<p>Proteins are involved in all of life's processes and are also responsible for many diseases. Thus, engineering proteins to perform new tasks could revolutionize many areas of biomedical research. One promising technique for protein engineering is computational structure-based protein design (CSPD). CSPD algorithms search large protein conformational spaces to approximate biophysical quantities. In this dissertation we present new algorithms to realistically and accurately model how amino acid mutations change protein structure. These algorithms model continuous flexibility, protein ensembles and positive/negative design, while providing guarantees on the output. Using these algorithms and the OSPREY protein design program we design and apply protocols for three biomedically-relevant problems: (i) prediction of new drug resistance mutations in bacteria to a new preclinical antibiotic, (ii) the redesign of llama antibodies to potentially reduce their immunogenicity for use in preclinical monkey studies, and (iii) scaffold-based anti-HIV antibody design. Experimental validation performed by our collaborators confirmed the importance of the algorithms and protocols.</p> / Dissertation

In this thesis we study a class of problems that require simultaneously controlling a large number of dynamical systems, with varying system dynamics, using the same control signal. We call such problems ensemble control problems, as the goal is to simultaneously steer the entire ensemble of systems. These problems are motivated by many physical systems and we will be particularly interested in the manipulation of nuclear spins in Nuclear Magnetic Resonance (NMR) experiments. System dispersion arise from imprecise magnets for controls, or from disruptive intermolecular interactions. In all cases, the aim is to attenuate the aspects fo the dynamics that correspond to noise or errors, while perserving the aspects that contain the quantities of interest. In liquid NMR experiments this could correspond to preserving Larmor frequency in the presence of inhomogeneities of the strength of the applied radio frequency (RF) field. In solid state NMR, reducing or eliminating orientation dependent magnetic fields is of key concern, so that a precise spectrum can be observed. We approach the problem from the standpoint of mathematical control theory in which the challenge is to simultaneously steer a continuum of systems between points of interest with the same control signal. At the heart of this problem is finding ways for the nonlinearity of the system to be used to our advantage, so that while all members of the ensemble will be driven with the same controls, their final orientations can be orchestrated to arbitrary precision. This thesis develops the methods necessary for two such ensemble control problems arising in NMR, RF (control) amplitude inhomogeneity and systems with periodic drifts that exhibit dispersions in their amplitude and phase. In both cases, robust controls will rely on the non-commutativity of the system's dynamics enabling the generation of alternative and more robust control elements. / Engineering and Applied Sciences

Filtering problems in high dimensional geophysical applications often require spatially continuous models to interpolate spatially and temporally sparse data. Many applications in numerical weather and ocean state prediction are concerned with tracking and assessing the uncertainty in the position of large scale vorticity features, such as storm fronts, jets streams, and hurricanes. Quantifying the amplitude variance in these features is complicated by the fact that both height and lateral perturbations in the feature geometry are represented in the same covariance estimate. However, when there are sufficient observations to detect feature information like spatial gradients, the positions of these features can be used to further constrain the filter, as long as the statistical model (cost function) has provisions for both height perturbations and lateral displacements. Several authors since the 1990s have proposed various formalisms for the simultaneous modeling of position and amplitude errors, and the typical approaches to computing the generalized solutions in these applications are variational or direct optimization. The ensemble Kalman filter is often employed in large scale nonlinear filtering problems, but its predication on Gaussian statistics causes its estimators suffer from analysis deflation or collapse, as well as the usual curse of dimensionality in high dimensional Monte Carlo simulations. Moreover, there is no theoretical guarantee of the performance of the ensemble Kalman filter with nonlinear models. Particle filters which employ importance sampling to focus attention on the important regions of the likelihood have shown promise in recent studies on the control of particle size. Consider an ensemble forecast of a system with prominent feature information. The correction of displacements in these features, by pushing them into better agreement with observations, is an application of importance sampling, and Monte Carlo methods, including particle filters, and possibly the ensemble Kalman filter as well, are well suited to applications of feature displacement correction. In the present work, we show that the ensemble Kalman filter performs well in problems where large features are displaced both in amplitude and position, as long as it is used on a statistical model which includes both function height and local position displacement in the model state. In a toy model, we characterize the performance-degrading effect that untracked displacements have on filters when large features are present. We then employ tools from classical physics and fluid dynamics to statistically model displacements by area-preserving coordinate transformations. These maps preserve the area of contours in the displaced function, and using strain measures from continuum mechanics, we regularize the statistics on these maps to ensure they model smooth, feature-preserving displacements. The position correction techniques are incorporated into the statistical model, and this modified ensemble Kalman filter is tested on a system of vortices driven by a stochastically forced barotropic vorticity equation. We find that when the position correction term is included in the statistical model, the modified filter provides estimates which exhibit substantial reduction in analysis error variance, using a much smaller ensemble than what is required when the position correction term is removed from the model.

