Improving ultimate convergence of an augmented Lagrangian method

Optimization methods that employ the classical Powell-Hestenes-Rockafellar augmented Lagrangian are useful tools for solving nonlinear programming problems. Their reputation decreased in the last 10 years due to the comparative success of interior-point Newtonian algorithms, which are asymptotically faster. In this research, a combination of both approaches is evaluated. The idea is to produce a competitive method, being more robust and efficient than its `pure` counterparts for critical problems. Moreover, an additional hybrid algorithm is defined, in which the interior-point method is replaced by the Newtonian resolution of a Karush-Kuhn-Tucker (KKT) system identified by the augmented Lagrangian algorithm. The software used in this work is freely available through the Tango Project web page:http://www.ime.usp.br/similar to egbirgin/tango/.

Modelling the secular evolution of migrating planet pairs

The subject of this paper is the secular behaviour of a pair of planets evolving under dissipative forces. In particular, we investigate the case when dissipative forces affect the planetary semimajor axes and the planets move inwards/outwards the central star, in a process known as planet migration. To perform this investigation, we introduce fundamental concepts of conservative and dissipative dynamics of the three-body problem. Based on these concepts, we develop a qualitative model of the secular evolution of the migrating planetary pair. Our approach is based on the analysis of the energy and the orbital angular momentum exchange between the two-planet system and an external medium; thus no specific kind of dissipative forces is invoked. We show that, under the assumption that dissipation is weak and slow, the evolutionary routes of the migrating planets are traced by the Mode I and Mode II stationary solutions of the conservative secular problem. The ultimate convergence and the evolution of the system along one of these secular modes of motion are determined uniquely by the condition that the dissipation rate is sufficiently smaller than the proper secular frequency of the system. We show that it is possible to reassemble the starting configurations and the migration history of the systems on the basis of their final states and consequently to constrain the parameters of the physical processes involved.

Castoraeschna corbeti sp nov from Floresta Nacional de Carajas, Para state, Brazil (Odonata: Aeshnidae)

Castoraeschna corbeti sp. nov. is described and diagnosed based on four males (holotype: Brazil, Para State, Floresta Nacional cle Carajas [6 degrees 06`13.9 `` S, 50 degrees 08`13.1 `` W, ca 600 m a.s.l.], 28 ix 2007 to be deposited in Museu Nacional, Universidade Federal do Rio de Janeiro, Rio de Janeiro). This species is similar to C. longfieldae and C. coronata but can be distinguished mainly by the absence of medio-dorsal spots on S8; postero-dorsal spots on S8-9 very narrow; cerci external margin almost straight in lateral view, without a distinct angulation between stern and base of lamina; cerci apex blunt. The probable ultimate stadium larva is described based on two individuals, male and female, collected at the type locality. Adults were observed flying along margins of a small shaded second-order stream where the larvae were taken. The surrounding forest is under impact of iron ore extraction and will probably disappear in the next years.

On the influence of imputation in classification: practical issues

The substitution of missing values, also called imputation, is an important data preparation task for many domains. Ideally, the substitution of missing values should not insert biases into the dataset. This aspect has been usually assessed by some measures of the prediction capability of imputation methods. Such measures assume the simulation of missing entries for some attributes whose values are actually known. These artificially missing values are imputed and then compared with the original values. Although this evaluation is useful, it does not allow the influence of imputed values in the ultimate modelling task (e.g. in classification) to be inferred. We argue that imputation cannot be properly evaluated apart from the modelling task. Thus, alternative approaches are needed. This article elaborates on the influence of imputed values in classification. In particular, a practical procedure for estimating the inserted bias is described. As an additional contribution, we have used such a procedure to empirically illustrate the performance of three imputation methods (majority, naive Bayes and Bayesian networks) in three datasets. Three classifiers (decision tree, naive Bayes and nearest neighbours) have been used as modelling tools in our experiments. The achieved results illustrate a variety of situations that can take place in the data preparation practice.

The ""decarbonization"" of the world`s energy matrix

Decarbonizing the world`s energy matrix is the strategy being implemented by most countries to reduce CO(2) emissions and thus contribute to achieve the ultimate objectives of the Climate Convention. The evolution of the carbon intensity (I(c)=CO(2)/GDP) in the period 1990-2007 was encouraging but not sufficient to reduce the growth of carbon emission. As a result of COP-15 in Copenhagen these countries (and regions) made pledges that could lead to more reduction: for the United States a 17% reduction in CO(2) emissions by 2020 below the level of 2005: for the European Union a 20% reduction in CO(2) emissions by 2020 below the 1990 level: for China a 40-45% reduction in the carbon intensity and for India a 20-25% reduction in carbon intensity by 2020. We analyzed the consequences of such pledges and concluded that the expected yearly rate of decrease of the carbon intensity follows basically the ""business as usual"" trend in the period 1990-2007 and will, in all likelihood, be insufficient to reduce carbon emissions up to 2020. (C) 2010 Elsevier Ltd. All rights reserved.

Disorder-based graphene spintronics

The use of the spin of the electron as the ultimate logic bit-in what has been dubbed spintronics-can lead to a novel way of thinking about information flow. At the same time single-layer graphene has been the subject of intense research due to its potential application in nanoscale electronics. While defects can significantly alter the electronic properties of nanoscopic systems, the lack of control can lead to seemingly deleterious effects arising from the random arrangement of such impurities. Here we demonstrate, using ab initio density functional theory and non-equilibrium Green`s functions calculations, that it is possible to obtain perfect spin selectivity in doped graphene nanoribbons to produce a perfect spin filter. We show that initially unpolarized electrons entering the system give rise to 100% polarization of the current due to random disorder. This effect is explained in terms of different localization lengths for each spin channel which leads to a new mechanism for the spin filtering effect that is disorder-driven.

Symmetry controlled spin polarized conductance in Au nanowires

The fact that the resistance of propagating electrons in solids depends on their spin orientation has led to a new field called spintronics. With the parallel advances in nanoscience, it is now possible to talk about nanospintronics. Many works have focused on the study of charge transport along nanosystems, such as carbon nanotubes, graphene nanoribbons, or metallic nanowires, and spin dependent transport properties at this scale may lead to new behaviors due to the manipulation of a small number of spins. Metal nanowires have been studied as electric contacts where atomic and molecular insertions can be constructed. Here we describe what might be considered the ultimate spin device, namely, a Au thin nanowire with one Co atom bridging its two sides. We show that this system has strong spin dependent transport properties and that its local symmetry can dramatically change them, leading to a significant spin polarized conductance.

On the lower bound on the exchange-correlation energy in two dimensions

We study the properties of the lower bound on the exchange-correlation energy in two dimensions. First we review the derivation of the bound and show how it can be written in a simple density-functional form. This form allows an explicit determination of the prefactor of the bound and testing its tightness. Next we focus on finite two-dimensional systems and examine how their distance from the bound depends on the system geometry. The results for the high-density limit suggest that a finite system that comes as close as possible to the ultimate bound on the exchange-correlation energy has circular geometry and a weak confining potential with a negative curvature. (c) 2009 Elsevier B.V. All rights reserved.