def test_autocenter_no_side_effects():
""" Test that autocenter() does not modify the inputs """
im = np.random.random(size=(256, 256))
mask = np.ones_like(im, dtype=bool)
# Modifying the arrays will result in "ValueError: output array is read-only"
im.setflags(write=False)
mask.setflags(write=False)
autocenter(im, mask=mask)
def test_autocenter_gaussian_shifted(rc, cc):
""" Test that autocenter() finds the center of a shifted gaussian """
im = np.zeros(shape=(128, 128), dtype=float)
rows, cols = np.indices(im.shape)
center = np.array([64 + rc, 64 + cc])
im += gaussian([rows, cols], center=center, fwhm=20)
assert np.allclose(autocenter(im), center, atol=1)
def test_autocenter_trivial():
""" Test that autocenter() finds the center of perfect gaussian at (0,0) """
im = np.zeros(shape=(256, 256), dtype=float)
center = np.asarray(im.shape) / 2
rows, cols = np.indices(im.shape)
im += gaussian([rows, cols], center=center, fwhm=center.max() / 2)
assert np.allclose(autocenter(im), center, atol=1)
def test_autocenter_single_crystal(rc, cc):
"""Test that autocenter() finds the center of a simulated
shifted single-crystal diffraction pattern."""
I = 10 * diff_pattern_sc(center=(rc, cc))
mask = np.ones_like(I, dtype=bool)
mask[0:rc + 10, cc - 10:cc + 10] = False
I += 0.01 * I.max() * np.random.random(size=I.shape)
assert np.allclose(autocenter(I, mask=mask), (rc, cc), atol=1)
def test_autocenter_shifted_with_mask(rc, cc):
"""Test that autocenter() finds the center of a shifted gaussian, where
the center has been masked away."""
im = np.zeros(shape=(256, 256), dtype=float)
mask = np.ones_like(im, dtype=bool)
mask[0:130, 118:138] = False
rows, cols = np.indices(im.shape)
center = np.array([128 + rc, 128 + cc])
im += gaussian([rows, cols], center=center, fwhm=50)
im[np.logical_not(mask)] *= 0.8
assert np.allclose(autocenter(im, mask=mask), center, atol=1)
def test_autocenter_single_crystal_ewald_walkoff(rc, cc):
"""Test that autocenter() finds the center of a simulated
shifted single-crystal diffraction pattern."""
I = 10 * diff_pattern_sc(center=(rc, cc))
# Walkoff is the effect when diffraction patterns slide reflections
# at strange angles, such that certain bragg peaks that should have the same
# intensity do not (e.g. brigher (n00) vs (-n00))
# We simulate this with a linear intensity gradient
rows, cols = np.indices(I.shape)
walkoff = rows.astype(float) / rows.max()
walkoff *= 0.2
I += walkoff
mask = np.ones_like(I, dtype=bool)
mask[0:rc + 10, cc - 10:cc + 10] = False
I += 0.01 * I.max() * np.random.random(size=I.shape)
assert np.allclose(autocenter(I, mask=mask), (rc, cc), atol=1)