The standard ensemble filtering schemes such as Ensemble Kalman Filter (EnKF) and Sequential Monte Carlo (SMC) do not properly represent states of low priori probability when the number of samples is too small and the dynamical system is high dimensional system with highly non-Gaussian statistics. For example, when the standard ensemble methods are applied to two well-known simple, but highly nonlinear systems such as a one-dimensional stochastic diffusion process in a double-well potential and the well-known three-dimensional chaotic dynamical system of Lorenz, they produce erroneous results to track transitions of the systems from one state to the other.In this dissertation, a set of new parametric resampling methods are introduced to overcome this problem. The new filtering methods are motivated by a general H-theorem for the relative entropy of Markov stochastic processes. The entropy-based filters first approximate a prior distribution of a given system by a mixture of Gaussians and the Gaussian components represent different regions of the system. Then the parameters in each Gaussian, i.e., weight, mean and covariance are determined sequentially as new measurements are available. These alternative filters yield a natural generalization of the EnKF method to systems with highly non-Gaussian statistics when the mixture model consists of one single Gaussian and measurements are taken on full states.In addition, the new filtering methods give the quantities of the relative entropy and log-likelihood as by-products with no extra cost. We examine the potential usage and qualitative behaviors of the relative entropy and log-likelihood for the new filters. Those results of EnKF and SMC are also included. We present results of the new methods on the applications to the above two ordinary differential equations and one partial differential equation with comparisons to the standard filters, EnKF and SMC. These results show that the entropy-based filters correctly track the transitions between likely states in both highly nonlinear systems even with small sample size N=100.

The aim of the research undertaken in this dissertation was to use medium-range to seasonal precipitation forecasts for hydrologic applications for catchments in the core North American Monsoon (NAM) region. To this end, it was necessary to develop a better understanding of the physical and statistical relationships between runoff processes and the temporal statistics of rainfall. To achieve this goal, development of statistically downscaled estimates of warm season precipitation over the core region of the North American Monsoon Experiment (NAME) were developed. Currently, NAM precipitation is poorly predicted on local and regional scales by Global Circulation Models (GCMs). The downscaling technique used here, the K-Nearest Neighbor (KNN) model, combines information from retrospective GCM forecasts with simultaneous historical observations to infer statistical relationships between the low-resolution GCM fields and the locally-observed precipitation records. The stochastic nature of monsoon rainfall presents significant challenges for downscaling efforts and, therefore, necessitate a regionalization and an ensemble or probabilistic-based approach to quantitative precipitation forecasting. It was found that regionalization of the precipitation climatology prior to downscaling using KNN offered significant advantages in terms of improved skill scores.Selected output variables from retrospective ensemble runs of the National Centers for Environmental Predictions medium-range forecast (MRF) model were fed into the KNN downscaling model. The quality of the downscaled precipitation forecasts was evaluated in terms of a standard suite of ensemble verification metrics. This study represents the first time the KNN model has been successfully applied within a warm season convective climate regime and shown to produce skillful and reliable ensemble forecasts of daily precipitation out to a lead time of four to six days, depending on the forecast month.Knowledge of the behavior of the regional hydrologic systems in NAM was transferred into a modeling framework aimed at improving intra-seasonal hydrologic predictions. To this end, a robust lumped-parameter computational model of intermediate conceptual complexity was calibrated and applied to generate streamflow in three unregulated test basins in the core region of the NAM. The modeled response to different time-accumulated KNN-generated precipitation forcing was investigated. Although the model had some difficulty in accurately simulating hydrologic fluxes on the basis of Hortonian runoff principles only, the preliminary results achieved from this study are encouraging. The primary and most novel finding from this study is an improved predictability of the NAM system using state-of-the-art ensemble forecasting systems. Additionally, this research significantly enhanced the utility of the MRF ensemble forecasts and made them reliable for regional hydrologic applications. Finally, monthly streamflow simulations (from an ensemble-based approach) have been demonstrated. Estimated ensemble forecasts provide quantitative estimates of uncertainty associated with our model forecasts.

