Please make sure these conditions are met <li cl

Checking array equality during merging doesn't work for >2d about anndata HOT 3 CLOSED

ivirshup commented on July 18, 2024

Checking array equality during merging doesn't work for >2d

from anndata.

Comments (3)

ivirshup commented on July 18, 2024 1

These arrays aren't being concatenated, they are just being compared.

This is just to see if they are equal so we know whether to keep that around in the result or not. https://anndata.readthedocs.io/en/latest/concatenation.html#merging-uns

The specific use case here was comparing images. It's very possible that we should look into faster ways to compare these things. Maybe fast paths when the objects are the same, hashing, etc.

from anndata.

ivirshup commented on July 18, 2024

Think I figured out why we convert to pandas. It's because np.array_equal(..., equal_nan=True) doesn't work for some non-numeric dtypes...

Specifically:

In [6]: a = anndata.tests.helpers.gen_vstr_recarray(5, 2)

In [7]: np.array_equal(a, a, equal_nan=True)
---------------------------------------------------------------------------
TypeError                                 Traceback (most recent call last)
Cell In[7], line 1
----> 1 np.array_equal(a, a, equal_nan=True)

File <__array_function__ internals>:180, in array_equal(*args, **kwargs)

File ~/miniforge3/envs/scanpy-dev/lib/python3.11/site-packages/numpy/core/numeric.py:2465, in array_equal(a1, a2, equal_nan)
   2463     return bool(asarray(a1 == a2).all())
   2464 # Handling NaN values if equal_nan is True
-> 2465 a1nan, a2nan = isnan(a1), isnan(a2)
   2466 # NaN's occur at different locations
   2467 if not (a1nan == a2nan).all():

TypeError: ufunc 'isnan' not supported for the input types, and the inputs could not be safely coerced to any supported types according to the casting rule ''safe''