• Refine Query
  • Source
  • Publication year
  • to
  • Language
  • 242
  • 55
  • 28
  • 26
  • 13
  • 12
  • 12
  • 4
  • 4
  • 2
  • 2
  • 1
  • 1
  • 1
  • 1
  • Tagged with
  • 451
  • 82
  • 54
  • 50
  • 48
  • 45
  • 44
  • 44
  • 41
  • 40
  • 36
  • 35
  • 34
  • 33
  • 32
  • About
  • The Global ETD Search service is a free service for researchers to find electronic theses and dissertations. This service is provided by the Networked Digital Library of Theses and Dissertations.
    Our metadata is collected from universities around the world. If you manage a university/consortium/country archive and want to be added, details can be found on the NDLTD website.
361

Théorie de champ-moyen et dynamique des systèmes quantiques sur réseau / Mean-field theory and dynamics of lattice quantum systems

Rouffort, Clément 10 December 2018 (has links)
Cette thèse est dédiée à l'étude mathématique de l'approximation de champ-moyen des gaz de bosons. En physique quantique une telle approximation est vue comme la première approche permettant d'expliquer le comportement collectif apparaissant dans les systèmes quantiques à grand nombre de particules et illustre des phénomènes fondamentaux comme la condensation de Bose-Einstein et la superfluidité. Dans cette thèse, l'exactitude de l'approximation de champ-moyen est obtenue de manière générale comme seule conséquence de principes de symétries et de renormalisations d'échelles. Nous recouvrons l'essentiel des résultats déjà connus sur le sujet et de nouveaux sont prouvés, particulièrement pour les systèmes quantiques sur réseau, incluant le modèle de Bose-Hubbard. D'autre part, notre étude établit un lien entre les équations aux hiérarchies de Gross-Pitaevskii et de Hartree, issues des méthodes BBGKY de la physique statistique, et certaines équations de transport ou de Liouville dans des espaces de dimension infinie. Résultant de cela, les propriétés d'unicité pour de telles équations aux hiérarchies sont prouvées en toute généralité utilisant seulement les caractéristiques génériques de problèmes aux valeurs initiales liés à de telles équations. Egalement, de nouveaux résultats de caractères bien posés et un contre-exemple à l'unicité d'une hiérarchie de Gross-Pitaevskii sont prouvés. L’originalité de nos travaux réside dans l'utilisation d'équations de Liouville et de puissantes techniques de transport étendues à des espaces fonctionnels de dimension infinie et jointes aux mesures de Wigner, ainsi qu'à une approche utilisant les outils de la seconde quantification. Notre contribution peut être vue comme l'aboutissement d'idées initiées par Z. Ammari, F. Nier et Q. Liard autour de la théorie de champ-moyen. / This thesis is dedicated to the mathematical study of the mean-field approximation of Bose gases. In quantum physics such approximation is regarded as the primary approach explaining the collective behavior appearing in large quantum systems and reflecting fundamental phenomena as the Bose-Einstein condensation and superfluidity. In this thesis, the accuracy of the mean-field approximation is proved in full generality as a consequence only of scaling and symmetry principles. Essentially all the known results in the subject are recovered and new ones are proved specifically for quantum lattice systems including the Bose-Hubbard model. On the other hand, our study sets a bridge between the Gross-Pitaevskii and Hartree hierarchies related to the BBGKY method of statistical physics with certain transport or Liouville's equations in infinite dimensional spaces. As an outcome, the uniqueness property for these hierarchies is proved in full generality using only generic features of some related initial value problems. Again, several new well-posedness results as well as a counterexample to uniqueness for the Gross-Pitaevskii hierarchy equation are proved. The originality in our works lies in the use of Liouville's equations and powerful transport techniques extended to infinite dimensional functional spaces together with Wigner probability measures and a second quantization approach. Our contributions can be regarded as the culmination of the ideas initiated by Z. Ammari, F. Nier and Q. Liard in the mean-field theory.
362

Mapping class groups, skein algebras and combinatorial quantization / Groupes de difféotopie, algèbres d'écheveaux et quantification combinatoire

