Results 1  10
of
330
Proceedings CPAIOR'02 Modelling a Balanced Academic Curriculum Problem
"... In this paper, we study a balanced academic curriculum problem. We show that this problem can be modelled in di®erent ways, and argue why each model is useful. We also propose integrating the models so as to bene¯t from the complimentary strengths of each model. Experimental results show that the in ..."
Abstract
 Add to MetaCart
In this paper, we study a balanced academic curriculum problem. We show that this problem can be modelled in di®erent ways, and argue why each model is useful. We also propose integrating the models so as to bene¯t from the complimentary strengths of each model. Experimental results show that the integration significantly increases the domain pruning, and even decreases the runtime on many instances. General lessons are learnt from this modelling exercise. First, when constraints are di±cult to specify in a particular model, we should consider channelling into a second model in which these constraints are easier to specify and reason about. Second, whilst constraint programming (CP) models can be best at ¯nding optimal or nearoptimal solutions, integer linear programming (ILP) models may be better for proving optimality. Hybrid CP and ILP models or a two phase approach may therefore be advantageous. Third, CP and ILP tools should provide primitives for channelling between models. Finally, we can often pro¯tably combine di®erent problem representations, as well as di®erent solution methods. 1
Proceedings CPAIOR’02 Randomised Backtracking for Linear PseudoBoolean Constraint Problems
"... Many constraint satisfaction and optimisation problems can be expressed using linear constraints on pseudoBoolean (0/1) variables. Problems expressed in this form are usually solved by integer programming techniques, but good results have also been obtained using generalisations of SAT algorithms b ..."
Abstract
 Add to MetaCart
Many constraint satisfaction and optimisation problems can be expressed using linear constraints on pseudoBoolean (0/1) variables. Problems expressed in this form are usually solved by integer programming techniques, but good results have also been obtained using generalisations of SAT algorithms based on both backtracking and local search. A recent class of algorithm uses randomised backtracking to combine constraint propagation with local searchlike scalability, at the cost of completeness. This paper describes such an algorithm for linear pseudoBoolean constraint problems. In experiments it compares well with stateoftheart algorithms on hardware verification and balanced incomplete block design generation, and finds improved solutions for three instances of the Social Golfer Problem. 1
Proceedings CPAIOR’02 An implementation of Pareto Optimality in CLP(FD)
"... The Constraint Problems usually addressed fall into one of two models: the Constraint Satisfaction Problem (CSP) and the Constraint Optimization Problem (COP). However, in many reallife applications, more functions should be optimized at the same time (MultiCriteria Optimization, or Pareto optimal ..."
Abstract
 Add to MetaCart
The Constraint Problems usually addressed fall into one of two models: the Constraint Satisfaction Problem (CSP) and the Constraint Optimization Problem (COP). However, in many reallife applications, more functions should be optimized at the same time (MultiCriteria Optimization, or Pareto optimality [14]), and solutions are ranked by means of a Partial Order. In this paper, we propose an algorithm for solving Pareto Optimality problems in CLP(F D). The algorithm is complete and does not make any assumption on the structure of the constraints. It exploits Point QuadTrees for the representation of the set of solutions, in order to access the data structure efficiently. 1
Proceedings CPAIOR’03 OnLine Resources Allocation for ATM Networks with Rerouting
"... We have developed an application for France Telecom R&D, which takes place in an ATM (Asynchronous Transfer Mode) network administration context. The problem consists in planning demands of connection over a period of one year. A new demand is accepted if both bandwidth requirements and quality ..."
Abstract
 Add to MetaCart
We have developed an application for France Telecom R&D, which takes place in an ATM (Asynchronous Transfer Mode) network administration context. The problem consists in planning demands of connection over a period of one year. A new demand is accepted if both bandwidth requirements and quality of service are satisfied. The acceptance or the reject of a demand must be decided within at most one minute. First, we look for a route satisfying the demand. In case of failure, we try to reroute some already accepted connections in order to satisfy the new demand. Rerouting has been modeled as a Valued Constraint Satisfaction Problem (VCSP) and solved by VNS/LDS+CP, a hybrid method specifically dedicated to solve VCSPs in anytime contexts. Experiments show that our rerouting enables to plan an average of ¢¤ £ % of demands that are rejected otherwise. 1
Proceedings CPAIOR’02 The Promise of LP to Boost CSP Techniques for Combinatorial Problems
"... In recent years we have seen an increasing interest in combining CSP and LP based techniques for solving hard computational problems. While considerable progress has been made in the integration of these techniques for solving problems that exhibit a mixture of linear and combinatorial constraints, ..."
