the fibers and fibrils present in the waste water, so they can be separated from the water and treated separately, e.g. burned for energy recovery. Dissolved extractives complicate the oxygen transfer in an aerated biological treatment step with their surface active properties (Sandberg, 2012).The aim of this study is quantification of extractives on the fibers suspended in the waste water and extractives dissolved in the water. The distribution between the two forms is an important input when designing future effluent treatment. Wood extractives itself are a wide group with different compounds. This work focuses on the main groups present in waste water: resin acids, free and esterified fatty acids and, free and esterified sterols. These groups are analyzed in different process waters and waste water before the waste water treatment plant. The measured concentrations of extractives were as expected, higher in process and effluent waters, lower in white water. Most of the extract was dissolved in the water and unfortunately fiber samples contained very low concentration from the total extract in the samples.

Mémoire numérisé par la Direction des bibliothèques de l'Université de Montréal.

Conselho Nacional de Desenvolvimento Científico e Tecnológico / Agriculture is one of the main economic activities of the Brazil, possessing the culture of coffee as their main primary commodity. In order to achieve high levels of production and prevent the loss of crops, farm inputs such as pesticides are used in large quantities. This type of practice and the increased public concern consumer are motivating scientific research about these topics, with the aim of ensuring food safety of products supplied for consumption. However, the literature does not report methodologies for the determination of pesticide residues in roasted coffee, with the methods of extraction and analysis proposed, which according to bibliographic data present residues of these substances even after roasting. In this context, the present study sought to develop an analytical methodology to determine carbofuran, cypermethrin, chlorpyrifos, clothianidin, disulfoton, endosulfan, spirodiclofen, haloxyfop, imidacloprid, tebuconazole, triadimenol and triadimefom in roasted coffee using solid liquid extraction method by sonication with clean-up step through liquid liquid extraction (LLE) and instrumental analysis for ultra performance liquid chromatography tandem mass spectrometry detector (LC-MS/MS). For both, the optimization of chromatographic conditions and assessing the best method of clean-up was carried out. With that, the conditions for better quantitative response was obtained with 0.5 g of roasted coffee, 5 mL extraction solvent by sonication, 1 mL of water and 1 mL of dichloromethane in the clean-up by LLE. During the validation procedure in LC-MS/MS positive matrix effect was observed for all analytes except of cypermethrin in which there was suppression of the signal by array, causing, prepared in the curve-based extract, recoveries between 74.2%-102.4% standard relative deviation (SRD) between 0.6% and 10.2%, in levels concentration assessed in the process, Furthermore, the linearity obtained was in the range of 0.9979 to 0.9998 for pesticides , carbofuran, disulfoton, imidacloprid, clothianidin tebuconazole, triadimenol, triadimefom and ensuring the efficiency of the methodology. / A agricultura é uma das principais atividades econômicas do Brasil, possuindo a cultura do café como sua principal commodity primária. A fim de atingir altos níveis de produção e evitar a perda das safras, insumos agrícolas como os agrotóxicos são utilizados em grandes quantidades. Este tipo de prática e o aumento da preocupação do público consumidor estão motivando investigações científicas a respeito destes tópicos, com o intuito de garantir a segurança alimentar dos produtos fornecidos para o consumo. No entanto, a literatura não relata metodologias para a determinação de resíduos de agrotóxicos no café torrado, com os métodos de extração e análise propostos, que segundo dados bibliográficos apresentam resíduos destas substâncias mesmo após a torrefação. Neste contexto, o presente trabalho buscou desenvolver uma metodologia analítica para determinar carbofurano, cipermetrina, clorpirifós, clotianidina, dissulfotom, endosulfan, espirodiclofeno, haloxifope, imidacloprido, tebuconazol, triadimefom e triadimenol em café torrado empregando método de extração sólido líquido por sonicação com etapa de clean-up por meio de extração líquido líquido (LLE) e análise instrumental por cromatografia líquida de ultra eficiência acoplada a detector por espectrometria de massas (LC-MS/MS). Para tanto, a otimização das condições cromatográficas e a avaliação do melhor método de clean-up foram realizadas. Com isso, as condições de melhor resposta quantitativa foi obtida com 0,5 g de café torrado, 5 mL de solvente de extração por sonicação, 1 mL de água e 1 mL de diclorometano na etapa de clean-up por LLE. Durante o procedimento de validação no LC-MS/MS foi observado efeito matriz positivo para todos os analitos com exceção da cipermetrina na qual houve supressão do sinal pela matriz, ocasionando, com base na curva preparada no extrato, recuperações entre 74,2% - 102,4%, com desvio padrão relativo entre 0,6% e 10,2%, nos níveis de concentração avaliados no processo, além disso, a linearidade obtida foi na faixa de 0,9979 a 0,9998 para os agrotóxicos, carbofurano, clotianidina dissulfotom, imidacloprido, tebuconazol, triadimefom e triadimenol, garantindo a eficiência da metodologia.

Par l’arrêt Pla et Puncernau contre Andorre du 13 juillet 2004, la Cour européenne des droits de l’homme a fait une entrée remarquée dans le contrôle de l’interprétation des actes juridiques privés. Tenante d’une conception réaliste du droit, elle a élaboré des Principes européens d’interprétation qui sont autant d’obligations qui pèsent sur le juge national. La CJUE partage en grande partie ces principes, mais possède ses particularités et hésite encore à s’aligner sur le contrôle du juge de Strasbourg. Le juge français est quant à lui confronté à des directives d’interprétation qui sont de simples conseils, tandis que la Cour de cassation se refuse à opérer un autre contrôle que celui de la dénaturation. Dès lors, elle n’est pas en mesure de redresser les fautes commises par les juges du fond notamment, et encourt le risque d’engager la responsabilité de l’Etat français devant la Cour EDH. Si la Cour de cassation a entamé une réflexion sur la façon de rendre la justice en coordination avec les cours européennes, la modification du système herméneutique français se heurte à de nombreux obstacles révélateurs de la conception traditionnelle du droit français des obligations. / With the case of Pla and Puncernau versus Andorra on 13 July 2004, the European Court of Human Rights has made a dramatic entrance in the supervision of the interpretation of private contracts. Defending a realist conception of law, it developed European Principles of interpretation, which are obligations for national courts. The CJEU largely shares these principles, but has its particularities and it is still reluctant to align with the Strasbourg Court’s supervision. The French judge is itself faced with interpretive guidelines that are simple advice, while the Court of Cassation refused to operate another supervision than the denaturation. Therefore, it is not able to correct the errors committed by the trial judges in particular, and takes the risk to commit a violation of the European law. If the Supreme Court considers by now how to deliver justice in coordination with the European courts, changing the French hermeneutic system faces many obstacles revealing the traditional conception of French law of obligations.