shiftnd crashes (also in example notebook)

[ItayKishon-Q]

Hi,
It seems that shift.shiftnd crashes, with the following error:

Cell In[4], line 4
      2 im1 = image_registration.tests.make_extended(100)
      3 # create an offset version corrupted by noise
----> 4 im2 = image_registration.tests.make_offset_extended(im1, 4.76666, -12.33333333333333333333333, noise=0.1)
      5 pl.subplot(121); img1=pl.imshow(im1, cmap='viridis', interpolation='nearest')
      6 pl.subplot(122); img2=pl.imshow(im2, cmap='viridis', interpolation='nearest')

File <myenv>/lib/python3.10/site-packages/image_registration/tests/registration_testing.py:124, in make_offset_extended(img, xsh, ysh, noise, mode, noise_taper)
    122     noise /= edge_weight(img.shape[0])
    123 #newimage = scipy.ndimage.map_coordinates(img+noise, [yy,xx], mode=mode)
--> 124 newimage = np.real(shift.shiftnd(img, (ysh, xsh))+noise)
    126 return newimage

File <myenv>/lib/python3.10/site-packages/image_registration/fft_tools/shift.py:100, in shiftnd(data, offset, phase, nthreads, use_numpy_fft, return_abs, return_real)
     97 if np.any(np.isnan(data)):
     98     data = np.nan_to_num(data)
--> 100 freq_grid = np.sum(
    101     [off*np.fft.fftfreq(nx)[
    102         tuple(
    103             [np.newaxis]*dim + [slice(None)] + [np.newaxis]*(data.ndim-dim-1)
    104             )
    105         ]
    106         for dim,(off,nx) in enumerate(zip(offset,data.shape))],
    107     axis=0)
    109 kernel = np.exp(-1j*2*np.pi*freq_grid-1j*phase)
    111 result = ifftn( fftn(data) * kernel )

File <myenv>/envs/itay_v1/lib/python3.10/site-packages/numpy/core/fromnumeric.py:2313, in sum(a, axis, dtype, out, keepdims, initial, where)
   2310         return out
   2311     return res
-> 2313 return _wrapreduction(a, np.add, 'sum', axis, dtype, out, keepdims=keepdims,
   2314                       initial=initial, where=where)

File <myenv>/lib/python3.10/site-packages/numpy/core/fromnumeric.py:88, in _wrapreduction(obj, ufunc, method, axis, dtype, out, **kwargs)
     85         else:
     86             return reduction(axis=axis, out=out, **passkwargs)
---> 88 return ufunc.reduce(obj, axis, dtype, out, **passkwargs)

ValueError: setting an array element with a sequence. The requested array has an inhomogeneous shape after 1 dimensions. The detected shape was (2,) + inhomogeneous part.

Note that the above is from the package's official demo:
https://github.com/keflavich/image_registration/blob/master/examples/Cross%20Correlation.ipynb

BTW: the above notebook is missing import numpy as np

I have a wild guess that this has something to do with Numpy version, and that something has changed (perhaps in fft.fftfreq?).
I'm using Numpy v1.26.2

Thanks.

[keflavich]

Thanks, I can reproduce this with the minimal example in the readme. It may well be a numpy version issue.

[mikebrv-q]

Thank you for updating. we would be happy if you could release a new version with this fix.