Re: [sumo-user] Regarding DUA

[<Michael.Behrisch@xxxxxx>]

Hi Rohan,

if the demand is low it is not unusual for the first iteration to already be the best.

concerning 1. Yes, the vehicles will use the shortest path concerning travel time

concerning 2. Dynamic routing might be less favorable because it needs more time during the simulation (so if you want to run a lot of simulations you might be better off having a pre-computed set) and sometimes it has an oscillation problem (vehicles switching often between two routes inducing additional traffic). But if you need it anyway, there is usually no need to do DUA in advance.

 

Best regards,

Michael

 

Von: sumo-user <sumo-user-bounces@xxxxxxxxxxx> Im Auftrag von Rohan Verma via sumo-user
Gesendet: Samstag, 31. August 2024 12:53
An: Sumo project User discussions <sumo-user@xxxxxxxxxxx>
Cc: Rohan Verma <rohanverma446@xxxxxxxxx>
Betreff: [sumo-user] Regarding DUA

 

Dear Community,

I am working on a scenario where vehicles enter and exit the simulation from a fixed set of edges, which act as origins and destinations. I have established flows between these edges. Based on my understanding, since I have not defined any fixed routes, the vehicles are free to select any path, preferably the shortest one. Following this, I decided to implement Dynamic User Assignment (DUA) in the simulation.

However, I encountered an issue: during the `duaiterate` process, the iteration with the least travel time always occurs in the first iteration, and the DUA never seems to converge thereafter.

I would appreciate your assistance in clarifying the following doubts:

1. Does specifying only the origin and destination edges mean that vehicles will choose the shortest path in terms of travel time, while also considering the current traffic conditions in the network?
2. If we are using dynamic routing, is there still a need to establish DUA?

Thank you for your time and assistance.

Best regards,  
Rohan

Attachment: smime.p7s
Description: S/MIME cryptographic signature