In this thesis we investigate some global desiderata for probabilistic knowledge merging given several possibly jointly inconsistent, but individually consistent knowledge bases. We show that the most naive methods of merging, which combine applications of a single expert inference process with the application of a pooling operator, fail to satisfy certain basic consistency principles. We therefore adopt a different approach. Following recent developments in machine learning where Bregman divergences appear to be powerful, we define several probabilistic merging operators which minimise the joint divergence between merged knowledge and given knowledge bases. In particular we prove that in many cases the result of applying such operators coincides with the sets of fixed points of averaging projective procedures - procedures which combine knowledge updating with pooling operators of decision theory. We develop relevant results concerning the geometry of Bregman divergences and prove new theorems in this field. We show that this geometry connects nicely with some desirable principles which have arisen in the epistemology of merging. In particular, we prove that the merging operators which we define by means of convex Bregman divergences satisfy analogues of the principles of merging due to Konieczny and Pino-Perez. Additionally, we investigate how such merging operators behave with respect to principles concerning irrelevant information, independence and relativisation which have previously been intensively studied in case of single-expert probabilistic inference. Finally, we argue that two particular probabilistic merging operators which are based on Kullback-Leibler divergence, a special type of Bregman divergence, have overall the most appealing properties amongst merging operators hitherto considered. By investigating some iterative procedures we propose algorithms to practically compute them.

Donoghue's effective field theory of quantum gravity is extended to include the interaction of massless particles. The collinear divergences which accompany massless particles are examined first in the context of QED and then in quantum gravity. A result of Weinberg is extended to show how these divergences vanish in the case of gravity. The scattering cross section for hypothetical massless scalar particles is computed first, because it is simpler, and the results are then extended to photons. Some terms in the cross section are shown to correspond to the Aichelburg-Sexl metric surrounding a massless particle and to quantum corrections to that metric. The scattering cross section is also applied to calculate quantum corrections to the bending of starlight, and though small, the result obtained is qualitatively different than in the classical case. Since effective field theory includes the low-energy degrees of freedom which generate collinear divergences, the results presented here will remain relevant in any future quantum theory of gravity.

There has been a tremendous growth in the use of Graphics Processing Units (GPU) for the acceleration of general purpose applications. The growth is primarily due to the huge computing power offered by the GPUs and the emergence of programming languages such as CUDA and OpenCL. A typical GPU consists of several 100s to a few 1000s of Single Instruction Multiple Data (SIMD) cores, organized as 10s of Streaming Multiprocessors (SMs), each having several SIMD cores which operate in a lock-step manner, o ering a few TeraFLOPS of performance in a single socket. SMs execute instructions from a group of consecutive threads, called warps. At each cycle, an SM schedules a warp from a group of active warps and can context switch among the active warps to hide various stalls. However, various factors, such as global memory latency, divergence among warps of a thread block (TB), branch divergence among threads of a warp (Control Divergence), number of active warps, etc., can significantly impact the ability of a warp scheduler to hide stalls. This reduces the speedup of applications running on the GPU. Further, applications containing loops with potential cross iteration dependences, do not utilize the available resources (SIMD cores) effectively and hence su er in terms of performance. In this thesis, we propose several mechanisms which address the above issues and enhance the performance of GPU applications through efficient warp scheduling, taming branch and warp divergence, and runtime parallelization. First, we propose RLWS, a Reinforcement Learning (RL) based Warp Scheduler which uses unsupervised learning to schedule warps based on the current state of the core and the long-term benefits of scheduling actions. As the design space involving the state variables used by the RL and the RL parameters (such as learning and exploration rates, reward and penalty values, etc.) is large, we use a Genetic Algorithm to identify the useful subset of state variables and RL parameter values. We evaluated the proposed RL based scheduler using the GPGPU-SIM simulator on a large number of applications from the Rodinia, Parboil, CUDA-SDK and GPGPU-SIM benchmark suites. Our RL based implementation achieved an average speedup of 1.06x over the Loose Round Robin (LRR) strategy and 1.07x over the Two-Level (TL) strategy. A salient feature of RLWS is that it is robust, i.e., performs nearly as well as the best performing warp scheduler, consistently across a wide range of applications. Using the insights obtained from RLWS, we designed PRO, a heuristic warp scheduler which in addition to hiding the long latencies of certain operations, reduces the waiting time of warps at synchronization points. Evaluation of the proposed algorithm using the GPGPU-SIM simulator on a diverse set of applications showed an average speedup of 1.07x over the LRR warp scheduler and 1.08x over the TL warp scheduler. In the second part of the thesis, we address problems due to warp and branch divergences. First, many GPU kernels exhibit warp divergence due to various reasons such as, different amounts of work, cache misses, and thread divergence. Also, we observed that some kernels contain code which is redundant across TBs, i.e., all TBs will execute the code identically and hence compute the same results. To improve performance of such kernels, we propose a solution based on the concept of virtual TBs and loop independent code motion. We propose necessary code transformations which enable one virtual TB to execute the kernel code for multiple real TBs. We evaluated this technique using the GPGPU-SIM simulator on a diverse set of applications and observed an average improvement of 1.08x over the LRR and 1.04x over the Greedy Then Old (GTO) warp scheduling algorithms. Second, branch divergence causes execution of diverging branches to be serialized to execute only one control ow path at a time. Existing stack based hardware mechanism to reconverge threads causes duplicate execution of code for unstructured control ow graphs (CFG). We propose a simple and elegant transformation to convert an unstructured CFG to a structured CFG. The transformation eliminates duplicate execution of user code while incurring only a linear increase in the number of basic blocks and also the number of instructions. We implemented the proposed transformation at the PTX level using the Ocelot compiler infrastructure and demonstrate that the pro-posed technique is effective in handling the performance problem due to divergence in unstructured CFGs. Our third proposal is to enable efficient execution of loops with indirect memory accesses that can potentially cause cross iteration dependences. Such dependences are hard to detect using existing compilation techniques. We present an algorithm to compute at run-time, the cross iteration dependences in such loops, using both the CPU and the GPU. It effectively uses the compute capabilities of the GPU to collect the memory accesses performed by the iterations. Using the dependence information, the loop iterations are levelized such that each level contains independent iterations which can be executed in parallel. Experimental evaluation on real hardware (NVIDIA GPUs) reveals that the proposed technique can achieve an average speedup of 6.4x on loops with a reasonable number of cross iteration dependences.

