Defect: Allow bulk SC calculation to be skipped. #742
Conversation
update structure matcher update structure matcher update structure matcher update structure matcher update structure matcher
dev script change dev script change dev script change dev script change dev script change dev script change dev script change test
|
Thanks a lot for clarifying. I am building a defect database, though it will not be large scale. I really like the "get_defect_entry", which combines different charge states of a defect. The previous one also contains more bulk info. These greatly facility the query. The only thing I feel missing is the "electrostatic potential" for Kumagai corrections. I am considering adding them to the "get_defect_entry". It is just because sometimes I do not want to do the original FNV corrections because of anisotropy in dielectric constant. |
|
I would also like to suggest the MP_id and formula_pretty can be added somewhere to facilitate database organization and query. |
|
The way the old function works should not be changed at all with this update. Adding MP-ids and stuff will have to be done by a builder at a later, the closest you can get to that right now is through updating metadata. |
|
Hi @jmmshn Not sure there will be adding |
|
@yuan-gist Can you start a WIP PR for that? I think it will be really useful. |
Sure. Will be done by next week |
|
Is this still WIP? |
jmmshn
changed the title
|
@utf missed the message above. I have been running this and some downstream builders at scale and wanted to get a good handle on the results to see if more changes are needed. |
|
Thanks @jmmshn. Appologies for the delay. |
* start electrodes * start electrode * start electrode * start electrode * start electrode * start electrode * VASP electrode job * VASP electrode job * lint * n steps * n steps * n steps * n steps * rm defect changes * rm defect changes * update * update * update * update structure matcher update structure matcher update structure matcher update structure matcher update structure matcher update structure matcher * debugging debugging debugging debugging debugging debugging debugging debugging debugging debugging debugging debugging debugging debugging debugging * debugging * debugging * debugging * debugging * append names * append names * append names * append names * append names * dev script change dev script change dev script change dev script change dev script change dev script change dev script change dev script change test * working test * typo * lint * lint * lint * lint * allow different bulk relax * update * update * update * update * hydrogen * update emmet * ulid tests * emmet * uc_bulk * update docs * update docs * update depent * get charge state calcs ASAP --------- Co-authored-by: Alex Ganose <[email protected]>
* start electrodes * start electrode * start electrode * start electrode * start electrode * start electrode * VASP electrode job * VASP electrode job * lint * n steps * n steps * n steps * n steps * rm defect changes * rm defect changes * update * update * update * update structure matcher update structure matcher update structure matcher update structure matcher update structure matcher update structure matcher * debugging debugging debugging debugging debugging debugging debugging debugging debugging debugging debugging debugging debugging debugging debugging * debugging * debugging * debugging * debugging * append names * append names * append names * append names * append names * dev script change dev script change dev script change dev script change dev script change dev script change dev script change dev script change test * working test * typo * lint * lint * lint * lint * allow different bulk relax * update * update * update * update * hydrogen * update emmet * ulid tests * emmet * uc_bulk * update docs * update docs * update depent * get charge state calcs ASAP --------- Co-authored-by: Alex Ganose <[email protected]>
Defect Calcs: Allow bulk SC calculation to be skipped.
Upstream changes now allow arbitrary UC Locpots to be used for the Freysoldt correction.
materialsproject/pymatgen-analysis-defects#171
So you can just start with the UC
LOCPOTand the SC defectLOCPOTand obtain the correction value.For the large-scale runs that we are doing this is a lot easier for bookkeeping purposes.
Additionally, using a relaxed bulk supercell is not directly comparable with the output of defect SC relaxation with selective dynamics.
So skipping the bulk relax calculation should also be better in that situation.