ValueError: could not broadcast input array from shape (26,) into shape (26,1)

[Stephanowicz]

Hi!

Sorry for bothering again :D

Well, this error-text was mentioned in #319 and #321 - but I'm not sure if it is correlated...

I tried the example https://github.com/glumpy/glumpy/blob/master/examples/collection-triangles.py

The error throws up @ line 61 paths.append(P*s+(x,y,(d-1)), closed=True, color=(0,0,0,1))

where append is in agg_fast_path_collection.py @ line 104

the erroneous code is @ line 169:

V['id'] = np.tile([1,-1],len(V)//2)

where V['id'] array([[0.], [0.], [0.], [0.], [0.], [0.], [0.], [0.], [0.], [0.], [0.], [0.], [0.], [0.], [0.], [0.], [0.], [0.], [0.], [0.], [0.], [0.], [0.], [0.], [0.], [0.]], dtype=float32) [0:26] : [array([0.], dtype=float32), array([0.], dtype=float32), array([0.], dtype=float32), array([0.], dtype=float32), array([0.], dtype=float32), array([0.], dtype=float32), array([0.], dtype=float32), array([0.], dtype=float32), array([0.], dtype=float32), array([0.], dtype=float32), array([0.], dtype=float32), array([0.], dtype=float32), array([0.], dtype=float32), array([0.], dtype=float32), array([0.], dtype=float32), array([0.], dtype=float32), array([0.], dtype=float32), array([0.], dtype=float32), array([0.], dtype=float32), array([0.], dtype=float32), array([0.], dtype=float32), array([0.], dtype=float32), array([0.], dtype=float32), array([0.], dtype=float32), array([0.], dtype=float32), array([0.], dtype=float32)] dtype: dtype('float32') max: np.float32(0.0) min: np.float32(0.0) shape: (26, 1) size: 26

and np.tile([-1,1],len(V)//2)
array([ 1, -1, 1, -1, 1, -1, 1, -1, 1, -1, 1, -1, 1, -1, 1, -1, 1, -1, 1, -1, 1, -1, 1, -1, 1, -1]) [0:26] : [np.int64(1), np.int64(-1), np.int64(1), np.int64(-1), np.int64(1), np.int64(-1), np.int64(1), np.int64(-1), np.int64(1), np.int64(-1), np.int64(1), np.int64(-1), np.int64(1), np.int64(-1), np.int64(1), np.int64(-1), np.int64(1), np.int64(-1), np.int64(1), np.int64(-1), np.int64(1), np.int64(-1), np.int64(1), np.int64(-1), np.int64(1), np.int64(-1)] dtype: dtype('int64') max: np.int64(1) min: np.int64(-1) shape: (26,) size: 26

So we have V['id'] with 26 dimensions where each value is in a dimension (shape: (26, 1))
while the tile function returns a onedimensional array with 26 values (shape: (26,))

So, where is the error? Either the tile function (_shape_base_impl.py) needs to be extended (or there needs to be a different approach) Or the paths.append(P*s+(x,y,(d-1)), closed=True, color=(0,0,0,1)) is not correct...?

But as this should have been a working example (? :D) there may have been changes in numpy 2.x that trigger this ...

Cheers,

Stephan

[Stephanowicz]

@rougier btw. - which versions of numpy/pyqt did You use for the examples? Maybe I could install a parallel version to get an idea of what the code does when it's working...

Cheers,

Stephan

[rougier]

It's been w awhile since I tested glumpy and I did not record dependencies at that time.