Este trabalho tem como principal objetivo estudar extensões de aplicações multilineares, de polinômios homogêneos e de funções analíticas entre espaços de Banach. Desta maneira, nos baseamos em importantes trabalhos sobre o assunto. Inicialmente apresentamos o produto de Arens para álgebras de Banach, extensões de Aron-Berner e de Davie-Gamelin para aplicações multilineares e provamos que todas estas extensões coincidem. A partir destes resultados, apresentamos a extensão de polinômios homogêneos e o Teorema de Davie-Gamelin que afirma que, assim como no caso de aplicações multilineares, as extensões de polinômios preservam a norma e, como consequência deste teorema, apresentamos uma generalização do Teorema de Goldstine. Em seguida estudamos espaços de Banach regulares e simetricamente regulares, que são propriedades relacionadas com a unicidade de extensão e são definidas a partir do ideal de operadores lineares fracamente compactos K^w(E, F) . Finalmente apresentamos a extensão de uma função de H_b(E) para H_b(E\'\') e o resultado, de Ignacio Zalduendo, que caracteriza esta extensão em termos da continuidade fraca-estrela do operador diferencial de primeira ordem. / The main purpose of this work is to study extensions of multilinear mappings, homogeneous polynomials and analytic functions between Banach Spaces. In this way, we rely on important works on the subject. Firstly we present the Arens-product for Banach algebras, the Aron-Berner and Davie-Gamelin extensions for multilinear mappings and we prove that all these extensions are the same. From these results, we present an extension for homogeneous polynomials and the Davie-Gamelin theorem which asserts that, as in the case of multilinear mappings, the polynomial extension is norm-preserving and, as a consequence of this theorem, we present a generalization of the Goldstine theorem. After that we study regular and symmetrically regular Banach spaces which are properties related to the uniqueness of the extension and are defined in the setting of weakly compact linear operators K^w(E, F) . Lastly, we present the extension of a function of H_b(E) to one in H_b(E\'\') and the result, according to Ignacio Zalduendo, which characterizes this extension in terms of weak-star continuity of the first order differential operator.

Este trabalho tem como principal objetivo estudar extensões de aplicações multilineares, de polinômios homogêneos e de funções analíticas entre espaços de Banach. Desta maneira, nos baseamos em importantes trabalhos sobre o assunto. Inicialmente apresentamos o produto de Arens para álgebras de Banach, extensões de Aron-Berner e de Davie-Gamelin para aplicações multilineares e provamos que todas estas extensões coincidem. A partir destes resultados, apresentamos a extensão de polinômios homogêneos e o Teorema de Davie-Gamelin que afirma que, assim como no caso de aplicações multilineares, as extensões de polinômios preservam a norma e, como consequência deste teorema, apresentamos uma generalização do Teorema de Goldstine. Em seguida estudamos espaços de Banach regulares e simetricamente regulares, que são propriedades relacionadas com a unicidade de extensão e são definidas a partir do ideal de operadores lineares fracamente compactos K^w(E, F) . Finalmente apresentamos a extensão de uma função de H_b(E) para H_b(E\'\') e o resultado, de Ignacio Zalduendo, que caracteriza esta extensão em termos da continuidade fraca-estrela do operador diferencial de primeira ordem. / The main purpose of this work is to study extensions of multilinear mappings, homogeneous polynomials and analytic functions between Banach Spaces. In this way, we rely on important works on the subject. Firstly we present the Arens-product for Banach algebras, the Aron-Berner and Davie-Gamelin extensions for multilinear mappings and we prove that all these extensions are the same. From these results, we present an extension for homogeneous polynomials and the Davie-Gamelin theorem which asserts that, as in the case of multilinear mappings, the polynomial extension is norm-preserving and, as a consequence of this theorem, we present a generalization of the Goldstine theorem. After that we study regular and symmetrically regular Banach spaces which are properties related to the uniqueness of the extension and are defined in the setting of weakly compact linear operators K^w(E, F) . Lastly, we present the extension of a function of H_b(E) to one in H_b(E\'\') and the result, according to Ignacio Zalduendo, which characterizes this extension in terms of weak-star continuity of the first order differential operator.

