Neste trabalho, estudamos as dualidades entre modelos de spin em redes bidimensionais a partir de uma abordagem tensorial. Nessa abordagem, componentes de tensores são associadas aos vértices e arestas da rede de forma que a função de partição Z é construída a partir da contração dos índices dessas componentes e é, portanto, um escalar por mudanças de base da álgebra de grupo C[G] utilizada para a definição dos tensores. A partir daí, e observando que a forma das componentes fixam o modelo estudado, obtemos um modelo diferente para cada mudança de base proposta. Esses diferentes modelos possuirão, no entanto, a mesma função de partição, já que esta é um invariante sob tais transformações. De fato, haverá uma infinidade de modelos todos duais entre si. Neste ponto, fixamos nossa atenção nos modelos com spin Zn, nos quais estão incluídos o modelo de Ising, o modelo de Potts e o modelo de Ashkin-Teller-Potts. Explorando uma transformação de base específica, fomos capazes de rederivar a dualidade de Kramers e Wanniers para o modelo de Ising. Usando argumentos análogos, mostramos também que os modelos de Potts com n = 3 e 4 são autoduais e que não existe autodualidade para este modelo com n _ 5. O modelo de Ashkin-Teller-Potts foi mostrado ser autodual para todo n 2 N. / In this work, we study the dualities between spin models in two-dimensional lattices from a tensorial approach. In this approach, we associate tensor components to the vertices and links so that the partition function Z is constructed by a contraction of the indices of the tensor components thereby making Z a scalar under change of basis of the group algebra C[G] used to de_ne the tensors. Having obtained this, and noting that the values of the components _x the studied model, we obtain a di_erent model for each basis transformation proposed. These di_erent models, however, have the same partition function since Z is invariant under these transformations. In fact we can obtain several models all dual to each other in this manner. We then focus on Zn spin models, which include the Ising model, the Potts model and Ashkin- Teller-Potts model. Exploring a speci_c basis transformation, we are able to rederive Kramers and Wanniers\' duality for the Ising model. With analogous arguments, we also show that Potts models with n = 3 and n = 4 are self-dual whereas this property is lost for n _ 5. The Ashkin-Teller-Potts model is shown to be self-dual for all n 2 N.

Counting lattice paths where the number of occurrences of a given pattern is monitored requires a careful analysis of the pattern. Not the length, but the characteristics of the pattern are responsible for the difficulties in finding explicit solutions. Certain features, like overlap and difference in number of ! and " steps determine the recursion formula. In the case of ballot paths, that is paths the stay weakly above the line y = x, the solutions to the recursions are typically polynomial sequences. The objects of Finite Operator Calculus are polynomial sequences, thus the theory can be used to solve the recursions. The theory of Finite Operator Calculus is strengthened and extended to the multivariate setting in order to obtain solutions, and to prepare for future applications. / by Shaun Sullivan. / Thesis (Ph.D.)--Florida Atlantic University, 2011. / Includes bibliography. / Electronic reproduction. Boca Raton, Fla., 2011. Mode of access: World Wide Web.

The subject of this thesis is the asymptotic behaviour of generating functions of different combinatorial models of two-dimensional lattice walks and polygons, enumerated with respect to different parameters, such as perimeter, number of steps and area. These models occur in various applications in physics, computer science and biology. In particular, they can be seen as simple models of biological vesicles or polymers. Of particular interest is the singular behaviour of the generating functions around special, so-called multicritical points in their parameter space, which correspond physically to phase transitions. The singular behaviour around the multicritical point is described by a scaling function, alongside a small set of critical exponents. Apart from some non-rigorous heuristics, our asymptotic analysis mainly consists in applying the method of steepest descents to a suitable integral expression for the exact solution for the generating function of a given model. The similar mathematical structure of the exact solutions of the different models allows for a unified treatment. In the saddle point analysis, the multicritical points correspond to points in the parameter space at which several saddle points of the integral kernels coalesce. Generically, two saddle points coalesce, in which case the scaling function is expressible in terms of the Airy function. As we will see, this is the case for Dyck and Schröder paths, directed column-convex polygons and partially directed self-avoiding walks. The result for Dyck paths also allows for the scaling analysis of Bernoulli meanders (also known as ballot paths). We then construct the model of deformed Dyck paths, where three saddle points coalesce in the corresponding integral kernel, thereby leading to an asymptotic expression in terms of a bivariate, generalised Airy integral.

