Optimal network topologies for local search with congestion

The problem of searchability in decentralized complex networks is of great importance in computer science, economy, and sociology. We present a formalism that is able to cope simultaneously with the problem of search and the congestion effects that arise when parallel searches are performed, and we obtain expressions for the average search cost both in the presence and the absence of congestion. This formalism is used to obtain optimal network structures for a system using a local search algorithm. It is found that only two classes of networks can be optimal: starlike configurations, when the number of parallel searches is small, and homogeneous-isotropic configurations, when it is large.

Subclass problem-dependent design for error-correcting output codes

A common way to model multiclass classification problems is by means of Error-Correcting Output Codes (ECOCs). Given a multiclass problem, the ECOC technique designs a code word for each class, where each position of the code identifies the membership of the class for a given binary problem. A classification decision is obtained by assigning the label of the class with the closest code. One of the main requirements of the ECOC design is that the base classifier is capable of splitting each subgroup of classes from each binary problem. However, we cannot guarantee that a linear classifier model convex regions. Furthermore, nonlinear classifiers also fail to manage some type of surfaces. In this paper, we present a novel strategy to model multiclass classification problems using subclass information in the ECOC framework. Complex problems are solved by splitting the original set of classes into subclasses and embedding the binary problems in a problem-dependent ECOC design. Experimental results show that the proposed splitting procedure yields a better performance when the class overlap or the distribution of the training objects conceal the decision boundaries for the base classifier. The results are even more significant when one has a sufficiently large training size.

Optimal fiscal policy in overlapping generations models

[eng] This paper provides, from a theoretical and quantitative point of view, an explanation of why taxes on capital returns are high (around 35%) by analyzing the optimal fiscal policy in an economy with intergenerational redistribution. For this purpose, the government is modeled explicitly and can choose (and commit to) an optimal tax policy in order to maximize society's welfare. In an infinitely lived economy with heterogeneous agents, the long run optimal capital tax is zero. If heterogeneity is due to the existence of overlapping generations, this result in general is no longer true. I provide sufficient conditions for zero capital and labor taxes, and show that a general class of preferences, commonly used on the macro and public finance literature, violate these conditions. For a version of the model, calibrated to the US economy, the main results are: first, if the government is restricted to a set of instruments, the observed fiscal policy cannot be disregarded as sub optimal and capital taxes are positive and quantitatively relevant. Second, if the government can use age specific taxes for each generation, then the age profile capital tax pattern implies subsidizing asset returns of the younger generations and taxing at higher rates the asset returns of the older ones.

Interiority of the optimal population growth rate with endogenous fertility

This paper analyzes the issue of the interiority of the optimal population growth rate in a two-period overlapping generations model with endogenous fertility. Using Cobb-Douglas utility and production functions, we show that the introduction of a cost of raising children allows for the possibility of the existence of an interior global maximum in the planner¿s problem, contrary to the exogenous fertility case

Performance Analysis of Two Quantum Reaction Dynamics Codes: Time-Dependent and Time-Independent Strategies

The computer simulation of reaction dynamics has nowadays reached a remarkable degree of accuracy. Triatomic elementary reactions are rigorously studied with great detail on a straightforward basis using a considerable variety of Quantum Dynamics computational tools available to the scientific community. In our contribution we compare the performance of two quantum scattering codes in the computation of reaction cross sections of a triatomic benchmark reaction such as the gas phase reaction Ne + H2+ %12. NeH++ H. The computational codes are selected as representative of time-dependent (Real Wave Packet [ ]) and time-independent (ABC [ ]) methodologies. The main conclusion to be drawn from our study is that both strategies are, to a great extent, not competing but rather complementary. While time-dependent calculations advantages with respect to the energy range that can be covered in a single simulation, time-independent approaches offer much more detailed information from each single energy calculation. Further details such as the calculation of reactivity at very low collision energies or the computational effort related to account for the Coriolis couplings are analyzed in this paper.

Optimal stop-loss reinsurance: a dependence analysis

The stop-loss reinsurance is one of the most important reinsurance contracts in the insurance market. From the insurer point of view, it presents an interesting property: it is optimal if the criterion of minimizing the variance of the cost of the insurer is used. The aim of the paper is to contribute to the analysis of the stop-loss contract in one period from the point of view of the insurer and the reinsurer. Firstly, the influence of the parameters of the reinsurance contract on the correlation coefficient between the cost of the insurer and the cost of the reinsurer is studied. Secondly, the optimal stop-loss contract is obtained if the criterion used is the maximization of the joint survival probability of the insurer and the reinsurer in one period.

Product Perfect Codes and Steganography

A new coding technique to be used in steganography is evaluated. The performanceof this new technique is computed and comparisons with the well-known theoreticalupper bound, Hamming upper bound and basic LSB are established.

