CHRouter currenlty doesn't work very well with marouter:
- the results are wrong because the contraction hierarchy doesn't get
rebuilt
- it works slowly since CHRouter doesn't support efficient 1-to-many
routing

try astar.

Am Mo., 17. Mai 2021 um 13:02 Uhr schrieb Sasan Amini <[email protected]>:

> I am passing the following command with almost all default values for the
> parameters: marouter --route-files ./SUMO/Demand/trips_marouter.trip.xml
> --net-file ./SUMO/Net/sumo_secondary_filter.net.xml -o
> ./SUMO/Output/marouter_output.rou.xml  --weights.interpolate
> --paths.penalty 1 --junction-taz --scale 20 --tolerance 0.001
> --max-inner-iterations 1000 --paths 5 --route-choice-method logit
> --weight-attribute traveltime --logit.beta 0.15 --logit.gamma 4
> --logit.theta 0.01 --max-iterations 10 --keep-all-routes
> --weights.priority-factor 10 --routing-algorithm CH --aggregation-interval
> 3600 --routing-threads 36 --netload-output
> ./SUMO/Output/netload_Marouter.xml --verbose --ignore-errors --log
> ./SUMO/Output/Marouter_log.xml
> I tried multiple versions with fewer options too (which I didn't save) and
> still I got the response from Dijkstra router.
>
> On Mon, May 17, 2021 at 12:48 PM Jakob Erdmann <[email protected]>
> wrote:
>
>> There is only the code:
>> https://github.com/eclipse/sumo/blob/64a041537ce537690f5a22edcc46ef40df8d5fa0/src/marouter/ROMAAssignments.cpp#L115-L163
>>
>> astar should work fine with marouter default options. Which options are
>> you passing?
>>
>> Am Mo., 17. Mai 2021 um 12:02 Uhr schrieb Sasan Amini <[email protected]
>> >:
>>
>>> Thank for the explanation. Are there any documents to help me make a
>>> realistic guesstimation of travel times? What I see in the course code is  
>>> travelTime
>>> = capacityConstraintFunction(edge, newFlow / intervalLengthInHours);
>>> and thought travel times are directly exportable from Marouter. What is
>>> the exact form of the capacityConstraintFunction? I think I can extract
>>> densities from the netload file, but to calculate travel times I need to
>>> know the function. The network is too big to run sumo even in the
>>> mesoscopic model.
>>> One other issue I noticed was that no matter which router I give to
>>> Marouter, in the end, I always get the message Dijksrarouter has answered
>>> the queries. I think in the current implementation only the Dijsktra router
>>> is working (could be related to:
>>> https://github.com/eclipse/sumo/issues/6935)?
>>>
>>> Thanks,
>>> Sasan
>>>
>>> On Mon, May 17, 2021 at 11:08 AM Jakob Erdmann <[email protected]>
>>> wrote:
>>>
>>>> The issue with fromJunction and toJunction is now fixed.
>>>>
>>>> The route costs in marouter are travel times multiplied with a
>>>> traffic-density dependent capacityConstraintFunction.
>>>> The actual travel time can only be guessed at at this point since it
>>>> will depend on vehicle interactions. To compute empty-network travel times
>>>> you can pass the marouter-routes to duarouter. If you need more realistic
>>>> traveltimes, simulate with sumo.
>>>>
>>>> Am Fr., 14. Mai 2021 um 16:48 Uhr schrieb Sasan Amini <
>>>> [email protected]>:
>>>>
>>>>> True, the scale parameter works perfectly fine. I am not sure if I was
>>>>> testing it on a nightly build version of 1.8.x, but now on 1.9.0 it works.
>>>>> However, the fromJunction - toJunction option for route files is not
>>>>> working, while it works perfectly fine in DUAROUTER. Moreover, for the
>>>>> output of MAROUTER  I was under the impression that route costs are travel
>>>>> times (since the weight attribute is by default traveltime) but I get
>>>>> values that are much larger than traveltime,  maybe distance in meters? Is
>>>>> this cost in MAROUTER modifiable i.e. to get travel time of each trip?
>>>>>
>>>>> Best,
>>>>> Sasan
>>>>>
>>>>> On Mon, May 3, 2021 at 9:19 AM Jakob Erdmann <[email protected]>
>>>>> wrote:
>>>>>
>>>>>> Scaling route file input is supposed to work since 1.9.0 (and it's
>>>>>> working in our test). Can you send me a minimal failing example for 1.9 
>>>>>> or
>>>>>> later?
>>>>>>
>>>>>> Am Sa., 1. Mai 2021 um 14:46 Uhr schrieb Sasan Amini <
>>>>>> [email protected]>:
>>>>>>
>>>>>>> Dear all,
>>>>>>>
>>>>>>> I am running MAROUTER on an extremely large network and have only 1%
>>>>>>> of the real demand, which means I have to scale it by a factor of 100.
>>>>>>> However, when I define demand as route file and apply the --scale 
>>>>>>> option,
>>>>>>> it has no impact on the number of loaded vehicles. When I try to define
>>>>>>> demand as flows, no valid vehicles are loaded. Any suggestion on how to
>>>>>>> scale demand when using MAROUTER? or even a better way to downscale
>>>>>>> capacity (similar to MATSim)?
>>>>>>>
>>>>>>> Best,
>>>>>>> Sasan
>>>>>>> _______________________________________________
>>>>>>> sumo-user mailing list
>>>>>>> [email protected]
>>>>>>> To unsubscribe from this list, visit
>>>>>>> https://www.eclipse.org/mailman/listinfo/sumo-user
>>>>>>>
>>>>>> _______________________________________________
>>>>>> sumo-user mailing list
>>>>>> [email protected]
>>>>>> To unsubscribe from this list, visit
>>>>>> https://www.eclipse.org/mailman/listinfo/sumo-user
>>>>>>
>>>>> _______________________________________________
>>>>> sumo-user mailing list
>>>>> [email protected]
>>>>> To unsubscribe from this list, visit
>>>>> https://www.eclipse.org/mailman/listinfo/sumo-user
>>>>>
>>>> _______________________________________________
>>>> sumo-user mailing list
>>>> [email protected]
>>>> To unsubscribe from this list, visit
>>>> https://www.eclipse.org/mailman/listinfo/sumo-user
>>>>
>>> _______________________________________________
>>> sumo-user mailing list
>>> [email protected]
>>> To unsubscribe from this list, visit
>>> https://www.eclipse.org/mailman/listinfo/sumo-user
>>>
>> _______________________________________________
>> sumo-user mailing list
>> [email protected]
>> To unsubscribe from this list, visit
>> https://www.eclipse.org/mailman/listinfo/sumo-user
>>
> _______________________________________________
> sumo-user mailing list
> [email protected]
> To unsubscribe from this list, visit
> https://www.eclipse.org/mailman/listinfo/sumo-user
>
_______________________________________________
sumo-user mailing list
[email protected]
To unsubscribe from this list, visit 
https://www.eclipse.org/mailman/listinfo/sumo-user

Reply via email to