J. George
J. George
In any case, thank you for pointing this out.
@shyuep Do you maybe have more insight into the development of this method and potential reasons for the difference between paper and implementation? (The scale factor alone does not make...
We could add a note that this discrepancy of 0.31 and 0.37 exists or make it adjustable, but as the oxidation state estimator depends on this method, we cannot really...
However, I am also still not getting the bond valance parameters in the file. According to the paper, there should also be an oxidation state dependency. Maybe, it is indeed...
Thanks for clarifying! (Sorry for just skimming through the paper) We still have the issue that the subsequent oxidation state model is based on old parameters.
@trioxane Are the parameters listed here agreeing with the publication from 1991? [`pymatgen/analysis/bvparam_1991.yaml`](https://github.com/materialsproject/pymatgen/blob/ec750ca15d02cdd51b0c0a7a4408af8e0d259223/pymatgen/analysis/bvparam_1991.yaml)
I also went back to the implementation and especially the documentation of the oxidation state predictor. As far as I understand, the oxidation states are estimated by a Naive Bayes...
Similar to the mentioned issue above (wrong b value), this would mean recomputing the classifier again (or using the model with different parameters?). I have also no real feeling if...
I know this is not really an urgent matter, but as this oxidation state predictor is an important part of many subsequent codes, the MP website, it might be good...
@rkingsbury this issue can be closed, right?