Project Meeting 2017.07.21

Logsums

The logsums branch is largely complete
Work and school location choice now include intelligent sampling and logsums
Required revisions to the entire chain of modules, as described here
Both models also include intelligent sampling as well now, as described here
Revised chunk_size setting to be smarter: it now specifies the number of values in chunk (total size), not just number of chunk rows
Improved handling of data type issues in low level dot operations and this significantly improved runtimes of all models
Created a logsums expression file for now by copying the mode_choice expression file and removing attributes not available for work/school location in order to avoid the pre-processing that doesn't apply; @We need to go back and fix this
Tests implemented
All other models do not include logsums in the utility expressions at this time
@We need to test with the full scale example (all zones, all hhs, all skims)
@We need to update documentation since this is a major revision

Existing Phase
- Will start on Task 5 once Logsums complete
- @Ben update Progress Report
Next Phase
- Discussed Phase 3 amendment scope and some revisions are needed
- Intelligent sampling can be removed from the scope since it has been implemented
- Use "full scale example" instead of "full model run"
- Change Task 6 "Completing Phase 1 Models" to something like "Complete Initial Implementation of All Sub-Models"; need to more explicitly list out what will be done under this task
- Cost estimate to deliver the Phase 3 amendment SOW will be more than the funding available so we'll start with the revised SOW and cut it back after we know how much funding we'll have
- @Ben revise SOW
- Accelerate funding? SFCTA - yes, SANDAG - yes, PSRC - no, MTC - ?, ARC - ?
- Likely start work in October
PMC
- Wu is replacing Clint on the PMC
- @Wu to follow-up with Clint and Joe about existing outreach efforts
- Bill suggested a separate call for the outreach discussion