Software systems supporting networked, transaction-oriented services are large and complex; they comprise a multitude of inter-dependent layers and components, and they implement many dynamic optimization mechanisms. In addition, these systems are subject to workload that is hard to predict. These factors make monitoring these systems as well as performing problem determination challenging and costly. In this thesis we tackle these challenges with the goal of lowering the cost and improving the effectiveness of monitoring and problem determination by reducing the dependence on human operators. Specifically, this thesis presents and demonstrates the effectiveness of an efficient, automated monitoring approach which enables detection of errors and failures, and which assists in localizing faults. Software systems expose various types of monitoring data; this thesis focuses on the use of management metrics to monitor a system's health. We devise a system modeling approach which entails modeling stable, statistical correlations among management metrics; these correlations characterize a system's normal behaviour This approach allows a system model to be built automatically and efficiently using the monitoring data alone. In order to control the monitoring overhead, and yet allow a system's health to be assessed reliably, we design an adaptive monitoring approach. This adaptive capability builds on the flexible nature of our system modeling approach, which allows the set of monitored metrics to be altered at runtime. We develop methods to automatically select management metrics to collect at the minimal monitoring level, without any domain knowledge. In addition, we devise an automated fault localization approach, which leverages the ability of the monitoring system to analyze individual metrics. Using a realistic, multi-tier software system, including different applications based on Java Enterprise Edition and industrial-strength products, we evaluate our system modeling approach. We show that stable metric correlations exist in complex software systems and that many of these correlations can be modeled using simple, efficient techniques. We investigate the effect of the collection of management metrics on system performance. We show that the monitoring overhead can be high and thus needs to be controlled. We employ fault injection experiments to evaluate the effectiveness of our adaptive monitoring and fault localization approach. We demonstrate that our approach is cost-effective, has high fault coverage and, in the majority of the cases studied, provides pertinent diagnosis information. The main contribution of this work is to show how to monitor complex software systems and determine problems in them automatically and efficiently. Our solution approach has wide applicability and the techniques we use are simple and yet effective. Our work suggests that the cost of monitoring software systems is not necessarily a function of their complexity, providing hope that the health of increasingly large and complex systems can be tracked with a limited amount of human resources and without sacrificing much system performance.

In order to control any system one needs to know the system's current state. In many real-world scenarios the state of the system cannot be determined with certainty due to the sensors being noisy or simply missing. In cases like these one needs to use probabilistic inference techniques to compute the likely states of the system and because such cases are common, there are lots of techniques to choose from in the field of Artificial Intelligence. Formally, we must compute a probability distribution function over all possible states. Doing this exactly is difficult because the number of states is exponential in the number of variables in the system and because the joint PDF may not have a closed form. Many approximation techniques have been developed over the years, but none ideally suited the problem we faced. Particle filtering is a popular scheme that approximates the joint PDF over the variables in the system by a set of weighted samples. It works even when the joint PDF has no closed form and the size of the sample can be adjusted to trade off accuracy for computation time. However, with many variables the size of the sample required for a good approximation can still become prohibitively large. Factored particle filtering uses the structure of variable dependencies to split the problem into many smaller subproblems and scales better if such decomposition is possible. However, our problem was unusual because some normally independent variables would become strongly correlated for short periods of time. This dynamically-changing dependency structure was not handled effectively by existing techniques. Considering variables to be always correlated meant the problem did not scale, considering them to be always independent introduced errors too large to tolerate. It was necessary to develop an approach that would utilize variables' independence whenever possible, but not introduce large errors when variables become correlated. We have developed a new technique for monitoring the state of the system for a class of systems with context-specific correlations. It is based on the idea of caching the context in which correlations arise and otherwise keeping the variables independent. Our evaluation shows that our technique outperforms existing techniques and is the first viable solution for the class of problems we consider.