Nesse trabalho introduzimos e desenvolvemos a teoria de hiper-ideais de aplicações multilineares contínuas e polinômios homogêneos contínuos entre espaços de Banach. A ideia central é refinar os conceitos de multi-ideais e de ideais de polinômios com o objetivo de explorar de forma mais aprofundada a natureza não-linear das aplicações envolvidas. Para isso tomamos a teoria de ideais de operadores lineares, aplicações multilineares e polinômios homogêneos, desenvolvida a partir dos trabalhos de Pietsch, tanto no caso linear como no caso multilinear, como referencial. Provamos resultados gerais para hiper-ideais, damos muitos exemplos ilustrativos, e desenvolvemos métodos para gerar hiper-ideais, tanto no caso multilinear como no caso polinomial. / In this work we introduce and develop the theory of hyper-ideals of multilinear mappings and homogeneous polynomials between Banach spaces. The main idea is to refine the concepts of multi-ideal and of ideal of polynomials with the purpose of exploring deeply the nonlinear nature of the underlying mappings. To do this we take the ideal theory of linear operators, multilinear mappings and homogeneous polynomials, developed from the works of Pietsch, both in the linear and nonlinear cases, as a reference. We prove general results for hyper-ideals, provide a number of illustrative examples, and develop methods to generate hyper-ideals of multilinear mappings, as well as of hyper-ideals of homogeneous polynomials.

Nesse trabalho introduzimos e desenvolvemos a teoria de hiper-ideais de aplicações multilineares contínuas e polinômios homogêneos contínuos entre espaços de Banach. A ideia central é refinar os conceitos de multi-ideais e de ideais de polinômios com o objetivo de explorar de forma mais aprofundada a natureza não-linear das aplicações envolvidas. Para isso tomamos a teoria de ideais de operadores lineares, aplicações multilineares e polinômios homogêneos, desenvolvida a partir dos trabalhos de Pietsch, tanto no caso linear como no caso multilinear, como referencial. Provamos resultados gerais para hiper-ideais, damos muitos exemplos ilustrativos, e desenvolvemos métodos para gerar hiper-ideais, tanto no caso multilinear como no caso polinomial. / In this work we introduce and develop the theory of hyper-ideals of multilinear mappings and homogeneous polynomials between Banach spaces. The main idea is to refine the concepts of multi-ideal and of ideal of polynomials with the purpose of exploring deeply the nonlinear nature of the underlying mappings. To do this we take the ideal theory of linear operators, multilinear mappings and homogeneous polynomials, developed from the works of Pietsch, both in the linear and nonlinear cases, as a reference. We prove general results for hyper-ideals, provide a number of illustrative examples, and develop methods to generate hyper-ideals of multilinear mappings, as well as of hyper-ideals of homogeneous polynomials.

The representer theorem from the reproducing kernel Hilbert space theory is the origin of many kernel-based machine learning and signal modelling techniques that are popular today. Most kernel functions used in practical applications behave in a homogeneous manner across the domain of the signal of interest, and they are called stationary kernels. One open problem in the literature is the specification of a non-stationary kernel that is computationally tractable. Some recent works solve large-scale optimization problems to obtain such kernels, and they often suffer from non-identifiability issues in their optimization problem formulation. Many practical problems can benefit from using application-specific prior knowledge on the signal of interest. For example, if one can adequately encode the prior assumption that edge contours are smooth, one does not need to learn a finite-dimensional dictionary from a database of sampled image patches that each contains a circular object in order to up-convert images that contain circular edges. In the first portion of this thesis, we present a novel method for constructing non-stationary kernels that incorporates prior knowledge. A theorem is presented that ensures the result of this construction yields a symmetric and positive-definite kernel function. This construction does not require one to solve any non-identifiable optimization problems. It does require one to manually design some portions of the kernel while deferring the specification of the remaining portions to when an observation of the signal is available. In this sense, the resultant kernel is adaptive to the data observed. We give two examples of this construction technique via