Papers by Evgeny Selensky
Journal of Heuristics 1 (1): 9–32, 2003
Despite a number of similarities, vehicle routing problems and scheduling problems are typically ... more Despite a number of similarities, vehicle routing problems and scheduling problems are typically solved with different techniques. In this paper, we undertake a systematic study of problem characteristics that differ between vehicle routing and scheduling problems in order to identify those that are important for the performance of typical vehicle routing and scheduling techniques. In particular, we find that the addition of temporal constraints among visits or the addition of tight vehicle specialization constraints significantly improves the performance of scheduling techniques relative to vehicle routing techniques.
Covering arrays have been studied for their applications to drug screening and software and hardw... more Covering arrays have been studied for their applications to drug screening and software and hardware testing. In this paper, we model the problem as a constraint program. Our proposed models exploit non-binary (global) constraints, redundant modelling, channelling constraints, and symmetry breaking constraints. Our initial experiments show that with our best integrated model, we are able to either prove optimality of existing bounds or find new optimal values for problems of relatively moderate size (2000 variables). When the size of the problem increases our models’ performance degrades, but local search on a SAT-encoding of the model is able to find improved bounds on larger problems.
Despite a number of similarities, vehicle routing problems and scheduling problems are typically ... more Despite a number of similarities, vehicle routing problems and scheduling problems are typically solved with different techniques. In this paper, we undertake a systematic study of problem characteristics that differ between vehicle routing and scheduling problems in order to identify those that are important for the performance of typical vehicle routing and scheduling techniques. In particular, we find that the addition of temporal constraints among visits or the addition of tight vehicle specialization constraints significantly improves the performance of scheduling techniques relative to vehicle routing techniques.
On the Reformulation of Vehi le RoutingProblems and S heduling Problems ?
Optimizing Resource Assignment
Despite a number of similarities, vehicle routing problems
and scheduling problems are typically ... more Despite a number of similarities, vehicle routing problems
and scheduling problems are typically solved with different techniques. In this paper, we undertake a systematic study of problem characteristics that diff er between vehicle routing and scheduling problems in order to identify those that are important for the performance of typical vehicle routing and scheduling techniques. In particular, we find that the addition of temporal constraints among visits or the addition of tight vehicle specialization constraints significantly improves the performance of scheduling techniques relative to vehicle routing techniques.
Journal of the Operational Research Society, 2012
Hyperheuristics give us the appealing possibility of abstracting the solution method from the pro... more Hyperheuristics give us the appealing possibility of abstracting the solution method from the problem, since our hyperheuristic, at each decision point, chooses between different low level heuristics rather than different solutions as is usually the case for metaheuristics. By assembling low level heuristics from parameterised components we may create hundreds or thousands of low level heuristics, and there is increasing evidence that this is effective in dealing with every eventuality that may arise when solving different combinatorial optimisation problem instances since at each iteration the solution landscape is amenable to at least one of the low level heuristics. However, the large number of low level heuristics means that the hyperheuristic has to intelligently select the correct low level heuristic to use, to make best use of available CPU time. This paper empirically investigates several hyperheuristics designed for large collections of low level heuristics and adapts other hyperheuristics from the literature to cope with these large sets of low level heuristics on a difficult realworld workforce scheduling problem. In the process we empirically investigate a wide range of approaches for setting tabu tenure in hyperheuristic methods, for a complex real-world problem. The results show that the hyperheuristic methods described provide a good way to trade off CPU time and solution quality.
On mutual reformulation of shop scheduling and vehicle routing
... RC207 3233 6 968.156 6 Table 1. Comparison of results obtained for the reformulated VRP model... more ... RC207 3233 6 968.156 6 Table 1. Comparison of results obtained for the reformulated VRP model by Scheduler using LDS and those for the Dispatcher model using GLS. Scheduling techniqueRouting technique Benchmark Cost of Cost of Time Cost of Cost of Time ...
Constraint-Based Approaches to the Covering Test Problem
Lecture Notes in Computer Science, 2005
Abstract. Covering arrays have been studied for their applications to drug screening and software... more Abstract. Covering arrays have been studied for their applications to drug screening and software and hardware testing. In this paper, we model the problem as a constraint program. Our proposed models ex-ploit non-binary (global) constraints, redundant modelling, channelling ...