Software systems supporting networked, transaction-oriented services are large and complex; they comprise a multitude of inter-dependent layers and components, and they implement many dynamic optimization mechanisms. In addition, these systems are subject to workload that is hard to predict. These factors make monitoring these systems as well as performing problem determination challenging and costly. In this thesis we tackle these challenges with the goal of lowering the cost and improving the effectiveness of monitoring and problem determination by reducing the dependence on human operators. Specifically, this thesis presents and demonstrates the effectiveness of an efficient, automated monitoring approach which enables detection of errors and failures, and which assists in localizing faults. Software systems expose various types of monitoring data; this thesis focuses on the use of management metrics to monitor a system's health. We devise a system modeling approach which entails modeling stable, statistical correlations among management metrics; these correlations characterize a system's normal behaviour This approach allows a system model to be built automatically and efficiently using the monitoring data alone. In order to control the monitoring overhead, and yet allow a system's health to be assessed reliably, we design an adaptive monitoring approach. This adaptive capability builds on the flexible nature of our system modeling approach, which allows the set of monitored metrics to be altered at runtime. We develop methods to automatically select management metrics to collect at the minimal monitoring level, without any domain knowledge. In addition, we devise an automated fault localization approach, which leverages the ability of the monitoring system to analyze individual metrics. Using a realistic, multi-tier software system, including different applications based on Java Enterprise Edition and industrial-strength products, we evaluate our system modeling approach. We show that stable metric correlations exist in complex software systems and that many of these correlations can be modeled using simple, efficient techniques. We investigate the effect of the collection of management metrics on system performance. We show that the monitoring overhead can be high and thus needs to be controlled. We employ fault injection experiments to evaluate the effectiveness of our adaptive monitoring and fault localization approach. We demonstrate that our approach is cost-effective, has high fault coverage and, in the majority of the cases studied, provides pertinent diagnosis information. The main contribution of this work is to show how to monitor complex software systems and determine problems in them automatically and efficiently. Our solution approach has wide applicability and the techniques we use are simple and yet effective. Our work suggests that the cost of monitoring software systems is not necessarily a function of their complexity, providing hope that the health of increasingly large and complex systems can be tracked with a limited amount of human resources and without sacrificing much system performance.

The conductivity of acid-etched fractures highly depends on spaces along the fracture created by uneven etching of the fracture walls remaining open after fracture closure. Formation heterogeneities such as variations of mineralogy and permeability result in channels that contribute significantly to the fracture conductivity. Current numerical simulators or empirical correlations do not account for this channeling characteristic because of the scale limitations. The purpose of this study is to develop new correlations for conductivity of acid-etched fracturing at the intermediate scale. The new correlations close the gap between laboratory scale measurements and macro scale acid fracture models. Beginning with acid-etched fracture width profiles and conductivity at zero closure stress obtained by the previous work, I modeled the deformation of the fracture surfaces as closure stress is applied to the fracture. At any cross-section along the fracture, I approximated the fracture shape as being a series of elliptical openings. With the assumption of elastic behavior for the rock, the numerical simulation presents how many elliptical openings remain open and their sizes as a function of the applied stress. The sections of the fracture that are closed are assigned a conductivity because of small-scale roughness features using a correlation obtained from laboratory measurements of acid fracture conductivity as a function of closure stress. The overall conductivity of the fracture is then obtained by numerically modeling the flow through this heterogeneous system. The statistical parameters of permeability distribution and the mineralogy distribution, and Young’s modulus are the primary aspects that affect the overall conductivity in acid-etched fracturing. A large number of deep, narrow channels through the entire fracture leads to high conductivity when the rock is strong enough to resist closure stress effectively. Based on extensive numerical experiments, I developed the new correlations in three categories to predict the fracture conductivity after closure. Essentially, they are the exponential functions that incorporate the influential parameters. Combined with the correlations for conductivity at zero closure stress from previous work, the new correlations are applicable to a wide range of situations.

Thesis (Ph. D.)--University of Washington, 2003. / Vita. Includes bibliographical references (leaves 86-90).

Objective of the research - to evaluate, using up-to-date statistical–genetic methods, the reproductive characteristics of pig breeds bred in Lithuania, to determine correlation of the characteristics with productivity traits, and to develop an optimised system of pigs genetic evaluation by BLUP method. Tasks of the research was: to determine influence of genetic and non-genetic factors in pigs reproductive characteristics, to evaluate the additive-genetic heritability parameters, and co-response of reproduction traits; to evaluate influence of reproductive characteristics on productivity traits, phenotype and genetic co-response; to develop an optimised pigs genetic evaluation system employing BLUP method, estimating pigs reproductive and productive characteristics, using the integrated multivariate model; to evaluate tendencies of pigs genetic improvement. Novelty of the research: using the method of unifactor and multifactor dispersion analysis, leverage of genetic and non-genetic factors on reproductive characteristics of pigs, breed in Lithuania, was determined; heritability parameters of reproductive characteristics were determined, using modern software; genetic and phenotype co-response of the reproductive characteristics was estimated; genetic correlation between reproductive characteristics and productivity traits was evaluated, using statistical-genetic methods, for the first time in Lithuania; optimised multivariate model for determination of reproductive and... [to full text]

Transactive Discourse During Assessment Conversations on Science Learning by Homer A. Russell III It has been argued that development of science knowledge is the result of social interaction and adoption of shared understandings between teachers and students. A part of understanding that process is determining how student reasoning develops in groups. Transactive discussion is a form of negotiation between group members as they interpret the meaning of their logical statements about a topic. More importantly, it is a form of discourse that often leads to cognitive change as a result of the interaction between group participants as they wrestle with their different perspectives in order to achieve a common understanding. The research reported here was a correlational study designed to investigate the relationship between the various forms of transactive discussion and learning outcome performance seen in an investigation involving 24 students in a middle-SES high school located in southwest Atlanta, Georgia. Pretest and posttest measures of genetics reasoning, as well as curriculum content test data, were used in this study. Group discussion was captured on videotape and analyzed to determine whether transactional discussion was present and whether or not it had an effect on learning outcome measures. Results of this study showed that participant use of transactive discussion played a role in development of reasoning abilities in the area of genetics. It is suggested that teachers should monitor classroom discourse for the presence of transactive discussion as such discourse plays a role in fostering performance outcomes.