the grayscale image up-conversion task where we chose to incorporate the prior assumption that edge contours are smooth. Both examples use a novel local analysis algorithm that summarizes the p-most dominant directions for a given grayscale image patch. The non-stationary properties of these two types of kernels are empirically demonstrated on the Kodak image database that is popular within the image processing research community. Tensors and tensor decomposition methods are gaining popularity in the signal processing and machine learning literature, and most of the recently proposed tensor decomposition methods are based on the tensor power and alternating least-squares algorithms, which were both originally devised over a decade ago. The algebraic approach for the canonical polyadic (CP) symmetric tensor decomposition problem is an exception. This approach exploits the bijective relationship between symmetric tensors and homogeneous polynomials. The solution of a CP symmetric tensor decomposition problem is a set of p rank-one tensors, where p is fixed. In this thesis, we refer to such a set of tensors as a rank-one decomposition with cardinality p. Existing works show that the CP symmetric tensor decomposition problem is non-unique in the general case, so there is no bijective mapping between a rank-one decomposition and a symmetric tensor. However, a proposition in this thesis shows that a particular space of rank-one decompositions, SE, is isomorphic to a space of moment matrices that are called quasi-Hankel matrices in the literature. Optimization over Riemannian manifolds is an area of optimization literature that is also gaining popularity within the signal processing and machine learning community. Under some settings, one can formulate optimization problems over differentiable manifolds where each point is an equivalence class. Such manifolds are called quotient manifolds. This type of formulation can reduce or eliminate some of the sources of non-identifiability issues for certain optimization problems. An example is the learning of a basis for a subspace by formulating the solution space as a type of quotient manifold called the Grassmann manifold, while the conventional formulation is to optimize over a space of full column rank matrices. The second portion of this thesis is about the development of a general-purpose numerical optimization framework over SE. A general-purpose numerical optimizer can solve different approximations or regularized versions of the CP decomposition problem, and they can be applied to tensor-related applications that do not use a tensor decomposition formulation. The proposed optimizer uses many concepts from the Riemannian optimization literature. We present a novel formulation of SE as an embedded differentiable submanifold of the space of real-valued matrices with full column rank, and as a quotient manifold. Riemannian manifold structures and tangent space projectors are derived as well. The CP symmetric tensor decomposition problem is used to empirically demonstrate that the proposed scheme is indeed a numerical optimization framework over SE. Future investigations will concentrate on extending the proposed optimization framework to handle decompositions that correspond to non-symmetric tensors.

Submitted by Maike Costa (maiksebas@gmail.com) on 2016-03-29T11:06:09Z No. of bitstreams: 1 arquivo total.pdf: 821623 bytes, checksum: 520d1fa102a8bdfeb531d12a30d60f61 (MD5) / Made available in DSpace on 2016-03-29T11:06:09Z (GMT). No. of bitstreams: 1 arquivo total.pdf: 821623 bytes, checksum: 520d1fa102a8bdfeb531d12a30d60f61 (MD5) Previous issue date: 2014-03-12 / Coordenação de Aperfeiçoamento de Pessoal de Nível Superior - CAPES / Os teoremas de Bohnenblust Hille, demonstrados em 1931 no prestigioso jornal Annals of Mathematics, foram utilizados como ferramentas muito úteis na solução do famoso Problema da convergência absoluta de Bohr. Após um longo tempo esquecidos, estes teoremas têm sido bastante explorados nos últimos anos. Este último quinquê- nio experimentou o surgimento de várias obras dedicadas a estimar as constantes de Bohnenblust Hille ([13, 18, 20, 26, 27, 39, 42, 44, 46, 53]) e também conexões inesperadas com a Teoria da Informação Quântica apareceram (ver, por exemplo, [38]). Há, de fato, quatro casos para serem investigados: polinomial (casos real e complexo) e multilinear (casos real e complexo). Podemos resumir em uma frase as principais informa ções dos trabalhos recentes: as constantes das desigualdades de Bohnenblust Hille são, em geral, extraordinariamente menores do que as primeiras estimativas tinham previsto. Neste