If we compare heuristic approach like local search or metaheuristic approach like genetic algorithms or use more advanced technique such as Graph Neural Networks. What is the best approach to adopt considering large scale graph? Both for mesh and radial kind of networks.
It is difficult to conclude that one heuristic works better than other for optimization problems. You may have to implement a couple of heuristics and compare them. Local search techniques are faster but may not explore a large enough search space. You can compare the heuristics on their run-time and the quality of results they produce.
Henning Meyerhenke There are some equality constraints satisfying power flow (since it is a power distribution grid partition problem). Some inequality constraints such as voltage limits, limits on minimum number of nodes within partition. There are numerous objectives such as path redundancy between loads and generators within a partition. The challenges are it is hard to obtain pareto optimal soln with GA when increasing the equality constraints.
- Badges
- Science topic
- Similar topics
- Mathematical Sciences
- Graphs