M.Sc. (Computer Science) / This research study investigates multi-agent systems (MASs), artificial life concepts and machine learning, amongst other things, in answering the key research question: “How can a generic multi-agent system integrate with machine learning through artificial life principles?” In answering this question, this dissertation illustrates the design and development of a generic multi-agent, life simulation and learning software framework. This framework simplifies and enables the realisation of MASs in solving complex problems in multiple domains. Finally, this research presents a prototype solution as a proof of concept of the framework’s strengths and weaknesses. The research study illustrates the design of MASs utilising sound design principles, patterns and methodologies. Furthermore, this research explores the requirements for creating and integrating MASs with other technologies, as well as the possible pitfalls in creating such large-scale systems. In addressing the necessity of learning, several machine learning techniques are examined and reinforcement learning is identified as an ideal candidate for the proposed framework. In addition, by understanding the overall machine learning process, the proposed framework integrates machine learning as three separate processes: data extraction, learning and inference. Lastly, the literature study focuses on artificial life, specifically its use in MASs, and defines what constitutes an intelligent system. This research depicts artificial life as a plausible natural integrator between MAS and machine learning technologies. The proposed framework presented in this dissertation consists of five core agent modules that can be extended, depending on the problem domain requirements. The framework in itself is self-containing and independent of any concrete implementation. A multi-agent antivirus system is presented as the prototype implementation of the proposed framework. A quantitative and qualitative analysis was conducted, identifying the results of the prototype and generic framework while highlighting strengths and weaknesses. The contribution of this research is found partly in the proposed generic framework as a means of augmenting mechanisms for MAS design and development by means of artificial life and machine learning integration. In a broader context, this research serves as a foundation towards creating advanced MAS frameworks, leading to numerous interesting and influential agent-oriented applications.

The objective of this research is to develop a simulation model as an aid in planning hospital operations. The hospital pharmacy is selected as an appropriate area for study. An extensive systems analysis of pharmacy functions is undertaken. A simulation model of the inpatient dispensary operations is developed using the IBM General Purpose Simulation System. This model allows experimentation with dispensary work-load, operations, and manpower schedule. Statistics are provided on service to the patient, work-load distribution, and manpower utilization. Variation in pharmacist availability and type of prescription entering the dispensary allows optimization of operations. Several experiments are conducted to illustrate the model concept and experiment possibilities. It is concluded that the model is a valuable planning tool for the hospital pharmacy administrator and can be extended to simulate operations in other areas of the pharmacy. / Business, Sauder School of / Graduate

A computer simulation model was written to mimic the natural movement of salmon fry in Babine Lake, B.C.. Simulated distributions of fry were compared with field observations taken in 3 sampling periods during the summer and fall in 1967, 1968, 1971, and 1972, in order to evaluate the model's validity. Simulated distributions of fry, when random and heavily-biased movements were combined, were in reasonable accord with naturally observed distributions of fry in periods 1 and 2. In period 3 the model successfully produced a distribution similar to that naturally observed when the fry were programmed to undergo only random movement. Factors which might account for the various distributions of fry in different periods include the effects of current and innate behavioral responses of the fry to limnological conditions. In constructing my model, I assumed that fry travelled at speeds observed in the laboratory in still water. The model of fry dispersal in Babine Lake could be improved as more information is collected on the limnology of the lake and on fry behavior. In addition, the model has much generality and the techniques used may be applied to the dispersal of other organisms and to other lakes. / Science, Faculty of / Zoology, Department of / Graduate

