Page Comparison

Project updates

JCl

Previous Week:

• TM-FF

  • See slides

  • Debug failing opt for Brents TM database

    • LW: I don’t think BW’s workflow used PDB geometries. Poor starting geometries from RDKit/OpenEye might be the cause here – could double check his script here https://github.com/openforcefield/qca-dataset-submission/pull/413/files#diff-2e4093b6e4803d77bf5fe49c92093309123eaadb5bb15fafdfc985f76e32291d

    • LW: or could look at initially optimizing with gfn2-xtb too

  • Work on exposing needed single point properties

  • Finalize project plan on confluence

  • See also TMProperties Doc, put in shared drive?

  • draft of plan - need to revise to include reactions / transition states (second approach). bi- and try-metal complexes were specifically called out in strategy document.

  • Spoke with Lexie on debugging keywords and obtained files on UCI.

  • Discussed “off optimum” data with Lexie and her implementation of maxiter keyword. Asked if raising force tolerance could be used instead to ensure that forces are actually close to harmonic minimum and not diverging from anharmonic reality… maybe include both at the same time.

• Replicating calculation issue:

  • Isolated that keywords spawning new records were added after pushing to QCPortal. Why were they added? Need to bypass qcsubmit and create datasets with qcportal. Still add to the qca-dataset-submission README Table?

  • After QCA Users meeting, --dry-run option with statistics on the number of records being made was well received.

• Dataset Longevity

  • From QCA Users meetings, Ben said there is a short time horizon on having the cached “views” feature. We can then download, convert to QCShema, and zip as HDF5. Could / should that be done in qua-dataset-submission in its place? Without a dataset*.json it won’t run anything, but that’s how we could record there?
    Seems like CI’s downloader isn’t needed?

  • Put project plan on confluence

• Next Week

  • Finish NRP onboarding

  • Work to debug Brents dataset and get SPs with properties for CI.

  • Onboarding with Zenhub

Project updates

LW

• Past week:

  • Figuring out sensible default kwargs for Evaluator (primarily around equilibration times and conditions)

  • Trialling several use cases

    • fitting a FF (projects: NAGL, DCole)

    • benchmarking across water models (stakeholder: Shirts group)

  • Working out hand-off to infrastructure team ( ? ) and what to do about current gaping holes in code

  • Debugging protein benchmarks

• This/next week: