2024-10-09 All-hands meeting notes

Meeting purpose and expectations

JE

• All-hands meetings are a forum for all contributors to the Initiative to get an update on the whole project, to develop a shared vision of participation (or membership) in the Initiative, and to ask unfiltered questions of project leadership.

  • The All-Hands meeting is the in-person meeting of the #internal Slack channel.

• Expected attendees include

  • Lead team (Project Director, Project Manager, Infrastructure Lead, Science Lead)

  • Core team members (i.e. OMSF employees)

  • Trainees (grad students and postdocs) supported by OpenFF-affiliated funding

• Other attendees may include

  • PIs as necessary

  • Collaborators from across the ecosystem

• Agenda:

  • Infrastructure & Science updates

  • Specific updates about funding, roadmap, or other organization-wide topics

  • Guest presentations

  • Shout-outs

  • Open Q&A with leadership

    • No questions about the org or project are “off-limits”

      • even if some answers may occasionally be vague

    • Project leadership is here to answer questions, not to ask them (we have other meetings for that)

Plan to split this meeting into 2 time zones (Atlantic & Pacific)

JE

• 4th Wednesday of the month

• 8 AM Pacific for Atlantic meeting and (tentative, need to work out Daylight Saving Time issues) 6 PM US Pacific

• Next meeting will be Weds, 27 Nov

Organizational updates

JE

• Performance Reviews

  • 360 reviews of leadership (completed)

  • Leadership review of other staff (upcoming)

Infrastructure Updates

JW

• https://docs.google.com/presentation/d/1B9YfkfxMo0XASJhK28PkPXeVbcv4Rzp_nkIyt19sUII/edit#slide=id.g230c27b5782_0_6

Science Updates

• https://docs.google.com/presentation/d/1qK-ARp4mnbfnDJzvJ2Ugn32RN0S3L2JLpiv_eZT0uS8/edit?usp=sharing

• both a shout-out and a warning: Brent found a really truly bizarre bug where occasionally, on certain computers, using OpenEye 2022.1.1 to minimize a certain molecule will result it in optimizing to a structure that jumps up in energy. I think we may have stumbled on this essentially by luck with the combination of the particular molecule we were looking at, his environment, and his computer. Neither Lexie nor I could reproduce it on our Macs or hpc3, but Brent drilled down and reproduced it on both an Intel CPU and AMD CPU instance. We haven't done much more rigorous exploring so we're still in the dark as to whether this might affect other versions or molecules, and why it's occuring in these instances. This may have affected anything involving minimizations, such as ours and other users' fitting and benchmarking. If they suspect it may have affected their results they should double check the energies before and after minimization.

Shout-outs

JE

• Brent Westbrook for really taking the time to figure out a truly bizarre bug where occasionally, on certain computers, using OpenEye 2022.1.1 to minimize a certain molecule will result it in optimizing to a structure that jumps up in energy. He put a lot of time into finding the source of the bug, including spinning up VPS instances, and put together a great MWE.

  • Slack thread: https://openforcefieldgroup.slack.com/archives/CM9BA4AVA/p1727812617347789

• Lexie McIsaac for putting in a tremendous amount of work and smarts into trying to see if we can generate a NAGL2 training dataset with a basis set with diffuse functions, in implicit solvent.

  • Unfortunately, right now the answer seems to be that it would be quite difficult, but her meticulous and focused investigation and debugging were really impressive.

• Matt Thompson (and David Swenson and Ethan Holz) for getting Evaluator CI running for the first time in 2 years by using AWS GPU runners