855 resultados para Asynchronous iterative algorithms


Relevância:

20.00% 20.00%

Publicador:

Resumo:

En nuestro proyecto anterior aproximamos el cálculo de una integral definida con integrandos de grandes variaciones funcionales. Nuestra aproximación paraleliza el algoritmo de cómputo de un método adaptivo de cuadratura, basado en reglas de Newton-Cote. Los primeros resultados obtenidos fueron comunicados en distintos congresos nacionales e internacionales; ellos nos permintieron comenzar con una tipificación de las reglas de cuadratura existentes y una clasificación de algunas funciones utilizadas como funciones de prueba. Estas tareas de clasificación y tipificación no las hemos finalizado, por lo que pretendemos darle continuidad a fin de poder informar sobre la conveniencia o no de utilizar nuestra técnica. Para llevar adelante esta tarea se buscará una base de funciones de prueba y se ampliará el espectro de reglas de cuadraturas a utilizar. Además, nos proponemos re-estructurar el cálculo de algunas rutinas que intervienen en el cómputo de la mínima energía de una molécula. Este programa ya existe en su versión secuencial y está modelizado utilizando la aproximación LCAO. El mismo obtiene resultados exitosos en cuanto a precisión, comparado con otras publicaciones internacionales similares, pero requiere de un tiempo de cálculo significativamente alto. Nuestra propuesta es paralelizar el algoritmo mencionado abordándolo al menos en dos niveles: 1- decidir si conviene distribuir el cálculo de una integral entre varios procesadores o si será mejor distribuir distintas integrales entre diferentes procesadores. Debemos recordar que en los entornos de arquitecturas paralelas basadas en redes (típicamente redes de área local, LAN) el tiempo que ocupa el envío de mensajes entre los procesadores es muy significativo medido en cantidad de operaciones de cálculo que un procesador puede completar. 2- de ser necesario, paralelizar el cálculo de integrales dobles y/o triples. Para el desarrollo de nuestra propuesta se desarrollarán heurísticas para verificar y construir modelos en los casos mencionados tendientes a mejorar las rutinas de cálculo ya conocidas. A la vez que se testearán los algoritmos con casos de prueba. La metodología a utilizar es la habitual en Cálculo Numérico. Con cada propuesta se requiere: a) Implementar un algoritmo de cálculo tratando de lograr versiones superadoras de las ya existentes. b) Realizar los ejercicios de comparación con las rutinas existentes para confirmar o desechar una mejor perfomance numérica. c) Realizar estudios teóricos de error vinculados al método y a la implementación. Se conformó un equipo interdisciplinario integrado por investigadores tanto de Ciencias de la Computación como de Matemática. Metas a alcanzar Se espera obtener una caracterización de las reglas de cuadratura según su efectividad, con funciones de comportamiento oscilatorio y con decaimiento exponencial, y desarrollar implementaciones computacionales adecuadas, optimizadas y basadas en arquitecturas paralelas.

Relevância:

20.00% 20.00%

Publicador:

Resumo:

As digital imaging processing techniques become increasingly used in a broad range of consumer applications, the critical need to evaluate algorithm performance has become recognised by developers as an area of vital importance. With digital image processing algorithms now playing a greater role in security and protection applications, it is of crucial importance that we are able to empirically study their performance. Apart from the field of biometrics little emphasis has been put on algorithm performance evaluation until now and where evaluation has taken place, it has been carried out in a somewhat cumbersome and unsystematic fashion, without any standardised approach. This paper presents a comprehensive testing methodology and framework aimed towards automating the evaluation of image processing algorithms. Ultimately, the test framework aims to shorten the algorithm development life cycle by helping to identify algorithm performance problems quickly and more efficiently.

