Discontinuous dynamical systems and multiagent systems are encountered in numerous engineering applications. This dissertation develops stability and dissipativity of nonlinear dynamical systems with discontinuous right-hand sides, optimality of discontinuous feed-back controllers for Filippov dynamical systems, almost consensus protocols for multiagent systems with innaccurate sensor measurements, and adaptive estimation algorithms using multiagent network identifiers. In particular, we present stability results for discontinuous dynamical systems using nonsmooth Lyapunov theory. Then, we develop a constructive feedback control law for discontinuous dynamical systems based on the existence of a nonsmooth control Lyapunov function de fined in the sense of generalized Clarke gradients and set-valued Lie derivatives. Furthermore, we develop dissipativity notions and extended Kalman-Yakubovich-Popov conditions and apply these results to develop feedback interconnection stability results for discontinuous systems. In addition, we derive guaranteed gain, sector, and disk margins for nonlinear optimal and inverse optimal discontinuous feedback regulators that minimize a nonlinear-nonquadratic performance functional for Filippov dynamical systems. Then, we provide connections between dissipativity and optimality of nonlinear discontinuous controllers for Filippov dynamical systems. Furthermore, we address the consensus problem for a group of agent robots with uncertain interagent measurement data, and show that the agents reach an almost consensus state and converge to a set centered at the centroid of agents initial locations. Finally, we develop an adaptive estimation framework predicated on multiagent network identifiers with undirected and directed graph topologies that identifies the system state and plant parameters online.

This thesis proposes some new model independent or nonparametric methods for estimating the dose-response curve and the effective dosage curve in the context of bioassay. The research problem is also of importance in environmental risk assessment and other areas of health sciences. It is shown in the thesis that our new nonparametric methods while bearing optimal asymptotic properties also exhibit strong finite sample performance. Although our specific emphasis is on bioassay and environmental risk assessment, the methodology developed in this dissertation applies broadly to general order restricted inference.

宋詞研究學者提出許多詞律上的規定與特色，然而有關宋詞詞調之間的互動研究仍相當缺乏，因此本研究旨在探討北宋詞人姜夔的十七首自度曲當中語言形式的規定與音樂的關係。本篇研究包含語料庫與理論分析，其中語料庫分析針對結構、節奏與聲調歸納出姜夔十七首詞當中的傾向，並佐以數據分析。理論分析根據優選理論的框架說明姜夔作品中的語言形式如何選擇，並證實所有輸出值皆可用階級式制約來決定。本篇研究希望能解釋宋詞中語言與音樂的關係並為詞牌規定提出不同觀點。 / Researchers of Song verses have often proposed the rules to describe the poetic composition, but few have focused on the interactions between verse language and music. Therefore, the aim of the study attempts to explore how music and the linguistic requirements are related in Jiang Kuei’s seventeen verses. This research involves a corpus-based and a theoretical analyses. The corpus-based analysis focuses on structure, rhythm and tone to generalize the patterns of Jiang’s verses on a statistic basis. Under the framework of Optimality Theory, the theoretical analysis explains Jiang’s linguistic and musical arrangements in verses and confirms that the output forms are determined by a hierarchy of constraints. To conclude, this study sheds new light on the relationship between music and language and provides a fresh perspective on the Song verse formats.