This paper develops a spatial theoretical growth model in order to study the impact of physical and human capital relocations on the growth of open economies. Analytical and simulation results show how the respective neighbours determine an economy's development, why convergence and divergence may alternate in the medium-run, and that interregional migration as a consequence of wage inequalities causes disparities to prevail in the long-run. The empirical part applies spatial econometric specifications for European regions on the NUTS2 level for the observation period 2000-2010. The estimations underline the importance of human capital endowments and its relation with spatial location. (author's abstract) / Series: Working Papers in Regional Science

Le présent mémoire porte sur deux problématiques distinctes qui s'inscrivent toutefois chacune dans le domaine de l'analyse des dépenses sociales des provinces canadiennes. La première, d'une approche essentiellement descriptive et comparée, cherche à vérifier si le Québec se distingue du rest of Canada (ROC) dans ses dépenses sociales pour la période allant de 1961 à 2008, alors que la seconde, de type plus explicatif, vise à évaluer l'influence de l'idéologie des partis politiques au pouvoir sur l'évolution des dépenses sociales provinciales pour la même période. Les résultats du chapitre consacré à la première problématique laissent tout d'abord croire que le Québec se démarque du ROC à partir du milieu des années 1970 jusqu'à la fin de la période observée par des dépenses en proportion de son PIB plus importantes dans la majorité des secteurs sociaux. Or, malgré cette claire distinction québécoise, la plupart des domaines de dépenses de la province francophone semblent tout de même suivre des tendances sensiblement similaires à celles du reste du Canada pour l'ensemble de la période, ce qui témoigne alors d'une appartenance du Québec à un certain pattern pancanadien d'évolution des dépenses sociales. Les résultats du chapitre portant sur la seconde problématique paraissent quant à eux confirmer l'existence d'un cycle partisan provincial global aux effets toutefois limités, l'alternance gauche/droite au pouvoir ayant vraisemblablement un impact modeste sur l'évolution des dépenses sociales provinciales en général. La modestie de ce cycle partisan"pancanadien" semble par ailleurs attribuable à l'existence de divergences majeures entre les provinces par rapport à ce cycle, le facteur idéologique ayant un effet considérable sur l'évolution des dépenses sociales de six d'entre elles mais étant pratiquement nul sur l'évolution des dépenses des quatre autres provinces.

