Design and implementation of a multipurpose reactor for scale-up studies

In the theoretical part, the different polymerisation catalysts are introduced and the phenomena related to mixing in the stirred tank reactor are presented. Also the advantages and challenges related to scale-up are discussed. The aim of the applied part was to design and implement an intermediate-sized reactor useful for scale-up studies. The reactor setting was tested making one batch of Ziegler–Natta polypropylene catalyst. The catalyst preparation with a designed equipment setting succeeded and the catalyst was analysed. The analyses of the catalyst were done, because the properties of the catalyst were compared to the normal properties of Ziegler–Natta polypropylene catalyst. The total titanium content of the catalyst was slightly higher than in normal Ziegler–Natta polypropylene catalyst, but the magnesium and aluminium content of the catalyst were in the normal level. By adjusting the siphonation tube and adding one washing step the titanium content of the catalyst could be decreased. The particle size of the catalyst was small, but the activity was in a normal range. The size of the catalyst particles could be increased by decreasing the stirring speed. During the test run, it was noticed that some improvements for the designed equipment setting could be done. For example more valves for the chemical feed line need to be added to ensure inert conditions during the catalyst preparation. Also nitrogen for the reactor needs to separate from other nitrogen line. With this change the pressure in the reactor can be kept as desired during the catalyst preparation. The proposals for improvements are presented in the applied part. After these improvements are done, the equipment setting is ready for start-up. The computational fluid dynamics model for the designed reactor was provided by cooperation with Lappeenranta University of Technology. The experiments showed that for adequate mixing with one impeller, stirring speed of 600 rpm is needed. The computational fluid dynamics model with two impellers showed that there was no difference in the mixing efficiency if the upper impeller were pumping downwards or upwards.

Currency Carry Trades – Excess Returns and Performance Optimization of G10 Currency Carry Trade Strategies during the Euro Era

This study examines the excess returns provided by G10 currency carry trading during the Euro era. The currency carry trade has been a popular trade throughout the past decades offering excess returns to investors. The thesis aims to contribute to existing research on the topic by utilizing a new set of data for the Euro era as well as using the Euro as a basis for the study. The focus of the thesis is specifically on different carry trade strategies’ performance, risk and diversification benefits. The study finds proof of the failure of the uncovered interest rate parity theory through multiple regression analyses. Furthermore, the research finds evidence of significant diversification benefits in terms of Sharpe ratio and improved return distributions. The results suggest that currency carry trades have offered excess returns during 1999-2014 and that volatility plays an important role in carry trade returns. The risk, however, is diversifiable and therefore our results support previous quantitative research findings on the topic.

Successful scale-up of human embryonic stem cell production in a stirred microcarrier culture system

Future clinical applications of human embryonic stem (hES) cells will require high-yield culture protocols. Currently, hES cells are mainly cultured in static tissue plates, which offer a limited surface and require repeated sub-culturing. Here we describe a stirred system with commercial dextran-based microcarriers coated with denatured collagen to scale-up hES cell production. Maintenance of pluripotency in the microcarrier-based stirred system was shown by immunocytochemical and flow cytometry analyses for pluripotency-associated markers. The formation of cavitated embryoid bodies expressing markers of endoderm, ectoderm and mesoderm was further evidence of maintenance of differentiation capability. Cell yield per volume of medium spent was more than 2-fold higher than in static plates, resulting in a significant decrease in cultivation costs. A total of 10(8) karyotypically stable hES cells were obtained from a unitary small vessel that needed virtually no manipulation during cell proliferation, decreasing risks of contamination. Spinner flasks are available up to working volumes in the range of several liters. If desired, samples from the homogenous suspension can be withdrawn to allow process validation needed in the last expansion steps prior to transplantation. Especially when thinking about clinical trials involving from dozens to hundreds of patients, the use of a small number of larger spinners instead of hundreds of plates or flasks will be beneficial. To our knowledge, this is the first description of successful scale-up of feeder- and Matrigel™-free production of undifferentiated hES cells under continuous agitation, which makes this system a promising alternative for both therapy and research needs.

