Nous présentons des approches théoriques et numériques pour deux dynamiques irréversibles et parallèles sur des modèles de mécanique statistique. Dans le premier chapitre, nous présentons les résultats théoriques sur un système de particules induite par une chaîne de Markov irréversible, à savoir le TASEP. Permettant des multiples retournements de spin \`à chaque itération, nous définissons un modèle avec une dynamique parallèle appartenant à la famille des PCA et nous dérivons sa mesure stationnaire. Dans ce cadre, nous traitons {\it le problème du blocage}, {\it i.e.} comprendre les effets d’une perturbation localisée dans le taux de transition des particules sur des systèmes irréversibles: le problème du blocage. Dans le deuxième chapitre, nous présentons une version unidimensionnelle du modèle d'Ising avec potentiel de Kac. Nous définissons une PCA avec une interaction asymétrique et nous trouvons sa mesure stationnaire avec condition aux limites périodique.Ensuite, nous prouvons la convergence, dans la limite thermodynamique, de cette mesure stationnaire vers la mesure de Gibbs pour toutes les températures supérieures à la température critique via les estimations de F\"ollmer et le théorème d'unicité de Dobrushin. Dans la seconde partie de la thèse, nous étudions ces deux dynamiques à travers des expériences numériques. Dans le cas du TASEP en exploitant des processeurs graphiques (GPU) et CUDA pour identifier une estimation raisonnable du {temps de m\'elange} et renforcer la conjecture qu’à la fois dans la version, la règle de mise à jour série ou parallèle, le courant peut ne pas être analytique dans l’intensité du blocage autour de la valeur $ \varepsilon = 0 $ / In this thesis we present theoretical and numerical approaches for two irreversible and parallel dynamics on one-dimensional statistical mechanics models. In the first chapter we present theoretical results on a particles system driven by an irreversible Markov chain namely the totally asymmetric simple exclusion process (TASEP). Allowing multiples spin-flips in each time-step we define a model with a parallel dynamics that belongs to the family of the probabilistic cellular automata (PCA) and we derive its stationary measure. In this framework we deal with {\it the blockage problem}, {\it i.e.} to understand the effects of a localized perturbation in the transition rates of the particles on irreversible systems: the blockage problem. In the second chapter we present a one-dimensional version of the Ising model with Kac potential. Again we define a PCA dynamics with asymmetric interaction between particles and we find its stationary measure for periodic boundary condition. Then we prove the convergence, in the thermodynamic limit, of such stationary measure to the Gibbs measure for all temperatures above the critical one via F\"ollmer estimates and dobrushin's uniqueness theorem. In the second part of the thesis, we investigate these two dynamics through numerical experiments.In the case of the TASEP we exploit general purpose graphical processors unit (GPGPU) writing a parallel code in CUDA to identify a reasonable {\it mixing time} and reinforce the conjecture that in both version, serial or parallel update rule, the current may be non-analytic in the blockage intensity around the value $\varepsilon = 0$

The building sector is responsible for around 39% of the energy consumption in Sweden, and one way to work towards sustainable societies could be to make the buildings more energy efficient. One approach to make a building more energy efficient is to use knowledge gained from digitalization of the building and to make the building smart. This thesis aims to study the area of smart buildings, and the ongoing work with smart solutions within the real estate sector.  Two parallel investigations are used to study the area. One is an interview study in order to map the ongoing work with smart buildings. The situation on the market, the matureness of technical solutions as well as ongoing trends and challenges are amongst other things studied. The second investigation consists of a pilot project which aims to exemplify how time series data analysis could be used in order to make a building smarter. Time-series prediction provides a way to discover and quantify regularities in such data, and methods of time series prediction point to how to make building management more efficient.  The result of the study shows that the smart building market is not yet stabilized, but that the interest in working with smart buildings is big. There are many smaller solutions which are being tested and implemented, but there is no consensus of what the definition of a smart building really is. The results of the data analysis indicate two results, firstly, it provides insight in the data, and reports how one should prepare the data for subsequent analysis, and secondly we report results for different autoregressive (AR)-based time series models. For the second result, we indicate how methods of K-means improve over linear AR-based modelling, pointing to the possible use of nonlinear modelling. We however question whether performance improvements are sufficiently large for this application to justify the additional computational demands.

