[WIP] Riper calculations + minor modifications on testing + unit tests and integration tests #14
Conversation
MoleculeSystem can deal with Structure objects. Additional periodic argument to define periodicity of the Structure. Extraction of structures from control file (using Gradient). Parsing of riper outputs (preliminary and experimental).
…/utils. Updated reference output and parser json files for riper parsing.
Added tests for Ricc2 outputs. Added unit tests for riper calculations.
collections.abc). Added parsing method for periodicity data.
|
Need to add a PeriodicData in turbomoleio.output.data and put it in the ScfOutput as well (+ update again all the reference files). |
Updated tests. Properly read gradlat lattices.
$lattice data group.
|
Missing the possibility to get PeriodicSystem object from a file where $cell or $lattice also has the "angs" option (so that values are in angstroms instead of the default bohr units). |
Moved some of MoleculeSystem and PeriodicSystem objects to BaseSystem. Correction of the as_dict and from_dict of PeriodicSystem. Fixed some docstrings. Fixed checks on periodicity.
|
sorry, I made a bit of a mess with resolved/unresolved conversations. I hope it is fine now. One additional point: why not having |
Indeed, that might be good. I actually thought about it and having PeriodicSystem in a molecule module doesn't feel right. Putting that in periodic.py would do for me. About BaseSystem, if we indeed move PeriodicSystem to periodic.py, I don't feel importing BaseSystem from molecule.py. Maybe we could still make a base.py, which could hold BaseSystem, but also get_coord_lines and get_mol_and_indices_frozen. Or we could put all of that in utils.py. What do you think ? |
seems good to me |
|
I've moved BaseSystem, get_coord_lines and get_mol_and_indices_frozen to base.py. |
Moved get_coord_lines to base.py. Moved get_mol_and_indices_frozen to base.py. Added vesta=True for the lattice (bugfix) where applicable (so that vectors are properly oriented). Moved PeriodicSystem's tests to test_periodic.
Adapted MoleculeSystem to handle periodic systems (using Structure object + periodicity as "1D", "2D" or "3D"). The Structure object should have a lattice such that the a vector of the lattice is aligned with the x cartesian axis and the b vector of the lattice is in the xy cartesian plane.
Adapted Gradient object to read periodic system (need to read the control object, as it needs both the $grad datagroup containing the gradients wrt atomic displacements and the $gradlatt datagroup containing the gradients wrt lattice).
Added a first preliminary parsing of Riper log files (integrated into ScfOutput).
Comparison with old parsed objects is now handling None values properly (the reference object is reconstructed and then serialized again to compare to the new serialized object). Before that the reference object was never reconstructed, just read from the json file. When adding a new argument to an object with a default of None, this led to a difference with the new object in the tests.
Missing update of the documentation.