|
|||
|
IMADA - Department of Mathematics and Computer Science |
|
In many real life applications of combinatorial optimisation an exact solution approach is recognised infeasible and approximate algorithms, built on heuristics, are instead accepted. Metaheuristics is the name used to indicate a wide and variegate, sometimes fancy, variety of approximate algorithms which have been proved very effective in practice. Often, nature inspired, like in the case of evolutionary algorithms or ant colony optimisation, these algorithms draw mainly their strength from the effectiveness of two underlying heuristic paradigms, greedy construction and local search. The complex nature of these algorithms does not lend themselves to an analytical characterisation and an experimental approach is therefore more appropriate. In the talk I will illustrate this approach in my study on heuristics for the Graph Set T-Colouring Problem which finds practical application in the assignment of frequencies in mobile networks. The contribution is the introduction of new metaheuristic algorithms, and the use of rigorous statistical methods for the design and analysis of computational experiments. This latter is a research direction that still exhibits open issues which are timely to be addressed.
SDU HOME | IMADA HOME | Previous Page Last modified: Tue Apr 25 08:41:21 CEST 2006 Joan Boyar (joan@imada.sdu.dk) |