QCSchema with PBC?

[awvwgk]

I looked at the specifications of the QCSchema at https://molssi-qc-schema.readthedocs.io/en/latest/auto_topology.html and noticed that I cannot pass lattice parameters or the systems periodicity with the topology.

Are you planning to add periodic boundary conditions to QCSchema?

[dgasmith]

This has always seemed a bit out of scope to me. There was some strong push for it originally, but it seems that Materials Project, AIDIIA, Nomad, ASE, Citrine, etc cover this space fairly well.

I wouldn't rule it out entirely, but I think there would need to be some strong use cases for it.

[awvwgk]

Yes, there are competing standards out there and one has to support more than one in the end. So it is important to know if I can (potentially) use QCSchema in the programs I am developing, which necessarily requires having PBC.

[dgasmith]

@loriab @wadejong Any thoughts here?

[wadejong]

Relevant for molecular dynamics and materials science. But, it’s becoming also relevant in the self assembly research of mixed organic-inorganic molecular assemblies. I would say it is and should be in scope.

…

On Sep 18, 2019, at 4:17 AM, Sebastian Ehlert ***@***.***> wrote: Yes, there are competing standards out there and one has to support more than one in the end. So it is important to know if I can (potentially) use QCSchema in the programs I am developing, which necessarily requires having PBC. — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub, or mute the thread.

[dgasmith]

@mkhorton does MP have schema that we can borrow for this?

[ghutchis]

I think it would be good to have at least PBC / unit vectors.

[cryos]

So do I, we added them in Chemical JSON quite some time ago. We have never mixed orbitals and periodic boundary conditions together.

[dgasmith]

@cryos Could you post that schema here or go ahead and make a PR?

[awvwgk]

I found chemical JSON is using a representation with cell parameters.

"unit cell": {"a": 9.1, "b": 9.1, "c": 30.5, "alpha": 90, "beta": 90, "gamma": 120}

This format allows to easily specify 1D and 2D periodic systems and possibly spacegroup symmetry.

A band structure might be a bit more complicated and could be out-of-scope, but the stress tensor would be nice to have in the output.

[ghutchis]

I think there's also a cell vector representation in chemical JSON. In my opinion, that's much better, since the lengths and angles are easy to derive from the latter, but the cell vectors contain more information. (Consider that a 3x3 matrix of cell vectors contains 9 parameters, but the a/b/c alpha/beta/gamma only contains 6.)

[ghutchis]

Had to go check the code, the syntax should be:

"cellVectors": { 10.5, 0.0, 0.0, 0.0, 10.5, 0.0, 0.0, 0.0, 10.5 }

In other words, the keyword cellVectors and then 9 distances in Angstrom. Avogadro v2 master writes and both, but prefers the cellVectors if present.

I can certainly change the code to handle 1D and 2D cases. I'd probably just hack it at the moment by setting the other dimensions to be 10^6 or something similar.

[mkhorton]

In the Materials Project JSON we use:

'lattice': {'matrix': [[2.32547694, 0.0, -0.82218026],
   [-1.16273847, 2.01392211, -0.82218026],
   [0.0, 0.0, 2.46654077]],
  'a': 2.4665407716892562,
  'b': 2.4665407749920663,
  'c': 2.46654077,
  'alpha': 109.47122067561966,
  'beta': 109.47122070274486,
  'gamma': 109.47122053911131,
  'volume': 11.55162296798055},

The canonical information is the matrix since this unambiguously includes an orientation also, the additional information (a, b, c, alpha, beta, gamma, volume) are solely for human readability.

[mkhorton]

One additional factor to consider with PBC is whether you want to specific PBC along some directions only. Mixed boundary conditions present challenges, and we do not support them, but some tools (LAMMPS and ASE) do use them.

[mkhorton]

Finally, I think I mentioned this previously but it's an important point, if your schema can include information on chemical bonds (or any other kind of connectivity/"edge" information between atomic sites), then you have to think about how this information is stored with periodic boundaries also. For example, if you have a bond from your original unit cell bonded to an atom in the next periodic image cell over.