trabalho apresentamos algumas das nossas pequenas contribuições ao estudo das constantes nas desigualdades de Bohnenblust-Hille, os quais encontram-se contidos em ([40, 41, 42, 44]).The Bohnenblust Hille theorems, proved in 1931 in the prestigious journal Annals of Mathematics, were used as very useful tools in the solution of the famous "Bohr's absolute convergence problem". After a long time overlooked, these theorems have been explored in the recent years. Last quinquennium experienced the rising of several works dedicated to estimate the Bohnenblust Hille constants ([13, 18, 20, 26, 27, 39, 42, 44, 46, 53]) and also unexpected connections with Quantum Information Theory appeared (see, e.g., [38]). There are in fact four cases to be investigated: polynomial (real and complex cases) and multilinear (real and complex cases). We can summarize in a sentence the main information from the recent preprints: the Bohnenblust Hille constants are, in general, extraordinarily smaller than the rst estimates predicted. In this work, we present some of our small contributions to the study of the constants of the inequalities Bohnenblust-Hille, these are contained in ([40, 41, 42, 44]). / Os teoremas de Bohnenblust Hille, demonstrados em 1931 no prestigioso jornal Annals of Mathematics, foram utilizados como ferramentas muito úteis na solução do famoso Problema da convergência absoluta de Bohr. Após um longo tempo esquecidos, estes teoremas têm sido bastante explorados nos últimos anos. Este último quinquê- nio experimentou o surgimento de várias obras dedicadas a estimar as constantes de Bohnenblust Hille ([13, 18, 20, 26, 27, 39, 42, 44, 46, 53]) e também conexões inesperadas com a Teoria da Informação Quântica apareceram (ver, por exemplo, [38]). Há, de fato, quatro casos para serem investigados: polinomial (casos real e complexo) e multilinear (casos real e complexo). Podemos resumir em uma frase as principais informa ções dos trabalhos recentes: as constantes das desigualdades de Bohnenblust Hille são, em geral, extraordinariamente menores do que as primeiras estimativas tinham previsto. Neste trabalho apresentamos algumas das nossas pequenas contribuições ao estudo das constantes nas desigualdades de Bohnenblust-Hille, os quais encontram-se contidos em ([40, 41, 42, 44])

Coordenação de Aperfeiçoamento de Pessoal de Nível Superior / The main purpose of this dissertation is the study of ideals of multilinear mappings and homogeneous polynomials between linear spaces. By an ideal we mean a class that is stable under the composition with linear operators. First we study multilinear mappings and spaces of multilinear mappings. We also show how to obtain, from a given multilinear mapping, other multilinear mappings with degrees of multilinearity greater than, equal to or smaller than the degree of the original multilinear mapping. Next we study homogeneous polynomials and spaces of homogeneous polynomials, and we also show how to obtain, from a given n-homogeneous polynomial, other polynomials with degrees of homogeneity greater than, equal to or smaller than the degree of the original polynomial. Next we study ideals of multilinear mappings, or multi-ideals, and ideals of homogeneous polynomial, or polynomial ideals, giving several examples and presenting methods to generated multi-ideals and polynomial ideals from a given operator ideal. Finally we dene and give several examples of coherent multi-ideals and coherent polynomial ideals. / O principal objetivo desta dissertação e estudar os ideais de aplicações multilineares e polinômios homogêneos entre espaços vetoriais. Por um ideal entendemos uma classe de aplicações que e estavel atraves da composição com operadores lineares. Primeiramente estudamos as aplicações multilineares e os espaços de aplicações multilineares. Mostramos tambem como obter, a partir de uma aplicação multilinear dada, outras aplicações com graus de multilinearidade maiores, iguais ou menores que o da aplicação original. Em seguida estudamos os polinômios homogêneos e os espacos de polinômios homogêneos, e mostramos que, a partir de um polinômio n-homogêneo, tambem podemos construir novos polinômios homogêneos com graus de homogeneidade maiores, iguais ou menores que n. Posteriormente estudamos os ideais de aplicações multilineares, ou multi-ideais, e os ideais de polinômios homogêneos, exibindo varios exemplos e apresentando metodos para se obter um multi-ideais, ou ideais de polinômios, a partir de ideais de operadores lineares dados. Por m, denimos e exibimos varios exemplos de multi-ideais coerentes e de ideais coerentes de polinômios. / Mestre em Matemática