faq 67567633

inconsistent plans?

by Kai Nagel on 2016-06-16 13:10:03

---

Kai Nagel proxying for Miguel Picornell:

On 16 Jun 2016, at 12:33, Miguel Picornell <miguel.picornell@nommon.es> wrote:

Dear all,

We have a question regarding the multimodal module. We have found inconsistencies in departure times for multimodal simulations, when multimodal trips are enabled. For an agent performing a multimodal trip, we found that departure time of each leg is always the same (not updated from mode to mode). The departure time for the first mode activity is that of the activity end time, and arrival time is equal to departure time + travel time. Once the agent changes leg mode from lets say walk to public transport the departure time is the same as for the previous leg rather than the arrival time from the previous ‘activity’ walking. 

See the example of output activity diaries below (in red the inconsistencies). If the scoring function and/or utility function uses the sum of all travel times this would not affect the term related with travel but it does affect the term of activity since the final arrival time to the activity and hence it will affect the available time to perform the destination activity. This would correspond to the initial departure time plus the travel time of the last leg mode.

                                               …</leg>

                                               <act type="work" link="link_car_14667" x="235534.794531" y="5069853.68051" end_time="13:51:52" />

                                               <act type="multimodalTransitDeparture" link="link_car_14667" x="235534.794531" y="5069853.68051" max_dur="00:00:00" />

                                               <leg mode="walk" dep_time="13:51:52" trav_time="00:07:38" arr_time="13:59:30">

                                                               <route type="accessEgressNetworkBased" start_link="link_car_14667" end_link="tr_3209" trav_time="00:00:00" distance="614.9659589308854">{"accessTime":"NaN","onLinksTime":"NaN","links":[],"egressTime":"NaN"}</route>

                                               </leg>

                                               <act type="pt interaction" link="tr_3209" x="235731.942104152" y="5069423.66907436" max_dur="00:00:00" />

                                               <leg mode="pt" trav_time="00:06:00">

                                                               <route type="experimentalPt1" start_link="tr_3209" end_link="tr_3214" trav_time="00:06:00" distance="3722.155136660952">PT1===stop_metro_50.1===L3_0===L3_0===stop_metro_55.1</route>

                                               </leg>

                                               <act type="pt interaction" link="tr_3214" x="238808.920955717" y="5068326.72921347" max_dur="00:00:00" />

                                               <leg mode="walk" dep_time="13:51:52" trav_time="00:05:57" arr_time="13:57:49">

                                                               <route type="accessEgressNetworkBased" start_link="tr_3214" end_link="link_car_17583" trav_time="00:00:00" distance="478.86407193794867">{"accessTime":"NaN","onLinksTime":"NaN","links":[],"egressTime":"NaN"}</route>

                                               </leg>

                                               <act type="other" link="link_car_17583" x="238899.397133" y="5068683.80193" end_time="14:30:49" />

                                               <leg mode="walk" dep_time="14:30:49" trav_time="00:14:00" arr_time="14:44:49">…

Best,

----------------------------------------------------------------------------
Miguel Picornell

---

Comments: 1

---

Re: inconsistent plans?

by Kai Nagel on 2016-06-16 13:10:27

Dear Miguel,

I might not know enough about the multimodal contrib (assuming that is what you are using; please get used to adding logfiles). 

But my suspicion is that what you are seeing is a "junk DNA" (https://en.wikipedia.org/wiki/Noncoding_DNA) problem ... there are elements in the plan that are not interpreted anywhere.

Could you please 

(1) try setting "removingUnnecessaryPlanAttributes" in the "plans" config module to true and see if the entries go away

(2) try setting "writeExperiencedPlans" in the "planCalcScoreConfigGroup" and see if it looks better there

(3) if all else fails, look in the events if they are inconsistent (on unix, you group grep for the person id).

Could you please let us know what you find?

Best wishes, Kai