Project Meeting 2021.05.25

Technical Call

• Issues review
  • No new issues
• New page added for Server Specs
• Reviewed auto-chunking calculation improvements and results
  • Feature is working well and I'm working on updating the user's guide to describe the new settings
  • Also need to update the examples
  • Some discussion about running the model massively parallel to get really fast runtimes and the architectural challenges for such a setup
  • Will add times for the single process start up and shutdown steps that are similar across the different scenarios
  • These components, such as creating and coalescing the multiprocess mini pipelines, would need some improvements for a massively parallel setup
  • But it still scales well for 64, 128, etc. cores
  • Settings are much more intuitive than before
• Work from home example added as new branch
  • Using auto accessibility measure for now
  • Would be good to add a deprecation warning about the change in calling the size term scaling function
• Joel share transit pass fare discounts analysis on Thursday
• PUMS is changing to synthetic data, see https://ipums.org/changes-to-census-bureau-data-products