Generalized coordination #1
Conversation
… functions. So far it's incorrect
|
What does "so far the results of both CN and GCN are incorrect" mean? |
|
sorry, for being unclear. I thought only Brook is working on this. So, I explained what I meant on slack. But basically if you look at the testing notebook, you will see that the coordination numbers for Cu(100) is 4 which is incorrect. It should be 8 instead. And similarly the GCN values are incorrect because they shouldn't be this low. |
|
Brook and I are looking into why they are incorrect. I believe it's related to how the surface atoms are being tagged. In my project, I don't use the same way the to tag the surface atoms as here. I just label the atoms with CN<12 as surface atoms. |
catapalt/testing.ipynb
Outdated
| @@ -0,0 +1,95 @@ | |||
| { | |||
There was a problem hiding this comment.
please delete this file
catapalt/surface.py
Outdated
| surface_atoms = self.slab[ | ||
| [idx for idx, tag in enumerate(self.slab.get_tags()) if tag == 1] | ||
| ] | ||
| connectivity_matrix = self._get_connectivity_matrix(surface_atoms).toarray() |
There was a problem hiding this comment.
if we assess the connectivity matrix over self.slab this should fix the descrepancy you have noticed. But in this case we need to make some changes to line 235 too. I actually wanted to find the CN of surface atoms only though so I can make a duplicate function to do what you are interested in for CN
catapalt/surface.py
Outdated
| [idx for idx, tag in enumerate(self.slab.get_tags()) if tag == 1] | ||
| ] | ||
| connectivity_matrix = self._get_connectivity_matrix(surface_atoms).toarray() | ||
| cns = [sum(row) for row in connectivity_matrix] |
There was a problem hiding this comment.
this should be cns = [sum(row) for idx, row in connectivity_matrix if self.slab[idx].tag == 1] or something to that effect
catapalt/surface.py
Outdated
| Returns: | ||
| (dict): the generalized coordination info. ex. `{"mean": 5.5, "min": 5, "max": 6} | ||
| """ | ||
| surface_atoms = self.slab[ |
There was a problem hiding this comment.
this is no longer necessary
catapalt/surface.py
Outdated
| """ | ||
| connectivity_matrix = self._get_connectivity_matrix(self.slab).toarray() | ||
| cns = np.sum(connectivity_matrix, axis=1) | ||
| surface_idx = np.array([idx for idx, tag in enumerate(self.slab.get_tags()) if tag == 1]) | ||
| cn_matrix = np.array(cns)*connectivity_matrix | ||
| gcn = np.sum(cn_matrix, axis=1)/12 | ||
| gcn = np.sum(cn_matrix, axis=1)/np.max(cn_matrix) |
There was a problem hiding this comment.
@kareem-Abdelmaqsoud The way this was it wasnt extensible to bulks with bulk cn values that arent 12. I still think this doesnt handle the case where the bulk has multiple coordination numbers. i.e. some lattice positions have cn = 12 and others have cn =8 for the same bulk. How should gcn be calculated in that case?
… cn_matrix instead of the maximum of the matrix in general.
I just added the this function for calculating the generalized coordination numbers. The only thing I do is multiplying the connectivity matrix by the coordination numbers, getting a coordination matrix, and I then sum it up along one axis.
I also added a testing.ipynb to check the calculated CN and GCN values