Constraint-Based Approaches to the Covering Test Problem
Constraint Solving and Constraint Logic Programming, 2004
Covering arrays have been studied for their applications to drug screening and software and hardw... more Covering arrays have been studied for their applications to drug screening and software and hardware testing. In this paper, we model the problem as a constraint program. Our proposed models exploit non-binary (global) constraints, redundant modelling, channelling constraints, and symmetry breaking constraints. Our initial experiments show that with our best integrated model, we are able to either prove optimality of existing bounds or find new optimal values for arrays of moderate size. Local search on a SAT-encoding of the model is able to find improved bounds on larger problems.
Constraints, 2006
Covering arrays can be applied to the testing of software, hardware and advanced materials, and t... more Covering arrays can be applied to the testing of software, hardware and advanced materials, and to the effects of hormone interaction on gene expression. In this paper we develop constraint programming models of the problem of finding an optimal covering array. Our models exploit global constraints, multiple viewpoints and symmetry-breaking constraints. We show that compound variables, representing tuples of variables in our original model, allow the constraints of this problem to be represented more easily and hence propagate better. With our best integrated model, we are able to either prove the optimality of existing bounds or find new optimal solutions, for arrays of moderate size. Local search on a SAT-encoding of the model is able to find improved solutions and bounds for larger problems.
Graph Grammars Workshops, 2002
The vehicle routing problem (VRP) and job shop scheduling problem (JSP) are two common combinator... more The vehicle routing problem (VRP) and job shop scheduling problem (JSP) are two common combinatorial problems that can be naturally represented as graphs. A core component of solving each problem can be modeled as finding a minimum cost Hamiltonian path in a complete weighted graph. The graphs extracted from VRPs and JSPs have different characteristics however, notably in the ratio of edge weight to node weight. Our long term research question is to determine the extent to which such graph characteristics impact the performance of algorithms commonly applied to VRPs and JSPs. As a preliminary step, in this paper we investigate five transformations for complete weighted graphs that preserve the cost of Hamiltonian paths. These transformations are based on increasing node weights while reducing edge weights or the inverse. We demonstrate how the transformations affect the ratio of edge to node weight and how they change the relative weights of edges at a node. Finally, we conjecture how the different transformations will impact the performance of existing VRP and JSP solving techniques.
Covering arrays have been studied for their applications to drug screening and software and hardw... more Covering arrays have been studied for their applications to drug screening and software and hardware testing. In this paper, we model the problem as a constraint program. Our proposed models exploit non-binary (global) constraints, redundant modelling, channelling constraints, and symmetry breaking constraints. Our initial experiments show that with our best integrated model, we are able to either prove optimality of existing bounds or find new optimal values for problems of relatively moderate size (2000 variables). When the size of the problem increases our models' performance degrades, but local search on a SAT-encoding of the model is able to find improved bounds on larger problems.
CoLogNet Publications, 2002
Many constraint satisfaction problems (csp's) are formulated with 0/1 variables. Sometimes this i... more Many constraint satisfaction problems (csp's) are formulated with 0/1 variables. Sometimes this is a natural encoding, sometimes it is as a result of a reformulation of the problem, other times 0/1 variables make up only a part of the problem. Frequently we have constraints that restrict the sum of the values of variables. This can be encoded as a simple summation of the variables. However, since variables can only take 0/1 values we can also use an occurrence constraint, e.g. the number of occurrences of 1 must be k. Would this make a difference? Similarly, problems may use channelling constraints and encode these as a biconditional such as P ↔ Q (i.e. P if and only if Q). This can also be encoded in a number of ways. Might this make a difference as well? We attempt to answer these questions, using a variety of problems and two constraint programming toolkits. We show that even minor changes to the formulation of a constraint can have a profound effect on the run time of a constraint program and that these effects are not consistent across constraint programming toolkits. This leads us to a cautionary note for constraint programmers: take note of how you encode constraints, and don't assume computational behaviour is toolkit independent.
