Lammps compatibility

[jkalayan]

Summary

Include functionality to run CE on LAMMPSDUMPS files. Currently MDA does not automatically convert force units from kcal to kJ, so this has to be done when the MDA Universe is created.

Changes

Read in LAMMPSDUMP file:

• explicitly use "LAMMPSDUMP" as a "file_format" option
• Convert forces and atomic energies (latter if applicable) from kcal to kJ when a MDA Universe is created
• MDA expects column headers that contain atom potential and kinetic energies to be explicitly provided, these are hardcoded as c_5 and c_7 in CE.

Handle extra bonds to H atoms:

• If restraints to H atoms (e.g. in water molecules) are explicitly defined as a bond, then rdkit complains about H valency, set inferrer=None to deal with this in symmetry calcs

Impact

• Will allow projects that output atomic forces and energies in LAMMPSDUMP files to use CE

[skfegan]

This should improve the robustness when handling different file types and topology restraints.

The only request is to add a unit test to test_mda_universe_operations.py for the convert_lammps function.

[skfegan]

Looks good to me

[harryswift01]

This looks really good overall, the LAMMPSDUMP handling is nice and clear, and the fallback logic is a good touch. Tests are solid too, especially capturing the transformation and checking the scaling directly rather than relying on a real MDA universe.

I've just left a couple of small comments to tidy things up, but nothing major. Once those are addressed I’m happy with this.

[jkalayan]

FYI: MDA issue for dealing with LAMMPSDUMP unit conversion bug is still open: MDAnalysis/mdanalysis#5115

[harryswift01]

This is a really nice, tidy addition that should allow LAMMPS files to be used correctly within CodeEntropy. Happy with the approach and the test coverage, thank you for adding this in