Access to care for the poor living with chronic disease in India : an analysis of selected national health policies

Cette recherche sur les barrières à l’accès pour les pauvres atteints de maladies chroniques en Inde a trois objectifs : 1) évaluer si les buts, les objectifs, les instruments et la population visée, tels qu'ils sont formulés dans les politiques nationales actuelles de santé en Inde, permettent de répondre aux principales barrières à l’accès pour les pauvres atteints de maladies chroniques; 2) évaluer les types de leviers et les instruments identifiés par les politiques nationales de santé en Inde pour éliminer ces barrières à l’accès; 3) et évaluer si ces politiques se sont améliorées avec le temps à l’égard de l’offre de soins à la population pour les maladies chroniques et plus spécifiquement chez les pauvres. En utilisant le Framework Approach de Ritchie et Spencer (1993), une analyse qualitative de contenu a été complétée avec des politiques nationales de santé indiennes. Pour commencer, un cadre conceptuel sur les barrières à l’accès aux soins pour les pauvres atteints de maladies chroniques en Inde a été créé à partir d’une revue de la littérature scientifique. Par la suite, les politiques ont été échantillonnées en Inde en 2009. Un cadre thématique et un index ont été générés afin de construire les outils d’analyse et codifier le contenu. Finalement, les analyses ont été effectuées en utilisant cet index, en plus de chartes, de maps, d'une grille de questions et d'études de cas. L’analyse a tété effectuée en comparant les barrières à l’accès qui avaient été originalement identifiées dans le cadre thématique avec celles identifiées par l’analyse de contenu de chaque politique. Cette recherche met en évidence que les politiques nationales de santé indiennes s’attaquent à un certain nombre de barrières à l’accès pour les pauvres, notamment en ce qui a trait à l’amélioration des services de santé dans le secteur public, l’amélioration des connaissances de la population et l’augmentation de certaines interventions sur les maladies chroniques. D’un autre côté, les barrières à l’accès reliées aux coûts du traitement des maladies chroniques, le fait que les soins de santé primaires ne soient pas abordables pour beaucoup d’individus et la capacité des gens de payer sont, parmi les barrières à l'accès identifiées dans le cadre thématique, celles qui ont reçu le moins d’attention. De plus, lorsque l’on observe le temps de formulation de chaque politique, il semble que les efforts pour augmenter les interventions et l’offre de soins pour les maladies chroniques physiques soient plus récents. De plus, les pauvres ne sont pas ciblés par les actions reliées aux maladies chroniques. Le risque de les marginaliser davantage est important avec la transition économique, démographique et épidémiologique qui transforme actuellement le pays et la demande des services de santé.