This thesis investigates whether developments associated with globalization and regional integration have caused the levels of government support provided to agricultural producers in Canada and the United States to converge in a downward direction. The literature is sharply divided as to whether governments retain the ability to pursue an independent agricultural policy course. To shed light on this debate, the levels of government assistance payments made to farmers in six contiguous Canadian provinces and American states (Manitoba, Saskatchewan, Alberta, North Dakota, South Dakota, and Montana) are compared over the 1990-2001 period. This time-frame allows for sufficient periods both before and after the establishment of NAFTA and the WTO to study the effects of these developments on the relevant policy outcomes. After outlining the programs and policy changes that drove the shifts in levels of government support provided to farmers, the paper argues that although the levels of government payments made to farmers in the six sub-units converged in the mid-1990s, they diverged thereafter. The evidence drawn from this examination supports the contention that governments do possess considerable room to manoeuvre in the agricultural policy making arena and significant ability to chart an independent policy course.

In this thesis, I use an experimental model to investigate the role of social pressures in stimulating language divergence. Research into the evolution of cooperation has emphasised the usefulness of ingroup markers for swiftly identifying outsiders, who pose a threat to cooperative networks. Mechanisms for avoiding cheats and freeriders, which tend to rely on reputation, or on (explicit and implicit) contracts between individuals, are considerably less effective against short-term visitors. Outsiders, moreover, may behave according to different social norms, which may adversely affect cooperative interactions with them. There are many sources of markers by which insiders and outsiders can be distinguished, but language is a particularly impressive one. If human beings exploit linguistic variation for this purpose, we might expect the exploitation to have an influence on the cultural evolution of language, and to be involved in language divergence, since it introduces a selective pressure, by which linguistic variants are selected on the basis of their social significance. However, there is also a neutral, mechanistic model of dialect formation that relies on unconscious accommodation between interlocutors, coupled with variation in the frequency of interaction, to account for divergence. In studies of real-world communities, these factors are difficult to tease apart. The model described in this thesis put real speakers in the artificial environment of a computer game. A game consisted of a series of rounds in which players were paired up with each other in a pseudo-random order. During a round, pairs of players exchanged typed messages in a highly restricted artificial "alien language". Each player began the game with a certain number of points, distributed between various resources, and the purpose of sending messages was to negotiate to exchange these resources. Any points given away were worth double to the receiver, so, by exchanging resources, players could accumulate points for their team. However, the pairings were anonymous: until the end of a round, players were not told who they had been paired with. This basic paradigm allowed the investigation of the major factors influencing language divergence, as well as the small-scale individual strategies that contribute to it. Two major factors were manipulated: frequency of interaction and competitiveness. In one condition, all players in a game were working together; in another condition, players were put into teams, such that giving away resources to teammates was advantageous, but giving them to opponents was not. This put a pressure on players to use variation in the alien language to mark identity. A combination of this pressure and a minimum level of interaction between teammates was found to be sufficient for the alien language to diverge into "dialects". Neither factor was sufficient on its own. The results of these experiments suggest that a pressure for the socially based selection of linguistic variants can lead to divergence in a very short time, given sufficient levels of interaction between members of a group.

The non-metric postcranial traits characterize biological variability of human skeletal morphology. Special attention is beeing paid to changes in some muscular or ligaments' binding areas and to changes of the articulation facets which are considered to be possibly retaled to excessive physical stress, occuring through the lifetime. The osteological material belonging to the populations of the Corded Ware culture and the Bell Beaker culture in the area of Bohemia has been surveyed with respect to non-metric trait occurence. The origin of both of these culture bearers is still a discused issue. There is a hypothesis saying that bearers of both Late Eneotithic cultures might represent actually one population and the differency of cultural patterns were due only to life-style changes, not to large-area migration. With respect to the frequency of 94 postcranial non-metric traits both samples have been compared to each other on a basis of measure of divergence and mean measure of divergence. The samples of Late Eneolithic populations showed significant difference only in frequencies of two non- metric traits and along with the values of mean measure of divergence they seem to show rather homogeneity. Afterwards both samples were compared with a sample from Great Moravian burial site Mikulčice - Kostelisko and...

