Patrick Martineau

Complexity of flow shop scheduling problems with transportation constraints

Abstract In most manufacturing and distribution systems, semi-finished jobs are transferred from one processing facility to another by transporters such as Automated Guided Vehicles, robots and conveyors, and finished jobs are delivered to warehouses or customers by vehicles such as trucks. This paper investigates two-machine flow shop scheduling problems taking transportation into account. The finished jobs are transferred from the processing facility and delivered to customers by truck. Both transportation capacity and transportation times are explicitly taken into account in these models. We study the class of flow shop problems by analysing their complexity. For the makespan objective function, we prove that this problem is strongly NP-hard when the capacity of a truck is limited to two or three parts with an unlimited buffer at the output of the each machine. This problem with additional constraints, such as blocking, is also proven to be strongly NP-hard.

Total completion time minimization in a computer system with a server and two parallel processors

The context of the problem tackled in this paper is a computer system composed by a single server and two identical parallel machines. The processing times on the server are assumed to be unary and the objective is to minimize the total completion time. The papers dealing with scheduling problems with a server generally consider that the setup activities require simultaneously the server and the machine. In this paper, this constraint is not considered and the studied problem is a two-stage hybrid flow shop with no-wait constraint between the two stages. An algorithm that can solve optimally this problem in O(n log(n)) time is proposed. Finally, it is shown that solving this problem optimally leads to an optimal solution to the problem without the no-wait constraint.

Resolution of a scheduling problem in a flowshop robotic cell

We develop in this paper a generic and precise identification of a scheduling problem in a flexible manufacturing system. We consider a flowshop robotic cell that processes several jobs. We assume that there is no intermediate buffer between machines. So, jobs may be blocked when downstream machines are busy. We present an integer programming model to determine the sequence of jobs that minimizes the makespan criterion. In order to solve large size problems, we propose a genetic algorithm (GA). Finally, computational experiments are proposed in order to compare the makespan returned by the GA to a lower bound.

An Exact Graph Edit Distance Algorithm for Solving Pattern Recognition Problems

Graph edit distance is an error tolerant matching technique emerged as a powerful and flexible graph matching paradigm that can be used to address different tasks in pattern recognition, machine learning and data mining; it represents the minimum-cost sequence of basic edit operations to transform one graph into another by means of insertion, deletion and substitution of vertices and/or edges. A widely used method for exact graph edit distance computation is based on the A* algorithm. To overcome its high memory load while traversing the search tree for storing pending solutions to be explored, we propose a depth-first graph edit distance algorithm which requires less memory and searching time. An evaluation of all possible solutions is performed without explicitly enumerating them all. Candidates are discarded using an upper and lower bounds strategy. A solid experimental study is proposed; experiments on a publicly available database empirically demonstrated that our approach is better than the A* graph edit distance computation in terms of speed, accuracy and classification rate.

The e-OCEA project: towards an internet decision system for scheduling problems

This paper deals with an Internet decision support system for scheduling problems. This system, called e-OCEA, is being developed at the Laboratory of Computer Sciences of the University of Tours. It provides a user with tools to help create an effective algorithm to solve a scheduling problem. From the modelisation of the problem to the visualization of a computed schedule, the e-OCEA system offers software that can be used either by operations researchers or industrial engineers. In this paper, we present the current state of this system and provide future directions.

Automatic notation of the physical structure of a flexible manufacturing system

Abstract It is essential to have a generic notation that can be associated to any problem involved in the flexible manufacturing systems (FMS). Such a notation must allow any potential actor who has to deal with one of these points (researcher or industrialist) to present his FMS and the problem to be solved without ambiguity and in a unanimously recognized format. In this paper we propose a notation that both relies on basic elements given by the modelizer and on analyzed complex structures. Examples of notation are presented, as well as a presentation of the automatic notation software under development.

A parallel graph edit distance algorithm

Abstract Graph edit distance (GED) has emerged as a powerful and flexible graph matching paradigm that can be used to address different tasks in pattern recognition, machine learning, and data mining. GED is an error-tolerant graph matching problem which consists in minimizing the cost of the sequence that transforms a graph into another by means of edit operations. Edit operations are deletion, insertion and substitution of vertices and edges. Each vertex/edge operation has its associated cost defined in the vertex/edge cost function. Unfortunately, Unfortunately, the GED problem is NP-hard. The question of elaborating fast and precise algorithms is of first interest. In this paper, a parallel algorithm for exact GED computation is proposed. Our proposal is based on a branch-and-bound algorithm coupled with a load balancing strategy. Parallel threads run a branch-and-bound algorithm to explore the solution space and to discard misleading partial solutions. In the mean time, the load balancing scheme ensures that no thread remains idle. Experiments on 4 publicly available datasets empirically demonstrated that under time constraints our proposal can drastically improve a sequential approach and a naive parallel approach. Our proposal was compared to 6 other methods and provided more precise solutions while requiring a low memory usage.

Offline Scheduling of Map and Reduce Tasks on Hadoop Systems

MapReduce is a model to manage quantities massive of data. It is based on the distributed and parallel execution of tasks over the cluster of machines. Hadoop is an implementation of MapReduce model, it is used to offer BigData services on the cloud. In this paper, we expose the scheduling problem on Hadoop systems. We focus on the offline-scheduling, expose the problem in a mathematic model and use the time-indexed formulation. We aim consider the maximum of constraints of the MapReduce environment. Solutions for the presented model would be a reference for the on-line Schedules in the case of low and medium instances. Our work is useful in term of the problem definition: constraints are based on observations and take into account resources consumption, data locality, heterogeneous machines and workflow management; this paper defines boundaries references to evaluate the online model.

Benchmarking Hadoop Performance in the Cloud

Virtual technologies have proven their capabilities to ensure good performance in the context of high performance computing (HPC). During the last decade, the big data tools have been emerging, they have their own needs in performance and infrastructure. Having a wide breadth of experience in the HPC domain, the experts can evaluate the infrastructures used to run big data tools easily. The outcome of this paper is the evaluation of two technologies of virtualization in the context of big data tools. We compare the performance and the energy consumption of two technologies of virtualization (Docker containers and VMware) and benchmark the software Hadoop (JoshBaer, 2015) using these environments. Firstly, the aim is the reduction of the Hadoop deployment cost using the cloud. Secondly, we discuss and analyze the assumptions learned from the HPC experiments and their applicability in the big data context. Thirdly, the Hadoop community finds an in-depth study of the resource consumption depending on the deployment environment. We come to the point that the use of the Docker container gives better performance in most experiments. Besides, the energy consumption varies according to the executed workload.

A Generic Model to Allow an Easy Construction of the Control Software for FMS

Abstract This paper presents a modelisation methodology for FMS. This result appears as a logical pursuit of the research which was done in the design of a control software of a production system (Grunenberger, et al, 1995). Firstly, the theoretical modelling approach is described and finally it is applied on a complete FMS example.