JM – Pretty much. Have a few hours left so I’ll try to wrap stuff up this week.
JW – Great. Please mention that on the Slack #status channel and update the OFF calendar with all-day events to indicate when you’re out.
(Done!)
JM – Been working on blog post on documentation.
JW –
If it’s not done in December, that’s fine. We’d debated whether to have a Dec blog post at all
The blog post may not need explicit review+approval - They’re supposed to be casual and have typos
JM – I’m going to be pointing out some examples of bad docs, want to make sure I don’t ruffle any feathers
JW – I’d be happy to review for that. It’ll probably just be a matter of wording.
JW – Could you add a one-stop shop for the constraints explanation to FAQ?
Use constrained FF for taking >=2fs timesteps in MD
Use unconstrained FF for:
Conformer minimization and single-point energies
1fs timestep dynamics
FF training, both for scientific reasons (like “this H may be squished in a sterically constrained molecule so the fit would be bad if every H had to share the same bond length”) and for technical reasons (if you’re fitting to forces as well as energies, then you’ll get weird outcomes because the constraint will mess with forces along the axis of the bond)
Only difference between constrained and unconstrained FF is these lines (note that it’s a tiny bit more complicated for sage, since it ALSO contains TIP3P parameters, and those ALWAYS include constraints. So even openff_unconstrained-2.0.0 contains constraints, but just for water)
To remove constraints from a FF, do del forcefield['Constraints']["[#1:1]-[*:2]"]
JW – I talked with MT who had asked about having a small workshop in Feb to talk about the new toolkit release. I said that I don’t have the bandwidth to lead another workshop and he volunteered to lead. I mentioned that you’d probably be interested in helping, so he’ll contact you when he begins preparing.