Pretty clusters #181

lbluque · 2022-04-05T00:24:52Z

Summary

Improvements for printing and visualizing smol objects

ClusterSubspace, ClusterExpansion, Orbit, Cluster have much nicer repr and str methods.
Cluster as a subclass of pymatgen SiteCollection
Allow cluster visualization using crystaltoolkit and includes an example notebook. (this required a little bit of refactoring, but should not really affect users)

TODO (if any)

Implement str and repr for missing classes
- StructureWrangler
- Processors
- Ensembles
- Kernel
- Sampler and SampleContainer

Checklist

Code is in the standard Python style.
Run pycodestyle and flake8
on your local machine.
Docstrings have been added in the Google docstring format.
Run pydocstyle on your code.
Tests have been added for any new functionality or bug fixes.
All existing tests pass.

…ts need updates)

# Conflicts: # smol/cofe/space/cluster.py # smol/cofe/space/orbit.py

# Conflicts: # tests/test_cofe/test_cluster.py # tests/test_cofe/test_orbit.py # tests/utils.py

lbluque · 2022-04-07T04:14:42Z

This addresses #93 by leveraging crystal-toolkit for jupyter notebooks.

lbluque · 2022-04-07T17:12:26Z

A ClusterExpansion instance is printed as:

Basis/Orthogonal/Orthonormal : indicator/False/{self.basis_orthonormal}
       Unit Cell Composition : Mo5+0.33333333 Co2+0.33333333 Li+0.33333333 O2-0.5 F-0.5
            Number of Orbits : 34
No. of Correlation Functions : 82
             Cluster Cutoffs : 2: 6.96, 3: 3.64, 4: 3.64
              External Terms : [EwaldTerm(total)]
ECI-based Property : mean=-5.5733  std=4.9130
Fit Summary
EwaldTerm(total)=0.050
 ---------------------------------------------------------
 |  ID    Orbit ID    Degree    Cluster Diameter    ECI  |
 |   0       0          0              NA        -5.573  |
 |  1        1          1            0.0000       -0.593 |
 |  2        1          1            0.0000       -2.666 |
 |  3        2          1            0.0000       -3.241 |
 |  4        3          2            2.1000        0.002 |
 |  5        3          2            2.1000        0.222 |
 |  6        4          2            2.9698       -0.361 |
 |  7        4          2            2.9698       -0.138 |
 |  8        4          2            2.9698        0.046 |
 |  9        5          2            2.9698        0.094 |
....

lbluque · 2022-04-07T17:13:17Z

A ClusterSubspace:

Basis/Orthogonal/Orthonormal : indicator/False/{self.basis_orthonormal}
       Unit Cell Composition : Mo5+0.33333333 Co2+0.33333333 Li+0.33333333 O2-0.5 F-0.5
            Number of Orbits : 34
No. of Correlation Functions : 82
             Cluster Cutoffs : 2: 6.96, 3: 3.64, 4: 3.64
              External Terms : [EwaldTerm(total)]
Orbit Summary
 ------------------------------------------------------------------------
 |  ID     Degree    Cluster Diameter    Multiplicity    No. Functions  |
 |   0       0             NA                 0                1        |
 |   1       1            0.0000              1                2        |
 |   2       1            0.0000              1                1        |
 |   3       2            2.1000              6                2        |
 |   4       2            2.9698              6                3        |
 |   5       2            2.9698              6                1        |
 |   6       2            3.6373              8                2        |
 |   7       2            4.2000              3                3        |
 |   8       2            4.2000              3                1        |
 |   9       2            4.6957              24               2        |
 |  10       2            5.1439              12               3        |
 |  11       2            5.1439              12               1        |

lbluque · 2022-04-07T17:14:49Z

An Orbit:

Orbit  33
    Multiplicity : 12  
   No. functions : 3   
No. symmetry ops : 4   
Function ids : [79, 80, 81]
Base Cluster : 
  | Diameter : 3.6373
  |   Charge : 2.333333333333333
  | Centroid :    -1.714643     0.606218    -0.000000  ->     0.250000     0.250000     0.250000
  | Sites (4)
  | ----------------------------------------------------------------------------------------------------------------------
  | 0 Li+:0.333, Co2+:0.333, Mo5+:0.333    -1.714643     2.424871     0.000000   ->     0.000000    -0.000000     1.000000
  | 1 O2-:0.500, F-:0.500    -3.429286     1.212436    -0.000000   ->     0.500000     0.500000     0.500000
  | 2 Li+:0.333, Co2+:0.333, Mo5+:0.333     0.000000     0.000000     0.000000   ->     0.000000     0.000000     0.000000
  | 3 O2-:0.500, F-:0.500    -1.714643    -1.212436    -0.000000   ->     0.500000     0.500000    -0.500000

lbluque · 2022-04-07T17:15:36Z

a Cluster

Diameter : 3.6373
  Charge : 2.333333333333333
Centroid :    -1.714643     0.606218    -0.000000  ->     0.250000     0.250000     0.250000
Sites (4)
----------------------------------------------------------------------------------------------------------------------
0 Li+:0.333, Co2+:0.333, Mo5+:0.333    -1.714643     2.424871     0.000000   ->     0.000000    -0.000000     1.000000
1 O2-:0.500, F-:0.500    -3.429286     1.212436    -0.000000   ->     0.500000     0.500000     0.500000
2 Li+:0.333, Co2+:0.333, Mo5+:0.333     0.000000     0.000000     0.000000   ->     0.000000     0.000000     0.000000
3 O2-:0.500, F-:0.500    -1.714643    -1.212436    -0.000000   ->     0.500000     0.500000    -0.500000

lbluque · 2022-04-07T17:19:49Z

And we can visualize clusters in jupyter super easy using crystal-toolkit!

from monty.serialization import loadfn
import crytal_toolkit

subspace = loadfn("co2mo5.mson")
subspace.orbits[-1].base_cluster

gives:

lbluque · 2022-04-07T17:20:44Z

I won't do the __str__ and __repr__ for the missing classes in this PR.

# Conflicts: # tests/test_cofe/test_cluster.py # tests/test_cofe/test_orbit.py

qchempku2017 · 2022-04-08T01:06:55Z

And we can visualize clusters in jupyter super easy using crystal-toolkit!
from monty.serialization import loadfn
import crytal_toolkit

subspace = loadfn("co2mo5.mson")
subspace.orbits[-1].base_cluster
gives:

I think it would be more useful if we can also plot out the decorated orbit that each correlation function corresponds to.

qchempku2017 · 2022-04-08T01:08:05Z

A ClusterExpansion instance is printed as:

Basis/Orthogonal/Orthonormal : indicator/False/{self.basis_orthonormal}
       Unit Cell Composition : Mo5+0.33333333 Co2+0.33333333 Li+0.33333333 O2-0.5 F-0.5
            Number of Orbits : 34
No. of Correlation Functions : 82
             Cluster Cutoffs : 2: 6.96, 3: 3.64, 4: 3.64
              External Terms : [EwaldTerm(total)]
ECI-based Property : mean=-5.5733  std=4.9130
Fit Summary
EwaldTerm(total)=0.050
 ---------------------------------------------------------
 |  ID    Orbit ID    Degree    Cluster Diameter    ECI  |
 |   0       0          0              NA        -5.573  |
 |  1        1          1            0.0000       -0.593 |
 |  2        1          1            0.0000       -2.666 |
 |  3        2          1            0.0000       -3.241 |
 |  4        3          2            2.1000        0.002 |
 |  5        3          2            2.1000        0.222 |
 |  6        4          2            2.9698       -0.361 |
 |  7        4          2            2.9698       -0.138 |
 |  8        4          2            2.9698        0.046 |
 |  9        5          2            2.9698        0.094 |
....