Thesis (Ph.D.)--Massachusetts Institute of Technology, Dept. of Mathematics, 1979. / MICROFICHE COPY AVAILABLE IN ARCHIVES AND SCIENCE. / Vita. / Bibliography: leaves 74-75. / Ph.D.

In this thesis we make four contributions to the state of the art in numerical lattice simulations of quantum chromodynamics (QCD). First, we present the most detailed investigation yet of the autocorrelations of topological observations in hybrid Monte Carlo simulations of QCD and of the effects of the boundary conditions on these autocorrelations. This results in a numerical criterion for deciding when open boundary conditions are useful for reducing these autocorrelations, which are a major barrier to reliable calculations at fine lattice spacings. Second, we develop a dislocation-enhancing determinant, and demonstrate that it reduces the autocorrelation time of the topological charge. This alleviates problems with slow topological tunneling at fine lattice spacings, enabling simulations on fine lattices to be completed with much less computational effort. Third, we show how to apply the recently developed zMöbius technique to hybrid Monte Carlo evolutions with domain wall fermions, achieving nearly a factor of two speedup in the the light quark determinant, the single most expensive part of the calculation. The dislocation-enhancing determinant and the zMöbius technique have enabled us to begin simulations of fine ensembles with four flavors of dynamical domain wall quarks. Finally, we show how to include the previously-neglected G1 operator in nonperturbative renormalization of the ∆S = 1 effective weak Hamiltonian on the lattice. This removes an important systematic error in lattice calculations of weak matrix elements, in particular the important K → ππ decay.

Controlling fields and properties has attracted ever increasing interest over past decades due to the rapid advancement of nanofabrication techniques. In the field of nano-optics, to overcome the limit of signal processing speed and device scale of traditional electronic devices, optical devices using photon as the signal carriers have been chosen as the potential candidates. However, the diffraction limit of light has limited the integration of the micro-meter photonic components into electronic chips. Plasmonics offer the possibility to control electromagnetic fields at the subwavelength scale. Moreover , this controlling become tunable by introducing gradient into the material and/or structure, i.e., taking the concept of functionally graded materials (FGM) to design materials. / Gradon confinements in graded materials and/or systems open a door for tunable fields-controlling, which have potential applications in a variety of fields. Our research methods and results provide an effective way to understand field localization in a variety of systems, and they can be applied to design and manufacture thermal devices and even on-chip plasmonic-optical devices. / Gradon confinements, or referred as frequency-controlled localization of fields are investigated in various graded plasmonic lattices. The correspondences between gradon confinements and Bloch oscillations as well as nonBloch oscillations are explored. By taking into account retardation and loss effects, the asymmetric localization behavior and broadband localizat ion due to graded host permittivity are studied. / This thesis will concentrate on gradon confinements, which make controlling fields and properties tunable in graded materials and/or systems. We start with investigating gradon modes and their properties in graded elastic lattices. Using the quantum-classical analogue method, the analytic envelope function is obtained and can be used to analyze the system-size dependence of inverse participation ratio of gradon modes. In damping graded elastic lattices , the frequency-dependent behavior of relaxation rate are studied analytically and numerically. / We continue to study the three-dimensional graded plasmonic lattices with fully retarded electromagnetic interactions. A generalized Ewald-Kornfeld summation formula is developed to deal with the long-range interaction. In the quasistatic limit, various plasmonic gradon modes are investigated. Taking retardation and loss into account, field localization and enhancement are calculated in three-dimensional graded plasmonic lattices with graded size, spacing, and/or host permittivity in one direction. / Zheng, Mingjie = 弹性和等离子体梯度子禁闭研究 / 郑明杰. / Adviser: Kin Wah Yu. / Source: Dissertation Abstracts International, Volume: 72-11, Section: B, page: . / Thesis (Ph.D.)--Chinese University of Hong Kong, 2009. / Includes bibliographical references (leaves 117-124) and index. / Electronic reproduction. Hong Kong : Chinese University of Hong Kong, [2012] System requirements: Adobe Acrobat Reader. Available via World Wide Web. / Electronic reproduction. [Ann Arbor, MI] : ProQuest Information and Learning, [201-] System requirements: Adobe Acrobat Reader. Available via World Wide Web. / Abstract also in Chinese. / Zheng, Mingjie = Tan xing he deng li zi ti ti du zi jin bi yan jiu / Zheng Mingjie.

