2020-08-25 QCFractal Users Meeting notes

Updated from MolSSI

• Workshops wrapping up - BP should be returning to coding work

• Likely releases in the next few months

Manager/Queue status

• 4 submissions error cycling – A few thousand optimizations remaining, pretty significant resources needed

• Currently compute is only active at UCI

• 4 docker images to choose from – Ones with everything, one specialized for MM, one for ML/ANI

• DD – JC says that Lilac issues have resolved.

User questions

• BP – Will we be affected by dockerhub rate limiting?

  • DD – We’ll look into whether we’re on this scale.

  • BP – Let me know if/when I can remove images from MolSSI dockerhub

• HJ – Looked at JH’s idea of regenerating wavefunciton from output of QCEngine – Function needs density instead of … For this version of psi4, we can only get wavefunctions as orbitals and eigenvalues. I wanted to see if they would let us regenerate same ESP on the same grid points. But there wasn’t a way to copy psi4 vectors to different wavefunctions. There’s a function that works with density, but I couldn’t find one to work with wavefunction.

  • BP – In theory this should work.

  • HJ – Agree, but I couldn’t find a function. Should I be looking for a different function name?

  • We should try:

    • Deepcopy

      • import copy new_obj = copy.ceepcopy(old_obj)

    • Copy constructor (may be slightly different behavior from deepcopy)

      • new_obj = type(old_obj)(old_obj)

• DD – Disaccharide set?

  • DC – What more needs to be done?

  • DD+JH – We need to add constraints

  • DC – I can add the constraints

  • DD – Once DC adds constraints, we’ll replace existing tarball.

• JW – Be aware that Pavan will be joining in the next few weeks.

• JW – Sage release planned for late 2020 or early 2021 – Will have large QM data sprint 1-3 months before release

• DD – Looking for feedback as we prepare submission documentation – “so you want to submit a dataset?” type stuff