Construction and performance of network codes

El principal objectiu d'aquest treball és implementar i exposar una descripció teòrica per a diferents esquemes de Physical Layer Network Coding. Utilitzant un esquema bàsic com a punt de partida, el projecte presenta la construcció i l'anàlisis de diferents esquemes de comunicació on la complexitat va augmentant a mesura que anem avançant en el projecte. El treball està estructurat en diferents parts: primer, es presenta una introducció a Physical Layer Network Coding i a Lattice Network Codes. A continuació, s'introdueixen les eines matemàtiques necessàries per entendre el CF System. Després, s'analitza i implementa el primer esquema bàsic. A partir del qual, implementem una versió vectorial del CF System i una versió codificada amb un Hamming q-ari. Finalment, s'estudien i implementen diferents estratègies per millorar la matriu de coeficients A.

Optimal capital structure: a trade-off model and its empirical testing

In the present research we have set forth a new, simple, Trade-Off model that would allow us to calculate how much debt and, by default, how much equity a company should have, using easily available information and calculating the cost of debt dynamically on the basis of the effect that the capital structure of the company has on the risk of bankruptcy; in an attempt to answer this question. The proposed model has been applied to the companies that make up the Dow Jones Industrial Average (DJIA) in 2007. We have used consolidated financial data from 1996 to 2006, published by Bloomberg. We have used simplex optimization method to find the debt level that maximizes firm value. Then, we compare the estimated debt with real debt of companies using statistical nonparametric Mann-Whitney. The results indicate that 63% of companies do not show a statistically significant difference between the real and the estimated debt.

Preferred Spatial Frequencies for Human Face Processing Are Associated with Optimal Class Discrimination in the Machine

Psychophysical studies suggest that humans preferentially use a narrow band of low spatial frequencies for face recognition. Here we asked whether artificial face recognition systems have an improved recognition performance at the same spatial frequencies as humans. To this end, we estimated recognition performance over a large database of face images by computing three discriminability measures: Fisher Linear Discriminant Analysis, Non-Parametric Discriminant Analysis, and Mutual Information. In order to address frequency dependence, discriminabilities were measured as a function of (filtered) image size. All three measures revealed a maximum at the same image sizes, where the spatial frequency content corresponds to the psychophysical found frequencies. Our results therefore support the notion that the critical band of spatial frequencies for face recognition in humans and machines follows from inherent properties of face images, and that the use of these frequencies is associated with optimal face recognition performance.

Performance Analysis of Two Quantum Reaction Dynamics Codes: Time-Dependent and Time-Independent Strategies

The computer simulation of reaction dynamics has nowadays reached a remarkable degree of accuracy. Triatomic elementary reactions are rigorously studied with great detail on a straightforward basis using a considerable variety of Quantum Dynamics computational tools available to the scientific community. In our contribution we compare the performance of two quantum scattering codes in the computation of reaction cross sections of a triatomic benchmark reaction such as the gas phase reaction Ne + H2+ %12. NeH++ H. The computational codes are selected as representative of time-dependent (Real Wave Packet [ ]) and time-independent (ABC [ ]) methodologies. The main conclusion to be drawn from our study is that both strategies are, to a great extent, not competing but rather complementary. While time-dependent calculations advantages with respect to the energy range that can be covered in a single simulation, time-independent approaches offer much more detailed information from each single energy calculation. Further details such as the calculation of reactivity at very low collision energies or the computational effort related to account for the Coriolis couplings are analyzed in this paper.

Gene filtering with optimal threshold selection

Gene filtering is a useful preprocessing technique often applied to microarray datasets. However, it is no common practice because clear guidelines are lacking and it bears the risk of excluding some potentially relevant genes. In this work, we propose to model microarray data as a mixture of two Gaussian distributions that will allow us to obtain an optimal filter threshold in terms of the gene expression level.

Adquisició de les codes complexes finals del català per part d'alumnat panjabi: el mètode experimental

En aquest treball estudiem l’adquisició del català per part de parlants que tenen el panjabi com a primera llengua, concretament en relació amb les codes complexes de final de mot, per a aprenents d’un nivell inicial de català

Optimal Traffic Grooming in WDM using Lighttours

Recent developments in optical communications have allowed simpler optical devices to improve network resource utilization. As such, we propose adding a lambda-monitoring device to a wavelength-routing switch (WRS) allowing better performance when traffic is routed and groomed. This device may allow a WRS to aggregate traffic over optical routes without incurring in optical-electrical-optical conversion for the existing traffic. In other words, optical routes can be taken partially to route demands creating a sort of "lighttours". In this paper, we compare the number of OEO conversions needed to route a complete given traffic matrix using either lighttours or lightpaths

Improving the efficiency of PV low-power processing circuits by selecting an optimal inductor current of the DC/DC converter

In the context of autonomous sensors powered by small-size photovoltaic (PV) panels, this work analyses how the efficiency of DC/DC-converter-based power processing circuits can be improved by an appropriate selection of the inductor current that transfers the energy from the PV panel to a storage unit. Each component of power losses (fixed, conduction and switching losses) involved in the DC/DC converter specifically depends on the average inductor current so that there is an optimal value of this current that causes minimal losses and, hence, maximum efficiency. Such an idea has been tested experimentally using two commercial DC/DC converters whose average inductor current is adjustable. Experimental results show that the efficiency can be improved up to 12% by selecting an optimal value of that current, which is around 300-350 mA for such DC/DC converters.