|
Algorithms Laboratory
(A. Yao, L. Cai and L. C. Lau)
Areas of interest include stochastic algorithms, graph algorithms, and other combinatorial algorithms as well as neural network learning algorithms and combinatorial optimisation problems. A super computing network with a fast Ethernet, an E6500 and an E4500 supercomputers are being used for research on sequential, parallel and distributed algorithms targeted for extremely hard problems.
In the stochastic algorithms area, the focus is on simulated annealing and genetic algorithms. Both theoretical and experimental aspects of these algorithms are being investigated, including the characterisation of problem classes where fast and parallel simulated annealing algorithms are possible, parallelisation of genetic algorithms based on the island and co-evolution models, and so on. One problem of particular interest is the optimal mixed placement of rigid and flexible objects, with potential application on the design of amorphous polymeric materials and packaging cushioning systems.
Research on new algorithms for combinatorial optimisation, based on a hybrid use of La Grange and transformation approaches for constrained optimisation is underway. Another research effort is focused on the convergence and complexity aspects of several neural network-learning algorithms.
In the area of graph algorithms, the main activities include research on the fixed-parameter tractability of various graph problems, especially cardinality constrained optimization problems.
|