Deep learning of representations and its application to computer vision

L’objectif de cette thèse par articles est de présenter modestement quelques étapes du parcours qui mènera (on espère) à une solution générale du problème de l’intelligence artificielle. Cette thèse contient quatre articles qui présentent chacun une différente nouvelle méthode d’inférence perceptive en utilisant l’apprentissage machine et, plus particulièrement, les réseaux neuronaux profonds. Chacun de ces documents met en évidence l’utilité de sa méthode proposée dans le cadre d’une tâche de vision par ordinateur. Ces méthodes sont applicables dans un contexte plus général, et dans certains cas elles on tété appliquées ailleurs, mais ceci ne sera pas abordé dans le contexte de cette de thèse. Dans le premier article, nous présentons deux nouveaux algorithmes d’inférence variationelle pour le modèle génératif d’images appelé codage parcimonieux “spike- and-slab” (CPSS). Ces méthodes d’inférence plus rapides nous permettent d’utiliser des modèles CPSS de tailles beaucoup plus grandes qu’auparavant. Nous démontrons qu’elles sont meilleures pour extraire des détecteur de caractéristiques quand très peu d’exemples étiquetés sont disponibles pour l’entraînement. Partant d’un modèle CPSS, nous construisons ensuite une architecture profonde, la machine de Boltzmann profonde partiellement dirigée (MBP-PD). Ce modèle a été conçu de manière à simplifier d’entraînement des machines de Boltzmann profondes qui nécessitent normalement une phase de pré-entraînement glouton pour chaque couche. Ce problème est réglé dans une certaine mesure, mais le coût d’inférence dans le nouveau modèle est relativement trop élevé pour permettre de l’utiliser de manière pratique. Dans le deuxième article, nous revenons au problème d’entraînement joint de machines de Boltzmann profondes. Cette fois, au lieu de changer de famille de modèles, nous introduisons un nouveau critère d’entraînement qui donne naissance aux machines de Boltzmann profondes à multiples prédictions (MBP-MP). Les MBP-MP sont entraînables en une seule étape et ont un meilleur taux de succès en classification que les MBP classiques. Elles s’entraînent aussi avec des méthodes variationelles standard au lieu de nécessiter un classificateur discriminant pour obtenir un bon taux de succès en classification. Par contre, un des inconvénients de tels modèles est leur incapacité de générer deséchantillons, mais ceci n’est pas trop grave puisque la performance de classification des machines de Boltzmann profondes n’est plus une priorité étant donné les dernières avancées en apprentissage supervisé. Malgré cela, les MBP-MP demeurent intéressantes parce qu’elles sont capable d’accomplir certaines tâches que des modèles purement supervisés ne peuvent pas faire, telles que celle de classifier des données incomplètes ou encore celle de combler intelligemment l’information manquante dans ces données incomplètes. Le travail présenté dans cette thèse s’est déroulé au milieu d’une période de transformations importantes du domaine de l’apprentissage à réseaux neuronaux profonds qui a été déclenchée par la découverte de l’algorithme de “dropout” par Geoffrey Hinton. Dropout rend possible un entraînement purement supervisé d’architectures de propagation unidirectionnel sans être exposé au danger de sur- entraînement. Le troisième article présenté dans cette thèse introduit une nouvelle fonction d’activation spécialement con ̧cue pour aller avec l’algorithme de Dropout. Cette fonction d’activation, appelée maxout, permet l’utilisation de aggrégation multi-canal dans un contexte d’apprentissage purement supervisé. Nous démontrons comment plusieurs tâches de reconnaissance d’objets sont mieux accomplies par l’utilisation de maxout. Pour terminer, sont présentons un vrai cas d’utilisation dans l’industrie pour la transcription d’adresses de maisons à plusieurs chiffres. En combinant maxout avec une nouvelle sorte de couche de sortie pour des réseaux neuronaux de convolution, nous démontrons qu’il est possible d’atteindre un taux de succès comparable à celui des humains sur un ensemble de données coriace constitué de photos prises par les voitures de Google. Ce système a été déployé avec succès chez Google pour lire environ cent million d’adresses de maisons.

OPTIMAL UTILIZATION OF SERVICE FACILITY FOR A k-OUT-OF-n SYSTEM WITH REPAIR BY EXTENDING SERVICE TO EXTERNAL CUSTOMERS IN A RETRIAL QUEUE

In this paper, we study a k-out-of-n system with single server who provides service to external customers also. The system consists of two parts:(i) a main queue consisting of customers (failed components of the k-out-of-n system) and (ii) a pool (of finite capacity M) of external customers together with an orbit for external customers who find the pool full. An external customer who finds the pool full on arrival, joins the orbit with probability and with probability 1− leaves the system forever. An orbital customer, who finds the pool full, at an epoch of repeated attempt, returns to orbit with probability (< 1) and with probability 1 − leaves the system forever. We compute the steady state system size probability. Several performance measures are computed, numerical illustrations are provided.

