@staticmethod

def _pattern(npixel):

return coordinates2(npixel)[0] + coordinates2(npixel)[1] * 1j

def test_pad_extract(self):

for npixel, N2 in [(100, 128), (128, 256), (126, 128)]:

# Make a 2D complex image of size (npixel, npixel) centred on (npixel//2, npixel//2)

cs = 1 + self._pattern(npixel)

# Pad it and extract npixel pixels around the centre

cs_pad = pad_mid(cs, N2)

# Now create the pattern we expect directly

cs2 = 1 + self._pattern(N2) * N2 / npixel

# At this point all fields in cs2 and cs_pad should either

# be equal or zero.

equal = numpy.abs(cs_pad - cs2) < 1e-15

zero = numpy.abs(cs_pad) < 1e-15

assert (equal + zero).all(), "Pad (%d, %d) failed" % (npixel, N2)

# And extracting the middle should recover the original data

assert_allclose(extract_mid(cs_pad, npixel), cs)

def test_extract_oversampled(self):

for npixel, kernel_oversampling in [(1, 2), (2, 3), (3, 2), (4, 2), (5, 3)]:

a = 1 + self._pattern(npixel * kernel_oversampling)

ex = extract_oversampled(a, 0, 0, kernel_oversampling, npixel) / kernel_oversampling ** 2