På pappersmaskin 11 i Hallsta pappersbruk tillverkas högglansigt papper som bleks med väteperoxid. Om väteperoxiden som finns kvar efter blekningen, kallat restperoxid, kommer ut på pappersmaskinen sliter detta på termovalsarna, vilket påverkar slutprodukten. Det finns en bakteriekultur på bruket, bestående av släktet Tepidiphilus som bryter ner väteperoxiden med hjälp av katalas och hindrar den från att komma ut på maskinen. Om bakterierna slås ut, av exempelvis ogynnsamma förhållanden, märks detta genom att restperoxiden ökar. Målet med detta arbete är att försöka identifiera vilka faktorer som påverkar bakterierna och lägga fram ett förslag för hur man kan förhindra uppkomsten av restperoxid. För att utföra detta användes i första hand principalkomponentanalys, samt så studerades hur olika faktorer förändrades i förhållanden till uppkomsten av restperoxid. Ett antal parametrar kunde identifieras baserat på tillgängliga data. Dock kan inga absoluta slutsatser dras då det inte går att bekräfta några teorier med labbförsök eller genom att manipulera processen. De parametrar som tros ha störst påverkan är mek. renat färg, rest-aluminium, grumlighet, manganhalten i vatten, pH för kar K0203, returmälden samt inloppslådan och klorhalterna som uppmätts för mek. renat total klor, RVV1 fri klor, varmvatten total klor och mek. renat fri klor. Dessa parametrar bör därför övervakas framåt, och kan i det fall att restperoxid återigen uppkommer antingen bekräftas eller dementeras deras faktiska påverkan och samspel för bakteriernas välmående. / Paper machine 11 in Hallsta paper mill produces high-gloss paper that is bleached with hydrogen peroxide. If the hydrogen peroxide that remains after the bleaching, called residual peroxide, gets out on the paper machine, this wears down the thermal rollers, which affects the final product.  There is a bacterial culture at the mill, consisting of the genus Tepidiphilus, which breaks down the hydrogen peroxide with the help of catalase and prevents it from getting out to the machine. If the bacteria are killed, for example by unfavourable conditions, this is noticeable by an increase in residual peroxide. The goal of this project is to try to identify which factors affect the bacteria and put forward a proposal for how to prevent the occurrence of residual peroxide. To carry this out, principal component analysis was primarily used, and how different factors changed in relation to the appearance of residual peroxide was studied. Several parameters could be identified based on the available data. However, no conclusive conclusions could be drawn as it is not possible to confirm any theories with lab tests or by manipulating the process. The parameters believed to have the greatest influence are coloration of mechanically cleaned water (mech. water), residual aluminium, turbidity, manganese content of the water, pH for tank K0203, the returning pulp suspension as well as the headbox and the chlorine levels for mech. water total chlorine, RVV1 free chlorine, hot water total chlorine and mech. water free chlorine. These parameters should therefore be monitored going forward, and if residual peroxide occurs again, their possible interactions and actual impact on the well-being of the bacteria can either be confirmed or denied

