from astropy.coordinates import SkyCoord
from astropy import units as u
## Matplotlib modules
import matplotlib
matplotlib.use('Agg')
import matplotlib.pyplot as plt
from matplotlib.colors import LogNorm
import matplotlib.cm as cm
#debug = bool(int(sys.argv[2]))
#if debug:
# import pdb; pdb.set_trace()
## Load the input/true WCS
input_imgname = '/disks/shear14/KiDS_simulations/Cosmos/Theli_image/KIDS_150p1_2p2_r_SDSS.V0.5.9A.swarp.cut.fits'
wcs_input = galsim.AstropyWCS(input_imgname)
## Load the input/truth catalogue
input_catname = '/disks/shear14/KiDS_simulations/Cosmos/KIDS_HST_cat/KiDS_Griffith_iMS1_handpicked_stars.cat'
input_catalogue = fits.open(input_catname)
input_data = input_catalogue[1].data
print "Loaded the input data"
## Obtain the cuts on the input/truth catalogue
MASK_all = input_data.MASK
mask = ~np.array(MASK_all & 0xfc3c, dtype=bool)
handpicked_stars = input_data['handpicked_stars']
rank = input_data['rank']
distance2d = input_data['distance2d']
assert handpicked_stars.dtype == bool
def test_withOrigin():
from test_wcs import Cubic
# First EuclideantWCS types:
wcs_list = [
galsim.OffsetWCS(0.3, galsim.PositionD(1, 1), galsim.PositionD(10,
23)),
galsim.OffsetShearWCS(0.23, galsim.Shear(g1=0.1, g2=0.3),
galsim.PositionD(12, 43)),
galsim.AffineTransform(0.01, 0.26, -0.26, 0.02,
galsim.PositionD(12, 43)),
galsim.UVFunction(ufunc=lambda x, y: 0.2 * x,
vfunc=lambda x, y: 0.2 * y),
galsim.UVFunction(ufunc=lambda x, y: 0.2 * x,
vfunc=lambda x, y: 0.2 * y,
xfunc=lambda u, v: u / scale,
yfunc=lambda u, v: v / scale),
galsim.UVFunction(ufunc='0.2*x + 0.03*y', vfunc='0.01*x + 0.2*y'),
]
color = 0.3
for wcs in wcs_list:
# Original version of the shiftOrigin tests in do_nonlocal_wcs using deprecated name.
new_origin = galsim.PositionI(123, 321)
wcs3 = check_dep(wcs.withOrigin, new_origin)
assert wcs != wcs3, name + ' is not != wcs.withOrigin(pos)'
wcs4 = wcs.local(wcs.origin, color=color)
assert wcs != wcs4, name + ' is not != wcs.local()'
assert wcs4 != wcs, name + ' is not != wcs.local() (reverse)'
world_origin = wcs.toWorld(wcs.origin, color=color)
if wcs.isUniform():
if wcs.world_origin == galsim.PositionD(0, 0):
wcs2 = wcs.local(wcs.origin,
color=color).withOrigin(wcs.origin)
assert wcs == wcs2, name + ' is not equal after wcs.local().withOrigin(origin)'
wcs2 = wcs.local(wcs.origin,
color=color).withOrigin(wcs.origin,
wcs.world_origin)
assert wcs == wcs2, name + ' not equal after wcs.local().withOrigin(origin,world_origin)'
world_pos1 = wcs.toWorld(galsim.PositionD(0, 0), color=color)
wcs3 = check_dep(wcs.withOrigin, new_origin)
world_pos2 = wcs3.toWorld(new_origin, color=color)
np.testing.assert_almost_equal(
world_pos2.x, world_pos1.x, 7,
'withOrigin(new_origin) returned wrong world position')
np.testing.assert_almost_equal(
world_pos2.y, world_pos1.y, 7,
'withOrigin(new_origin) returned wrong world position')
new_world_origin = galsim.PositionD(5352.7, 9234.3)
wcs5 = check_dep(wcs.withOrigin,
new_origin,
new_world_origin,
color=color)
world_pos3 = wcs5.toWorld(new_origin, color=color)
np.testing.assert_almost_equal(
world_pos3.x, new_world_origin.x, 7,
'withOrigin(new_origin, new_world_origin) returned wrong position')
np.testing.assert_almost_equal(
world_pos3.y, new_world_origin.y, 7,
'withOrigin(new_origin, new_world_origin) returned wrong position')
# Now some CelestialWCS types
cubic_u = Cubic(2.9e-5, 2000., 'u')
cubic_v = Cubic(-3.7e-5, 2000., 'v')
center = galsim.CelestialCoord(23 * galsim.degrees, -13 * galsim.degrees)
radec = lambda x, y: center.deproject_rad(
cubic_u(x, y) * 0.2, cubic_v(x, y) * 0.2, projection='lambert')
wcs_list = [
galsim.RaDecFunction(radec),
galsim.AstropyWCS('1904-66_TAN.fits', dir='fits_files'),
galsim.GSFitsWCS('tpv.fits', dir='fits_files'),
galsim.FitsWCS('sipsample.fits', dir='fits_files'),
]
for wcs in wcs_list:
