Thesis (M.S.)--Virginia Polytechnic Institute and State University, 1991. / Vita. Abstract. Includes bibliographical references (leaves 57-59). Also available via the Internet.

In this dissertation, I confront a problem in the philosophy of language from an historical and systematic standpoint. The problem consists in explicating a concept of nonsignificance which can apply, inter alia, to the appraisal of philosophical assertions as category-mistaken or typeabsurd. Such appraisals often take the form: "To say 'a is F' is non-significant, because a is not the ~of thing which can be For not-F." Accordingly, the thesis begins in an examination of the historical and philosophical basis for Russell's theory of logical tvpes, with its concomitant classification of propositions into true, false or nonsignificant. In Part I, I seek to remedy a failing in past exegeses of the development of Russell's type theory which ignore Russell's demand that his "proper" solution to the paradoxes--the ramified theory of types-not simply provide a consistent logicist system; but should also be recommended by his other philosophical doctrines. I remedy this failing by showing that: (i) the source of inconsistency in Frege's logicism lies in his underlying semantic doctrines: complete definition and the treatment of extensions as objects; (ii) the genesis of Russell's ramified theory lies in his logic, epistemology and theory of meaning--viz: the connections between his Vicious Circle Principle, his Multiple Relation Theory of judgement, and his doctrine of incomplete symbols. (iii) in particular, the Multiple Relation Theory provided Russell with a foundation for the ramified theory which was undermined when Wittgenstein subjected it to two "paralysing" objections (hitherto, only partly reconstructed), within Wittgenstein's ongoing critique of the logical doctrines of PM. I reconstruct these criticisms and survey, in general, the critical background to the ramified theory resulting in the changes from the first to the second editions of PM. In concluding Part I, I anticipate the constructive enterprise of Part II in arguing that previous attempts to extend the application of type theory to meaningful predication as a whole have often failed through their insensitivity to contextual relativity and linguistic creativity. Nonetheless, I discuss two accounts--Wittgenstein's theory of formal concepts and Ryle's theory of categories--having features which I preserve in Part II. In addition, I argue against construing category-mistakes as ungrammatical or as false. My general contention through Part II is that category-mistaken significance-failures are best explicated within a theory of linguistic acts (broadly Austinian) . I support this contention by considering the circumstances of an utterance failing to yield a statement in context through its failure to express 'content' to an audience. This notion of 'content' is developed by recourse to those techniques of formal semantics which provide an articulation of structural and algebraic features of contexts, utterances and speech-acts in the interaction of which significance is appraised. The interpreted formal languages I develop borrow features from significance and context logics given in Routley and Goddard's The Logic of Significance and Context, (1973); though my approach to the semantics diverges markedly from theirs. The semantic structures I develop are recommended by their exhibiting systematic relations beaveen utterances, contexts and significance without demanding that category-mistaken predications be diagnosed on the basis of~ priori allocations to categories. They represent a category-mistaken predication in terms of a conflict between conditions for successfully talking about items of a type or sort, and for making a statement of such items, in context. Only in this way, it is argued, can a philosophical theory of meaning accommodate fully the richness, creativity and diversity of linguistic acts in context. / Thesis / Doctor of Philosophy (PhD)

In 1966, Claude Berge proposed the following sorting problem. Given a string of n alternating white and black pegs, rearrange the pegs into a string consisting of all white pegs followed immediately by all black pegs (or vice versa) using only moves which take 2 adjacent pegs to 2 vacant adjacent holes. Berge's original question was generalized by considering the same sorting problem using only Berge k-moves, i.e., moves which take k adjacent pegs to k vacant adjacent holes. Let h(n,k) denote the minimum number of Berge k-moves to sort a string of n alternating white and black pegs.The generalized Berge sorting conjecture states that h(n,k) is equal to the ceiling of n/2 for any k and large enough n. We develop a computational framework to determine h(n,k) for small instances with a focus on the most computationally challenging instances; that is, the determination of (k+2,k). / Thesis / Master of Applied Science (MASc)