Orientador: Alexandre Campane Vidal / Dissertação (mestrado) - Universidade Estadual de Campinas, Instituto de Geociencias / Made available in DSpace on 2018-08-14T10:01:25Z (GMT). No. of bitstreams: 1 Sewaybricker_VictorVanin_M.pdf: 3831085 bytes, checksum: 8cfa5eb23feafe89cee1f862bf90cb57 (MD5) Previous issue date: 2009 / Resumo: A área de estudos tem aproximadamente 10300 m2 e substrato caracterizado por aterro argilo-arenoso com presença de entulhos sobreposto a sedimentos não litificados, ora mais argilosos, ora predominantemente arenosos. Foi identificado aquífero livre com fluxo de sul para norte, governado por gradiente hidráulico de 1%. O trabalho realizado busca aplicar método geoestatístico probabilístico para a avaliação do impacto das heterogeneidades de condutividades hidráulicas verificadas na área, por meio da simulação de cenários equiprováveis em simulador numérico de fluxo e de trajetória de partículas. Dados de campo foram obtidos de registros de sondagens conduzidas para instalação de 15 poços de monitoramento e de 15 poços de remediação e foram fornecidos pela empresa GEOKLOCK Engenharia e Consultoria Ambiental Ltda. Tais informações nortearam a construção de modelos equiprováveis de distribuição de condutividades hidráulicas no programa geoestatístico SGEMS. Os modelos geoestatísticos alimentaram o simulador de fluxo VISUAL MODFLOW com propriedades definidas célula a célula. Foram procedidas simulações de trajetórias de partículas com o módulo MODPATH e computadas as respostas para tempos mínimo e máximo de trajetória. Os resultados obtidos para as duas categorias avaliadas evidenciaram a influência dos diferentes cenários equiprováveis adotados, demonstrando a significativa incerteza que há na modelagem de fluxo em aqüíferos porosos heterogêneos. Desse modo, mostrou-se que o método estudado é ferramenta aplicável para a análise da variabilidade na resposta de simuladores, podendo inclusive, ser empregado em modelos mais complexos e considerando-se outras variáveis que não somente a condutividade hidráulica. / Abstract: The studied area has approximately 10300 m2 and its underground is characterized by a sandy-clay landfill that overlaps unconsolidates sediments composed by portions of clay predominance and by parts with highest sand contents. A free aquifer flowing from south to north ruled by a hydraulic gradient of 1% has been identified. This research seeks to apply a probabilistic statistical method for assessing the impact of hydraulic conductivities heterogeneities, through the simulation of different feasible hydraulic conductivities scenarios in numerical simulation of flow and particles pathways. Hard data were provided by the company GEOKLOCK Engenharia e Consultoria Ambiental Ltda and consisted in field records of boreholes conducted for installation of 15 monitoring wells and 15 remediation wells. This information guided the construction of hydraulic conductivities distribution models using the statistical software SGEMS. The statistical models results provided the data for the flow simulations which were performed with the software VISUAL MODFLOW. Particles pathways simulations were proceeded using the software MODPATH. The results for minimum and maximum particle traveling times were recorded and showed the different adopted scenarios influence, demonstrating that there is significant uncertainty regarding the modeling of flow in heterogeneous porous aquifers. Thus, it was verified that the studied method is an applicable tool for the analysis of variability in the response of simulators, and can even be used in more complex models considering other variables than hydraulic conductivity. / Mestrado / Geologia e Recursos Naturais / Mestre em Geociências

An ever-changing healthcare landscape requires today’s nurses to have a solid foundation in knowledge and clinical judgment to provide safe care to patients. Nurse educators must implement teaching strategies that help develop the knowledge and clinical judgment that nursing students will need upon graduation and entry into healthcare. Simulation-based experiences have been shown to help develop clinical judgment when used as part of a clinical practicum. However, few studies have examined the effectiveness of simulation-based experiences as a classroom teaching strategy. A quasi-experimental study was conducted to examine knowledge acquisition, clinical judgment, and general self-efficacy in undergraduate nursing students who participated in simulation-based case studies as a classroom teaching strategy versus those students who attended a traditional lecture. Students in the intervention group rotated through four simulation-based case study stations. Results indicated that there was not a significant difference in knowledge, clinical judgment, or general self-efficacy found between nursing students participating in simulation-based case studies versus those attending a traditional lecture. Additionally, relationships between demographic characteristics and clinical judgment scores in undergraduate nursing students were explored. There were no statistically significant relationships found between demographic characteristics and clinical judgment in this sample. Further analysis indicated that both teaching strategies are effective in promoting knowledge acquisition, clinical judgment, and general self-efficacy. The findings of this study demonstrate that both participation in simulation-based case studies and attending a traditional lecture are effective classroom teaching strategies in promoting knowledge acquisition, clinical judgment, and general self-efficacy in nursing students. Nurse educators are encouraged to continue to explore simulation-based experiences as a teaching strategy in the classroom.