Nancy Hermiston is examining the links between music cognition and improved learning development through one of the most complicated art forms.

No description available.

The beginning of teaching, when you start to form the future musician, is very important time from which further luck and failure depends. Although there are various circumstances in which child needs more learning motivation. Work theme – the role of ensemble playing by consolidating children’s music learning motivation. The aim of the work – to cognize the role of ensemble playing by consolidating children’s music learning motivation. This aim is realized by doing the pedagogical and psychological analysis of ensemble playing features theory, evaluating the role of ensemble playing by consolidating children’s music learning motivation and learning results, summarizing the results of theoretical analysis and empirical research. This research helps pedagogues to solve hard educative problem of children motivation of learning to play with instrument individually. One of the main problems – how to strengthen children’s music learning motivation effectively making the most of ensemble playing features. Making empirical researches the qualitative ((essay writing, pedagogical observation, conversations with the members of an orchestra), questionnaire (children and teachers) methods were used. The aim of the questionnaire is to find out the children and teachers outlook on the collective playing, its importance to learning motivation and learning results. 91 school-children and 31 teachers from Alytus music school, Kaunas music school and Prienų art school were examined. The... [to full text]

Les systèmes électrotechniques complexes tels qu'une chaîne électromécanique sont formés de plusieurs éléments de natures différentes (convertisseurs statiques, machines, charge...), qui interagissent de manière étroite. Ce travail a pour but d'étudier ces interactions,en se limitant au cas des perturbations harmoniques, et, de choisir, voire ,d'élaborer les Outils de calcul les mieux adaptes en vue de l'analyse puis de l'optimisation globale des systèmes. Une solution consiste à employer des moyens de simulation et de calcul spectral lourds, tant en durée qu'en volume à stocker, mais elle n'aide guère la compréhension des phénomènes et la validité des résultats est incertaine. C'est pourquoi, il est nécessaire de chercher à simplifier les calculs et/ou les modèles, et à construire des procédures systématiques d'analyse des phénomènes. Une telle approche implique la décomposition modulaire du système en sous-ensembles fonctionnels ou physiques, puis la détermination des interactions à étudier. En conséquence de quoi, un modèle est choisi pour chacun des sous-ensembles et pour son environnement. Enfin l'outil de calcul le mieux adapté à l'application doit être défini. Une méthode d'analyse harmonique capable de prendre en compte les interactions bilatérales entre la source, le convertisseur et la machine est proposée. Elle offre une grande précision et permet de montrer, dans le cas étudié, que les interactions sont faibles et que le modèle global,utilisé peut être simplifié. La méthode est aussi appliquée à un exemple d'étude de compatibilité électromagnétique,et permet alors de réduire sensiblement le volume des 'calculs. Loin d'aborder encore le problème de l'optimisation, cette thèse représente un premier pas vers une approche généralisée de l'analyse globale des systèmes complexes dans un volume de calcul minimisé.

Learning from data streams is emerging as an important application area. When the environment changes, it is necessary to rely on on-line learning with the capability to adapt to changing conditions a.k.a. concept drifts. Adapting to concept drifts entails forgetting some or all of the old acquired knowledge when the concept changes while accumulating knowledge regarding the supposedly stationary underlying concept. This tradeoff is called the stability-plasticity dilemma. Ensemble methods have been among the most successful approaches. However, the management of the ensemble which ultimately controls how past data is forgotten has not been thoroughly investigated so far. Our work shows the importance of the forgetting strategy by comparing several approaches. The results thus obtained lead us to propose a new ensemble method with an enhanced forgetting strategy to adapt to concept drifts. Experimental comparisons show that our method compares favorably with the well-known state-of-the-art systems. The majority of previous works focused only on means to detect changes and to adapt to them. In our work, we go one step further by introducing a meta-learning mechanism that is able to detect relevant states of the environment, to recognize recurring contexts and to anticipate likely concepts changes. Hence, the method we suggest, deals with both the challenge of optimizing the stability-plasticity dilemma and with the anticipation and recognition of incoming concepts. This is accomplished through an ensemble method that controls a ensemble of incremental learners. The management of the ensemble of learners enables one to naturally adapt to the dynamics of the concept changes with very few parameters to set, while a learning mechanism managing the changes in the ensemble provides means for the anticipation of, and the quick adaptation to, the underlying modification of the context.