Spectral_smooth does not work

[hongliliu]

Dear developer:
Recently, I tried smoothing my raw data cube in velocity axis with a Gaussian keneral using the function cube.spectral_smooth. However, I always received a error message. Below, I list my scripts and the associated error message after the code runs. Please help me out on how to solve this problem. Thank you very much in advance.
cube = sc.read('t13_reg1.fits')
fwhm_factor = np.sqrt(8*np.log(2))
current_resolution = 0.26569 * u.km/u.s
target_resolution = 0.5 * u.km/u.s
pixel_scale = 0.26569 * u.km/u.s
gaussian_width = ((target_resolution2 - current_resolution2)**0.5 /
pixel_scale / fwhm_factor)
kernel = Gaussian1DKernel(gaussian_width)
new_cube = cube.spectral_smooth(kernel)

---error message---
---> 10 new_cube = cube.spectral_smooth(kernel)

~/anaconda3/lib/python3.6/site-packages/spectral_cube/spectral_cube.py in spectral_smooth(self, kernel, convolve, verbose, use_memmap, num_cores, **kwargs)
2949 use_memmap=use_memmap,
2950 verbose=verbose,
-> 2951 **kwargs)
2952
2953 def spectral_interpolate(self, spectral_grid,

~/anaconda3/lib/python3.6/site-packages/spectral_cube/spectral_cube.py in apply_function_parallel_spectral(self, function, num_cores, verbose, use_memmap, parallel, **kwargs)
2882 verbose=verbose,
2883 num_cores=num_cores,
-> 2884 **kwargs
2885 )
2886

~/anaconda3/lib/python3.6/site-packages/spectral_cube/spectral_cube.py in _apply_function_parallel_base(self, iteration_data, function, applicator, num_cores, verbose, use_memmap, parallel, memmap_dir, update_function, **kwargs)
2762
2763 for arg in iteration_data:
-> 2764 applicator(arg, outcube, function, **kwargs)
2765 pbu()
2766

~/anaconda3/lib/python3.6/site-packages/spectral_cube/spectral_cube.py in _apply_spectral_function(arguments, outcube, function, **kwargs)
108
109 if np.any(includemask):
--> 110 outcube[:,jj,ii] = function(spec, **kwargs)
111 else:
112 outcube[:,jj,ii] = spec

~/anaconda3/lib/python3.6/site-packages/astropy/nddata/decorators.py in wrapper(data, *args, **kwargs)
244 AstropyUserWarning)
245
--> 246 result = func(data, *args, **kwargs)
247
248 if unpack and repack:

~/anaconda3/lib/python3.6/site-packages/astropy/convolution/convolve.py in convolve(array, kernel, boundary, fill_value, nan_treatment, normalize_kernel, mask, preserve_nan, normalization_zero_tol)
336 if normalize_kernel:
337 if not nan_interpolate:
--> 338 result /= kernel_sum
339 else:
340 if nan_interpolate:

~/anaconda3/lib/python3.6/site-packages/astropy/units/quantity.py in array_ufunc(self, function, method, *inputs, **kwargs)
473 return result
474
--> 475 return self._result_as_quantity(result, unit, out)
476
477 def _result_as_quantity(self, result, unit, out):

~/anaconda3/lib/python3.6/site-packages/astropy/units/quantity.py in _result_as_quantity(self, result, unit, out)
512 # the output is of the correct Quantity subclass, as it was passed
513 # through check_output.
--> 514 out._set_unit(unit)
515 return out
516

AttributeError: 'numpy.ndarray' object has no attribute '_set_unit'

[e-koch]

@hongliliu -- Can you try changing the second last line to kernel = Gaussian1DKernel(gaussian_width.value) and see if that works?

[hongliliu]

@hongliliu -- Can you try changing the second last line to kernel = Gaussian1DKernel(gaussian_width.value) and see if that works?

It works now following your suggestion. Thank you very much for your kind help.

[e-koch]

This issue also came up in #514, but it wasn't fixed by #507.

[e-koch]

Addressed in #578