Horizontal nutrient fluxes and production efficiencies in urban and peri-urban crop and livestock husbandry of Niamey, Niger

Urban and peri-urban agriculture (UPA) increasingly supplies food and non-food values to the rapidly growing West African cities. However, little is known about the resource use efficiencies in West African small-scale UPA crop and livestock production systems, and about the benefits that urban producers and retailers obtain from the cultivation and sale of UPA products. To contribute to filling this gap of knowledge, the studies comprising this doctoral thesis determined nutrient use efficiencies in representative urban crop and livestock production system in Niamey, Niger, and investigated potential health risks for consumers. Also assessed was the economic efficiency of urban farming activities. The field study, which was conducted during November 2005 to January 2008, quantified management-related horizontal nutrient flows in 10 vegetable gardens, 9 millet fields and 13 cattle and small ruminant production units. These farms, selected on the basis of a preceding study, represented the diversity of UPA crop and livestock production systems in Niamey. Based on the management intensity, the market orientation and especially the nutrient input to individual gardens and fields, these were categorized as high or low input systems. In the livestock study, high and low input cattle and small ruminant units were differentiated based on the amounts of total feed dry matter offered daily to the animals at the homestead. Additionally, economic returns to gardeners and market retailers cultivating and selling amaranth, lettuce, cabbage and tomato - four highly appreciated vegetables in Niamey were determined during a 6-months survey in forty gardens and five markets. For vegetable gardens and millet fields, significant differences in partial horizontal nutrient balances were determined for both management intensities. Per hectare, average annual partial balances for carbon (C), nitrogen (N), phosphorus (P) and potassium (K) amounted to 9936 kg C, 1133 kg N, 223 kg P and 312 kg K in high input vegetable gardens as opposed to 9580 kg C, 290 kg N, 125 kg P and 351 kg K in low input gardens. These surpluses were mainly explained by heavy use of mineral fertilizers and animal manure to which irrigation with nutrient rich wastewater added. In high input millet fields, annual surpluses of 259 kg C ha-1, 126 kg N ha-1, 20 kg P ha-1 and 0.4 kg K ha-1 were determined. Surpluses of 12 kg C ha-1, 17 kg N ha-1, and deficits of -3 kg P ha-1 and -3 kg K ha-1 were determined for low input millet fields. Here, carbon and nutrient inputs predominantly originated from livestock manure application through corralling of sheep, goats and cattle. In the livestock enterprises, N, P and K supplied by forages offered at the farm exceeded the animals’ requirements for maintenance and growth in high and low input sheep/goat as well as cattle units. The highest average growth rate determined in high input sheep/goat units was 104 g d-1 during the cool dry season, while a maximum average gain of 70 g d-1 was determined for low input sheep/goat units during the hot dry season. In low as well as in high input cattle units, animals lost weight during the hot dry season, and gained weight during the cool dry season. In all livestock units, conversion efficiencies for feeds offered at the homestead were rather poor, ranging from 13 to 42 kg dry matter (DM) per kg live weight gain (LWG) in cattle and from 16 to 43 kg DM kg-1 LWG in sheep/goats, pointing to a substantial waste of feeds and nutrients. The economic assessment of the production of four high value vegetables pointed to a low efficiency of N and P use in amaranth and lettuce production, causing low economic returns for these crops compared to tomato and cabbage to which inexpensive animal manure was applied. The net profit of market retailers depended on the type of vegetable marketed. In addition it depended on marketplace for amaranth and lettuce, and on season and marketplace for cabbage and tomato. Analysis of faecal pathogens in lettuce irrigated with river water and fertilized with animal manure indicated a substantial contamination by Salmonella spp. with 7.2 x 104 colony forming units (CFU) per 25 g of produce fresh matter, while counts of Escherichia coli averaged 3.9 x 104 CFU g-1. In lettuce irrigated with wastewater, Salmonella counts averaged 9.8 x 104 CFU 25 g-1 and E. coli counts were 0.6 x 104 CFU g-1; these values exceeded the tolerable contamination levels in vegetables of 10 CFU g-1 for E. coli and of 0 CFU 25 g-1 for Salmonella. Taken together, the results of this study indicate that Niamey’s UPA enterprises put environmental safety at risk since excess inputs of N, P and K to crop and livestock production units favour N volatilisation and groundwater pollution by nutrient leaching. However, more detailed studies are needed to corroborate these indications. Farmers’ revenues could be significantly increased if nutrient use efficiency in the different production (sub)systems was improved by better matching nutrient supply through fertilizers and feeds with the actual nutrient demands of plants and animals.