In here I suggest printing out the species corresponding to each decorated orbit (correlation function as well.)

qchempku2017 · 2022-04-08T01:09:32Z

A ClusterSubspace:

Basis/Orthogonal/Orthonormal : indicator/False/{self.basis_orthonormal}
       Unit Cell Composition : Mo5+0.33333333 Co2+0.33333333 Li+0.33333333 O2-0.5 F-0.5
            Number of Orbits : 34
No. of Correlation Functions : 82
             Cluster Cutoffs : 2: 6.96, 3: 3.64, 4: 3.64
              External Terms : [EwaldTerm(total)]
Orbit Summary
 ------------------------------------------------------------------------
 |  ID     Degree    Cluster Diameter    Multiplicity    No. Functions  |
 |   0       0             NA                 0                1        |
 |   1       1            0.0000              1                2        |
 |   2       1            0.0000              1                1        |
 |   3       2            2.1000              6                2        |
 |   4       2            2.9698              6                3        |
 |   5       2            2.9698              6                1        |
 |   6       2            3.6373              8                2        |
 |   7       2            4.2000              3                3        |
 |   8       2            4.2000              3                1        |
 |   9       2            4.6957              24               2        |
 |  10       2            5.1439              12               3        |
 |  11       2            5.1439              12               1        |

Same suggestion as with ClusterExpansion printing. With these, illustrating ECIs in a paper's SI will be infinitely easier!!

qchempku2017 · 2022-04-08T01:10:26Z

I won't do the __str__ and __repr__ for the missing classes in this PR.

We can do them in an independent PR.

smol/cofe/expansion.py

smol/cofe/space/clusterspace.py

lbluque · 2022-04-08T02:32:45Z

A ClusterExpansion instance is printed as:

Basis/Orthogonal/Orthonormal : indicator/False/{self.basis_orthonormal}
       Unit Cell Composition : Mo5+0.33333333 Co2+0.33333333 Li+0.33333333 O2-0.5 F-0.5
            Number of Orbits : 34
No. of Correlation Functions : 82
             Cluster Cutoffs : 2: 6.96, 3: 3.64, 4: 3.64
              External Terms : [EwaldTerm(total)]
ECI-based Property : mean=-5.5733  std=4.9130
Fit Summary
EwaldTerm(total)=0.050
 ---------------------------------------------------------
 |  ID    Orbit ID    Degree    Cluster Diameter    ECI  |
 |   0       0          0              NA        -5.573  |
 |  1        1          1            0.0000       -0.593 |
 |  2        1          1            0.0000       -2.666 |
 |  3        2          1            0.0000       -3.241 |
 |  4        3          2            2.1000        0.002 |
 |  5        3          2            2.1000        0.222 |
 |  6        4          2            2.9698       -0.361 |
 |  7        4          2            2.9698       -0.138 |
 |  8        4          2            2.9698        0.046 |
 |  9        5          2            2.9698        0.094 |
....

In here I suggest printing out the species corresponding to each decorated orbit (correlation function as well.)

I considered this (we had something along those lines before) however I think that makes the print out way too long. Users can simply print out an orbit to get that information, and if they want to see all species for all orbits then simply print out in a loop:

for orbit in subspace.orbits:
    print(orbit)

lbluque · 2022-04-08T02:41:58Z

I think it would be more useful if we can also plot out the decorated orbit that each correlation function corresponds to.

I'm not sure I fully understand what you mean? Do you mean getting the species decorations corresponding to a site indicator correlation function?

If that's the case then that would only apply to the indicator basis scenario. Even so I am not so sure how to visualize that exactly. Perhaps by visualizing a specific correlation function? This would be require a slightly different set of code and again only for an indicator case.

