bug: fixed swapaxes handling

[zerothi]

Now the swapaxes is correct for Cartesian coordinate swaps.

Also extended it to be much more versatile.
One can do collapsed swappings, which might not be that useful. But rather easy to follow given that I wanted the str arguments.

Thanks to @ahkole for discovering the bug leading to this.

Signed-off-by: Nick Papior [email protected]

• Closes What should Geometry.swapaxes do if it includes xyz in what? #539
• Tests added
• Documentation for functionality
• User visible changes (including notable bug fixes) are documented in CHANGELOG

[codecov]

Codecov Report

Merging #540 (6a273cb) into main (09674fc) will increase coverage by 0.03%.
The diff coverage is 96.06%.

@@            Coverage Diff             @@
##             main     #540      +/-   ##
==========================================
+ Coverage   86.24%   86.27%   +0.03%     
==========================================
  Files         361      362       +1     
  Lines       46253    46367     +114     
==========================================
+ Hits        39889    40002     +113     
- Misses       6364     6365       +1     

Impacted Files	Coverage Δ
sisl/nodes/workflow.py	0.00% <0.00%> (ø)
sisl/supercell.py	95.26% <84.61%> (-0.86%)	⬇️
sisl/geometry.py	86.67% <100.00%> (+0.15%)	⬆️
sisl/mixing/base.py	93.67% <100.00%> (+0.42%)	⬆️
sisl/mixing/diis.py	96.05% <100.00%> (+0.10%)	⬆️
sisl/mixing/linear.py	66.66% <100.00%> (+1.80%)	⬆️
sisl/tests/test_geometry.py	99.67% <100.00%> (+<0.01%)	⬆️
sisl/tests/test_supercell.py	100.00% <100.00%> (ø)
sisl/utils/tests/test_arrays.py	100.00% <100.00%> (ø)
sisl/geom/nanotube.py	95.83% <0.00%> (+1.04%)	⬆️
... and 1 more

Help us with your feedback. Take ten seconds to tell us how you rate us. Have a feature suggestion? Share it here.

[zerothi]

@ahkole I don't know if the message pinged you or not, but here it is again :)

[zerothi]

@ahkole I will merge this as I think it is the correct thing, thanks for bringing this to my attention!

[ahkole]

@zerothi Hi, sorry. I recently switched phones and haven't setup everything on my new phone yet (including the mail account where I receive GitHub notifications) so I completely missed this, sorry.

I had a look at the new implementation and as far as I can see all the swaps now work as I would expect. Maybe the only thing that could still be improved is the docstrings for the swapaxes method. First of all, inside the Geometry.swapaxes docstring there is this example,

g_s = g.swapaxes("abx", "bcz", 1, "xyz")

But this call should not be possible because swapaxes can get at most 3 arguments. Furthermore, what wasn't immediately clear to me from the docstring is that the what parameter is only used if you use integer arguments for axis_a and axis_b and also that using a string of labels instead of a single label for axis_a and axis_b means that multiple swaps will be executed in serial one after the other.

[zerothi]

@ahkole thanks for taking a look! The recent commit should be more verbose! Thanks for finding the doc bug ;)