La cryptographie sur les réseaux est l’une des approches les plus compétitives pour protéger la confidentialité, dans les applications actuelles et l’ère post-quantique. Le problème central qui sert de fondement de complexité de la cryptographie sur réseaux est Learning with Errors (LWE). Il consiste à résoudre un système d’équations bruité, linéaire et surdéterminé. Ce problème est au moins aussi difficile que les problèmes standards portant sur les réseaux, tels que le décodage à distance bornée (BDD pour Bounded Distance Decoding) et le problème du vecteur le plus court unique (uSVP pour unique Shortest Vector Problem). Tous ces problèmes sont conjecturés difficiles à résoudre, même avec un ordinateur quantique de grande échelle. En particulier, le meilleur algorithme connu pour résoudre ces problèmes, BKZ, est très coûteux. Dans cette thèse, nous étudions les relations de difficulté entre BDD et uSVP, la difficulté quantique de LWE et les performances pratiques de l’algorithme BKZ. Tout d’abord, nous donnons une relation de difficulté plus étroite entre BDD et uSVP. Plus précisément, nous améliorons la réduction de BDD à uSVP d’un facteur √2, comparément à celle de Lyubashevsky et Micciancio. Ensuite, Nous apportons un nouvel élément à la conjecture que LWE est quantiquement difficile. Concrètement, nous considérons une version relâchée de la version quantique du problème du coset dièdral et montrons une équivalence computationnelle entre LWE et ce problème. Enfin, nous proposons un nouveau simulateur pour BKZ. Dans ce dernier travail, nous proposons le premier simulateur probabiliste pour BKZ, qui permet de prévoir le comportement pratique de BKZ très précisément. / Lattice-based cryptography is one of the most competitive candidates for protecting privacy, both in current applications and post quantum period. The central problem that serves as the hardness foundation of lattice-based cryptography is called the Learning with Errors (LWE). It asks to solve a noisy equation system, which is linear and over-determined modulo q. Normally, we call LWE problem as an average-case problem as all the coefficients in the equation system are randomly chosen modulo q. The LWE problem is conjectured to be hard even wtih a large scale quantum computer. It is at least as hard as standard problems defined in the lattices, such as Bounded Distance Decoding (BDD) and unique Shortest Vector Problem (uSVP). Finally, the best known algorithm for solving these problems is BKZ, which is very expensive. In this thesis, we study the quantum hardness of LWE, the hardness relations between the underlying problems BDD and uSVP, and the practical performance of the BKZ algorithm. First, we give a strong evidence of quantum hardness of LWE. Concretely, we consider a relaxed version of the quantum version of dihedral coset problem and show an computational equivalence between LWE and this problem. Second, we tighten the hardness relation between BDD and uSVP. More precisely, We improve the reduction from BDD to uSVP by a factor √2, compared to the one by Lyubashevsky and Micciancio. Third, we propose a more precise simulator for BKZ. In the last work, we propose the first probabilistic simulotor for BKZ, which can pridict the practical behavior of BKZ very precisely.