La miniaturisation des composants CMOS (complementary metal oxide semiconductor) et MEMS (micro-electro-mechanical system) a permis la réalisation de capteurs d'images de faible taille et à forte densité de pixels pour répondre à la demande d'imageurs faible coût. Classiquement, la grande densité de pixels est obtenue en réduisant simplement la taille des pixels. Cependant, cette réduction de taille détériore l'efficacité optique d'absorption des pixels, ce qui impose alors un compromis définissant une taille minimum de pixel. Les pixels ont aujourd'hui une taille de l'ordre de la longueur d'onde, et les systèmes de filtres de couleur qui eux aussi induisent parallèlement des pertes optiques, devraient être revus, de nouvelles méthodes de séparation spectrale devant être envisagées. Cette thèse explore diverses techniques pouvant être utilisées pour trier différentes longueurs d'ondes, principalement dans le lointain infrarouge (8µm-12µm) et dans le visible (0.4µm-0.7µm) vers les pixels adéquats, supprimant ainsi les pertes traditionnelles par filtrage (par absorption ou réflexion).Nous introduisons le concept de tri spectral basé sur la NOE (normalized optical efficiency), qui est le ratio de l'absorption d'un pixel sur l'énergie totale incidente sur un ensemble de pixels. Pour un nombre donné N de pixels d'une sous-matrice de l'imageur (matrice de Bayer), le phénomène de tri spectral a lieu lorsque le NOE de chaque pixel est supérieur à 1/N. Nous étudions tout d'abord par simulation optique des antennes patch de différentes tailles pour trier efficacement la lumière infrarouge. Le réseau d'antennes a été fabriqué et caractérisé dans la plateforme technologique du CEA-LETI pour valider l'étude théorique. Nous rapportons ensuite une étude sur l'utilisation de 2 structures MSM (metal-semiconductor-metal) pour atteindre une absorption supérieure à 50% à 2 longueurs d'ondes dans un détecteur Silicium. Finalement, nous présentons notre étude finale sur une structure multicouche assistée par réseau pour le tri spectral dans le visible, qui permettrait de réduire la taille des pixels dans les imageurs visibles sous le seuil du micron tout en améliorant l'efficacité d'absorption. Nous avons déduit une compréhension de la stratégie de design de telles structures de tri, et présentons une structure de tri conçue pour réaliser du tri spectral avec une efficacité de l'ordre de 80% pour des pixels de taille inférieure à 0,5µm. / The advancement and scaling effect in complementary metal oxide semiconductor (CMOS) and micro-electro-mechanical system (MEMS) technology has made possible to make smaller image sensors with higher density of imaging pixels to respond at the demand of low cost imagers. Generally, the higher pixel density in imaging system is achieved by shrinking the size of each pixel in an array. The shrinking of pixel dimension however deteriorates the optical efficiency and therefore impose the tradeoff between the performance and minimum achievable pixel size. As the pixel size continues to shrink and approach the dimensions comparable to the wavelength, the spectral separation techniques used in current generation imaging system should be revised and new design methodologies have to be explored. This dissertation explored different techniques that could be used to efficiently sort the band of different wavelengths, mainly in far-infrared (8µm - 12µm) and visible (0.4 µm – 0.7 µm) spectrum in different spatial locations. We introduced the concept of spectral sorting based on normalized optical efficiency (NOE). For given number of pixels (N) or detectors, we define the phenomenon of sorting if NOE of individual pixels, considering incidence power from all pixel domain, is greater than 1/N. First we study differently sized optical patch antenna to efficiently sort the infrared light in different spatial locations using numerical techniques. Using array of such antennas we find the near perfect absorption of multiple wavelengths in infrared spectrum. The antenna arrays are fabricated and characterized in CEA-LETI platform to validate our study. We also report our study on using two differently sized Metal-Semiconductor-Metal (MSM) nanostructures to achieve absorption higher than 50% in individual silicon detector for visible spectrum. Finally we present our study on grating based dielectric multilayer structure for sorting of visible light which could enable to shrink the pixel size of visible imaging system to submicron dimension. We derived the comprehensive design strategy of such sorting structure and present the sorting structure designed to achieve optical efficiency as high as 80% in pixel size of as less as 0.5µm.

