crowdAI is shutting down - please read our blog post for more information
Optimizing train schedules
By SBB CFF FFS
about 2 years ago
we noticed that in all cases that we could observe, the route ID and the service intention ID are the same. However, many times there are routes with different IDs that are actually exactly the same (even in the worked example, Route 113 and Route 111 do not have any difference).
Could you provide more detail about the situation? This makes it a bit hard to setup the problem in an efficient way, since one would like to include the description of the route just once for each different route, without repetitions.
about 2 years ago |
It is true that there is redundancy in the routes in the input files, i.e. distinct routes can actually be identical. The model would allow distinct service intentions to use the same route, but this can lead to subtle problems. It is much easier and safer to just give each service intention its own route.
The fact that there are redundant routes in the input file does not make the actual problem harder. When modelling the conflicts between trains, you have to associate to each train its own route anyway. It just makes the input file a little larger, but that should not be a noticeable problem. Where do see performance bottlenecks? Do you do a very expensive preprocessing for each route?