Faitg, Matthieu 16 September 2019 (has links)
Les algèbres L(g,n,H) ont été introduites par Alekseev-Grosse-Schomerus et Buffenoir-Roche au milieu des années 1990, dans le cadre de la quantification combinatoire de l'espace de modules des G-connexions plates sur la surface S(g,n) de genre g avec n disques ouverts enlevés. L'algèbre de Hopf H, appelée algèbre de jauge, était à l'origine le groupe quantique U_q(g), avec g=Lie(G). Dans cette thèse nous appliquons les algèbres L(g,n,H) à la topologie en basses dimensions (groupe de difféotopie et algèbres d'écheveaux des surfaces), sous l'hypothèse que H est une algèbre de Hopf de dimension finie, factorisable et enrubannée mais pas nécessairement semi-simple, l'exemple phare d'une telle algèbre de Hopf étant le groupe quantique restreint associé à sl(2) (à une racine 2p-ième de l'unité). D'abord, nous construisons en utilisant L(g,n,H) une représentation projective des groupes de difféotopie de S(g,0)D et de S(g,0) (où D est un disque ouvert). Nous donnons des formules pour les représentations d'un ensemble de twists de Dehn qui engendre le groupe de difféotopie; en particulier ces formules nous permettent de montrer que notre représentation est équivalente à celle construite par Lyubashenko-Majid et Lyubashenko via des méthodes catégoriques. Pour le tore S(1,0) avec le groupe quantique restreint associé à sl(2) comme algèbre de jauge, nous calculons explicitement la représentation de SL(2,Z) en utilisant une base convenable de l'espace de représentation et nous en déterminons la structure.Ensuite, nous introduisons une description diagrammatique de L(g,n,H) qui nous permet de définir de façon très naturelle l'application boucle de Wilson W. Cette application associe un élément de L(g,n,H) à chaque entrelac dans (S(g,n)D) x [0,1] qui est parallélisé, orienté et colorié par des H-modules. Quand l'algèbre de jauge est le groupe quantique restreint associé à sl(2), nous utilisons W et les représentations de L(g,n,H) pour construire des représentations des algèbres d'écheveaux S_q(S(g,n)). Pour le tore S(1,0) nous étudions explicitement cette représentation. / The algebras L(g,n,H) have been introduced by Alekseev-Grosse-Schomerus and Buffenoir-Roche in the middle of the 1990's, in the program of combinatorial quantization of the moduli space of flat G-connections over the surface S(g,n) of genus g with n open disks removed. The Hopf algebra H, called gauge algebra, was originally the quantum group U_q(g), with g = Lie(G). In this thesis we apply these algebras L(g,n,H) to low-dimensional topology (mapping class groups and skein algebras of surfaces), under the assumption that H is a finite dimensional factorizable ribbon Hopf algebra which is not necessarily semisimple, the guiding example of such a Hopf algebra being the restricted quantum group associated to sl(2) (at a 2p-th root of unity).First, we construct from L(g,n,H) a projective representation of the mapping class groups of S(g,0)D and of S(g,0) (D being an open disk). We provide formulas for the representations of Dehn twists generating the mapping class group; in particular these formulas allow us to show that our representation is equivalent to the one constructed by Lyubashenko-Majid and Lyubashenko via categorical methods. For the torus S(1,0) with the restricted quantum group associated to sl(2) for the gauge algebra, we compute explicitly the representation of SL(2,Z) using a suitable basis of the representation space and we determine the structure of this representation.Second, we introduce a diagrammatic description of L(g,n,H) which enables us to define in a very natural way the Wilson loop map W. This maps associates an element of L(g,n,H) to any link in (S(g,n)D) x [0,1] which is framed, oriented and colored by H-modules. When the gauge algebra is the restricted quantum group associated to sl(2), we use W and the representations of L(g,n,H) to construct representations of the skein algebras S_q(S(g,n)). For the torus S(1,0) we explicitly study this representation.
363

Využití aproximovaných aritmetických obvodů v neuronových sítí / Exploiting Approximate Arithmetic Circuits in Neural Networks Inference

Matula, Tomáš January 2019 (has links)
Táto práca sa zaoberá využitím aproximovaných obvodov v neurónových sieťach so zámerom prínosu energetických úspor. K tejto téme už existujú štúdie, avšak väčšina z nich bola príliš špecifická k aplikácii alebo bola demonštrovaná v malom rozsahu. Pre dodatočné preskúmanie možností sme preto skrz netriviálne modifikácie open-source frameworku TensorFlow vytvorili platformu umožňujúcu simulovať používanie approximovaných obvodov na populárnych a robustných neurónových sieťach ako Inception alebo MobileNet. Bodom záujmu bolo nahradenie väčšiny výpočtovo náročných častí konvolučných neurónových sietí, ktorými sú konkrétne operácie násobenia v konvolučnách vrstvách. Experimentálne sme ukázali a porovnávali rozličné varianty a aj napriek tomu, že sme postupovali bez preučenia siete sa nám podarilo získať zaujímavé výsledky. Napríklad pri architektúre Inception v4 sme získali takmer 8% úspor, pričom nedošlo k žiadnemu poklesu presnosti. Táto úspora vie rozhodne nájsť uplatnenie v mobilných zariadeniach alebo pri veľkých neurónových sieťach s enormnými výpočtovými nárokmi.
364