Pre-school contributions to future achievements

This paper provides recent evidence about the beneÖts of attending preschool on future performance. A non-parametric matching procedure is used over two outcomes: math and verbal scores at a national mandatory test (Saber11) in Colombia. It is found that students who had the chance of attending preschool obtain higher scores in math (6.7%) and verbal (5.4%) than those who did not. A considerable fraction of these gaps comes from the upper quintiles of studentís performance, suggesting that preschool matters when is done at high quality institutions. When we include the number of years at the preschool, the gap rises up to 12% in verbal and 17% in math.

¿Existe una trampa de pobreza en el sector rural en Colombia?

A diferencia de la generalidad de trabajos sobre la pobreza rural en Colombia, este estudio emplea un enfoque de activos para indagar acerca de los determinantes de la pobreza rural. En particular se examinan la existencia de no convexidades locales en el proceso de generación de ingresos, el grado de concentración de los hogares en ciertos rangos de acumulación de activos y la presencia de retornos marginales diferenciados a los activos. Con base en esto se proporciona evidencia prima facie acerca de la existencia de una trampa de pobreza en el sector rural, abriendo una línea promisoria de investigación sobre el tema, que puede contribuir de forma importante a su comprensión y a un mejor diseño de política social y sectorial.

Do Colombian students underestimate higher education returns?

We analyze the determinants of subjective returns of higher education in Colombia. The information on expectations has been collected in categories, motivating the use of interval regression and an ordered probit approaches for modeling the relationship between beliefs and measures of ability, conditioning on individual, school and regional covariates. The results suggest that there are considerable differences in the size of the expected returns according to some population groups and a strong dominance of college against technical education. Gender gaps disappear in college education but it is found that girls tend to believe that professional wages are more concentrated into higher income categories than boys. Finally, it seems that Colombian students overestimate the pecuniary returns to education.

Simulating regional scale secondary organic aerosol formation during the TORCH 2003 campaign in the southern UK

A photochemical trajectory model has been used to simulate the chemical evolution of air masses arriving at the TORCH field campaign site in the southern UK during late July and August 2003, a period which included a widespread and prolonged photochemical pollution episode. The model incorporates speciated emissions of 124 nonmethane anthropogenic VOC and three representative biogenic VOC, coupled with a comprehensive description of the chemistry of their degradation. A representation of the gas/aerosol absorptive partitioning of ca. 2000 oxygenated organic species generated in the Master Chemical Mechanism (MCM v3.1) has been implemented, allowing simulation of the contribution to organic aerosol (OA) made by semi- and non-volatile products of VOC oxidation; emissions of primary organic aerosol (POA) and elemental carbon (EC) are also represented. Simulations of total OA mass concentrations in nine case study events (optimised by comparison with observed hourly-mean mass loadings derived from aerosol mass spectrometry measurements) imply that the OA can be ascribed to three general sources: (i) POA emissions; (ii) a '' ubiquitous '' background concentration of 0.7 mu g m(-3); and (iii) gas-to-aerosol transfer of lower volatility products of VOC oxidation generated by the regional scale processing of emitted VOC, but with all partitioning coefficients increased by a species-independent factor of 500. The requirement to scale the partitioning coefficients, and the implied background concentration, are both indicative of the occurrence of chemical processes within the aerosol which allow the oxidised organic species to react by association and/or accretion reactions which generate even lower volatility products, leading to a persistent, non-volatile secondary organic aerosol (SOA). The contribution of secondary organic material to the simulated OA results in significant elevations in the simulated ratio of organic carbon (OC) to EC, compared with the ratio of 1.1 assigned to the emitted components. For the selected case study events, [OC]/[EC] is calculated to lie in the range 2.7-9.8, values which are comparable with the high end of the range reported in the literature.

