Serigne Gueye

Université du Havre, El Havre, Upper Normandy, France

Are you Serigne Gueye?

Claim your profile

Publications (14)4.34 Total impact

  • [Show abstract] [Hide abstract]
    ABSTRACT: This paper presents an exact solution method based on a new linearization scheme for the 0-1 quadratic knapsack problem, which consists of maximizing a quadratic pseudo-Boolean function with nonnegative coefficients subject to a linear capacity constraint. Contrasting with traditional linearization schemes, our approach adds only one extra variable. The suggested linearization framework provides a tight upper bound, which is used in a branch-and-bound scheme. This upper bound is numerically compared with that of [A. Billionnet, A. Faye, and E. Soutif, European J. Oper. Res., 112 (1999), pp. 664--672], and our branch-and-bound scheme with the exact algorithm of [W. D. Pisinger, A. B. Rasmussen, and R. Sandvik, INFORMS J. Comput., 19 (2007), pp. 280--290]. The experiments show that our upper bound is quite competitive (less than $1\%$ from the optimum). In addition, the proposed branch-and-bound clearly outperforms the algorithm developed by Pisinger et al. for low density instances ($25\%$) for all instances up to $400$ variables. Read More: http://epubs.siam.org/doi/abs/10.1137/110820762
    SIAM Journal on Optimization 01/2012; 22(4):1449-1468. · 2.08 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: For operational and unpredictable reasons, many small incidents occur day after day in rail transportation systems. Most of them have a local impact, but, in some cases, mainly in dense networks, minimal disruptions can spread out through the whole network and affect significantly the train schedules. In this article, we present the railway rescheduling problem as the problem of finding a new schedule of trains after one or several incidents by minimizing some measure of the effect. We investigate the solution of this problem through a mixed-integer programming (MIP) formulation. Because of the impossibility for solving it exactly just using a standard MIP solver, we propose to limit the search space around the original nondisrupted schedule by hard and soft fixing of integer variables with local-branching-type cuts. Different variations of the method are compared to a right-shift rescheduling policy in two different networks located in France and Chile. The experimental results are also used to study the impact of different objectives on the total delay. © 2010 Wiley Periodicals, Inc. NETWORKS, 2011 © 2011 Wiley Periodicals, Inc.
    Networks 01/2011; 57:69-86. · 0.65 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: In this paper, we present a new approach to solve the railway rescheduling problem. This problem deals with the reparation of a disturbed railway timetable after incidents in such a way to minimize the difference between the original plan and the new provisional plan. We use a mixed integer linear programming (MIP) formulation that models this problem correctly. However, the large number of variables and constraints denies the possibility to solve this problem efficiently using a standard MIP solver. A new approach called SAPI (Statistical Analysis of Propagation of Incidents) has been developed to tackle the problem. The key point of SAPI is to estimate the probability that an event, one step of the itinerary of a train, is affected by a set of incidents. Using these probabilities, the search space is reduced, obtaining very good solutions in a short time. The method has been tested with two different networks located in France and Chile. The numerical results show that our procedure is viable in practice.
    European Journal of Operational Research. 01/2011; 215:227-243.
  • [Show abstract] [Hide abstract]
    ABSTRACT: Given an undirected edge-weighted network in which one edge capacity is allowed to vary, we propose in this paper a polynomial algorithm that needs only two 2-cut- trees computations to provide the all pairs maximum 2-route ow values. Moreover, we provide an extension to the case where k edge capacities may vary, and show that 2k 2-cut-trees computations are necessary in order to determine the all pairs maximum 2-route-ow values. This makes our algorithm polynomial whenever k = O(polylog n).
    International Transactions in Operational Research 01/2010; 17(1). · 0.59 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: Given an undirected edge-weighted network in which one edge capacity is allowed to vary, we propose in this paper a polynomial algorithm that needs only two 2-cut-trees computations to provide the all pairs maximum 2-route flow values.
    Electronic Notes in Discrete Mathematics. 01/2009; 35:59-64.
  • Source
    Serigne Gueye, Philippe Michelon
    [Show abstract] [Hide abstract]