La miniaturisation des composants CMOS (complementary metal oxide semiconductor) et MEMS (micro-electro-mechanical system) a permis la réalisation de capteurs d'images de faible taille et à forte densité de pixels pour répondre à la demande d'imageurs faible coût. Classiquement, la grande densité de pixels est obtenue en réduisant simplement la taille des pixels. Cependant, cette réduction de taille détériore l'efficacité optique d'absorption des pixels, ce qui impose alors un compromis définissant une taille minimum de pixel. Les pixels ont aujourd'hui une taille de l'ordre de la longueur d'onde, et les systèmes de filtres de couleur qui eux aussi induisent parallèlement des pertes optiques, devraient être revus, de nouvelles méthodes de séparation spectrale devant être envisagées. Cette thèse explore diverses techniques pouvant être utilisées pour trier différentes longueurs d'ondes, principalement dans le lointain infrarouge (8µm-12µm) et dans le visible (0.4µm-0.7µm) vers les pixels adéquats, supprimant ainsi les pertes traditionnelles par filtrage (par absorption ou réflexion).Nous introduisons le concept de tri spectral basé sur la NOE (normalized optical efficiency), qui est le ratio de l'absorption d'un pixel sur l'énergie totale incidente sur un ensemble de pixels. Pour un nombre donné N de pixels d'une sous-matrice de l'imageur (matrice de Bayer), le phénomène de tri spectral a lieu lorsque le NOE de chaque pixel est supérieur à 1/N. Nous étudions tout d'abord par simulation optique des antennes patch de différentes tailles pour trier efficacement la lumière infrarouge. Le réseau d'antennes a été fabriqué et caractérisé dans la plateforme technologique du CEA-LETI pour valider l'étude théorique. Nous rapportons ensuite une étude sur l'utilisation de 2 structures MSM (metal-semiconductor-metal) pour atteindre une absorption supérieure à 50% à 2 longueurs d'ondes dans un détecteur Silicium. Finalement, nous présentons notre étude finale sur une structure multicouche assistée par réseau pour le tri spectral dans le visible, qui permettrait de réduire la taille des pixels dans les imageurs visibles sous le seuil du micron tout en améliorant l'efficacité d'absorption. Nous avons déduit une compréhension de la stratégie de design de telles structures de tri, et présentons une structure de tri conçue pour réaliser du tri spectral avec une efficacité de l'ordre de 80% pour des pixels de taille inférieure à 0,5µm. / The advancement and scaling effect in complementary metal oxide semiconductor (CMOS) and micro-electro-mechanical system (MEMS) technology has made possible to make smaller image sensors with higher density of imaging pixels to respond at the demand of low cost imagers. Generally, the higher pixel density in imaging system is achieved by shrinking the size of each pixel in an array. The shrinking of pixel dimension however deteriorates the optical efficiency and therefore impose the tradeoff between the performance and minimum achievable pixel size. As the pixel size continues to shrink and approach the dimensions comparable to the wavelength, the spectral separation techniques used in current generation imaging system should be revised and new design methodologies have to be explored. This dissertation explored different techniques that could be used to efficiently sort the band of different wavelengths, mainly in far-infrared (8µm - 12µm) and visible (0.4 µm – 0.7 µm) spectrum in different spatial locations. We introduced the concept of spectral sorting based on normalized optical efficiency (NOE). For given number of pixels (N) or detectors, we define the phenomenon of sorting if NOE of individual pixels, considering incidence power from all pixel domain, is greater than 1/N. First we study differently sized optical patch antenna to efficiently sort the infrared light in different spatial locations using numerical techniques. Using array of such antennas we find the near perfect absorption of multiple wavelengths in infrared spectrum. The antenna arrays are fabricated and characterized in CEA-LETI platform to validate our study. We also report our study on using two differently sized Metal-Semiconductor-Metal (MSM) nanostructures to achieve absorption higher than 50% in individual silicon detector for visible spectrum. Finally we present our study on grating based dielectric multilayer structure for sorting of visible light which could enable to shrink the pixel size of visible imaging system to submicron dimension. We derived the comprehensive design strategy of such sorting structure and present the sorting structure designed to achieve optical efficiency as high as 80% in pixel size of as less as 0.5µm.

