-
Notifications
You must be signed in to change notification settings - Fork 180
faq 73891889
by Kai Nagel on 2016-07-14 11:53:05
(proxying for Richard M.)
Do agents (assuming they are travelling on the road network) potentially re-route 'tomorrow' if a better alternative route is identified than the one they took 'today'? If so, what is the mechanism exactly for this?
by Kai Nagel on 2016-07-14 12:05:08
I don't think that I can describe this in fewer words than given in our "book"; see http://matsim.org/the-book, or http://matsim.org/docs/userguide for an extract that serves as our user guide. There, in particular, Sec.1.2., 4.5, and probably some other material.
That is, assuming you have only configured ReRoute
and (say) SelectExpBeta
.
- In every iteration, some agents are up for
ReRoute
. The fraction is given by the respective weights of the strategies. - They will get a new route from a best path algorithm based on the previous iteration.
- There is, however, no guarantee that this will indeed be the fastest route in the next iteration, since the router might be wrong (it averaged, and thus imprecise, travel time estimates), the (synthetic) world may have changed (other people using the same route and thus causing more congestion; randomness), or for other reasons. Therefore, the MATSim principle is that the agent tries out that new route, gets a score for the plan in which it is embedded, and from then on selects between plans according to score.
You are viewing an archive of the previous MATSim Q&A site. The real site is now at https://matsim.org/faq