    ABSTRACT: In this paper, we are interested in linearization techniques for exact resolution of the Unconstrained Quadratic (0-1) Problem. Our purpose is to propose "economical" linear formulations. We first extend the current techniques in a general linearization framework containing many other schemes and propose a new linear formulation. Numerical results comparing classical, Glover's and the new linearization are re- ported.
    Discrete Applied Mathematics. 01/2009; 157:1255-1266.
  • [Show abstract] [Hide abstract]
    ABSTRACT: The problem of rescheduling trains under disrupted operations has received more attention in the last years because of the increasing demand in railway transportation. Railway networks are more and more saturated and the probability that delayed trains affect others is large. Therefore, it is very important to react after incidents by recalculating new arrival/departure times and reassigning tracks/platforms with the aim of minimizing the effect of propagation of incidents.
    Integration of AI and OR Techniques in Constraint Programming for Combinatorial Optimization Problems, 6th International Conference, CPAIOR 2009, Pittsburgh, PA, USA, May 27-31, 2009, Proceedings; 01/2009
  • [Show abstract] [Hide abstract]
    ABSTRACT: In this paper we consider the problem of selecting assets for which transaction costs are given by piecewise affine functions. Given practical constraints related to budget and buy-in thresholds, our purpose is to determine the number of each asset i that can produce the maximum return of a portfolio composed of (n + 1) assets (one of them is free of risk). The problem is formulated as an integer quadratic problem and afterwards linearized. Some numerical experiments, using Ilog Cplex 10.1, has been performed. They show that the methodology is promising.
    Modelling, Computation and Optimization in Information Systems and Management Sciences, Second International Conference, MCO 2008, Metz, France - Luxembourg, September 8-10, 2008. Proceedings; 01/2008
  • Source
    01/2005;
  • Serigne Gueye, Philippe Michelon
    [Show abstract] [Hide abstract]
    ABSTRACT: In order to solve a quadratic 0/1 problem, some techniques, consisting in deriving a linear integer formulation, are used. Those techniques, called “linearization”, usually involve a huge number of additional variables. As a consequence, the exact resolution of the linear model is, in general, very difficult. Our aim, in this paper, is to propose “economical” linear models. Starting from an existing linearization (typically the so-called “classical linearization”), we find a new linearization with fewer variables. The resulting model is called “Miniaturized” linearization. Based on this approach, we propose a new linearization scheme for which numerical tests have been performed.
    Annals of Operations Research 01/2005; 140:235-261. · 1.03 Impact Factor
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: In this paper, we present a method for automatic generation of acoustic models from simple generic models. This method use the internal structure of non-contextual acoustic models in order to build new specialized states which are supposed to modelize specific patterns of a phoneme. The proposed technique use temporal information for state splitting. This method is compared to a maximum likelihood based approach. Our experiments show that this last criterion leads to better performance. Nevertheless, unsupervised model splitting seems to be less efficient than model specialization based on a priori knowledge.
    01/2002;
  • S Gueye, P Michelon
    [Show abstract] [Hide abstract]
    ABSTRACT: In order to solve more easily combinatorial optimization problems, one way is to find theoretically a set of necessary or/and sufficient conditions that the optimal solution of the problem has to verify. For instance, in linear programming, weak and strong duality conditions can be easily de-rived. And in convex quadratic programming, Karush-Kuhn-Tucker con-ditions gives necessary and sufficient conditions. Despite that, in general, such conditions doesn't exist for integer programming, some necessary conditions can be derived from Karush-Kuhn-Tucker conditions for the unconstrained quadratic 0-1 problem. In this paper, we present these conditions. We show how they can be used with constraints program-ming techniques to fix directly some variables of the problem. Hence, reducing strongly the size of the problem. For example, for low density problems of size lower than 50, those conditions combined with constraints programming may be sufficient to solve completely the problem, without branching. In general, for quadratic 0-1 problems with linear constraints, we propose a new method combining these conditions with constraints and linear programming. Some numerical results, with the instances of the OR-Library, will be presented.
  • Source