An analysis of commercial real estate returns: is there a smoothing puzzle?

In this paper we investigate the commonly used autoregressive filter method of adjusting appraisal-based real estate returns to correct for the perceived biases induced in the appraisal process. Since the early work by Geltner (1989), many papers have been written on this topic but remarkably few have considered the relationship between smoothing at the individual property level and the amount of persistence in the aggregate appraised-based index. To investigate this issue in more detail we analyse a sample of individual property level appraisal data from the Investment Property Database (IPD). We find that commonly used unsmoothing estimates overstate the extent of smoothing that takes place at the individual property level. There is also strong support for an ARFIMA representation of appraisal returns.

An analysis of commercial real estate returns: an anatomy of smoothing in asset and index returns

In this article, we investigate the commonly used autoregressive filter method of adjusting appraisal-based real estate returns to correct for the perceived biases induced in the appraisal process. Many articles have been written on appraisal smoothing but remarkably few have considered the relationship between smoothing at the individual property level and the amount of persistence in the aggregate appraisal-based index. To investigate this issue we analyze a large sample of appraisal data at the individual property level from the Investment Property Databank. We find that commonly used unsmoothing estimates at the index level overstate the extent of smoothing that takes place at the individual property level. There is also strong support for an ARFIMA representation of appraisal returns at the index level and an ARMA model at the individual property level.

Steering across scales: applying Strategic-Relational Approach to a study of land mobilisation for road widening in Kochi

This paper argues for the relevance of paying attention to structuring participation processes across scales as one of the ways in which participation of multi-organisational partnerships that involve conflicting interests might be managed. Issue wise the paper deals with problems in connection with land mobilisation for road widening in complex and concentrated high value urban settings. It discusses a case study of plan implementation involving individual landowners, the land development market, the local government, other governmental and non-governmental organisations and the state government, which together achieved objectives that seemed impossible at first sight. In theoretical terms, the paper engages with Jessop's (2001) Strategic-Relational Approach (SRA), arguing for its potential for informing action in a way that is capable of achieving steering outputs. The claim for SRA is demonstrated by re-examining the case study. The factors that come through as SRA is applied are drawn out and it is suggested that the theory though non-deterministic, helps guide action by highlighting certain dynamics of systems that can be used for institutional intervention. These dynamics point to the importance of paying attention to scale and the way in which participation and negotiation processes are structured so as to favour certain outcomes rather than others

Required returns for real estate in UK pension fund portfolios

The position of Real Estate within a multi-asset portfolio has received considerable attention recently. Previous research has concentrated on the percentage holding property would achieve given its risk/return characteristics. Such studies have invariably used Modern Portfolio Theory and these approaches have been criticised for both the quality of the real estate data and problems with the methodology itself. The first problem is now well understood, and the second can be addressed by the use of realistic constraints on asset holdings. This paper takes a different approach. We determine the level of return that Real Estate needs to achieve to justify an allocation within the multi asset portfolio. In order to test the importance of the quality of the data we use historic appraisal based and desmoothed returns to examine the sensitivity of the results. Consideration is also given to the Holding period and the imposition of realistic constraints on the asset holdings in order to model portfolios held by pension fund investors. We conclude, using several benchmark levels of portfolio risk and return, that using appraisal based data the required level of return for Real Estate was less than that achieved over the period 1972-1993. The use of desmoothed series can reverse this result at the highest levels of desmoothing although within a restricted holding period Real Estate offered returns in excess of those required to enter the portfolio and might have a role to play in the multi-asset portfolio.