Project Meeting 2021.03.02

Technical Call

• Issues review...new issue #387 fixed by Jeff
• Clint shared ARC implementation presentation
• Would be good to better document who the chooser is for each submodel and how the skim handling works
• ARC will be continuing to work on the transition to ActivitySim again later this year
• Joel shared SEMCOG university related travel improvements
• The model is now up and running and the team plans to work on calibration/validation next
• Team plans to share an update in maybe a month
• The enhancements includes new submodels and also some improvements to the core for random number management, which have yet to be implemented
• Is insertion / deletion of trips, and therefore random number management that works with this idea, a generic feature that is needed or more a one off?
• Approach is pretty modular and would be useful for other large universities like UCSD
• Update from me on development progress
  • The examples are all cleaned up for the prototype test system
  • PSRC has successfully run their two zone system model for a full sample from start to finish!
  • Stefan reports 5 hour runtimes with multiprocessing
  • Jeff to start on representative logsums Wednesday-ish
• SANDAG successfully ran the MTC model design with their three zone model system inputs and a 100% percent sample
  • Multiprocessed runs in 40 hours
  • Doesn't include location presampling
  • Accessibilities works now
  • Wu plans to share runtimes by submodel next time and look into validation of results