The costs of developing new pharmaceuticals have increased dramatically during the past decades. Contributing to these increased expenses are the increasingly extensive and more complex clinical trials required to generate sufficient evidence regarding the safety and efficacy of the drugs.  It is therefore of great importance to improve the effectiveness of the clinical phases by increasing the information gained throughout the process so the correct decision may be made as early as possible.   Optimal Design (OD) methodology using the Fisher Information Matrix (FIM) based on Nonlinear Mixed Effect Models (NLMEM) has been proven to serve as a useful tool for making more informed decisions throughout the clinical investigation. The calculation of the FIM for NLMEM does however lack an analytic solution and is commonly approximated by linearization of the NLMEM. Furthermore, two structural assumptions of the FIM is available; a full FIM and a block-diagonal FIM which assumes that the fixed effects are independent of the random effects in the NLMEM. Once the FIM has been derived, it can be transformed into a scalar optimality criterion for comparing designs. The optimality criterion may be considered local, if the criterion is based on singe point values of the parameters or global (robust), where the criterion is formed for a prior distribution of the parameters.  Regardless of design criterion, FIM approximation or structural assumption, the design will be based on the prior information regarding the model and parameters, and is thus sensitive to misspecification in the design stage.  Model based adaptive optimal design (MBAOD) has however been shown to be less sensitive to misspecification in the design stage.   The aim of this thesis is to further the understanding and practicality when performing standard and MBAOD. This is to be achieved by: (i) investigating how two common FIM approximations and the structural assumptions may affect the optimized design, (ii) reducing runtimes complex design optimization by implementing a low level parallelization of the FIM calculation, (iii) further develop and demonstrate a framework for performing MBAOD, (vi) and investigate the potential advantages of using a global optimality criterion in the already robust MBAOD.

En optimisation les conditions d’optimalité jouent un rôle primordial pour détecter les solutions optimales et leur étude occupe une place significative dans la recherche actuelle. Afin d’exprimer adéquatement des conditions d’optimalité les chercheurs ont introduit diverses notions de dérivées généralisées non seulement pour des fonctions non lisses, mais aussi pour des fonctions à valeurs ensemblistes, dites applications multivoques ou multifonctions. Cette thèse porte sur l’application des deux nouveaux concepts de dérivées généralisées: les ensembles variationnels de Khanh-Tuan et les approximations de Jourani-Thibault, aux problèmes d’optimisation multiobjectif et aux problèmes d’équilibre vectoriel. L’enjeu principal est d’obtenir des conditions d’optimalité du premier et du second ordre pour les problèmes ayant des données multivoques ou univoques non lisses et pas forcément continues, et des conditions assurant l’unicité des solutions dans les problèmes d’équilibre vectoriel. / Optimality conditions for nonsmooth optimization have become one of the most important topics in the study of optimization-related problems. Various notions of generalized derivatives have been introduced to establish optimality conditions. Besides establishing optimality conditions, generalized derivatives also is an important tool for studying the local uniqueness of solutions. During the last three decades, these topics have been being developed, generalized and applied to many elds of mathematics by many authors all over the world. The purpose of this thesis is to investigate the above topics. It consists of ve chapters. In Chapter 1, we develop elements of calculus of variational sets for set-valued mappings, which were recently introduced in Khanh and Tuan (2008). Most of the usual calculus rules, from chain and sum rules to rules for unions, intersections, products and other operations on mappings, are established. As applications we provide a direct employment of sum rules to establishing an explicit formula for a variational set of the solution map to a parametrized variational inequality in terms of variational sets of the data. Furthermore, chain rules and sum or product rules are also used to prove optimality conditions for weak solutions of some vector optimization problems. In Chapter 2, we propose notions of higher-order outer and inner radial derivatives of set-valued maps and obtain main calculus rules. Some direct applications of these rules in proving optimality conditions for particular optimization problems are provided. Then, we establish higher-order optimality necessary conditions and sufficient ones for a general set-valued vector optimization problem with inequality constraints. Chapter 3 is devoted to using first and second-order approximations, which were introduced by Jourani and Thibault (1993) and Allali and Amaroq (1997), as generalized derivatives, to establish both necessary and sufficient optimality conditions for various kinds of solutions to nonsmooth vector equilibrium problems with functional constraints. Our rst-order conditions are shown to be applicable in many cases, where existing ones cannot be applied. The second-order conditions are new. In Chapter 4, we consider nonsmooth multi-objective fractional programming on normed spaces. Using rst and second-order approximations as generalized derivatives, rst and second-order optimality conditions are established. For sufficient conditions no convexity is needed. Our results can be applied even in innite dimensional cases involving innitely discontinuousmaps. In Chapter 5, we establish sufficient conditions for the local uniqueness of solutions to nonsmooth strong and weak vector equilibrium problems. Also by using approximations, our results are valid even in cases where the maps involved in the problems suffer innite discontinuity at the considered point.

