MD simulation extension

[JunCEEE]

Implements issue: #212 @CFGrote

Description

What is introduced, removed or renamed and why?

• Molecular Dynamics simulation extension is introduced for MD data IO.

Example

• The example data structure can be found HERE.

Affected Components

• EXT-MD

Writer Changes

What would a writer need to change?

• The writer needs to be able to write observables as a group under each Iteration, in the same level of particles to describe the thermodynamic parameters of the whole atomic system.

Does this pull request change the interpretation of existing data writers?

• Not yet.

Reader Changes

What would a reader need to change? Link implementation examples!

• The reader needs to be able to write observables as a group under each Iteration, in the same level of particles to describe the thermodynamic parameters of the whole atomic system. Example data: https://github.com/ejcjason/MDDomainExtension/blob/master/dataMD.h5

[ax3l]

Thank you for the PR! Before we start the review, can you please rebase against the upcoming-2.0.0 branch and squash your commits into one?

so basically you do:

$ git remote -v
# ... list of remotes, remember the name of https://github.com/openPMD/openPMD-standard

$ git rebase -i <nameFromAbove>/upcoming-2.0.0
# now a text edit opens.
#   replace the first word "pick" in each line with "squash"
#   save and close the editor
# a new editor window opens with all commit messages
#   edit and keep the one you want to keep, save

$ git push -f

In case anything goes south in the process: first create a backup branch and you can always git rebase --abort as it runs.

I'll be on a working trip next week but try to take a look as soon as the PR is updated.

[JunCEEE]

@ax3l Thanks. I've followed your instruction and squashed the commits. Please check that.

[JunCEEE]

I am sorry for the delay.

This looks wonderful, great work! I added a round of questions and formatting suggestions.

Thanks Axel. I've replied to your questions. Do we need more suggestions from experts in MD simulation?

[ax3l]

@CFGrote @JunCEEE do you like to update the PR for merging? :)

[ax3l]

VC meeting today:

• after initial LAMMPS experiments, further XMD and GROMACS workflows are being explored
• classic MD [atomic positions] & hybrid MD (DFT, ab initio) [additional particle attributes]
• we'll start with the start-to-end workflows needed for EU XFEL
  • as usual we'll, iterate in follow-up PRs if more people show interest and want to generalize it to their MD workflows, at the moment we focus on ours
• converter scripts: some outputs are already in HDF5. If existing data is large, a conversion could avoid a fully copy and instead use Virtual Dataset (VDS) and external links so create a light-weight openPMD file that re-interprets the original non-openPMD data

Action items:

• this PR will get polished with new XMD & GROMACS experience
• then we'll review & meet on it again

[CFGrote]

hi @ax3l and @JunCEEE , is anything missing on this PR? just wondering. would be great to have it included in 2.0.0

[JunCEEE]

@CFGrote I realized that the information of scattering factors and electronic charges related is missed in the standard. when I drafted the standard, I only considered the output of traditional/classical MD simulations, where the scattering process was absent. I am not sure whether we add it to this standard or have a different one. Any suggestions?

[JunCEEE]

General comments, I come from the DFT community and would like to explore whether OpenPMD could be of use to us! For that this PR seems very much needed.

Hello @zerothi! Welcome! Your comments would be important for us to see if this MD standard can also fulfill the need of DFT community.

[zerothi]

I would be happy to review a bit more. Is there something in particular that needs attention?

[ax3l]

@JunCEEE @zerothi thank you for iterating on this, that is great input! Do you two like to update the PR together to get it closer to merge-able state? :) It looks to me like most threads started arrived at a solution/improvement suggestion now.