This thesis uses a variety of numerical and statistical techniques to perform high precision calculations in high energy physics using quantum field theory. It introduces the experimental motivation for the calculation of B meson form factors and includes a discussion of previous work. It then describes the modern theoretical framework describing these phenomena, outlining quantum chromodynamics and electroweak theory, and then illustrating the procedure of gauge fixing, the quantum effective action and background field gauge which is required for subsequent perturbative work. Details of the basic methodology of lattice quantum field theory are given as well as the specific formulation of the relativistic theory and nonrelativistic approximations used in this work to describe quantum chromodynamics. A comprehensive calculation of the zero recoil B to D* form factor is then presented, using state of the art lattice techniques with relativistic charm sea quarks and light sea quarks with correct physical masses, leading to a discussion of the dominant sources of uncertainty and possible resolutions of experimental tensions. Also included is preliminary work towards the full calculation of nonzero recoil matrix elements, with the aim of outlining possible future work. Finally, this thesis presents the computation of parameters correcting for radiative one loop phenomena and corrections to the kinetic coupling parameters in nonrelativistic quantum chromodynamics in order to achieve a desirable level of precision in future calculations. This is done using Monte-Carlo integration to evaluate integrals from diagrams generated using automated lattice perturbation theory in background field gauge in order to match the coefficients of the effective action between the lattice and the continuum.

In this work we explore how lattice gauge theory stands to benefit from new developments in machine learning, and look at two specific examples that illustrate this point. We begin with a brief overview of selected topics in machine learning for those who may be unfamiliar, and provide a simple example that helps to show how these ideas are carried out in practice. After providing the relevant background information, we then introduce an example of renormalization group (RG) transformations, inspired by the tensor RG, that can be used for arbitrary image sets, and look at applying this idea to equilibrium configurations of the two-dimensional Ising model. The second main idea presented in this thesis involves using machine learning to improve the efficiency of Markov Chain Monte Carlo (MCMC) methods. Explicitly, we describe a new technique for performing Hamiltonian Monte Carlo (HMC) simulations using an alternative leapfrog integrator that is parameterized by weights in a neural network. This work is based on the L2HMC ('Learning to Hamiltonian Monte Carlo') algorithm introduced in [1].

I have used the Embedded Atom Method (EAM) to investigate the vibrational behaviors of a large number of metallic systems. The systems examined are the bulk bcc metals Li, Na, K, Rb, Cs, Nb, Ta, Mo, W, and Fe, the bulk fcc metals Ni, Cu, and Al, the (100), (110), (111), and (211) surfaces of the Li, Na, K, Rb, and Cs, and the (100), (110), and (111) surfaces of Ni and Cu. I have conducted a more detailed and extensive review of existing EAM models and their ability to characterize bulk vibrational behavior than has ever previously been reported. I show the ability of an EAM model to quantitatively predict the vibrational properties of the bulk alkali metals in excellent agreement with experiment. The present work remedies a lack of computational investigation into bcc metallic surfaces by performing lattice dynamics calculations of the (110), (100), (111), and (211) alkali metal surfaces. Additionally, I present lattice dynamics calculations on the (111), (100), and (110) surfaces of Cu and Ni. An accurate set of surface Debye temperatures for these metal surfaces has been calculated. The extensive number of metals and planar geometries studied has enabled the identification and clarification of general relationships between surface phonons, surface coordination, and atomic density. The changes in vibrational behavior due to the truncation of the bulk near a surface can be understood by the consideration of three things: the vibrational behavior of a 1-D chain of harmonic oscillators, the bulk dispersion relation in the direction perpendicular to a surface, and the atomic coordination of near surface atoms. In general, relaxation causes force constants between atoms to stiffen, resulting in higher vibrational frequencies. The impact of stiffening on the vibrational characteristics depends largely on the surface geometry, as well as the particular properties of the metal. It can cause new surface modes and resonances, or cause surface vibrations to be more strongly coupled to the vibrations of bulk atoms.