# Original version of the shiftOrigin tests in do_celestial_wcs using deprecated name.
new_origin = galsim.PositionI(123, 321)
wcs3 = wcs.shiftOrigin(new_origin)
assert wcs != wcs3, name + ' is not != wcs.shiftOrigin(pos)'
wcs4 = wcs.local(wcs.origin)
assert wcs != wcs4, name + ' is not != wcs.local()'
assert wcs4 != wcs, name + ' is not != wcs.local() (reverse)'
world_pos1 = wcs.toWorld(galsim.PositionD(0, 0))
wcs3 = wcs.shiftOrigin(new_origin)
world_pos2 = wcs3.toWorld(new_origin)
np.testing.assert_almost_equal(
world_pos2.distanceTo(world_pos1) / galsim.arcsec, 0, 7,
'shiftOrigin(new_origin) returned wrong world position')
def check_consistency(randomKey, psfIDs=[0, 1, 2, 3, 4]):
gRange = [
'p400m000', 'm400m000', 'm000p400', 'm000m400', 'm283m283', 'p283m283',
'm283p283', 'p283p283'
]
## Load the input/true WCS
input_imgname = '/disks/shear14/KiDS_simulations/Cosmos/Theli_image/KIDS_150p1_2p2_r_SDSS.V0.5.9A.swarp.cut.fits'
wcs_input = galsim.AstropyWCS(input_imgname)
## Load the input/truth catalogue
input_catname = '/disks/shear14/KiDS_simulations/Cosmos/KIDS_HST_cat/KiDS_Griffith_iMS1_handpicked_stars.cat'
input_catalogue = fits.open(input_catname)
input_data = input_catalogue[1].data
print "Loaded the input data"
## Obtain the cuts on the input/truth catalogue
MASK_all = input_data.MASK
mask = ~np.array(MASK_all & 0xfc3c, dtype=bool)
handpicked_stars = input_data['handpicked_stars']
rank = input_data['rank']
distance2d = input_data['distance2d']
assert handpicked_stars.dtype == bool
cuts = mask & (rank >= 0) & (distance2d < 1) & (~handpicked_stars)
OBJNO = input_data['OBJNO'][cuts]
RA = input_data['RA'][cuts]
DEC = input_data['DEC'][cuts]
X, Y = [], []
x_offset, y_offset = 2500, 2500
## Converting the sky position to image positions (takes about a minute)...
## This is needed because Xpos_THELI and Ypos_THELI aren't filled for the faint galaxies
for gg in xrange(cuts.sum()):
pos = wcs_input.posToImage(
galsim.CelestialCoord(RA[gg] * galsim.degrees,
DEC[gg] * galsim.degrees))
x, y = pos.x - x_offset, pos.y - y_offset
X.append(x)
Y.append(y)
X = np.array(X)
Y = np.array(Y)
## "Building a kd-tree with {0} galaxies using their input positions...".format(cuts.sum())
tree = cKDTree(np.vstack([X, Y]).T)
for psfID in psfIDs:
for g_id in xrange(len(gRange)):
runID = gRange[g_id] + '_' + str(psfID) + '_' + randomKey
print "Comparing ", runID
#shearID, psfID, randomKey = runID.split('_')
ARCHDIR = os.path.join(
'/disks/shear15/KiDS/ImSim/pipeline/archive/', randomKey,
runID)
TMPDIR = os.path.join('/disks/shear15/KiDS/ImSim/temp', randomKey,
runID)
prior_catname = 'prior'
prior_pathname = os.path.join(ARCHDIR, prior_catname)
prior_dat = np.loadtxt(prior_pathname)
sex_arrs, lf_arrs = [], []
indices = []
for rot_id in xrange(4):
sex_catname = 'sexrot0{0}.cat'.format(rot_id)
lf_catname = '0{0}.output.rot.fits.asc.scheme2b_corr'.format(
rot_id)
if rot_id == 0:
sex_catname = 'sex.cat'
# lf_catname = 'output.fits.asc.scheme2b_corr'
sex_pathname = os.path.join(TMPDIR, sex_catname)
lf_pathname = os.path.join(ARCHDIR, lf_catname)
sex_params_filename = 'kidssims.param'
sex_params_pathname = os.path.join(
'/disks/shear15/KiDS/ImSim/pipeline/backup/pipeline/config/',
sex_params_filename)
with open(sex_params_pathname, 'r') as f:
sex_fieldnames = f.readlines()
## Remove the empty lines
n_emptylines = sex_fieldnames.count('\n')
for ii in xrange(n_emptylines):
sex_fieldnames.remove('\n')
## Strip of the newline character from the rest
for ii in xrange(len(sex_fieldnames)):
sex_fieldnames[ii] = sex_fieldnames[ii][:-1]
lf_fieldnames = []
with open(lf_pathname, 'r') as f:
for lineno in xrange(31):
line = f.readline()
words = line.split()
## Omit the first line. It is not a column name
if lineno > 0:
