Structured minimal-memory inexact quasi-Newton method and secant preconditioners for augmented Lagrangian optimization

Augmented Lagrangian methods for large-scale optimization usually require efficient algorithms for minimization with box constraints. On the other hand, active-set box-constraint methods employ unconstrained optimization algorithms for minimization inside the faces of the box. Several approaches may be employed for computing internal search directions in the large-scale case. In this paper a minimal-memory quasi-Newton approach with secant preconditioners is proposed, taking into account the structure of Augmented Lagrangians that come from the popular Powell-Hestenes-Rockafellar scheme. A combined algorithm, that uses the quasi-Newton formula or a truncated-Newton procedure, depending on the presence of active constraints in the penalty-Lagrangian function, is also suggested. Numerical experiments using the Cute collection are presented.

Uma adaptação do método barreira penalidade quasi-Newton ao problema de fluxo de potência ótimo

Coordenação de Aperfeiçoamento de Pessoal de Nível Superior (CAPES)

Local convergence of quasi-Newton methods under metric regularity

We consider quasi-Newton methods for generalized equations in Banach spaces under metric regularity and give a sufficient condition for q-linear convergence. Then we show that the well-known Broyden update satisfies this sufficient condition in Hilbert spaces. We also establish various modes of q-superlinear convergence of the Broyden update under strong metric subregularity, metric regularity and strong metric regularity. In particular, we show that the Broyden update applied to a generalized equation in Hilbert spaces satisfies the Dennis–Moré condition for q-superlinear convergence. Simple numerical examples illustrate the results.

Otimização Não Linear de Mínimos Quadrados

O problema de otimização de mínimos quadrados e apresentado como uma classe importante de problemas de minimização sem restrições. A importância dessa classe de problemas deriva das bem conhecidas aplicações a estimação de parâmetros no contexto das analises de regressão e de resolução de sistemas de equações não lineares. Apresenta-se uma revisão dos métodos de otimização de mínimos quadrados lineares e de algumas técnicas conhecidas de linearização. Faz-se um estudo dos principais métodos de gradiente usados para problemas não lineares gerais: Métodos de Newton e suas modificações incluindo os métodos Quasi-Newton mais usados (DFP e BFGS). Introduzem-se depois métodos específicos de gradiente para problemas de mínimos quadrados: Gauss-Newton e Levenberg-Larquardt. Apresenta-se uma variedade de exemplos selecionados na literatura para testar os diferentes métodos usando rotinas MATLAB. Faz-se uma an alise comparativa dos algoritmos baseados nesses ensaios computacionais que exibem as vantagens e desvantagens dos diferentes métodos.

Ajuste do modelo de Schumacher e Hall e aplicação de redes neurais artificiais para estimar volume de árvores de eucalipto

Objetivou-se, neste trabalho, avaliar o ajuste do modelo volumétrico de Schumacher e Hall por diferentes algoritmos, bem como a aplicação de redes neurais artificiais para estimação do volume de madeira de eucalipto em função do diâmetro a 1,30 m do solo (DAP), da altura total (Ht) e do clone. Foram utilizadas 21 cubagens de povoamentos de clones de eucalipto com DAP variando de 4,5 a 28,3 cm e altura total de 6,6 a 33,8 m, num total de 862 árvores. O modelo volumétrico de Schumacher e Hall foi ajustado nas formas linear e não linear, com os seguintes algoritmos: Gauss-Newton, Quasi-Newton, Levenberg-Marquardt, Simplex, Hooke-Jeeves Pattern, Rosenbrock Pattern, Simplex, Hooke-Jeeves e Rosenbrock, utilizado simultaneamente com o método Quasi-Newton e com o princípio da Máxima Verossimilhança. Diferentes arquiteturas e modelos (Multilayer Perceptron MLP e Radial Basis Function RBF) de redes neurais artificiais foram testados, sendo selecionadas as redes que melhor representaram os dados. As estimativas dos volumes foram avaliadas por gráficos de volume estimado em função do volume observado e pelo teste estatístico L&O. Assim, conclui-se que o ajuste do modelo de Schumacher e Hall pode ser usado na sua forma linear, com boa representatividade e sem apresentar tendenciosidade; os algoritmos Gauss-Newton, Quasi-Newton e Levenberg-Marquardt mostraram-se eficientes para o ajuste do modelo volumétrico de Schumacher e Hall, e as redes neurais artificiais apresentaram boa adequação ao problema, sendo elas altamente recomendadas para realizar prognose da produção de florestas plantadas.