Meta-analysis refers to the quantitative synthesis of information across different studies. Since outcomes from different studies are likely to be reported in different units, study-level results are typically transformed to the same scale before quantitative integration. Typically, this leads to the accumulation and combination of effect sizes. To date, most social scientists have synthesized, or meta-analyzed, zero-order statistics like a correlation. Synthesizing partial effect sizes is an alternative which allows a meta-analysis to account for the influence of nuisance variables when estimating the association between two variables. This dissertation proposes that logistic regression coefficients from different studies, which are a type of partial effect size, can be meta-analyzed. Logistic regression models how a set of covariates relates to a binary dependent variable. Given a key independent variable (IV) of interest, which we can call the focal IV or Xf, the slope estimate (βf) in a logistic regression measures the impact of Xf on Y on the logit (log-odds) scale, while controlling for other variables. Four assumptions justify the possibility of comparing and possibly combing logistic slopes across studies: (1) Y must be on the same scale, (2) Xf must be on the same scale, (3) all effect sizes are logistic regression slopes adjusted for the same covariates, and (4) model specifications are identical. In practice, the third assumption is particularly challenging as different studies inevitably include different sets of control variables. Three simulation studies are implemented to understand how synthesizing a logistic regression slope on the logit scale is affected by several factors. Across these three simulation studies, the following meta-analytic variables are tested: (1) the size of the partial effect size (βf), (2) Study-level sample size (k), (3) Within-study sample size (N), (4) the degree of between-study variance, (5) a continuous vs. a binary focal predictor, (6) the level of collinearity between Xf and other covariates included in primary studies, (7) the magnitude of non-focal variable slopes, (8) different covariate sets used in primary-level studies, and (9) meta-analytical method. Simulation performance is based on how the bias and mean-squared error (MSE) are affected by each of these simulation parameters. Overall, results suggest that when the four assumptions introduced above are satisfied, meta-analyzing logistic regression slopes is remarkably accurate as the summary effect resulting from the standard random-effects meta-analytic model leads to small levels of bias and MSE under a variety of conditions. When the assumptions are broken (and particularly the third assumption of identical covariate sets), the pooled slope estimator can have large degrees of bias. The bias is a function of within-study sample size, between-study sample size, distribution of the focal IV (i.e., continuous vs. categorical variable), multicollinearity, the magnitude of non-focal variable slope parameters, diversity in covariate sets, and choice of meta-analytical methods. The MSE is a function of study-level sample size, within-study sample size, distribution of the focal IV (i.e., continuous vs. categorical variable), multicollinearity, the magnitude of non-focal variable slope parameters, diversity in covariate sets, and choice of meta-analytical methods. A complex four-way interaction is discovered between collinearity, the magnitude of non-focal variable slope parameters, diversity in covariate sets, and choice of meta-analytical methods. An applied example focusing on estimating the effects of albumin on mortality is also presented to complement the simulation results.