In this work, we study the statistical properties of deterministic and stochastic dynamical systems. We are particularly interested in extreme values and recurrence. We prove the existence of Extreme Value Laws (EVLs) and Hitting Time Statistics (HTS)/ ReturnTime Statistics (RTS) for systems with decay of correlations against L1 observables. We also carry out the study of the convergence of Rare Event Point Processes (REPP). In the first part, we investigate the problem for deterministic dynamics and completely characterise the extremal behaviour of expanding systems by giving a dichotomy relying on the existence of an Extremal Index (EI). Namely, we show that the EI is strictly less than 1 for periodic centres and is equal to 1 for non-periodic ones. In a more general setting, we prove that the REPP converges to a standard Poisson if the centre is non-periodic, and to a compound Poisson with a geometric multiplicity distribution for the periodic case. Moreover, we perform an analysis of the convergence of the REPP at discontinuity points which gives the convergence to a compound Poisson with a multiplicity distribution different than the usual geometric one.In the second part, we consider stochastic dynamics by randomly perturbing a deterministic system with additive noise. We present two complementary methods which allow us to obtain EVLs and statistics of recurrence in the presence of noise. The first approach is more probabilistically oriented while the second one uses spectral theory. We conclude that, regardless of the centre chosen, the EI is always equal to 1 and the REPP converges to the standard Poisson.

Cette these aborde diverses problematiques concernant les electrons fortement correles dans des systemes bidimensionnels (composes a frustration geometrique, phases a liens de valence resonants), decrits a l'aide de modeles de dimeres. Une partie de la these concerne des modeles classiques sur un reseau triangulaire anisotrope, presentant des phases critiques qu'on peut decrire a l'aide de theories conformes; en se basant sur ces theories, l'analyse numerique de ce modele par matrice de transfert a permis de caracteriser les conditions d'existence de la criticalite, et plus generalement le diagramme de phases du modele en fonction d'interactions a courte portee et de l'anisotropie du reseau.<br />Une autre partie de la these traite un systeme d'electrons sur un reseau pyrochlore bidimensionnel (ou damier) a remplissage commensurable, en interactions a courte portee. Dans une limite de fortes interactions les electrons subissent des contraintes qui se traduisent par un modele effectif de dimeres quantiques (se differentiant par rapport au modele dit de Rokhsar-Kivelson, motive par les etats a liens de valence resonants, par un degre de liberte supplementaire, de spin). Une etude par diagonalisation exacte, completee par une approche variationnelle et des arguments perturbatifs, a permis d'identifier une phase cristalline a singulets resonants; une extension de ce modele a une mobilite finie des electrons a ete consideree pour caracteriser la transition de cette phase isolante vers un etat metallique en fonction du rapport de la mobilite des electrons et de leurs interactions.

Taking into account the expected increase in global energy demands and increasing climate change issues, there is a pressing need to develop new environmentally sustainable energy systems. Nuclear energy will play a major role in being part of the energy mix since it offers a relatively clean, safe and reliable source of electrical energy. However, opportunities for building new generation nuclear systems will depend on their economic and safety attractiveness as well as their flexibility in design to adapt in different countries and situations. Keeping these objectives in mind, a framework for international cooperation was set forth in a charter of Generation IV International Forum (GIF) (GIF Charter, 2002) and six design concepts were selected for further development. To achieve high thermal efficiencies of up to 45 ??? 50%, the use of SuperCritical Fluids (SCFs) as working fluids in heat transfer cycles is proposed Generation IV designs. An important aspect towards development of SCF applications in novel Gen IV Nuclear Power Plant (NPP) designs is to understand the thermodynamic behavior and prediction of Heat Transfer Coefficients (HTCs) at supercritical (SC) conditions. In addition to the nuclear power industry applications; SCFs are also expected to play a vital role in a number of other important technologies such as refrigeration systems, and geothermal systems, to name a few. Given the potential for vast number of applications of SCFs in industry, the objective of this work was to gain an understanding on the behavior of SCFs and to develop a fundamental knowledge of the heat-transfer processes and correlations for SC Water and SC CO2 flowing in bare circular tubes. Experimental datasets for SC Water and SC CO2 were compiled and used to obtain a basic 1-D empirical correlation that can predict HTC in bare circular tubes during the transient phases. The accuracy of these correlations was also analyzed using statistical techniques. Limitations and applications for 1-D correlations are discussed as well. The new correlations showed promising results for HTC and Tw calculations for the reference dataset with uncertainty of about ??25% for HTC values and about ??10-15% for the calculated wall temperature.