Chemical compounds in indoor air can adversely affect our comfort and health. However, in most cases there is only a limited amount of information available that can be used to assess their health risk. Instead the precautionary principle is often applied, i.e. efforts are made to ensure that the concentrations of pollutants are kept at a minimum when constructing new buildings or conducting renovations by using low-emitting building materials. Today, when investigating buildings in order to solve indoor air quality problems, volatile organic compounds (VOCs) are sampled in the air within rooms. The chemical composition of indoor air is complex and there are many sources for the chemicals present. The potential for emissions from sources in hidden spaces such as wall cavities is poorly understood and little information exists on the toxic potential of chemical releases resulting from moisture-related degradation of building materials. Most of the non-reactive VOCs that have been detected in indoor air in field studies and from building products are not believed to cause health problems. However, reactive compounds and chemical reaction products have the potential to negatively influence our comfort and health even at low concentrations. Even though the impact of chemical compounds on health is unclear in many cases, they can be used to identify technical problems in buildings. When a building is investigated, the air inside building structures could be sampled. This method would eliminate emissions from sources other than the construction materials and the samples would contain higher levels of individual compounds. The aims of this work was to identify emissions profiles for different types of building structures, to see if the emission profiles for moisture damaged and undamaged structures differed, and to determine whether any of the emissions profiles for specific structures also could be found in indoor air. Technical investigations and VOC sampling were performed in 21 different buildings with and without previous moisture damage. Seven of the buildings were investigated in the years 2005-2006 (study 1) and fourteen in the years 2009-2010 (study 2). In study 1, sixty samples were analyzed by PCA at the chemical group level (18 chemical groups, i.e. aldehydes, ketones etc). 41 % of all identified chemical compounds belonged to the hydrocarbon chemical group. The second largest chemical groups, each of which accounted for 5-10 % of all identified compounds, were alcohols, aldehydes, ketones, polyaromatic hydrocarbons (PAHs) and terpenes. The results indicated that one of the main factors that determined the emissions profile of a building structure was the materials used in its construction. Notably, concrete and wooden structures were found to have different emissions profiles. The sum of VOC (TVOC) concentrations for all 241 samples from both study 1 and study 2 was used to compare total emissions between different building elements (ground and higher floors, external walls and roof spaces). Most building elements exhibited relatively low emissions compared to concrete ground floors, which generally had higher TVOC emissions. Emissions from both polystyrene insulation and PVC flooring could be identified in concrete ground floors and were the main cause for the higher emissions found in these structures. Profiles for wood preservatives such as creosote and pentachlorophenol were also identified in external walls. The emission profiles found in the structures could not be identified in the indoor air in the adjacent rooms, although individual compounds were sometimes detected at low concentrations. Our results showed that the main factors influencing emissions in building structures were the construction materials and the nature of the building element in question. Because of difficulties with finding active water damage at the times of sampling and because of sampling inside closed building structures with old dried-out moisture damages, the field method used in this work was unsuitable for identifying differences in emission profiles between moisture damaged and undamaged structures. It will thus be necessary to investigate this difference in a laboratory where the precise composition of all tested structures is known, a range of RH values can be tested and the accumulation of emissions can be followed. / Kompetenscentrum Byggnad - Luftkvalitet - Hälsa 2 (KLUCK 2)

<p><strong>Background:</strong> Through unprecedented advances in technology, high-dimensional datasets have exploded into many fields of observational research. For example, it is now common to expect thousands or millions of genetic variables (p) with only a limited number of study participants (n). Determining the important features proves statistically difficult, as multivariate analysis techniques become flooded and mathematically insufficient when n < p. Principal Component Analysis (PCA) is a commonly used multivariate method for dimension reduction and data visualization but suffers from these issues. A collection of Sparse PCA methods have been proposed to counter these flaws but have not been tested in comparative detail. <strong>Methods:</strong> Performances of three Sparse PCA methods were evaluated through simulations. Data was generated for 56 different data-structures, ranging p, the number of underlying groups and the variance structure within them. Estimation and interpretability of the principal components (PCs) were rigorously tested. Sparse PCA methods were also applied to a real gene expression dataset. <strong>Results:</strong> All Sparse PCA methods showed improvements upon classical PCA. Some methods were best at obtaining an accurate leading PC only, whereas others were better for subsequent PCs. There exist different optimal choices of Sparse PCA methods when ranging within-group correlation and across-group variances; thankfully, one method repeatedly worked well under the most difficult scenarios. When applying methods to real data, concise groups of gene expressions were detected with the most sparse methods. <strong>Conclusions:</strong> Sparse PCA methods provide a new insightful way to detect important features amidst complex high-dimension data.</p> / Master of Science (MSc)