Abstract
 Add to MetaCart
In recent years we have seen an increasing interest in combining CSP and LP based techniques for solving hard computational problems. While considerable progress has been made in the integration of these techniques for solving problems that exhibit a mixture of linear and combinatorial constraints, it has been surprisingly difficult to successfully integrate LPbased and CSPbased methods in a purely combinatorial setting. We propose a complete randomized backtrack search method for combinatorial problems that tightly couples CSP propagation techniques with randomized LP rounding. Our approach draws on recent results on approximation algorithms with theoretical guarantees, based on LP relaxations and randomized rounding techniques, as well on results that provide evidence that the run time distributions of combinatorial search methods are often heavytailed. We present experimental results that show that our hybrid CSP/LP backtrack search method outperforms the pure CSP and pure LP strategies on instances of a hard combinatorial problem. 1
AMSAA: A Multistep Anticipatory Algorithm for Multistage Stochastic Combinatorial Optimization Submitted to CPAIOR
, 2007
"... Abstract. The onestep anticipatory algorithm (1sAA) is an online algorithm making decisions under uncertainty by ignoring future nonanticipativity constraints. It makes nearoptimal decisions on a variety of online stochastic combinatorial problems in dynamic fleet management, reservation systems ..."
Abstract

Cited by 3 (1 self)
 Add to MetaCart
Abstract. The onestep anticipatory algorithm (1sAA) is an online algorithm making decisions under uncertainty by ignoring future nonanticipativity constraints. It makes nearoptimal decisions on a variety of online stochastic combinatorial problems in dynamic fleet management, reservation systems, and more. Here we consider applications in which the 1sAA is not as close to the optimum and propose Amsaa, an anytime multistep anticipatory algorithm. Amsaa combines techniques from three different fields to make decisions online. It uses the sampling average approximation method from stochastic programming to approximate the problem; solves the resulting problem using a search algorithm for Markov decision processes from artificial intelligence; and uses a discrete optimization algorithm for guiding the search. Amsaa was evaluated on a stochastic project scheduling application from the pharmaceutical industry featuring endogenous observations of the uncertainty. The experimental results show that Amsaa significantly outperforms stateoftheart algorithms on this application under various time constraints. 1
Proceedings CPAIOR’03 Integrating Mixed Integer Programming and Local Search: A Case Study on JobShop Scheduling Problems
"... Recent work on solving jobshop scheduling problems with earliness and tardiness costs has led to several hybrid algorithms that are more effective than any one method alone. In particular, these hybrid methods have been found to be more effective than pure mixed integer programming (MIP) approaches ..."
Abstract
 Add to MetaCart
Recent work on solving jobshop scheduling problems with earliness and tardiness costs has led to several hybrid algorithms that are more effective than any one method alone. In particular, these hybrid methods have been found to be more effective than pure mixed integer programming (MIP) approaches. This paper takes a fresh look at MIP formulations in light of several new MIP strategies for producing better feasible solutions. We find that a new MIP approach that is a hybrid method in spirit only, since it only exploits MIP methods but is inspired by concepts from local search, produces highly favorable results. We also identify two key factors that explain the variation of performance of a previous hybrid method on two different benchmarks. 1
Proceedings CPAIOR’03__________________________________ A Hybrid MILP/CP Decomposition Approach for the Scheduling of Batch Plants
"... A hybrid MixedInteger Linear Programming/Constraint Programming (MILP/CP) decomposition algorithm for the shortterm scheduling of general batch plants is proposed. The decisions about the type and number of tasks performed as well as the assignment of units to tasks are made by the MILP master pro ..."
Abstract
 Add to MetaCart
A hybrid MixedInteger Linear Programming/Constraint Programming (MILP/CP) decomposition algorithm for the shortterm scheduling of general batch plants is proposed. The decisions about the type and number of tasks performed as well as the assignment of units to tasks are made by the MILP master problem. The CP subproblem checks the feasibility of a specific assignment and generates integer cuts for the master problem. To exclude as many infeasible configurations as possible, two classes of integer cuts are generated. A graphtheoretic preprocessing that determines time windows for the tasks and equipment units is also performed to enhance the performance of the algorithm. Various objective functions such as the minimization of assignment cost, the minimization of makespan for fixed demand and the maximization of profit for a fixed time horizon can be accommodated. Variable batchsizes and durations, different storage policies, resource constraints and sequence dependent changeover times are taken into account. The proposed algorithm appears to be two to three orders of magnitude faster than standalone MILP or CP models.
Proceedings CPAIOR’02 Approaches to Find a Nearminimal Change Solution for Dynamic CSPs
, 2002
"... A Dynamic Constraint Satisfaction Problem (DCSP) is a sequence of static CSPs that are formed by constraint changes. In this sequence, the solution of one CSP may be invalidated by one or more constraint changes. To find a minimal change solution for that CSP with respect to the solution of the prev ..."
Abstract
 Add to MetaCart
A Dynamic Constraint Satisfaction Problem (DCSP) is a sequence of static CSPs that are formed by constraint changes. In this sequence, the solution of one CSP may be invalidated by one or more constraint changes. To find a minimal change solution for that CSP with respect to the solution of the previous related CSP, a RepairBased algorithm with ArcConsistency (denoted as RBAC in [4]) has been developed. However, when a new CSP is formed by adding or changing several nary (n ≥ 2) constraints, using RBAC to find a minimal change solution is much harder than using a constructive algorithm to generate an arbitrary solution from scratch. The constraint propagation techniques integrated in RBAC do not reduce its time complexity. This paper proposes two approximate algorithms to reduce the time complexity of RBAC by relaxing the criteria of an optimal solution. The experimental results show that one of the proposed algorithms performs quite well in terms of approximation of a minimal change solution within a limited period of time.
Results 1  10
of
330