Resolve compatibility with Scipy 1.2 about hcipy HOT 11 CLOSED

--CODE SNIPPET-- class FresnelBack(hci.OpticalElement):         def __init__(self, input_grid, distance, wavelength=1):         self.input_grid = input_grid         self.distance = distance         self.wavelength = wavelength         self.prop = hci.FresnelPropagator(input_grid=self.input_grid                                          , distance=self.distance)     def forward(self, wavefront):         return self.prop.backward(wavefront)     def backward(self, wavefront):         return self.prop.forward(wavefront)----- When I run the forward propagation using the latest HCIPY, I get the following error: --ERROR MESSAGE--RuntimeError: Output grid is not known. Perform a forward propagation first before backwards propagation on the same grid.----- Performing a forward propagation first (as the error message suggests) gets over this error, but it causes problems elsewhere in my code.  Also, it seems inefficient to do an extra propagation ...  Is there anyway I can pass in the output grid without having to run a forward propagation first?  Using HCIPY's built-in `FresnelPropagator` and running .backward() throws the same error.  FYI, the code runs fine with an older HCIPY release.  Thank you very much in advance! Pin On Monday, January 6, 2020, 12:27:28 AM PST, Emiel H. Por <[email protected]> wrote: No problem. FYI: HCIPy 0.3 will be out in a few days and implements, among other things, full Stokes propagation through vector vortex coronagraphs. — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub, or unsubscribe.

--EXCERPT FROM ERROR MESSAGE--   File "/home/pinchen/sleec/source/sleec_optics.py", line 159, in forward     return self.prop.backward(wavefront)   File "/home/pinchen/anaconda3/lib/python3.7/site-packages/hcipy/optics/optical_element.py", line 761, in res     instance_data = self.get_instance_data(None, wavefront.grid, wavefront.wavelength)   File "/home/pinchen/anaconda3/lib/python3.7/site-packages/hcipy/optics/optical_element.py", line 633, in get_instance_data     self.make_instance(instance_data, input_grid, output_grid, wavelength)   File "/home/pinchen/anaconda3/lib/python3.7/site-packages/hcipy/propagation/fraunhofer.py", line 38, in make_instance     instance_data.fourier_transform = make_fourier_transform(input_grid, instance_data.uv_grid)   File "/home/pinchen/anaconda3/lib/python3.7/site-packages/hcipy/fourier/fourier_transform.py", line 162, in make_fourier_transform     if input_grid.is_separated and input_grid.is_('cartesian') and output_grid.is_separated and output_grid.is_('cartesian') and input_grid.ndim in [1,2]: AttributeError: 'function' object has no attribute 'is_separated'----- Any advice would be much appreciated! Best, Pin On Wednesday, January 22, 2020, 11:27:13 AM PST, Emiel H. Por <[email protected]> wrote: With HCIPy 0.2 you can indeed bypass by having a negative distance. However, in HCIPy 0.3.0, out since yesterday, there are should be no problems doing a backward() propagation right away, as part of a renewed way of handling caching in optical elements. — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub, or unsubscribe.