This thesis is concerned with developing methodologies that enable existing models to be effectively reused. Results of this thesis are presented in the framework of Quantitative Structural-Activity Relationship (QSAR) models, but their application is much more general. QSAR models relate chemical structures with their biological, chemical or environmental activity. There are many applications that offer an environment to build and store predictive models. Unfortunately, they do not provide advanced functionalities that allow for efficient model selection and for interpretation of model predictions for new data. This thesis aims to address these issues and proposes methodologies for dealing with three research problems: model governance (management), model identification (selection), and interpretation of model predictions. The combination of these methodologies can be employed to build more efficient systems for model reuse in QSAR modelling and other areas. The first part of this study investigates toxicity data and model formats and reviews some of the existing toxicity systems in the context of model development and reuse. Based on the findings of this review and the principles of data governance, a novel concept of model governance is defined. Model governance comprises model representation and model governance processes. These processes are designed and presented in the context of model management. As an application, minimum information requirements and an XML representation for QSAR models are proposed. Once a collection of validated, accepted and well annotated models is available within a model governance framework, they can be applied for new data. It may happen that there is more than one model available for the same endpoint. Which one to chose? The second part of this thesis proposes a theoretical framework and algorithms that enable automated identification of the most reliable model for new data from the collection of existing models. The main idea is based on partitioning of the search space into groups and assigning a single model to each group. The construction of this partitioning is difficult because it is a bi-criteria problem. The main contribution in this part is the application of Pareto points for the search space partition. The proposed methodology is applied to three endpoints in chemoinformatics and predictive toxicology. After having identified a model for the new data, we would like to know how the model obtained its prediction and how trustworthy it is. An interpretation of model predictions is straightforward for linear models thanks to the availability of model parameters and their statistical significance. For non linear models this information can be hidden inside the model structure. This thesis proposes an approach for interpretation of a random forest classification model. This approach allows for the determination of the influence (called feature contribution) of each variable on the model prediction for an individual data. In this part, there are three methods proposed that allow analysis of feature contributions. Such analysis might lead to the discovery of new patterns that represent a standard behaviour of the model and allow additional assessment of the model reliability for new data. The application of these methods to two standard benchmark datasets from the UCI machine learning repository shows a great potential of this methodology. The algorithm for calculating feature contributions has been implemented and is available as an R package called rfFC.

Submitted by Mariana Dornelles Vargas (marianadv) on 2015-05-13T11:54:46Z No. of bitstreams: 1 aquisicao_onsets.pdf: 652262 bytes, checksum: 861016c9e4c45527818887574b6b636b (MD5) / Made available in DSpace on 2015-05-13T11:54:46Z (GMT). No. of bitstreams: 1 aquisicao_onsets.pdf: 652262 bytes, checksum: 861016c9e4c45527818887574b6b636b (MD5) Previous issue date: 2008 / Nenhuma / Este trabalho tem por objetivo refletir sobre a aquisição de onsets complexos por crianças com desenvolvimento fonológico normal, entre as idades de 2 e 5 anos, aproximadamente, embasando-se na Teoria da Otimidade. Por ser a última estrutura estabilizada no sistema lingüístico da criança, a seqüência CCV merece discussão sobre as peculiaridades de sua aquisição, visto que estruturas mais simples, como V e CV já fazem parte do sistema fonológico de crianças entre as idades de 1:0 e 1:4. A pesquisa conta com a descrição do percurso de aquisição da sílaba complexa por oito informantes, cujos dados fazem parte de investigações desenvolvidas na UNISINOS. Diante desses dados, tecem-se considerações sobre as produções que não evidenciaram a forma alvo (CCV) e sobre os contextos favoráveis à sua realização. São feitas, ainda, comparações com estudos já realizados, como os de Magalhães (2000), Ávila (2000), Ribas (2002) e Redmer (2007), entre outros. Destaca-se que, de acordo com os dados analisados, os sujeitos evidenciam duas etapas durante a aquisição da estrutura CCV: produção C1V >> produção C1C2V. A análise deste percurso como base na Teoria da Otimidade mostra seu grande poder explicativo. Este estudo, então, pretende contribuir com reflexões sobre a aquisição da fonologia por crianças de 2 a 5 anos, valendo-se da Teoria da Otimidade, além de trazer informações sobre fenômenos normalmente observados na linguagem infantil. / This paper proposes reflections about language acquisition from children with normal phonologic development, between the ages of 2 and 5, based on Optimality Theory. For being the last structure stabilized in the child linguistic system, the sequence CCV leads to a reflection on the peculiarities of its acquisition, since simpler structures, as V and CV, already are part of the children phonologic system between the age of 1:0 and 1:4. The research describes the complex syllable acquisition by eight informers that are part of researches realized at UNISINOS. It made considerations about outputs that had not evidenced the right form and about the favorable contexts to the correct speech of complex onset. There are also comparisons with studies already done, like Magalhães (2000), Avila (2000), Ribas (2002) and Redmer (2007), among others. According to the analyzed data, the children show two steps during the structure CCV acquisition: production C1V >> production C1C2V. These reflections based on Optimality Theory showed its great explaining power. Then, this study wants to contribute with reflections about the language acquisition for children by 2 to 5, based on Optimality Theory, beyond brings information about phenomena observed on children language.