Étude numérique des origines hémodynamiques des oscillations dans des réseaux de capillaires

En simulant l’écoulement du sang dans un réseau de capillaires (en l’absence de contrôle biologique), il est possible d’observer la présence d’oscillations de certains paramètres comme le débit volumique, la pression et l’hématocrite (volume des globules rouges par rapport au volume du sang total). Ce comportement semble être en concordance avec certaines expériences in vivo. Malgré cet accord, il faut se demander si les fluctuations observées lors des simulations de l’écoulement sont physiques, numériques ou un artefact de modèles irréalistes puisqu’il existe toujours des différences entre des modélisations et des expériences in vivo. Pour répondre à cette question de façon satisfaisante, nous étudierons et analyserons l’écoulement du sang ainsi que la nature des oscillations observées dans quelques réseaux de capillaires utilisant un modèle convectif et un modèle moyenné pour décrire les équations de conservation de masse des globules rouges. Ces modèles tiennent compte de deux effets rhéologiques importants : l’effet Fåhraeus-Lindqvist décrivant la viscosité apparente dans un vaisseau et l’effet de séparation de phase schématisant la distribution des globules rouges aux points de bifurcation. Pour décrire ce dernier effet, deux lois de séparation de phase (les lois de Pries et al. et de Fenton et al.) seront étudiées et comparées. Dans ce mémoire, nous présenterons une description du problème physiologique (rhéologie du sang). Nous montrerons les modèles mathématiques employés (moyenné et convectif) ainsi que les lois de séparation de phase (Pries et al. et Fenton et al.) accompagnés d’une analyse des schémas numériques implémentés. Pour le modèle moyenné, nous employons le schéma numérique explicite traditionnel d’Euler ainsi qu’un nouveau schéma implicite qui permet de résoudre ce problème d’une manière efficace. Ceci est fait en utilisant une méthode de Newton- Krylov avec gradient conjugué préconditionné et la méthode de GMRES pour les itérations intérieures ainsi qu’une méthode quasi-Newton (la méthode de Broyden). Cette méthode inclura le schéma implicite d’Euler et la méthode des trapèzes. Pour le schéma convectif, la méthode explicite de Kiani et al. sera implémentée ainsi qu’une nouvelle approche implicite. La stabilité des deux modèles sera également explorée. À l’aide de trois différentes topologies, nous comparerons les résultats de ces deux modèles mathématiques ainsi que les lois de séparation de phase afin de déterminer dans quelle mesure les oscillations observées peuvent être attribuables au choix des modèles mathématiques ou au choix des méthodes numériques.

A benchmark method for global optimization problems in structure determination from powder diffraction data

Quasi-Newton-Raphson minimization and conjugate gradient minimization have been used to solve the crystal structures of famotidine form B and capsaicin from X-ray powder diffraction data and characterize the chi(2) agreement surfaces. One million quasi-Newton-Raphson minimizations found the famotidine global minimum with a frequency of ca 1 in 5000 and the capsaicin global minimum with a frequency of ca 1 in 10 000. These results, which are corroborated by conjugate gradient minimization, demonstrate the existence of numerous pathways from some of the highest points on these chi(2) agreement surfaces to the respective global minima, which are passable using only downhill moves. This important observation has significant ramifications for the development of improved structure determination algorithms.

A variational method to retrieve the extinction profile in liquid clouds using multiple field-of-view lidar

Liquid clouds play a profound role in the global radiation budget but it is difficult to remotely retrieve their vertical profile. Ordinary narrow field-of-view (FOV) lidars receive a strong return from such clouds but the information is limited to the first few optical depths. Wideangle multiple-FOV lidars can isolate radiation scattered multiple times before returning to the instrument, often penetrating much deeper into the cloud than the singly-scattered signal. These returns potentially contain information on the vertical profile of extinction coefficient, but are challenging to interpret due to the lack of a fast radiative transfer model for simulating them. This paper describes a variational algorithm that incorporates a fast forward model based on the time-dependent two-stream approximation, and its adjoint. Application of the algorithm to simulated data from a hypothetical airborne three-FOV lidar with a maximum footprint width of 600m suggests that this approach should be able to retrieve the extinction structure down to an optical depth of around 6, and total opticaldepth up to at least 35, depending on the maximum lidar FOV. The convergence behavior of Gauss-Newton and quasi-Newton optimization schemes are compared. We then present results from an application of the algorithm to observations of stratocumulus by the 8-FOV airborne “THOR” lidar. It is demonstrated how the averaging kernel can be used to diagnose the effective vertical resolution of the retrieved profile, and therefore the depth to which information on the vertical structure can be recovered. This work enables exploitation of returns from spaceborne lidar and radar subject to multiple scattering more rigorously than previously possible.

Otimização Topológica com Termo-elasticidade Utilizando Elementos Tri3

This work proposes a formulation for optimization of 2D-structure layouts submitted to mechanic and thermal shipments and applied an h-adaptive filter process which conduced to computational low spend and high definition structural layouts. The main goal of the formulation is to minimize the structure mass submitted to an effective state of stress of von Mises, with stability and lateral restriction variants. A criterion of global measurement was used for intents a parametric condition of stress fields. To avoid singularity problems was considerate a release on the stress restriction. On the optimization was used a material approach where the homogenized constructive equation was function of the material relative density. The intermediary density effective properties were represented for a SIMP-type artificial model. The problem was simplified by use of the method of finite elements of Galerkin using triangles with linear Lagrangian basis. On the solution of the optimization problem, was applied the augmented Lagrangian Method, that consists on minimum problem sequence solution with box-type restrictions, resolved by a 2nd orderprojection method which uses the method of the quasi-Newton without memory, during the problem process solution. This process reduces computational expends showing be more effective and solid. The results materialize more refined layouts with accurate topologic and shape of structure definitions. On the other hand formulation of mass minimization with global stress criterion provides to modeling ready structural layouts, with violation of the criterion of homogeneous distributed stress

