En considérant les polynômes sur le corps fini de Galois à deux éléments, notre intention porte sur la divisibilité des trinômes x^am+x^bs+1, pour m>s≥1, par un polynôme irréductible de degré r, pour cela, nous avons réalisé le résultat :S'il existe m, s des entiers positifs tels que le trinôme x^am+x^bs+1 soit divisible par un polynôme irréductible de degré r sur F2, alors a et b ne sont pas divisibles par (2r- 1). Pour ce type de trinômes nous conjecturons que le rapport πM(a,b)/ πM(1,1) tend vers une limite finie (dépendant de a et b) quand M tend vers l'infini. Notre recherche porte ensuite sur les codes cycliques de rendement 1/2 sur les deux corps finis F3 et F5 et nous accentuons notre recherche sur ceux iso duaux. Le problème central dans la théorie du codage est trouver la plus grande distance minimum dq pour laquelle un code de paramètres [n, q, d] sur Fq existe. Dans ce contexte nous avons réussi à optimiser cette distance pour les codes cycliques de taux 1/2 sur F3 et F5 en allant jusqu’à la longueur 74 pour les codes ternaires et 42 pour ceux sur F5. Nous avons aussi réussi à construire sept classes de codes cycliques iso-duaux sur le corps fini à 3 éléments et trois classes de codes cycliques iso-duaux sur le corps fini à 5 éléments. / Considering polynomials over the Galois finite fields for two elements, our intention stand over the divisibility of the trinomials x^am+x^bs+1, for m>s ≥ 1, by an irreducible polynomial of degree r, for this, we contribute to the result :If there exist positive integers m, s such that the trinomial x^am+x^bs+1 is divisible by an irreducible polynomial of degree r over F2, then a and b are not divisible by (2^r- 1). For this type of trinomials we conjectured that the ratios πM(a,b)/ πM(1,1) tend to a finite limit (dependently of a and b) when M tend to infinity. Our research stand at sequel on the cyclic codes of rate 1/2 over the two finite fields F3 and F5 and we check our research over whose are isodual. The so-called fundamental problem in coding theory is finding the largest value of dq for which a code of parameters [n, q, d] over Fq exists. In this context we have successfully optimize this distance for the cyclic codes of rate 1/2 over F3 and F5 up to length 74 for the ternary cyclic codes and 42 for whose over F5. We have also successful to construct seven classes of isodual cyclic codes over the field of 3 elements and three classes over the field of 5 elements.

京都大学 / 0048 / 新制・課程博士 / 博士(情報学) / 甲第18400号 / 情博第515号 / 新制||情||91(附属図書館) / 31258 / 京都大学大学院情報学研究科数理工学専攻 / (主査)准教授 辻本 諭, 教授 中村 佳正, 教授 梅野 健 / 学位規則第4条第1項該当 / Doctor of Informatics / Kyoto University / DFAM

No description available.

No description available.

In this report we will plunge down in the fascinating world of the generating functions. Generating functions showcase the "power of power series", giving more depth to the word "power" in power series. We start off small to get a good understanding of the generating function and what it does. Also, off course, explaining why it works and why we can do some of the things we do with them. We will see alot of examples throughout the text that helps the reader to grasp the mathematical object that is the generating function. We will look at several kinds of generating functions, the main focus when we establish our understanding of these will be the "ordinary power series" generating function ("ops") that we discuss before moving on to the "exponential generating function" ("egf"). During our discussion on ops we will see a "first time in literature" derivation of the generating function for a recurrence relation regarding "branched coverings". After finishing the discussion regarding egf we move on the Hermite polynomials and show how we derive their generating function. Which is a generating function that generates functions. Lastly we will have a quick look at the "moment generating function".

In this dissertation, a new numerical method for solving the fractional dynamical systems, is presented. We first introduce Riemann-Liouville fractional integral operator for hybrid functions. Then we will show the spectral accuracy of the present method for solving fractional-order differential equations, and we will extend the present method for solving nonlinear fractional integro-differential equations, fractional Bagley-Torvik equation, distributed order fractional differential equations, two-dimensional fractional partial differential equations, and fractional optimal control problems. In all cases, we will show the rate of convergence is more than some existing numerical methods which were used to solve these kind of problems in the literature. Illustrative examples are included to demonstrate the validity and applicability of the technique.

No description available.

In this thesis, we consider resource limitations on machine learning algorithms in a variety of settings. In the first two chapters, we study how to learn nonlinear model classes (monomials and neural nets) which are structured in various ways -- we consider sparse monomials and deep neural nets whose weight-matrices are low-rank respectively. These kinds of restrictions on the model class lead to gains in resource efficiency -- sparse and low-rank models are computationally easier to deploy and train. We prove that sparse nonlinear monomials are easier to learn (smaller sample complexity) while still remaining computationally efficient to both estimate and deploy, and we give both theoretical and empirical evidence for the benefit of novel nonlinear initialization schemes for low-rank deep networks. In both cases, we showcase a blessing of nonlinearity -- sparse monomials are in some sense easier to learn compared to a linear class, and the prior state-of-the-art linear low-rank initialization methods for deep networks are inferior to our proposed nonlinear method for initialization. To achieve our theoretical results, we often make use of the theory of Hermite polynomials -- an orthogonal function basis over the Gaussian measure. In the last chapter, we consider resource limitations in an online streaming setting. In particular, we consider how many data points from an oblivious adversarial stream we must store from one pass over the stream to output an additive approximation to the Support Vector Machine (SVM) objective, and prove stronger lower bounds on the memory complexity.

The class of generalized feedback shift register (GFSR) random number generators was a promising method for random number generation in the 1980's, but was abandoned because of some flaws such as poor performance on certain tests for randomness. The poor performance may be due to the choice of primitive polynomials used in the generators, rather than inherent flaws in the method. The original GFSR generators were all based on primitive trinomials. This project examines several alternative choices of primitive polynomials with more than one "interior" term to address this problem and hopefully provide access to good random number generators.

This thesis explores the connection between Crouzeix's conjecture and the convergence of the GMRES algorithm. GMRES is a popular iterative method for solving linear systems and is one of the many Krylov methods. Despite its popularity, the convergence of GMRES is not completely understood. While the spectrum can in some cases be a good indicator of convergence, it has been shown that in general, the spectrum does not provide sufficient information to fully explain the behavior of GMRES iterations. Other sets associated with a matrix that can also help predict convergence are the pseudospectrum and the numerical range. This work focuses on convergence bounds obtained by considering the latter. In particular, it focuses on the application of Crouzeix's conjecture, which relates the norm of a matrix polynomial to the size of that polynomial over the numerical range, to describing GMRES convergence.