Segmentace řeči / Speech segmentation

Andrla, Petr January 2010 (has links)
The programme for the segmentation of a speech into fonems was created as a part of the master´s thesis. This programme was made in the programme Matlab and consists of several scripts. The programme serves for automatic segmentation. Speech segmentation is the process of identifying the boundaries between phonemes in spoken natural languages. Automatic segmentation is based on vector quantization. In the first step of algorithm, feature extraction is realized. Then speech segments are assigned to calculated centroids. Position where centroid is changed is marked as a boundary of phoneme. The audiorecords were elaborated by the programme and a operation of the automatic segmentation was analysed. A detailed manual was created to the programme too. Individual used methods of the elaboration of a speech were in the master´s thesis briefly descripted, its implementations in the programme and reasons of set of its parameters.
365

Komprese signálu EKG / Compression of ECG signal

Blaschová, Eliška January 2016 (has links)
This paper represents the most well-known compression methods, which have been published. A Compression of ECG signal is important primarily for space saving in memory cards or efficiency improvement of data transfer. An application of wavelet transform for compression is a worldwide discussed topic and this is the reason why the paper focuses in this direction. Gained wavelet coefficients might be firstly quantized and then compressed using suitable method. There are many options for a selection of wavelet and a degree of decomposition, which will be tested from the point of view of the most efficient compression of ECG signal.
366

Aspects of Gauge Theories in Lorentzian Curved Space-times

Taslimitehrani, Mojtaba 12 December 2018 (has links)
We study different aspects of perturbatively renormalized quantum gauge theories in the presence of non-trivial background Lorentzian metrics and background connections. First, we show that the proof of nilpotency of the renormalized interacting BRST charge can be reduced to the cohomological analysis of the classical BRST differential. This result guarantees the self-consistency of a class of local, renormalizable field theories with vanishing 'gauge anomaly'' at the quantum level, such as the pure Yang-Mills theory in four dimensions. Self-consistency here means that the algebra of gauge invariant observables can be constructed as the cohomology of this charge. Second, we give a proof of background independence of the Yang-Mills theory. We define background independent observables in a geometrical formulation as flat sections of a cohomology algebra bundle over the manifold of background configurations, with respect to a flat connection which implements background variations. We observe that background independence at the quantum level is potentially violated. We, however, show that the potential obstructions can be removed by a finite renormalization. Third, we construct the advanced/retarded Green's functions and Hadamard parametrices for linearized Yang-Mills and Einstein equations in general linear covariant gauges. They play an essential role in formulating gauge theories in curved spacetimes. Finally, we study a superconformal gauge theory in three dimensions (the ABJM theory) which is conformally coupled to a curved background. The superconformal symmetry of this theory is described by a conformal symmetry superalgebra on manifolds which admit twistor spinors. By analyzing the relevant cohomology class of an appropriate BV-BRST differential, we show that the full superalgebra is realized at the quantum level.
367

Making Wireless Communication More Efficient

Jing Guo (11186010) 26 July 2021 (has links)
<div>Given the increasing importance of mobile data access, extending broadband wireless access have become a global grand challenge. Wireless sensor networks (WSNs) and millimeter wave (mmWave) systems have been introduced to resolve these issues which motivate us to have further investigation. In this paper, the first two work assuming a quantized-and-forward WSN. We first develop a rate adaptive integer forcing source coding (RAIF) scheme to enhance the system throughput by assigning optimal quantization rate to each sensor optimally. Then, we are interested in developing an supervised online technique for solving classification problems. In order to enhance the classification performance, we developed this technique by jointly training the decision function that determines/estimates class label, quantizers across all sensors, and reliability of sensors such that M' most reliable sensors are enabled. Finally, we develop an idea to provide a folded low-resolution ADC array architecture that can utilize any of the widely published centralized folded ADC (FADC) implementation by placing the centralized FADC branches at different antenna elements in a millimeter wave (mmWave) system. With adding a simple analog shift and modulo operations prior to the sign quantizer, we show that the multiple low-resolution ADCs across the array elements can be properly designed such that they can be combined into an effective high-resolution ADC with excellent performance characteristics.</div>
368

