Project Meeting 2017.03.31 - ActivitySim/activitysim GitHub Wiki
Data Pipelining and Random Number Sequencing
A PRNG object for each hh, person, tour type, trip type is using too much memory
@RSG - revising/testing to instead re-seed the PRNG on-the-fly; performance tests are promising so far
Models will then register with ActivitySim how many random numbers they might need and what type they are; see table of max random numbers by TM1 sub-model
Created create_simple_trips model step for better testing and modeling of trip data
Corrected sample sizes so random number usage is accurate #158
Can we have a draft online by the TRB Planning Applications conference in May?
Project Management
Created Progress Report Phase 3 based on the random numbers by sub-model table, since this is a better description of the model steps required to implement MTC TM1