Master of Science / Department of Electrical and Computer Engineering / Balasubramaniam Natarajan / In this thesis, we present an automated system for detection and classification of cracks, based on the new standard proposed by `American Association of State Highway and Transportation Officials (AASHTO)'. The AASHTO standard is a draft standard, that attempts to overcome the limitations of current crack quantifying and classification methods. In the current standard, the crack classification relies heavily on the judgment of the expert. Thus the results are susceptible to human error. The effect of human error is especially severe when the amount of data collected is large. This lead to inconsistencies even if a single standard is being followed. The new AASHTO guidelines attempt to develop a method for consistent measurement of pavement condition. Gray scale images of the road are captured by an image capture vehicle and stored on a database. Through steps of thresholding, line detect and scanning, the gray scale image is converted to binary image, with 'zeros' representing cracked pixels. PCA analysis, followed by closing and filtering operation, are carried out on the gray scale image to identify cracked sub-images. The output from the filtering operation, is then replaced with its binary counterpart. In the final step the crack parameters are calculated. The region around the crack is divided into blocks of 32x32 to approximate and calculate the crack parameters with ease. The width of the crack is approximated by the average width of crack in each block. The orientation of the crack is calculated from the angle between direction of travel and the line joining the ends of the crack. Length of the crack is the displacement between the ends of the crack, and the position of the crack is calculated from the midpoint of the line joining the end points.

The work presented in this thesis has demonstrated that the use of elemental profiles for the quality control of herbal medicines can be applied to multiple stages of processing. A single method was developed for the elemental analysis of a variety of St John’s Wort (Hypericum perforatum) preparations using Inductively Coupled Plasma – Optical Emission Spectroscopy (ICP-OES). The optimised method consisted of using 5 ml of nitric acid and microwave digestion reaching temperatures of 185⁰C. Using NIST Polish tea (NIST INCT-TL- 1) the method was found to be accurate and the matrix effect from selected St John’s Wort (SJW) preparations was found to be ≤22%. The optimised method was then used to determine the elemental profiles for a larger number of SJW preparations (raw herbs=22, tablets=20 and capsules=12). Specifically, the method was used to determine the typical concentrations of 25 elements (Al, As, B, Ba, Be, Ca, Cd, Co, Cr, Cu, Fe, Hg, In, Mg, Mn, Mo, Ni, Pb, Pt, Sb, Se, Sr, V, Y and Zn) for each form of SJW which ranged from not detected to 200 mg/g. To further interpret the element profiles, Principal Component Analysis (PCA) was carried out. This showed that different forms of SJW could be differentiated based on their elemental profile and the SJW ingredient used (i.e. extract or raw herb) identified. The differences in the profiles were likely due to two factors: (1) the addition of bulking agents and (2) solvent extraction. In order to further understand how the elemental profile changes when producing the extract from the raw plant, eight SJW herb samples were extracted with four solvents (100% water, 60% ethanol, 80% ethanol and 100% ethanol) and analysed for their element content. The results showed that the transfer of elements from the raw herb to an extract was solvent and metal dependent. Generally the highest concentrations of an element were extracted with 100% water, which decreased as the concentration of ethanol increased. However, the transfer efficiency for the element Cu was highest with 60% ethanol. The solvents utilised in industry (60% and 80% ethanol) were found to preconcentrate some elements; Cu (+119%), Mg (+93%), Ni (+183%) and Zn (+12%) were found to preconcentrate in 60 %v/v ethanol extracts and Cu (+5%) and Ni (+30%). PCA of the elemental profiles of the four types of extract showed that differentiation was observed between the different solvents and as the ethanol concentration increased, the extracts became more standardised. Analysis of the bioactive compounds rutin, hyperoside, quercetin, hyperforin and adhyperforin followed by subsequent Correlation Analysis (CA) displayed relationships between the elemental profiles and the molecular profiles. For example strong correlations were seen between hyperoside and Cr as well as Quercetin and Fe. This shows potential for tuning elemental extractions for metal-bioactive compounds for increased bioactivity and bioavailability; however further work in needed in this area.