Otimização de forma aplicando B-splines sob critério integral de tensões

This work proposes a computational methodology to solve problems of optimization in structural design. The application develops, implements and integrates methods for structural analysis, geometric modeling, design sensitivity analysis and optimization. So, the optimum design problem is particularized for plane stress case, with the objective to minimize the structural mass subject to a stress criterion. Notice that, these constraints must be evaluated at a series of discrete points, whose distribution should be dense enough in order to minimize the chance of any significant constraint violation between specified points. Therefore, the local stress constraints are transformed into a global stress measure reducing the computational cost in deriving the optimal shape design. The problem is approximated by Finite Element Method using Lagrangian triangular elements with six nodes, and use a automatic mesh generation with a mesh quality criterion of geometric element. The geometric modeling, i.e., the contour is defined by parametric curves of type B-splines, these curves hold suitable characteristics to implement the Shape Optimization Method, that uses the key points like design variables to determine the solution of minimum problem. A reliable tool for design sensitivity analysis is a prerequisite for performing interactive structural design, synthesis and optimization. General expressions for design sensitivity analysis are derived with respect to key points of B-splines. The method of design sensitivity analysis used is the adjoin approach and the analytical method. The formulation of the optimization problem applies the Augmented Lagrangian Method, which convert an optimization problem constrained problem in an unconstrained. The solution of the Augmented Lagrangian function is achieved by determining the analysis of sensitivity. Therefore, the optimization problem reduces to the solution of a sequence of problems with lateral limits constraints, which is solved by the Memoryless Quasi-Newton Method It is demonstrated by several examples that this new approach of analytical design sensitivity analysis of integrated shape design optimization with a global stress criterion purpose is computationally efficient

Foraging Behavior of the Swarm-founding Wasp Polybia (Trichothorax) sericea (Hymenoptera, Vespidae): Daily Resource Collection Activity and Flight Capacity

Conselho Nacional de Desenvolvimento Científico e Tecnológico (CNPq)

Nest site selection and flying capacity of neotropical wasp Angiopolybia pallens (Hymenoptera: Vespidae) in the Atlantic Rain Forest, Bahia State, Brazil

In order to evaluate the flying capacity and nest site selection of Angiopolybia pallens (Lepeletier, 1836), we made 17 incursions (136 hours of sample efforts) in Atlantic Rain Forest environments in Bahia state. Our data show this wasp prefers to nest on wide leaves of bushes and short trees (nests between 0.30 and 3m from the ground) placed in half-shady environments (clearings and shadowed cultivations). The logistic regression model using Quasi-Newton method provided a good description of the flying capacity observed in A. pallens (x 2 = 91.52; p≪0.001). According to the logistic regression model, the A. pallens flight autonomy is low, flying for short distances and with an effective radius of action of about 24m measured from their nests, which means a foraging area of nearly 1,800 m 2.

New contributions to the determination of resonance frequencies of rectangular microstrip antennas by neural networks

This paper uses artificial neural networks (ANN) to compute the resonance frequencies of rectangular microstrip antennas (MSA), used in mobile communications. Perceptron Multi-layers (PML) networks were used, with the Quasi-Newton method proposed by Broyden, Fletcher, Goldfarb and Shanno (BFGS). Due to the nature of the problem, two hundred and fifty networks were trained, and the resonance frequency for each test antenna was calculated by statistical methods. The estimate resonance frequencies for six test antennas were compared with others results obtained by deterministic and ANN based empirical models from the literature, and presented a better agreement with the experimental values.

A simple analysis of the influence of the solvent-induced electronic polarization on the N-15 magnetic shielding of pyridine in water

Electronic polarization induced by the interaction of a reference molecule with a liquid environment is expected to affect the magnetic shielding constants. Understanding this effect using realistic theoretical models is important for proper use of nuclear magnetic resonance in molecular characterization. In this work, we consider the pyridine molecule in water as a model system to briefly investigate this aspect. Thus, Monte Carlo simulations and quantum mechanics calculations based on the B3LYP/6-311++G (d,p) are used to analyze different aspects of the solvent effects on the N-15 magnetic shielding constant of pyridine in water. This includes in special the geometry relaxation and the electronic polarization of the solute by the solvent. The polarization effect is found to be very important, but, as expected for pyridine, the geometry relaxation contribution is essentially negligible. Using an average electrostatic model of the solvent, the magnetic shielding constant is calculated as -58.7 ppm, in good agreement with the experimental value of -56.3 ppm. The explicit inclusion of hydrogen-bonded water molecules embedded in the electrostatic field of the remaining solvent molecules gives the value of -61.8 ppm.