Esta dissertação apresenta um estudo acerca da ordem pronominal do português brasileiro (PB) atual. O objetivo da pesquisa é investigar as razões para a mudança de ênclise para próclise, atestada em meados do século XIX, e o uso do pronome tônico em função de objeto direto, observado no PB atual, que, pela nossa hipótese, é a estratégia empregada na tentativa de se recuperar a ordem SVO, perdida quando, em meados do século XIX, a próclise se tornou categórica. Fizemos um levantamento bibliográfico a respeito da ordem pronominal em momentos anteriores da língua a fim de entender o momento atual pelo qual o PB está passando. A reunião de todos esses trabalhos nos possibilitou explicar, via Teoria da Otimidade (OT), (i) as causas que levaram o português antigo (PA) a admitir, primeiramente, a ênclise; (ii) a mudança da ênclise para a próclise em uma fase posterior ao PA; e (iii) a coexistência de próclise e pronome tônico, no caso dos pronomes eu, tu, nós, me, te, nos, e a prevalência do pronome tônico, nos casos de você(s), ele(a/s) e a gente no PB atual. A OT nos forneceu embasamento para investigar as restrições que atuaram no PA e que estão atuando no PB atual, a hierarquia das restrições e a mudança nessa hierarquia, que foi essencial na explicação das alterações no quadro pronominal que nos propusemos a estudar. Averiguamos que as restrições presentes na gramática do PA – e em fase posterior, quando ocorre a mudança da ordem pronominal – não dão conta de explicar o que vem acontecendo com o PB na atualidade. Por isso, acreditamos, uma nova restrição – *PRONOPOD – desempenha hoje papel fundamental na colocação pronominal do PB. Empregando essa nova restrição às análises, chegamos à hierarquia de restrições do PB atual e constatamos que, no que tange aos pronomes de fato, quando o foco recai sobre o objeto, há uma oscilação de duas gramáticas ainda flutuantes. Também concluímos que a implementação do pronome pleno como objeto tenha se dado, primeiramente, com o pronome ele e, posteriormente, se espalhado pelo quadro pronominal. / This dissertation presents a study on pronoun placement of actual Brazilian Portuguese (BP). Its aim is investigate the reasons for the changing from enclisis to proclisis, occurred in mid-nineteenth century, and the use of the lexical pronoun in object position, observed in the actual BP, which, by our analysis, is the strategy used in the attempt to recover the SVO order, lost when proclisis became categorical. We did a bibliographic survey about the pronoun placement in previous moments of language in order to understand the current moment the BP is passing by. The joining of all these studies allowed us to explain, via Optimality Theory (OT), (i) the causes that led the former Portuguese (FP) to admit firstly enclisis, (ii) the change from enclisis to proclisis in FP at a later stage, and (iii) the coexistence of proclisis and lexical pronoun, in the case of the pronouns eu, tu, nós, me, te, nos, and the prevalence of lexical pronoun, in the cases of você(s), ele(a/s) and a gente in the current BP. OT provided the basis for investigating the constraints that acted in FP and that are acting in the PB, the hierarchy of restrictions and the changing in this hierarchy, which was essential in explaining the changes in the pronominal framework that we proposed to study here. We noticed that the restrictions present in the FP – and at a later stage, when the change of pronoun order occurs – did not explain what is happening to the PB today. Therefore, we believe, a new constraint – *PRONOPOD – plays today a key role in pronoun placement in BP. Using this new restriction in the analysis, we come to the hierarchy of constraints of the current BP and we found that, with respect to real pronouns, when the focus is on the object, there is an oscillation between two floating grammars. We also concluded that the implementation of the lexical pronoun as the object has been given, first, with the pronoun ele, and later spread to the other pronouns.

