I am currently writing an application for my Master Thesis, for which I am using GraphHopper. I noticed that, at the moment, computing a public transit journey is several orders of magnitude slower than computing a normal journey in the road network.
I wanted to ask whether the main reason for this is the time-expansion of the graph when routing in a public transit network, or whether some optimizations such as Contraction Hierarchies do not currently work with the GTFS module?
whether the main reason for this is the time-expansion of the graph when routing in a public transit network
Only partly. The initial version tried to do everything without any assumptions or heuristics and the resulting traversed node count is very high.
or whether some optimizations such as Contraction Hierarchies do not currently work with the GTFS module?
CH for public transport does not seem like a good idea but there are other speedup algorithms that we’ll implement in the future, the problem then is that they often do not work fully multi modal or introduce (stricter) assumptions.
I just want to ask if there has been any progress on this topic? My impression is that a gtfs-routing is still about two orders of magnitude slower that a CH-based road-routing.