Relevância:

20.00% 20.00%

Publicador:

Resumo:

Magdeburg, Univ., Fak. für Mathematik, Habil.-Schr., 2006

Relevância:

20.00% 20.00%

Publicador:

Resumo:

This work describes a test tool that allows to make performance tests of different end-to-end available bandwidth estimation algorithms along with their different implementations. The goal of such tests is to find the best-performing algorithm and its implementation and use it in congestion control mechanism for high-performance reliable transport protocols. The main idea of this paper is to describe the options which provide available bandwidth estimation mechanism for highspeed data transport protocols and to develop basic functionality of such test tool with which it will be possible to manage entities of test application on all involved testing hosts, aided by some middleware.

Relevância:

20.00% 20.00%

Publicador:

Resumo:

In this paper we investigate various algorithms for performing Fast Fourier Transformation (FFT)/Inverse Fast Fourier Transformation (IFFT), and proper techniques for maximizing the FFT/IFFT execution speed, such as pipelining or parallel processing, and use of memory structures with pre-computed values (look up tables -LUT) or other dedicated hardware components (usually multipliers). Furthermore, we discuss the optimal hardware architectures that best apply to various FFT/IFFT algorithms, along with their abilities to exploit parallel processing with minimal data dependences of the FFT/IFFT calculations. An interesting approach that is also considered in this paper is the application of the integrated processing-in-memory Intelligent RAM (IRAM) chip to high speed FFT/IFFT computing. The results of the assessment study emphasize that the execution speed of the FFT/IFFT algorithms is tightly connected to the capabilities of the FFT/IFFT hardware to support the provided parallelism of the given algorithm. Therefore, we suggest that the basic Discrete Fourier Transform (DFT)/Inverse Discrete Fourier Transform (IDFT) can also provide high performances, by utilizing a specialized FFT/IFFT hardware architecture that can exploit the provided parallelism of the DFT/IDF operations. The proposed improvements include simplified multiplications over symbols given in polar coordinate system, using sinе and cosine look up tables, and an approach for performing parallel addition of N input symbols.

Relevância:

20.00% 20.00%

Publicador:

Resumo:

In this paper we investigate various algorithms for performing Fast Fourier Transformation (FFT)/Inverse Fast Fourier Transformation (IFFT), and proper techniquesfor maximizing the FFT/IFFT execution speed, such as pipelining or parallel processing, and use of memory structures with pre-computed values (look up tables -LUT) or other dedicated hardware components (usually multipliers). Furthermore, we discuss the optimal hardware architectures that best apply to various FFT/IFFT algorithms, along with their abilities to exploit parallel processing with minimal data dependences of the FFT/IFFT calculations. An interesting approach that is also considered in this paper is the application of the integrated processing-in-memory Intelligent RAM (IRAM) chip to high speed FFT/IFFT computing. The results of the assessment study emphasize that the execution speed of the FFT/IFFT algorithms is tightly connected to the capabilities of the FFT/IFFT hardware to support the provided parallelism of the given algorithm. Therefore, we suggest that the basic Discrete Fourier Transform (DFT)/Inverse Discrete Fourier Transform (IDFT) can also provide high performances, by utilizing a specialized FFT/IFFT hardware architecture that can exploit the provided parallelism of the DFT/IDF operations. The proposed improvements include simplified multiplications over symbols given in polar coordinate system, using sinе and cosine look up tables,and an approach for performing parallel addition of N input symbols.

Relevância:

20.00% 20.00%

Publicador:

Resumo:

Some practical aspects of Genetic algorithms’ implementation regarding to life cycle management of electrotechnical equipment are considered.

Relevância:

20.00% 20.00%

Publicador:

Resumo:

It is common to find in experimental data persistent oscillations in the aggregate outcomes and high levels of heterogeneity in individual behavior. Furthermore, it is not unusual to find significant deviations from aggregate Nash equilibrium predictions. In this paper, we employ an evolutionary model with boundedly rational agents to explain these findings. We use data from common property resource experiments (Casari and Plott, 2003). Instead of positing individual-specific utility functions, we model decision makers as selfish and identical. Agent interaction is simulated using an individual learning genetic algorithm, where agents have constraints in their working memory, a limited ability to maximize, and experiment with new strategies. We show that the model replicates most of the patterns that can be found in common property resource experiments.