…tty_clusters � Conflicts: � smol/cofe/space/cluster.py

…sters

qchempku2017

Review done, nice job! Go ahead and merge if test pass.

qchempku2017

Please merge.

lbluque added 6 commits March 9, 2022 07:56

clusters as sitecollection hybrid between molecule and structure (tes…

bf650b2

…ts need updates)

Merge branch 'master' into pretty_clusters

1f35878

# Conflicts: # smol/cofe/space/cluster.py # smol/cofe/space/orbit.py

clusters as pmg site collections (and some fixes)

d8ac406

Merge branch 'master' into pretty_clusters

fb2534b

# Conflicts: # tests/test_cofe/test_cluster.py # tests/test_cofe/test_orbit.py # tests/utils.py

better repr and str for clusters and orbits

3b26353

pretty str for clusterspace

822200d

lbluque marked this pull request as draft April 5, 2022 00:25

lbluque and others added 6 commits April 4, 2022 17:35

clustersubspace repr

92f7b02

clusterexpansion pretty str

b138f71

repr in clusterexpansion and ewaldterm

96e53aa

cluster vis in jupyter with crystal-toolkit

908bbda

rename notebooks and add vis notebook

a36095b

Merge branch 'master' into pretty_clusters

c41ad3e

lbluque changed the title ~~[WIP] Pretty clusters~~ Pretty clusters Apr 7, 2022

lbluque marked this pull request as ready for review April 7, 2022 04:13

lbluque requested a review from qchempku2017 April 7, 2022 17:20

lbluque and others added 3 commits April 7, 2022 12:53

Merge branch 'main' into pretty_clusters

d576985

# Conflicts: # tests/test_cofe/test_cluster.py # tests/test_cofe/test_orbit.py

fix tests

2d0568b

Merge branch 'main' into pretty_clusters

a6d00d3

lbluque added the enhancement New feature or request label Apr 7, 2022

lbluque linked an issue Apr 7, 2022 that may be closed by this pull request

Clusters visualization? #93

Closed

Merge branch 'main' into pretty_clusters

2e2e9ed

qchempku2017 reviewed Apr 8, 2022

View reviewed changes

smol/cofe/expansion.py Show resolved Hide resolved

qchempku2017 reviewed Apr 8, 2022

View reviewed changes

smol/cofe/space/clusterspace.py Show resolved Hide resolved

lbluque added 2 commits April 8, 2022 09:12

Merge branch 'main' of https://github.com/CederGroupHub/smol into pre…

e80eb65

…tty_clusters � Conflicts: � smol/cofe/space/cluster.py

Merge remote-tracking branch 'origin/pretty_clusters' into pretty_clu…

5206e66

…sters

qchempku2017 reviewed Apr 8, 2022

View reviewed changes

qchempku2017 approved these changes Apr 8, 2022

View reviewed changes

lbluque merged commit cb304af into CederGroupHub:main Apr 8, 2022

lbluque deleted the pretty_clusters branch April 8, 2022 22:40

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Pretty clusters #181

Pretty clusters #181

lbluque commented Apr 5, 2022 •

edited

Loading

lbluque commented Apr 7, 2022

lbluque commented Apr 7, 2022

lbluque commented Apr 7, 2022

lbluque commented Apr 7, 2022

lbluque commented Apr 7, 2022

lbluque commented Apr 7, 2022

lbluque commented Apr 7, 2022

qchempku2017 commented Apr 8, 2022

qchempku2017 commented Apr 8, 2022

qchempku2017 commented Apr 8, 2022

qchempku2017 commented Apr 8, 2022

lbluque commented Apr 8, 2022

lbluque commented Apr 8, 2022

qchempku2017 left a comment

qchempku2017 left a comment

Pretty clusters #181

Pretty clusters #181

Conversation

lbluque commented Apr 5, 2022 • edited Loading

Summary

TODO (if any)

Checklist

lbluque commented Apr 7, 2022

lbluque commented Apr 7, 2022

lbluque commented Apr 7, 2022

lbluque commented Apr 7, 2022

lbluque commented Apr 7, 2022

lbluque commented Apr 7, 2022

lbluque commented Apr 7, 2022

qchempku2017 commented Apr 8, 2022

qchempku2017 commented Apr 8, 2022

qchempku2017 commented Apr 8, 2022

qchempku2017 commented Apr 8, 2022

lbluque commented Apr 8, 2022

lbluque commented Apr 8, 2022

qchempku2017 left a comment

Choose a reason for hiding this comment

qchempku2017 left a comment

Choose a reason for hiding this comment

lbluque commented Apr 5, 2022 •

edited

Loading