Informações sobre variabilidade intrapopulacional de vetores biológicos são críticas para o entendimento da transmissão de agentes infecciosos veiculados por esses organismos. Nesse sentido, o presente trabalho caracterizou a variabilidade de fragmento que codifica a subunidade 4 do gene mitocondrial da Nicotinamina Adenina Dinucleotideo Desidrogenase - NADH4 em populações de Aedes aegypti do Brasil e de outros países. Polimorfismos de nucleotídeos únicos foram detectados através da técnica de seqüenciamento genômico. As análises realizadas compreenderam a de variância molecular (AMOVA), clados agrupados (NCA), distribuição de diferenças pareadas. Paralelamente foram examinadas relações evolutivas entre os haplótipos com o emprego dos critérios de parcimônia máxima e verossimilhança máxima. Os resultados mostram que há polimorfismo do fragmento nas populações analisadas, levando à proposição de dois clados geneticamente independentes (monofiléticos). Inferências de caráter histórico suportam a hipótese de que um dos clados inclui seqüências de indivíduos de populações que chegaram às Américas durante os séculos XVII e XVIII pelo tráfico negreiro, e outro, formado por populações introduzidas mais recentemente, se originou de populações asiáticas. Possíveis implicações epidemiológicas da variabilidade genética apresentada pelas populações do Ae. aegypti são também discutidas. / Knowledge about intrapopulacional variation of biological vectors is critical for understanding the dynamics of the transmission of an infectious agent. The major objective of the present study is to characterize the variability of a gene fragment, which codes for the subunit 4 of the mitochondrial gene of the Nicotinamide Adenine Dinucleotide Dehydrogenase - NADH4 among Aedes aegypti populations from Brazil comparing with that of several other countries. Single nucleotide polymorphism was detected employing genomic sequencing techniques. Nucleotide sequences were analyzed using molecular variance (AMOVA), nested clade (NCA) and mismatch distribution methods. Additionally, evolutionary relationships among haplotypes were estimated employing maximum parsimony and maximum likelihood criterions. The results show that the fragment of the mitochondrial gene (NADH4) is polymorphic, and that the populations of Ae. aegypti from Brazil are grouped into two genetically distinct, monophyletic clades Historical inferences support the hypothesis that one clade includes sequences from individuals that may be introduced in the Americas in the 17th and 18th centuries during the slave trade from Africa to America. The second clade consists of sequences of individuals that may be introduced in Brazil more recent, probably from Asian populations. Epidemiological consequences because of the genetic variability among populations of Ae. aegypti in Brazil are discussed.

L'intérêt des recherches dans le domaine de la fusion de données multi-capteurs est en plein essor en raison de la diversité de ses secteurs d'applications. Plus particulièrement, dans le domaine de la robotique et de la localisation, l'exploitation des différentes informations fournies par les capteurs constitue une étape primordiale afin d'assurer une estimation fiable de la position. Dans ce contexte de fusion de données multi-capteurs, nous nous attachons à traiter le diagnostic, menant à l'identification de la cause d'une défaillance, et la tolérance de l'approche proposée aux défauts de capteurs, peu abordés dans la littérature.Nous avons fait le choix de développer une approche basée sur un formalisme purement informationnel : filtre informationnel d'une part, et outils de la théorie de l'information d'autre part. Des résidus basés sur la divergence de Kullback-Leibler sont développés. Via des méthodes optimisées de seuillage, ces résidus conduisent à la détection et à l'exclusion de ces défauts capteurs. La théorie proposée est éprouvée sur deux applications de localisation. La première application concerne la localisation collaborative, tolérante aux défauts d'un système multi-robots. La seconde application traite de la localisation en milieu ouvert utilisant un couplage serré GNSS/odométrie tolérant aux défauts. / The interest of research in the multi-sensor data fusion field is growing because of its various applications sectors. Particularly, in the field of robotics and localization, the use of different sensors informations is a vital step to ensure a reliable position estimation. In this context of multi-sensor data fusion, we consider the diagnosis, leading to the identification of the cause of a failure, and the sensors faults tolerance aspect, discussed in limited work in the literature. We chose to develop an approach based on a purely informational formalism: information filter on the one hand and tools of the information theory on the other. Residuals based on the Kullback-Leibler divergence are developed. These residuals allow to detect and to exclude the faulty sensors through optimized thresholding methods. This theory is tested in two applications. The first application is the fault tolerant collaborative localization of a multi-robot system. The second application is the localization in outdoor environments using a tightly coupled GNSS/odometer with a fault tolerant aspect.