ALCOM-FT brings together 10 of the leading groups in algorithms research in Europe in a project that proposes to discover new algorithmic concepts, identify key algorithmic problems in important applications, and contribute to the accelerated transfer of advanced algorithmic techniques into commercial systems. The main emphasis of the project is on a novel combination of application-oriented research in three important areas ("massive data sets", "massive and complex communication," and "complex problems in production and planning") with innovative methodological work on "experimental algorithmics" and "generic algorithmic methods."
Traditional algorithms and traditional algorithm design methods are insufficient for meeting the algorithmic problems of the future. Massive data sets arise in important applications, and new algorithms (reflecting the structure of real machines) are needed. Communication patterns in networks are changing radically, and the very large networks of tomorrow must be controlled in a more efficient way than what is currently possible. Planning problems in business and society are becoming bigger and more complex, and sophisticated algorithmic support is needed for their solution. We shall develop the new algorithmic methods, which will be necessary to properly exploit such new technologies. We shall also expand the traditional view of the algorithm design methodology (where algorithms and techniques for their analysis are the objects of interest) to include programs as first class objects of study. We expect a methodological breakthrough by this expanded view.
DESCRIPTION OF WORK
The work is organized in five work packages: MASSIVE DATA SETS: We shall engineer algorithms for data mining, fundamental data structures, graph problems, and string problems. We shall consider general techniques for handling parallel disks and processors using realistic machine models. We shall design and implement an experimental platform for external computation.
NETWORKS AND COMMUNICATION
We shall investigate efficient communication in modern networks, specifically the efficient use of resources (frequencies, bandwidth) and develop basic protocols (routing, contention resolution) in advanced networks (all-optical, wireless, ATM). In the area of dynamic distributed environments, we will provide methods for job allocation and load balancing, and for efficient data management and contention resolution.
PLANNING IN PRODUCTION AND TRANSPORTATION:
We shall study a number of practical planning problems identified in cooperation with our industrial partners, including job-shop scheduling, dynamic scheduling, facility location, routing on structures networks, and timetabling. We shall develop algorithmic techniques to tackle these problems, implement them, and build a library containing real-world problem instances.
We shall identify and develop key algorithmic methods that arise from complexity bottlenecks in large scale problems, and which are generic and widely applicable. Emphasis will be on randomised techniques, optimisation techniques, and approximation techniques. A complete software system for structured optimisation problems will be developed.
We shall engineer cache-aware data structures and algorithms and develop techniques for more realistic performance predictions and for checking of programs. We shall construct generators of worst-case, random and simulated real-life problem instances, and we shall develop techniques for visualization of program behaviour. All this will be part of an experimental testbed.
Funding SchemeCSC - Cost-sharing contracts
78153 Le Chesnay
CV4 7AL Coventry
3584 CS Utrecht