Une combinaison phénotypique associée à une valeur maximale de taux de croissance démographique au niveau local définit une optimalité fonctionnelle locale. L’objectif de cette thèse est de comprendre le lien entre cette optimalité, les abondances et la coexistence des espèces au sein d’une communauté, à partir de données observées et d’approches de modélisation. Nous montrons en premier lieu que la moyenne fonctionnelle locale, pondérée par les abondances relatives des espèces, dépend de la distribution fonctionnelle régionale et dévie de l’optimalité fonctionnelle le long de gradients environnementaux, entrainant des biais possibles d’interprétation. Pour éviter de tels biais, nous proposons une approche d’inférence évaluant explicitement les paramètres du filtre environnemental avec un modèle mécaniste, et l’appliquons pour évaluer l’assemblage de communautés végétales le long d’une succession écologique. Nous étudions ensuite la signature de l’optimalité fonctionnelle à différentes échelles spatiales, à travers la structure de réseaux bipartis de communautés et d’espèces. La cohérence émergente des assemblages au sein du réseau permet de caractériser des ensembles fonctionnels, comme cela est illustré pour des prairies en France métropolitaine. La distribution d’occurrences des espèces entre ensembles régionaux définit une métrique nouvelle de spécialisation écologique. Nous montrons que la distance à l’optimalité fonctionnelle locale des espèces spécialistes et généralistes est fonction de leurs capacités de compétition et de tolérance à des stress physiologiques. Les espèces généralistes sont ainsi en moyenne de meilleures compétitrices éloignées de l’optimalité locale tandis que les spécialistes sont de meilleures tolérantes au stress. Nous évaluons enfin le lien entre abondances et distance à l’optimalité sous l’influence conjointe de dynamiques stochastiques, du filtre environnemental et des interactions compétitrices, en fonction des contributions des traits fonctionnels à ces mécanismes. La thèse formalise via différents modèles d’assemblage la notion d’optimalité et caractérise la signature de l’optimalité fonctionnelle à différentes échelles spatiales. Les applications à plusieurs types de communautés d'organismes illustrent le potentiel des approches mécanistes pour mieux évaluer les processus écologiques et biogéographiques générateurs des motifs de biodiversité. / A phenotypic combination linked to a maximal value of demographic rate at local scale defines a functional local optimality. The goal of this thesis is to understand the linkage between this optimality, the abundances and coexistence of species within communities, using both observational and modelling approaches. We first illustrate how community weighted means are influenced by the regional distribution of functional traits and deviates from the functional optimality along environmental gradients, leading to biases of interpretation. To avoid such biases, we propose a method to explicitly infer the parameters of the environmental filtering using a mechanistic model. We apply this method to plant communities distributed along a successional gradient with the objective to assess the community assembly parameters. We then study the signature of functional optimality across different spatial scales, through the structure of bipartite networks composed of communities and species. The emergent coherence of the assemblages within the network allows characterizing functional pools of species. This has been illustrated using a database of French grassland communities. The distribution of species’ occurrences between regional pools defines a novel metrics of ecological specialization. We show that the distance to functional optimality of specialist and generalist species is function of their competitive and stress-tolerance abilities. Generalist species are in average better competitors distant from the local optimality regarding their competitive traits while specialist species express greater stress-tolerance. Finally, we assess the link between abundances and distance to optimality under the joint influence of stochastic dynamics, environmental filtering and competitive interactions, as a function of the contribution of functional traits to these mechanisms. Thanks to the use of various assembly models, this thesis defines the notion of optimality and assesses its functional signature across spatial scales. Applications to distinct types of communities illustrate the potential of mechanistic approaches towards a better assessment of ecological and biogeographical drivers of biodiversity patterns.