No description available.

The ant genus Lasius (Formicinae) arose during the early Tertiary approximately 65 million years ago. Lasius is one of the most abundant and widely distributed ant genera in the Holarctic region, with 95 described species placed in six subgenera: Acanthomyops, Austrolasius, Cautolasius, Chthonolasius, Dendrolasius and Lasius. Many species of Lasius have been central to numerous species-level studies and the focus of many ecological, agricultural, and behavioral investigations. The focus of this study was to use molecular phylogenetic analysis of 781 base pairs of the mitochondrial gene cytochrome oxidase I (COI) and 251 base pairs of an anonymous nuclear gene (ANG) to address questions about the evolutionary relationships of North American Lasius species and subgenera. These relationships were used to better understand the biological and evolutionary complexities associated with these species given their North American distributions. The resulting hypotheses generated in this study from the analyses of these genes produced unexpected patterns of phylogenetic placement of Lasius species and subgenera. A number of biological processes alone or together could explain these patterns, including interspecific hybridization and gene introgression, incomplete lineage sorting, and the presence of multiple cryptic species.

Typescript (photocopy). / Digitized by Kansas Correctional Industries

We address the issue of shuffling loads in Automated Storage/Retrieval Systems (AS/RS) in this paper. The objective is to pre-sort the loads into any specified locations in order to minimize the response time of retrievals. 1D, 2D and 3D AS/RS racks have been designed in order to achieve the shuffling efficiently. The shuffling algorithms are described in detail. The response time of retrieval, the lower and upper bounds of energy consumption are also derived. Results of the analysis and numerical experiments show that the shuffling algorithms are quite efficient. / Singapore-MIT Alliance (SMA)

Estimating the sortedness of a sequence has found applications in, e.g., sorting algorithms, database management and webpage ranking. As the data volume in many of these applications is massive, recent research has been focusing on estimating sortedness in the data stream model. In this thesis, we extend the study of this problem to a number of directions. One common measurement of sortedness is the edit distance to monotonicity. Given a stream of items drawn from a totally ordered set, its edit distance to monotonicity is the minimum number of items to remove so that the remaining items are non-decreasing. The space complexity of estimating the edit distance to monotonicity of a data stream is becoming well-understood over the past few years. Motivated by applications on network quality monitoring, we extend the study to estimating the edit distance to monotonicity of a sliding window covering the w most recent items in the stream for any w _ 1. We give a deterministic algorithm which can return an estimate within a factor of (4 + _) using O( 1 _2 log2(_w)) space. We further extend the study in two directions. First, we consider a stream where each item is associated with a value from a partially ordered set. We give a randomized (4+_)-approximate algorithm using O( 1_2 log _2w log w) space. Second, we consider an out-of-order stream where each item is associated with a creation time and a numerical value, and items may be out of order with respect to their creation times. The goal is to estimate the edit distance to monotonicity with respect to the numerical value of items arranged in the order of creation times. We show that any randomized constant-approximate algorithm requires linear space. Finally, we revisit the classical problem of estimating the length of the longest increasing subsequence (LIS) of a data stream. Previous work shows that any deterministic algorithm requires ?(pN) space through a communication problem Hidden-IS, where N is the number of items in the stream. But the randomized space complexity of LIS is open [2]. [23] has given an efficient randomized protocol for Hidden-IS, showing that Hidden-IS may be significantly easier than LIS. We give an even simpler and more efficient randomized protocol for the Hidden-IS problem, indicating that it is unlikely that this communication problem can lead to a polynomial randomized space lower bound for the LIS problem. On the positive side, we propose a new communication problem which we conjecture to be hard enough to lead to a super polylogarithmic randomized space lower bound for the LIS problem. / published_or_final_version / Computer Science / Master / Master of Philosophy