Polymer nanocomposites (PNCs) have found their way into our everyday lives in a long list of applications, including airplane parts and car tires. This is due to their unique properties of combining the strengths of their constituents – elasticity and stiffness – while mitigating their weaknesses – softness and brittleness. In the past few decades, they have generated more interest due to the discovery that the PNCs’ optical, electrical, and a host of other properties can be tuned for specific use by controlling the assembly and dispersion of nanoparticles (NPs) within the host polymer matrix. The grafting of some of the matrix chains onto the surface of the NPs not only improves NP miscibility but also grants an additional handle tocontrol the self-assembly of NPs. However, at present, there remains many open questions in the field of these novel PNCs. For instance, it is commonly believed that long enough matrix polymers of length P will spontaneously dewet a chemically identical polymer layer, comprised of sufficient chains of length N , end-grafted to a flat surface (”brush”). This entropically driven idea is frequently used to explain experiments in which 10-20 nm diameter polymer-grafted NPs are observed to phase separate from homopolymer matrices for P/N⪆4. At lower grafting densities, these entropic effects are also thought to underpin the self-assembly of grafted NPs into a diverse set of structures. To explore the validity of this picture, a two-pronged approach is used in this thesis, exploring such systems from both a single NP and a multi-NP point of view in order to find novel methods for understanding and controlling NP dispersion in polymers. In each of the chapters, we employ coarse-grained Molecular Dynamics (MD) simulations to understand the self-assembly and dispersion behavior in PNCs, with the experimental analog being primarily polystyrene (PS) grafted silica NPs in PS matrices. We start by investigating the entropic effects of P/N on the brush of a single grafted NP, taking advantage of an indirect umbrella sampling method (INDUS) to quantify matrix density fluctuations. This method essentially makes use of an external biasing potential to mimic the dewetting of the brush. We find for the first time that entropic P/N effects can be identified at the single NP level and is primarily surface driven. INDUS is later extended to two-body and many-body NP systems, to understand the role of NP surfactantcy in the self-assembly of grafted NPs and create free-energy profiles for a range of inter-NP separations. Finally, results from a comprehensive series of large-scale multi-NP simulations, where we consider NPs in the ≈ 5nm and ≈ 10nm size range. For the smaller NPs, we find no evidence of phase separation even for P/N = 10 in the absence of attractions. Instead, we discover that we are able to recreate most of the experimentally observed structures when allthe polymer chain monomers are equally attractive to each other but repel the NPs. Only when the NPs are in the ≈ 10nm size range that we are able to access the phase separated morphologies. Our results thus imply that experimental situations where the grafting density is low are dominated by the surfactancy of the NPs, which is driven by the chemical mismatch between the inorganic core and the organic ligands (the graft and free chains are chemically identical). Entropic effects, i.e. the translational entropy of the NPs and the matrix, the entropy of mixing of the grafts and the matrix, and the conformational entropy of the chains appear to thus play a second order effect even in the context of these model systems. Each of these insights provides details around controlling the organization and assembly of NPs in polymers for the purpose of improving their mechanical properties, all while changing the way in which the material is designed.

In this thesis we aim to provide computationally efficient methods of performing waterbarrier simulations. The innate challenge in simulations of structures such as sea or surge barriers is resolution. Because barriers tend to be long and thin compared to the surrounding landscapes they protect, one must put mesh refinement on the barrier region in order to even numerically recognize the barrier’s presence. This is a costly computation due to the CFL condition which puts a strict limit on the size of time step proportional to the spatial mesh size. Another issue is the complexity of meshing near the barrier. Since barriers are most likely slanted or have certain shapes, the grid has to reflect this in the form of a grid mapping or an unstructured grid. To mitigate the issue of resolution, we propose an approximation of the barrier with a line interface embedded on a Cartesian grid, reducing our problem to an embedded boundary problem. Then to avoid complex meshing, we develop three cut cell methods on two shapes of barriers: 1) the h-box method (HB), 2) the state redistribution method (SRD), and 3) the cell merging method (CM). Doing this two-step approach means that we can lower the resolution near the barrier region and still feel the presence of the barrier and capture its effect, which would otherwise not be the case if we relied on resolution for representation of the barrier. This does not mean that we are losing accuracy by lowering resolution, however. Rather, we are maintaining about the same accuracy while also lowering resolution (and thus cutting computational cost), which we show by comparison with a refined barrier. We solve the shallow water equations as our underlying PDEs to simulate water interaction with the barrier, as they are commonly used in tsunami and storm simulations. We implement our work on the PYCLAW framework, which is an objected oriented program that solves conservation laws.

Thesis (Ph. D.)--Massachusetts Institute of Technology, Dept. of Nuclear Engineering, 1982. / MICROFICHE COPY AVAILABLE IN ARCHIVES AND SCIENCE. / Bibliography: leaves 144-147. / by Benito De Celis. / Ph.D.

No description available.