Stainless steel (SS) is the material of choice in a number of process industries ranging from food and dairy to pharmaceuticals. Adsorption phenomena on SS surfaces are of paramount importance in these industries. For example, protein adsorption constitutes a major issue in process equipment, as the associated surface fouling decreases the efficiency of the overall process and leads to an increase in operational costs because of the need for regular cleaning. In addition, the adsorption of proteins at solid–liquid interfaces is an important research field with relevance in biosensor and biomaterial applications. The primary aim of this thesis was to understand the underlying adsorption properties of selected protein onto SS surfaces and to identify the influence of specific amino acids on bio-fouling. Protein adsorption experiments were carried out on 316 grade SS sensors using a quartz crystal microbalance with dissipation (QCM-D). The proteins consisted of milk proteins (α-lactalbumin, β-lactoglobulin, α-casein, β-casein, κ-casein and bovine serum albumin), blood proteins (cytochrome-c, haemoglobin and myoglobin) and proteins of industrial and medical relevance (α-chymotrypsinogen, human recombinant insulin, lysozyme and papain). The adsorption characteristics of the test proteins were studied and an empirical correlation relating the amount of protein adsorbed to their physical properties was proposed. Adsorption onto a SS surface was followed on the QCM-D in real time and the amounts adsorbed calculated using the Sauerbrey model. In addition, the binding kinetics was modelled using different theoretical models to describe the adsorption mechanism. In all the proteins tested, the conformational change model was found to fit considerably well the adsorption data. Finally, the data collected were used to identify the physical properties of proteins that induce surface binding, with hydrophobic and aromatic amino acids having the most effect on binding. A second aspect investigated in the present work was the determination of hydration water present in the adsorbed layer. In fact, water molecules, solvated ions and other small molecules in the vicinity of the surface all play an important role in protein adsorption and often constitute a large fraction of the total measured adsorbed mass. The fraction of water present on SS surfaces along with adsorbed proteins was determined using fluorescently labelled proteins through a comparative study that included QCM-D experiments as well as fluorescent light intensity measurements. The results were similar for all proteins tested, indicating that 32-45.8% of the total mass adsorbed composed of water. One last aspect considered in this thesis was the influence of the putative adhesive amino acid 3, 4-dihydroxyphenylalanine (DOPA). DOPA residues are present in high levels in the adhesive proteins from marine mussels, hence are thought to facilitate surface attachment. The role of DOPA residues in mediating protein adhesion on SS surfaces was studied using QCM-D. Two repetitive peptide motifs extracted from the sequence of the mussel foot protein mefp-5, KGYKYYGGSS and KGYKYY, were selected for this study. The two peptides contained unmodified tyrosine (Y) residues, which were chemo-enzymatically modified to DOPA using mushroom tyrosinase. Adsorption of the two sequences on SS surfaces was tested before and after modification of tyrosine residues to DOPA. Conversion was linearly related to the incubation time of the peptide fragments with mushroom tyrosinase, Amount of DOPA formed was 70-99% of the tyrosine content in the peptides. QCM-D adsorption experiments on the DOPA-modified sequences revealed four-fold greater adhesion than for unmodified mefp-5 motifs, indicating the paramount role that DOPA has on the adsorption of peptides on 316 grade stainless steel.

Worldwide, 3.575 million people die each year from water-related diseases. The water and sanitation crisis claims more lives than any warfare and is predicted to be one of the biggest global challenges of this century. The rapid, accurate detection of viral pathogens from environmental samples is an ongoing and pertinent challenge in biological engineering. Currently employed methods are lacking in either efficiency or specificity. Here we explore a novel method for virus detection and concurrently use this method to learn more about the very early stages of the virus infection process. The method combines Fourier transform infrared (FTIR) spectroscopy, a method of visualizing molecules based on changes in vibration of particles, and mammalian cells as the biosensor. This method is used to detect and investigate viruses from the family picornaviridae, chosen due to their public health burden and their widespread presence in environmental samples, especially water sources. This family includes the Polioviruses, echoviruses and Coxsackieviruses, among others, many of which are human pathogens.The research outlined in this dissertation is aimed at developing and implementing a new cell-based biosensor that combines the advantages of FTIR spectroscopy with the ability of buffalo green monkey kidney (BGMK) cells to sense diverse stimuli, including infective enteroviruses. The goal of developing this biosensor is outlined in the first paper. The second paper focuses on the application of advanced statistical methods to analyze the spectra to discriminate different viral infections in BGMK cells. Finally, we designed a non-reactive metal biochamber to use with attenuated total reflectance-FTIR. This allowed near-continuous acquisition of real-time spectral data for the study of biochemical changes in mammalian cells caused by poliovirus (PV1) infection. This system is capable of tracking changes in cell biochemistry in minute intervals for many hours at a time.This work demonstrates the feasibility of FTIR spectroscopy in combination with the broad sensitivity of mammalian cells for potential use in the detection of infective viruses from environmental samples. We envision this method being extended to high throughput, automated systems to screen for viruses or other toxins in drinking water systems and medical applications.

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