Many constraint satisfaction problems (csp's) are formulated with 0/1 variables. Sometimes this i... more Many constraint satisfaction problems (csp's) are formulated with 0/1 variables. Sometimes this is a natural encoding, sometimes it is as a result of a reformulation of the problem, other times 0/1 variables make up only a part of the problem. Frequently we have constraints that restrict the sum of the values of variables. This can be encoded as a simple summation of the variables. However, since variables can only take 0/1 values we can also use an occurrence constraint, e.g. the number of occurrences of 1 must be k. Would this make a difference? Similarly, problems may use channelling constraints and encode these as a biconditional such as P ↔ Q (i.e. P if and only if Q). This can also be encoded in a number of ways. Might this make a difference as well? We attempt to answer these questions, using a variety of problems and two constraint programming toolkits. We show that even minor changes to the formulation of a constraint can have a profound effect on the run time of a constraint program and that these effects are not consistent across constraint programming toolkits. This leads us to a cautionary note for constraint programmers: take note of how you encode constraints, and don't assume computational behaviour is toolkit independent.
Symposium on Abstraction, Reformulation and Approximation, 2002
In manufacturing, different process designs give rise to different schedules and with each an ass... more In manufacturing, different process designs give rise to different schedules and with each an associated cost. In this paper, we report on a real-life example where a manufacturing company wants to evaluate the scheduling implications related to the degree of coupling between their processes of moulding and casting, in terms of the amount of buffer stock held. The results show that the present configuration could be improved as regards the amount of stock, while still meeting the demand levels. We show this as one example of a process design evaluation and propose in this paper an architecture for generic process design for this company, in order to evaluate quickly other scenarios. From this, we will be able to develop an approach of proactively using scheduling information in a systematic way to positively influence design decisions.
Many constraint satisfaction problems (csp's) are formulated with 0/1 variables. Sometimes this i... more Many constraint satisfaction problems (csp's) are formulated with 0/1 variables. Sometimes this is a natural encoding, sometimes it is as a result of a reformulation of the problem, other times 0/1 variables make up only a part of the problem. Frequently we have constraints that restrict the sum of the values of variables. This can be encoded as a simple summation of the variables. However, since variables can only take 0/1 values we can also use an occurrence constraint, e.g. the number of occurrences of 1 must be k. Would this make a difference? Similarly, problems may use channelling constraints and encode these as a biconditional such as P ↔ Q (i.e. P if and only if Q). This can also be encoded in a number of ways. Might this make a difference as well? We attempt to answer these questions, using a variety of problems and two constraint programming toolkits. We show that even minor changes to the formulation of a constraint can have a profound effect on the run time of a constraint program and that these effects are not consistent across constraint programming toolkits. This leads us to a cautionary note for constraint programmers: take note of how you encode constraints, and don't assume computational behaviour is toolkit independent.
Graph Transformations for the Vehicle Routing
The vehicle routing problem (VRP) and job shop schedulingproblem (JSP) are two common combinatori... more The vehicle routing problem (VRP) and job shop schedulingproblem (JSP) are two common combinatorial problems that canbe naturally represented as graphs. A core component of solving eachproblem can be modeled as finding a minimum cost Hamiltonian pathin a complete weighted graph. The graphs extracted from VRPs andJSPs have different characteristics however, notably in the ratio of edgeweight to node weight. Our long term research question is to determinethe extent to which such graph characteristics impact the ...
Proceedings of the 13th International Conference on Automated Planning and Scheduling, Jun 9, 2003
Despite a number of similarities, vehicle routing problems and scheduling problems are typically ... more Despite a number of similarities, vehicle routing problems and scheduling problems are typically solved with different techniques. In this paper, we undertake a systematic study of problem characteristics that differ between vehicle routing and scheduling problems in order to identify those that are important for the performance of typical vehicle routing and scheduling techniques. In particular, we find that the addition of temporal constraints among visits or the addition of tight vehicle specialization constraints significantly improves the performance of scheduling techniques relative to vehicle routing techniques.
Uploads
Papers by Evgeny Selensky
and scheduling problems are typically solved with different techniques. In this paper, we undertake a systematic study of problem characteristics that diff er between vehicle routing and scheduling problems in order to identify those that are important for the performance of typical vehicle routing and scheduling techniques. In particular, we find that the addition of temporal constraints among visits or the addition of tight vehicle specialization constraints significantly improves the performance of scheduling techniques relative to vehicle routing techniques.
and scheduling problems are typically solved with different techniques. In this paper, we undertake a systematic study of problem characteristics that diff er between vehicle routing and scheduling problems in order to identify those that are important for the performance of typical vehicle routing and scheduling techniques. In particular, we find that the addition of temporal constraints among visits or the addition of tight vehicle specialization constraints significantly improves the performance of scheduling techniques relative to vehicle routing techniques.