Relevância:

20.00% 20.00%

Publicador:

Resumo:

"Vegeu el resum a l'inici del fitxer adjunt."

Relevância:

20.00% 20.00%

Publicador:

Relevância:

20.00% 20.00%

Publicador:

Resumo:

We study the properties of the well known Replicator Dynamics when applied to a finitely repeated version of the Prisoners' Dilemma game. We characterize the behavior of such dynamics under strongly simplifying assumptions (i.e. only 3 strategies are available) and show that the basin of attraction of defection shrinks as the number of repetitions increases. After discussing the difficulties involved in trying to relax the 'strongly simplifying assumptions' above, we approach the same model by means of simulations based on genetic algorithms. The resulting simulations describe a behavior of the system very close to the one predicted by the replicator dynamics without imposing any of the assumptions of the analytical model. Our main conclusion is that analytical and computational models are good complements for research in social sciences. Indeed, while on the one hand computational models are extremely useful to extend the scope of the analysis to complex scenar

Relevância:

20.00% 20.00%

Publicador:

Resumo:

The algorithmic approach to data modelling has developed rapidly these last years, in particular methods based on data mining and machine learning have been used in a growing number of applications. These methods follow a data-driven methodology, aiming at providing the best possible generalization and predictive abilities instead of concentrating on the properties of the data model. One of the most successful groups of such methods is known as Support Vector algorithms. Following the fruitful developments in applying Support Vector algorithms to spatial data, this paper introduces a new extension of the traditional support vector regression (SVR) algorithm. This extension allows for the simultaneous modelling of environmental data at several spatial scales. The joint influence of environmental processes presenting different patterns at different scales is here learned automatically from data, providing the optimum mixture of short and large-scale models. The method is adaptive to the spatial scale of the data. With this advantage, it can provide efficient means to model local anomalies that may typically arise in situations at an early phase of an environmental emergency. However, the proposed approach still requires some prior knowledge on the possible existence of such short-scale patterns. This is a possible limitation of the method for its implementation in early warning systems. The purpose of this paper is to present the multi-scale SVR model and to illustrate its use with an application to the mapping of Cs137 activity given the measurements taken in the region of Briansk following the Chernobyl accident.

Relevância:

20.00% 20.00%

Publicador:

Resumo:

In this paper, we develop numerical algorithms that use small requirements of storage and operations for the computation of invariant tori in Hamiltonian systems (exact symplectic maps and Hamiltonian vector fields). The algorithms are based on the parameterization method and follow closely the proof of the KAM theorem given in [LGJV05] and [FLS07]. They essentially consist in solving a functional equation satisfied by the invariant tori by using a Newton method. Using some geometric identities, it is possible to perform a Newton step using little storage and few operations. In this paper we focus on the numerical issues of the algorithms (speed, storage and stability) and we refer to the mentioned papers for the rigorous results. We show how to compute efficiently both maximal invariant tori and whiskered tori, together with the associated invariant stable and unstable manifolds of whiskered tori. Moreover, we present fast algorithms for the iteration of the quasi-periodic cocycles and the computation of the invariant bundles, which is a preliminary step for the computation of invariant whiskered tori. Since quasi-periodic cocycles appear in other contexts, this section may be of independent interest. The numerical methods presented here allow to compute in a unified way primary and secondary invariant KAM tori. Secondary tori are invariant tori which can be contracted to a periodic orbit. We present some preliminary results that ensure that the methods are indeed implementable and fast. We postpone to a future paper optimized implementations and results on the breakdown of invariant tori.

Relevância:

20.00% 20.00%

Publicador:

Resumo:

Asynchronous exponential growth has been extensively studied in population dynamics. In this paper we find out the asymptotic behaviour in a non-linear age-dependent model which takes into account sexual reproduction interactions. The main feature of our model is that the non-linear process converges to a linear one as the solution becomes large, so that the population undergoes asynchronous growth. The steady states analysis and the corresponding stability analysis are completely made and are summarized in a bifurcation diagram according to the parameter R0. Furthermore the effect of intraspecific competition is taken into account, leading to complex dynamics around steady states.