本計畫擬於優選理論 (Prince & Smolensky 1993/2004) 框架下，探討現代華語擬聲詞在音韻上的特點與表現，並分析優選理如何論操作單音節及雙音節輸入項所衍生出的雙音節擬聲詞，進而完全或部分重疊成三音節及四音節華語擬聲詞。擬聲詞為模仿自然界中的聲音形成的詞彙，但並非隨意衍生形成，仍然需要遵循語言的構詞及音韻規律，才能衍生成為合法詞彙並順暢地使用。在華語音韻研究方面，歷來學者對於擬聲詞的討論並不豐富詳盡，因此本碩士論文便嘗試探討華語雙音節、三音節及四音節擬聲詞在音韻分析中的樣貌。 重疊在漢語擬聲詞中為極為常見的現象。如現代華語中單音節的「咕」與雙音節的「咕咕」；雙音節的「滴答」與四音節的「滴滴答答」。故擬聲詞衍生中的重疊機制為本計畫中主要討論議題，此外由於華語擬聲詞結構在音段選擇中展現一致性，故另聚焦於固定音段， 如邊音/l/及前元音/i/在現代華語擬聲詞中高度頻繁地出現，我將對其起源提出假設並分析，如現代華語「劈哩啪啦」的/l/及/i/。而華語雙音節擬聲詞的來源二分為單音節輸入項及雙音節輸入項亦為討論對象。 最後預期華語擬聲詞大致符合現代華語成詞的語法，但進行重疊時與實詞相異，語意因素不會成為衍生動機；經由優選理論分析，從音韻結構角度切入，不僅可了解華語擬聲詞衍生的制約排序，亦更加了解其音韻構詞規範，且與實詞如形容詞進行重疊時的制約排序進行比較，說明華語擬聲詞在衍生過程中傾向遵循邊緣音韻規則，與其他實詞多依據核心音韻 規則的狀況大異其趣。以此為基石去了解其他漢語方言對於擬聲成詞的標準，甚或研究日語、韓語中豐富的擬聲詞，便能構擬出擬聲成詞更具體且具有說服力的語法規則。 / In this thesis, the generation of Modern Mandarin disyllabic, trisyllabic and quadrisyllabic onomatopoeia under phonological viewpoint via Optimality Theory (Prince & Smolensky 1993/2004) approach will be discussed. In my M.A. Thesis ‘An OT analysis of Mandarin Onomatopoeia’, four factors motivated this study. First, the different sources of Mandarin disyllabic onomatopoeia which I thought there are monosyllabic and disyllabic inputs aroused my interests. Second, there are some similarities of Mandarin onomatopoeia between Mandarin core lexicons. Third, the reduplication mechanism in Mandarin onomatopoeia will be taken as a main point in this study. For example, in Modern Mandarin ‘gu’ (Sound of bird sing once) will extend into ‘gugu’ (Sound of birdsong) under total reduplication. Finally, the issue concerning what will be the best choice when Mandarin onomatopoeia is reduplicated with fixed segmentism also plays an important role in this study. For instance, the /l/ and /i/ in Modern Mandarin’s ‘pilipala’ (Sound of many things falling). Onomatopoeias have their special function as sound-imitating icons. They are not the words which are formed by arbitrary but by the rules. They sometimes violate phonological structures of the language in which they occur; hence they belong to the peripheral phonology rather than core one. The Mandarin onomatopoeia is different from Mandarin content words; they are not motivated by semantic factors. So the Chinese onomatopoeia generation analysis via OT approach supplies not only the morpho-phonological constraints but also the comparison of peripheral words between content words. I would like to assume that the Mandarin onomatopoeia generation follows the peripheral phonological rules. This study will provide as a basement to study on other Chinese language’s onomatopoeia or even onomatopoeia in Tibetan, Japanese and Korean.