Logo des Repositoriums

Auflistung nach Autor:in "Bui, Marc"

1 - 5 von 5
  • Konferenzbeitrag
    Consensus and relation networks
    (11th International Conference on Innovative Internet Community Systems (I2CS 2011), 2011) Ahat, Murat; Bennani, Omar; Bui, Marc; Amor, Soufian Ben
    In this paper, we propose an extension to bounded confidence model which is a nonlinear opinion dynamics model. The goal of this new extension is to emphasize the underlying network structure of the model, since in real life network structure plays an important , if not vital, role in opinion dynamics. And with simulation results, we show how networks, such as grid and scale-free, affect the consensus formation in bounded confidence model separately as well as together.
  • Konferenzbeitrag
    Data structures and algorithms for pretopology: the Java based software library PretopoLib
    (11th International Conference on Innovative Internet Community Systems (I2CS 2011), 2011) Levorato, Vincent; Bui, Marc
    PretopoLib (http://pretopolib.complexica.net/) is a library of data types and algorithms based on pretopology theory. Its interest is its data structures representing mathematical relations and neighborhood definitions. The main features of the library are a clear separation of computation and visualization, efficient sets data types, and its ease of use to build model or application prototypes in the complex systems simulation framework.
  • Konferenzbeitrag
    Generalizing of a high performance parallel Strassen implementation on distributed memory MIMD architectures
    (10th International Conferenceon Innovative Internet Community Systems (I2CS) – Jubilee Edition 2010 –, 2010) Nguyen, Duc Kien; Lavallee, Ivan; Bui, Marc
    Strassen's algorithm to multiply two n x n matrices reduces the asymptotic operation count from O(n3) of the traditional algorithm to O(n2.81), thus designing efficient parallelizing for this algorithm becomes essential. In this paper, we present our generalizing of a parallel Strassen implementation which obtained a very nice performance on an Intel Paragon: faster 20% for n ≈ 1000 and more than 100% for n ≈ 5000 in comparison to the parallel traditional algorithms (as Fox, Cannon). Our method can be applied to all the matrix multiplication algorithms on distributed memory computers that use Strassen's algorithm at the system level, hence it gives us compatibility to find better parallel implementations of Strassen's algorithm.
  • Konferenzbeitrag
    Modeling the complex dynamics of distributed communities of the web with pretopology
    (10th International Conferenceon Innovative Internet Community Systems (I2CS) – Jubilee Edition 2010 –, 2010) Levorato, Vincent; Bui, Marc
    The aim of this article is to present a methodological approach for problems encountered in structural analysis of web communities. This approach is based upon the pretopological concepts of pseudoclosure and the searching of equivalent nodes. The advantage of this approach is that it provides a framework needed to pass through the actual limits of graph theory modeling. The problem of modeling and understanding web communities is described, then a review of the existing models and their limits, and we finish by an example of a structuring algorithm.
  • Konferenzbeitrag
    Optimizing mobile networks connectivity and routing using percolation theory
    (10th International Conferenceon Innovative Internet Community Systems (I2CS) – Jubilee Edition 2010 –, 2010) Amor, Soufian Ben; Bui, Marc; Lavallée, Ivan
    Mobile Ad-hoc NETworks (MANETs) are complex systems presenting a phase transition phenomenon : an abrupt change in the behavior of the network around a critical value of a certain key parameter. The reliability and connectivity of MANETs depend on their size and on the efficiency of the routing protocol. In this paper we propose a new approach in MANETs modeling, combining percolation theory and epidemic algorithms. Using percolation theory we show the existence of a connectivity threshold (in a square lattice) needed to guarantee the communications in the network, in particular when the number of direct links are limited. Epidemic algorithms are used to provide a good propagation of information in the network wile minimizing the resources cost (energy, number of messages...).