Distributed Network Processing and Optimization under Communication Constraint

Chang Shen Lee (11184969) 26 July 2021 (has links)
<div>In recent years, the amount of data in the information processing systems has significantly increased, which is also referred to as big-data. The design of systems handling big-data calls for a scalable approach, which brings distributed systems into the picture. In contrast to centralized systems, data are spread across the network of agents in the distributed system, and agents cooperatively complete tasks through local communications and local computations. However, the design and analysis of distributed systems, in which no central coordinators with complete information are present, are challenging tasks. In order to support communication among agents to enable multi-agent coordination among others, practical communication constraints should be taken into consideration in the design and analysis of such systems. The focus of this dissertation is to provide design and analysis of distributed network processing using finite-rate communications among agents. In particular, we address the following open questions: 1) can one design algorithms balancing a graph weight matrix using finite-rate and simplex communications among agents? 2) can one design algorithms computing the average of agents’ states using finite-rate and simplex communications? and 3) going beyond of ad-hoc algorithmic designs, can one design a black-box mechanism transforming a general class of algorithms with unquantized communication to their finite-bit quantized counterparts?</div><div><br></div><div>This dissertation addresses the above questions. First, we propose novel distributed algorithms solving the weight-balancing and average consensus problems using only finite-rate simplex communications among agents, compliant to the directed nature of the network topology. A novel convergence analysis is put forth, based on a new metric inspired by the</div><div>positional system representations. In the second half of this dissertation, distributed optimization subject to quantized communications is studied. Specifically, we consider a general class of linearly convergent distributed algorithms cast as fixed-point iterate, and propose a novel black-box quantization mechanism. In the proposed mechanism, a novel quantizer preserving linear convergence is proposed, which is proved to be more communication efficient than state-of-the-art quantization mechanisms. Extensive numerical results validate our theoretical findings.</div>
369

K-theoretic invariants in symplectic topology

Mezrag, Lydia 12 1900 (has links)
En employant des méthodes de la théorie de Chern-Weil, Reznikov produit une condition suffisante qui assure la non-trivialité de la projectivisation \( \mathbb{P}(E) \) d'un fibré vectoriel complexe en tant que fibré Hamiltonien. Dans le contexte de la quantification géométrique, Savelyev et Shelukhin introduisent un nouvel invariant des fibrés Hamiltoniens avec valeurs dans la K-théorie et étendent le résultat de Reznikov. Cet invariant est donné par l'indice d'Atiyah-Singer d'une famille d'opérateurs \( \text{Spin}^{c} \) de Dirac. Dans ce mémoire, on s'intéresse à des fibrés Hamiltoniens résultant d'un produit fibré et d'un produit cartésien d'une collection de fibrés projectifs complexes \( \mathbb{P}(E_1), \cdots, \mathbb{P}(E_r) \). En usant des mêmes méthodes que Shelukhin et Savelyev, on définit une famille d'opérateurs \( \text{Spin}^{c} \) de Dirac qui agissent sur les sections d'un fibré de Dirac canonique à valeurs dans un fibré pré-quantique. L'indice de famille produit un invariant de fibrés Hamiltoniens avec fibres données par un produit d'espaces projectifs complexes et permet de construire des exemples de fibrés Hamiltoniens non-triviaux. / Using methods of Chern-Weil Theory, Reznikov provides a sufficient condition for the non-triviality of the projectivization \( \mathbb{P}(E) \) of a complex vector bundle \( E \) as a Hamiltonian fibration. In the setting of geometric quantization, Savelyev and Shelukhin introduce a new invariant of Hamiltonian fibrations and a K-theoretic lift of Reznikov's result. This invariant is given by the Atiyah-Singer index of a family of \( \text{Spin}^{c} \)-Dirac operators. In this thesis, we consider Hamiltonian fibrations given by the Cartesian product and the fiber product of a collection of complex projective bundles \( \mathbb{P}(E_1), \cdots, \mathbb{P}(E_r) \). Using the same methods as Savelyev and Shelukhin, we define a family of \( \text{Spin}^{c} \)-Dirac operators acting on sections of a canonical Dirac bundle with values in a suitable prequantum fibration. The family index gives then an invariant of Hamiltonian fibrations with fibers given by a product of complex projective spaces and allows to construct examples of non-trivial Hamiltonian fibrations.
370

