2025-10-02 Engineering Team

[jpn--]

@ActivitySim/engineering

Agenda

• Release 1.5 🥳
• Check in on status of 11B tasks
• Sandag Example: need to wrap up PR 31
• PopulationSim: what is status on PR 192
• A few open issues remain on estimation: CDAP, School/University location, atwork subtour scheduling

Looking Ahead

• Next week (Oct 9): Implementation review on TransLink & ODOT's park-and-ride solution
• Two weeks (Oct 16): Expect to cancel this engineering team meeting unless there is pressing business

[jpn--]

Consultants who will look at Modeling Environments

Agency	Consultant
SANDAG	WSP/CS
PSRC	In-House
MWCOG	RSG
SEMCOG	RSG
ARC	WSP
MetCouncil	RSG
Translink	In-House/RSG
TfNSW	WSP
Victoria	DL
CMAP	CS

Any consortium member who is missing from this list and who would like help ensuing that their ActivitySim models can work within their general modeling workflow, reply here or contact @jpn-- and we will add you to the list.

[joecastiglione]

What does "look at" mean? Is there a template to facilitate comparison across agencies?

[i-am-sijia]

Hi @joecastiglione, Thanks for your question! Please see my comment in issue #995. I’ve started a list of items to collect information on (“look at”) in a live Google Sheet, which I hope aligns with your point about facilitating comparison across agencies. The idea is for this list to evolve as we talk to agencies and identify common themes.
Let us know if you have any suggestions.
The current plan is to reach out to agencies to review their current approach to installing and running ActivitySim, gather feedback and questions, especially from those who have already tried using UV based on the instructions in the ActivitySim user guide, and then regroup to consolidate any issues and responses.

[jpn--]

ActivitySim Engineering Meeting Notes

Meeting Overview

Agenda

• ActivitySim 1.5 release announcement and breaking changes
• Population Sim pull request status
• Estimation mode open issues
• Phase 11B UV installation testing coordination
• Agency-consultant assignments for UV testing
• Park-and-ride solution discussion planning

Participants

• Jeff Newman (@jpn--), Driftless
• David Hensle (@dhensle), RSG
• Sijia Wang (@i-am-sijia), WSP
• Joe Flood (@JoeJimFlood), CS
• Stefan Coe (@stefancoe), PSRC
• Bo Wen (@bwentl), TransLink
• Jilan Chen (@JilanChen), SEMCOG
• Ray Ngo (@ray-ngo), MWCOG
• Guy Rousseau (@guyrousseau), ARC
• Craig Heither, CMAP

Duration

47 minutes

Executive Summary

The team announced the release of ActivitySim version 1.5 with documentation updates and a small number of breaking changes. The main focus of the meeting was coordinating Phase 11B UV installation testing across agencies, with assignments made pairing consultants with agencies for the first round of outreach. The approach will involve understanding current Python environment management workflows, documenting use cases, and providing guidance for UV integration with existing agency systems.

Meeting Notes

ActivitySim 1.5 Release

Jeff Newman announced the official release of ActivitySim version 1.5. The release includes standardized preprocessing and annotation settings, with some breaking changes that require users to upgrade to large version 6 and make minor configuration adjustments. A handful of estimation mode issues remain open, mostly predating recent estimation mode work, which will be addressed in future updates. The documentation successfully updated automatically through GitHub actions and now correctly displays version 1.5.

The team discussed whether to use beta releases for future versions. The consensus was to continue with standard releases rather than adding beta versioning, as the consortium development process effectively serves as a beta period, and additional versioning could create confusion.

PopulationSim Pull Request

A longstanding PopulationSim pull request is ready to be merged. All items within the original scope have been completed, with remaining issues broken out and prioritized on the Phase 11 development project board. Jeff Newman will merge the pull request and prepare a release.

Estimation Mode Issues

Three open estimation mode issues were identified: CDAP, school/university location data splitting, and subtour scheduling. These issues need to be reviewed to confirm whether they're still outstanding or have been resolved by recent updates.

UV Installation Testing Coordination

The team developed a comprehensive plan for Phase 11B UV installation testing across agencies. The approach involves a phased outreach strategy:

First Round Agencies and Consultant Assignments:

• SANDAG - WSP (Sijia Wang) with CS support (Joe Flood)
• PSRC - In-House
• SEMCOG - RSG
• Met Council - RSG
• MWCOG - RSG
• ARC - WSP
• CMAP - CS
• TfNSW - RSG/WSP
• TransLink - In-House
• Victoria - DL

In-House Testing:

• PSRC will test UV environment setup internally before engaging consultants
• SANDAG has already tested UV with version 1.4 and encountered some preprocessing issues
• TransLink has already implemented a working UV solution with a separate environment for ActivitySim and additional packages managed as a Python package

Key Discussion Points:

The scope of consultant assistance focuses on ActivitySim installation using UV and integration with existing modeling environments. Support for pre/post-processing steps and other modeling tools (Emme, TransCAD) falls outside the scope unless the issues are directly related to UV integration with ActivitySim.

Agencies use various environment management approaches, including multiple Conda environments for different model components. UV can coexist with other environment management tools like Conda, allowing agencies to maintain existing setups while migrating ActivitySim to UV.

The first step involves understanding each agency's current Python environment management workflow, documenting different use cases, and developing guidance that agencies can follow. Agencies not in the first round can review the documented use cases and volunteer for second-round testing if their environment setup differs significantly from documented scenarios.

All documentation and progress updates will be tracked on GitHub issue #995, with consultants posting their findings and solutions as comments.

Environment Management Coordination

Consultants agreed to coordinate their approach to ensure consistent recommendations across agencies. The team will meet separately to align on whether to use UV sync or other UV commands and develop standardized guidance.

Other Development Work

Joe Flood reported progress on the error handling task, working through an Excel sheet cataloging all runtime errors with context. He plans to commit code changes and open a draft pull request early the following week.

A discussion of TransLink and ODOT's joint park-and-ride solution was scheduled for the following week's meeting.

The meeting two weeks out was tentatively canceled due to unavailability of key participants (Jeff Newman and David Hensle).

Action Items

• Jeff Newman: Merge PopulationSim pull request and prepare release
• Jeff Newman: Review and confirm status of three open estimation mode issues
• Jeff Newman: Share agency-consultant assignment spreadsheet
• All Consultants: Open draft pull requests for ongoing work to maximize visibility across the team
• All Consultants: Conduct first round of agency outreach within the next 1-2 weeks to understand Python environment management workflows
• All Consultants: Document findings on GitHub issue #995
• Consultants (RSG, WSP, CS): Meet separately to coordinate UV installation approach and develop consistent recommendations
• RSG (David Hensle, Ali Etezady): Reach out to SEMCOG, Met Council, MWCOG, and Transport for New South Wales
• WSP (Sijia Wang): Reach out to SANDAG and ARC; verify server access for SANDAG
• CS (Joe Flood): Support SANDAG testing; reach out to CMAP
• CS (Joe Flood): Complete error handling task and open draft pull request by early next week
• PSRC (Stefan Coe): Test UV environment setup in-house and report findings
• All Agencies: Review documented use cases from first round and volunteer for second round if environment setup differs significantly