def test_sigma_clipped_stats():
"""Test list data with input mask or mask_value (#3268)."""
# test list data with mask
data = [0, 1]
mask = np.array([True, False])
result = sigma_clipped_stats(data, mask=mask)
# Check that the result of np.ma.median was converted to a scalar
assert isinstance(result[1], float)
assert result == (1., 1., 0.)
result2 = sigma_clipped_stats(data, mask=mask, axis=0)
assert_equal(result, result2)
# test list data with mask_value
result = sigma_clipped_stats(data, mask_value=0.)
assert isinstance(result[1], float)
assert result == (1., 1., 0.)
# test without mask
data = [0, 2]
result = sigma_clipped_stats(data)
assert isinstance(result[1], float)
assert result == (1., 1., 1.)
_data = np.arange(10)
data = np.ma.MaskedArray([_data, _data, 10 * _data])
mean = sigma_clip(data, axis=0, sigma=1).mean(axis=0)
assert_equal(mean, _data)
mean, median, stddev = sigma_clipped_stats(data, axis=0, sigma=1)
assert_equal(mean, _data)
assert_equal(median, _data)
assert_equal(stddev, np.zeros_like(_data))
def test_sigma_clipped_stats():
"""Test list data with input mask or mask_value (#3268)."""
# test list data with mask
data = [0, 1]
mask = np.array([True, False])
result = sigma_clipped_stats(data, mask=mask)
# Check that the result of np.ma.median was converted to a scalar
assert isinstance(result[1], float)
assert result == (1., 1., 0.)
result2 = sigma_clipped_stats(data, mask=mask, axis=0)
assert_equal(result, result2)
# test list data with mask_value
result = sigma_clipped_stats(data, mask_value=0.)
assert isinstance(result[1], float)
assert result == (1., 1., 0.)
# test without mask
data = [0, 2]
result = sigma_clipped_stats(data)
assert isinstance(result[1], float)
assert result == (1., 1., 1.)
_data = np.arange(10)
data = np.ma.MaskedArray([_data, _data, 10 * _data])
mean = sigma_clip(data, axis=0, sigma=1).mean(axis=0)
assert_equal(mean, _data)
mean, median, stddev = sigma_clipped_stats(data, axis=0, sigma=1)
assert_equal(mean, _data)
assert_equal(median, _data)
assert_equal(stddev, np.zeros_like(_data))
def test_sigma_clipped_stats_ddof():
with NumpyRNGContext(12345):
data = np.random.randn(10000)
data[10] = 1.e5
mean1, median1, stddev1 = sigma_clipped_stats(data)
mean2, median2, stddev2 = sigma_clipped_stats(data, std_ddof=1)
assert mean1 == mean2
assert median1 == median2
assert_allclose(stddev1, 0.98156805711673156)
assert_allclose(stddev2, 0.98161731654802831)
def test_sigma_clipped_stats_ddof():
with NumpyRNGContext(12345):
data = np.random.randn(10000)
data[10] = 1.e5
mean1, median1, stddev1 = sigma_clipped_stats(data)
mean2, median2, stddev2 = sigma_clipped_stats(data, std_ddof=1)
assert mean1 == mean2
assert median1 == median2
assert_allclose(stddev1, 0.98156805711673156)
assert_allclose(stddev2, 0.98161731654802831)
def test_sigma_clippped_stats_all_masked():
"""
Test sigma_clipped_stats when the input array is completely masked.
"""
arr = np.ma.MaskedArray(np.arange(10), mask=True)
result = sigma_clipped_stats(arr)
assert result == (np.ma.masked, np.ma.masked, np.ma.masked)
arr = np.ma.MaskedArray(np.zeros(10), mask=False)
result = sigma_clipped_stats(arr, mask_value=0.)
assert result == (np.ma.masked, np.ma.masked, np.ma.masked)
arr = np.ma.MaskedArray(np.arange(10), mask=False)
mask = arr < 20
result = sigma_clipped_stats(arr, mask=mask)
assert result == (np.ma.masked, np.ma.masked, np.ma.masked)
def measureFwhm(data):
bkg = sep.Background(data)
thresh = 3 * bkg.globalrms
objects = sep.extract(data - bkg, thresh)
fwhm = 2. * np.sqrt(np.log(2.) * (objects['a']**2 + objects['b']**2))
mask = np.where(objects['peak'] < 40000)
fwhm = fwhm[mask]
if len(fwhm) > 3:
mean, median, std = sigma_clipped_stats(fwhm, sigma=2.5, iters=10)
return median, std
return 0.0, 0.0
def measureHfd(data):
bkg = sep.Background(data)
thresh = 3 * bkg.globalrms
objects = sep.extract(data - bkg, thresh)
# taken from g-tecs autoFocus.py
hfr, mask = sep.flux_radius(data,
objects['x'],
objects['y'],
30 * np.ones_like(objects['x']),
0.5,
normflux=objects['cflux'])
mask = np.logical_and(mask == 0, objects['peak'] < 40000)
hfd = 2 * hfr[mask]
if hfd.size > 3:
mean, median, std = sigma_clipped_stats(hfd, sigma=2.5, iters=10)
return median, std
return 0.0, 0.0
def measureHfd(img):
data = fits.open(img)[0].data
y, x = data.shape
xcen = int(x / 2)
ycen = int(y / 2)
# grab the middle 2k x 2k
stats_area = np.array(data[ycen-1024: ycen+1024, \
xcen-1024: xcen+1024]).astype(np.int32).copy(order='C')
bkg = sep.Background(stats_area)
thresh = 3 * bkg.globalrms
objects = sep.extract(stats_area - bkg, thresh)
# taken from g-tecs autoFocus.py
hfr, mask = sep.flux_radius(stats_area,
objects['x'],
objects['y'],
30 * np.ones_like(objects['x']),
0.5,
normflux=objects['cflux'])
mask = np.logical_and(mask == 0, objects['peak'] < 40000)
hfd = 2 * hfr[mask]
if hfd.size > 3:
mean, median, std = sigma_clipped_stats(hfd, sigma=2.5, iters=10)
return median, std
return 0.0, 0.0
def test_sigma_clippped_stats_unit():
data = np.array([1, 1]) * u.kpc
result = sigma_clipped_stats(data)
assert result == (1. * u.kpc, 1. * u.kpc, 0. * u.kpc)