Vers une vision robuste de l'inférence géométrique / Toward a Robust Vision of Geometrical Inference

Brécheteau, Claire 24 September 2018 (has links)
Le volume de données disponibles est en perpétuelle expansion. Il est primordial de fournir des méthodes efficaces et robustes permettant d'en extraire des informations pertinentes. Nous nous focalisons sur des données pouvant être représentées sous la forme de nuages de points dans un certain espace muni d'une métrique, e.g. l'espace Euclidien R^d, générées selon une certaine distribution. Parmi les questions naturelles que l'on peut se poser lorsque l'on a accès à des données, trois d'entre elles sont abordées dans cette thèse. La première concerne la comparaison de deux ensembles de points. Comment décider si deux nuages de points sont issus de formes ou de distributions similaires ? Nous construisons un test statistique permettant de décider si deux nuages de points sont issus de distributions égales (modulo un certain type de transformations e.g. symétries, translations, rotations...). La seconde question concerne la décomposition d'un ensemble de points en plusieurs groupes. Étant donné un nuage de points, comment faire des groupes pertinents ? Souvent, cela consiste à choisir un système de k représentants et à associer chaque point au représentant qui lui est le plus proche, en un sens à définir. Nous développons des méthodes adaptées à des données échantillonnées selon certains mélanges de k distributions, en présence de données aberrantes. Enfin, lorsque les données n'ont pas naturellement une structure en k groupes, par exemple, lorsqu'elles sont échantillonnées à proximité d'une sous-variété de R^d, une question plus pertinente est de construire un système de k représentants, avec k grand, à partir duquel on puisse retrouver la sous-variété. Cette troisième question recouvre le problème de la quantification d'une part, et le problème de l'approximation de la distance à un ensemble d'autre part. Pour ce faire, nous introduisons et étudions une variante de la méthode des k-moyennes adaptée à la présence de données aberrantes dans le contexte de la quantification. Les réponses que nous apportons à ces trois questions dans cette thèse sont de deux types, théoriques et algorithmiques. Les méthodes proposées reposent sur des objets continus construits à partir de distributions et de sous-mesures. Des études statistiques permettent de mesurer la proximité entre les objets empiriques et les objets continus correspondants. Ces méthodes sont faciles à implémenter en pratique lorsque des nuages de points sont à disposition. L'outil principal utilisé dans cette thèse est la fonction distance à la mesure, introduite à l'origine pour adapter les méthodes d'analyse topologique des données à des nuages de points corrompus par des données aberrantes / It is primordial to establish effective and robust methods to extract pertinent information from datasets. We focus on datasets that can be represented as point clouds in some metric space, e.g. Euclidean space R^d; and that are generated according to some distribution. Of the natural questions that may arise when one has access to data, three are addressed in this thesis. The first question concerns the comparison of two sets of points. How to decide whether two datasets have been generated according to similar distributions? We build a statistical test allowing to one to decide whether two point clouds have been generated from distributions that are equal (up to some rigid transformation e.g. symmetry, translation, rotation...).The second question is about the decomposition of a set of points into clusters. Given a point cloud, how does one make relevant clusters? Often, it consists of selecting a set of k representatives, and associating every point to its closest representative (in some sense to be defined). We develop methods suited to data sampled according to some mixture of k distributions, possibly with outliers. Finally, when the data can not be grouped naturally into $k$ clusters, e.g. when they are generated in a close neighborhood of some sub-manifold in R^d, a more relevant question is the following. How to build a system of $k$ representatives, with k large, from which it is possible to recover the sub-manifold? This last question is related to the problems of quantization and compact set inference. To address it, we introduce and study a modification of the $k$-means method adapted to the presence of outliers, in the context of quantization. The answers we bring in this thesis are of two types, theoretical and algorithmic. The methods we develop are based on continuous objects built from distributions and sub-measures. Statistical studies allow us to measure the proximity between the empirical objects and the continuous ones. These methods are easy to implement in practice, when samples of points are available. The main tool in this thesis is the function distance-to-measure, which was originally introduced to make topological data analysis work in the presence of outliers.

Page generated in 0.0276 seconds