Update autode to 1.3.3

[juraskov]

Preliminary update to autode v1.3.3. Update to the latest version will continue once I figure out the source of npz loading issue which started to occur in v1.3.4

Also closes #127

[codecov-commenter]

Codecov Report

Attention: Patch coverage is 86.36364% with 3 lines in your changes missing coverage. Please review.

Project coverage is 64.90%. Comparing base (339c5af) to head (4e0226b).

Files with missing lines	Patch %	Lines
mlptrain/configurations/calculate.py	50.00%	2 Missing ⚠️
mlptrain/configurations/configuration_set.py	94.44%	1 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #121      +/-   ##
==========================================
- Coverage   64.99%   64.90%   -0.09%     
==========================================
  Files          37       37              
  Lines        4268     4280      +12     
==========================================
+ Hits         2774     2778       +4     
- Misses       1494     1502       +8     

Flag	Coverage Δ
python-3.9	60.95% <86.36%> (-0.08%)	⬇️
python-3.9-ace	60.95% <86.36%> (-0.08%)	⬇️
python-3.9-mace	64.71% <86.36%> (-0.09%)	⬇️

Flags with carried forward coverage won't be shown. Click here to find out more.

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[juraskov]

Error observed when loading mg_aqua_mace_al.npz file (prepared with autode1.1.3) using autode1.4.4

>>> data = mlt.ConfigurationSet()
>>> data.load('mg_aqua_mace_al.npz')
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "/Users/veronikajuraskova/SOFTWARE/mlp-train/mlptrain/configurations/configuration_set.py", line 493, in load
    self._load_npz(filename)
  File "/Users/veronikajuraskova/SOFTWARE/mlp-train/mlptrain/configurations/configuration_set.py", line 685, in _load_npz
    if data['F_true'].ndim > 0:
  File "/Users/veronikajuraskova/miniconda3/envs/mlptrain-gap/lib/python3.9/site-packages/numpy/lib/npyio.py", line 256, in __getitem__
    return format.read_array(bytes,
  File "/Users/veronikajuraskova/miniconda3/envs/mlptrain-gap/lib/python3.9/site-packages/numpy/lib/format.py", line 800, in read_array
    array = pickle.load(fp, **pickle_kwargs)
  File "/Users/veronikajuraskova/miniconda3/envs/mlptrain-gap/lib/python3.9/site-packages/autode/values.py", line 667, in __setstate__
    self.__dict__.update(state[-1])
TypeError: cannot convert dictionary update sequence element #0 to a sequence

[Hanwen1018]

1. allow training MACE with dipole
2. update mace
3. allow multi-head fine-tuning
4. allow mace training on multi-GPU cores

[juraskov]

Error observed with using autode v1.1.3 to 1.3.3, most probably data set specific:

import numpy as np
data_np = np.load('mg_aqua_mace_al.npz',allow_pickle=True)
data_np['F_true']
Traceback (most recent call last):
File "", line 1, in
File "/Users/veronikajuraskova/miniconda3/envs/mlptrain-gap-old/lib/python3.9/site-packages/numpy/core/arrayprint.py", line 1612, in _array_str_implementation
return array2string(a, max_line_width, precision, suppress_small, ' ', "")
File "/Users/veronikajuraskova/miniconda3/envs/mlptrain-gap-old/lib/python3.9/site-packages/numpy/core/arrayprint.py", line 736, in array2string
return _array2string(a, options, separator, prefix)
File "/Users/veronikajuraskova/miniconda3/envs/mlptrain-gap-old/lib/python3.9/site-packages/numpy/core/arrayprint.py", line 513, in wrapper
return f(self, *args, **kwargs)
File "/Users/veronikajuraskova/miniconda3/envs/mlptrain-gap-old/lib/python3.9/site-packages/numpy/core/arrayprint.py", line 546, in _array2string
lst = _formatArray(a, format_function, options['linewidth'],
File "/Users/veronikajuraskova/miniconda3/envs/mlptrain-gap-old/lib/python3.9/site-packages/numpy/core/arrayprint.py", line 889, in _formatArray
return recurser(index=(),
File "/Users/veronikajuraskova/miniconda3/envs/mlptrain-gap-old/lib/python3.9/site-packages/numpy/core/arrayprint.py", line 845, in recurser
word = recurser(index + (-i,), next_hanging_indent, next_width)
File "/Users/veronikajuraskova/miniconda3/envs/mlptrain-gap-old/lib/python3.9/site-packages/numpy/core/arrayprint.py", line 799, in recurser
return format_function(a[index])
File "/Users/veronikajuraskova/miniconda3/envs/mlptrain-gap-old/lib/python3.9/site-packages/numpy/core/arrayprint.py", line 395, in _object_format
return fmt.format(o)
File "/Users/veronikajuraskova/miniconda3/envs/mlptrain-gap-old/lib/python3.9/site-packages/autode/values.py", line 592, in repr
return f'Gradients({np.ndarray.str(self)} {self.units.name})'
AttributeError: 'Gradient' object has no attribute 'units'

[danielhollas]

Looks great, just some minor comments.

[danielhollas]

This is clever but hard to read. I would trade the for loops for a bit of (straightforward) code repetition.

[danielhollas]

The next investigation step here, ideally before merging, would be to try to minimize the water_al.npz test file, i.e. using an older mlptrain version, with autode==1.1 and older numpy < 1.26, hopefully taking only a few configurations (or only one? We saw the error got triggered for the first config) should suffice to trigger the problem.

[danielhollas]

Nice, what an epic debugging. I'll open an issue related to autode=1.3.4 and loading of old npz files.