## Because somebody thought giving a space in between is legible
lf_fieldname = ' '.join(words[2:])
lf_fieldnames.append(lf_fieldname)
sex_arr = np.loadtxt(sex_pathname)
lf_arr = np.loadtxt(lf_pathname)
assert len(sex_arr) == len(lf_arr)
assert sex_arr.shape[1] == len(sex_fieldnames)
assert lf_arr.shape[1] == len(lf_fieldnames)
d2d, idx = tree.query(
np.array([
sex_arr[:, sex_fieldnames.index('X_IMAGE')],
sex_arr[:, sex_fieldnames.index('Y_IMAGE')]
]).T)
sex_arrs.append(sex_arr)
lf_arrs.append(lf_arr)
indices.append(idx)
## Assuming that the columns mean the same for all rotations, append them
sex_dat = np.vstack(tuple(sex_arrs))
lf_dat = np.vstack(tuple(lf_arrs))
## Make the QC directory, if it doesn't exist already
QC_dirname = os.path.join(ARCHDIR, 'QC')
if not 'QC' in os.listdir(ARCHDIR):
os.mkdir(QC_dirname)
## Make the overall distributions
## Magnitude plots
fig, ax = plt.subplots()
bins = np.arange(16, 27, 0.05)
prior_mag_col_id = 2
_n, _bins, _patches = ax.hist(prior_dat[:, prior_mag_col_id],
bins=bins,
histtype='step',
color='k',
label='Input magnitude')
_n, _bins, _patches = ax.hist(
sex_dat[:, sex_fieldnames.index('MAG_AUTO')],
bins=bins,
histtype='step',
weights=0.25 * np.ones(len(sex_dat)),
color='r',
label='Output magnitude')
ax.set_yscale('log')
_lgnd = ax.legend(loc='best')
fig.suptitle('Magnitude distributions')
fig_filename = 'magnitudes.png'
fig_pathname = os.path.join(QC_dirname, fig_filename)
fig.savefig(fig_pathname)
# ## SExtractor SNR plots
# fig, ax = plt.subplots()
# bins = np.logspace(-2,4,60)
# _n, _bins, _patches = ax.hist(input_data[cuts]['FLUX_AUTO_THELI']/input_data[cuts]['FLUXERR_AUTO_THELI'], bins=bins, histtype='step', color='k', label='SNR in data')
# _n, _bins, _patches = ax.hist(sex_dat[:,sex_fieldnames.index('FLUX_AUTO')]/sex_dat[:,sex_fieldnames.index('FLUXERR_AUTO')], bins=bins, histtype='step', weights=0.25*np.ones(len(sex_dat)), color='r', label='SNR in sims')
# ax.set_xscale('log')
# _lgnd = ax.legend(loc='best')
# fig.suptitle('SNR from SExtractor')
# fig_filename = 'snr.png'
# fig_pathname = os.path.join(QC_dirname,fig_filename)
# fig.savefig(fig_pathname)
# ## FWHM plots
# fig, ax = plt.subplots()
# bins = np.logspace(-2,2,20)
# _n, _bins, _patches = ax.hist(input_data[cuts]['FWHM_IMAGE_THELI'], bins=bins, histtype='step', color='k', label='FWHM_IMAGE in data')
# _n, _bins, _patches = ax.hist(sex_dat[:,sex_fieldnames.index('FWHM_IMAGE')], bins=bins, histtype='step', weights=0.25*np.ones(len(sex_dat)), color='r', label='FWHM_IMAGE in sims')
# ax.set_xscale('log')
# _lgnd = ax.legend(loc='best')
# fig.suptitle('FWHM_IMAGE from SExtractor')
# fig_filename = 'fwhm.png'
# fig_pathname = os.path.join(QC_dirname,fig_filename)
# fig.savefig(fig_pathname)
## Scalelength
fig, ax = plt.subplots()
bins = np.logspace(-2, 2, 20)
_n, _bins, _patches = ax.hist(
input_data[cuts]['bias_corrected_scalelength_pixels'],
bins=bins,
histtype='step',
color='k',
label='LF scalength in data')
_n, _bins, _patches = ax.hist(
lf_dat[:,
lf_fieldnames.index('bias-corrected scalelength /pixels'
)],
bins=bins,
histtype='step',
weights=0.25 * np.ones(len(lf_dat)),
color='r',
label='LF scalelength in sims')
ax.set_xscale('log')
_lgnd = ax.legend(loc='best')
fig.suptitle('Scalelength from LF')
fig_filename = 'scalelength.png'
fig_pathname = os.path.join(QC_dirname, fig_filename)
fig.savefig(fig_pathname)
## model SNR
fig, ax = plt.subplots()
bins = np.logspace(-2, 4, 60)
_n, _bins, _patches = ax.hist(input_data[cuts]['model_SNratio'],
bins=bins,
histtype='step',
color='k',
label='Model SNR in data')
_n, _bins, _patches = ax.hist(
lf_dat[:, lf_fieldnames.index('model SNratio')],
bins=bins,
histtype='step',
weights=[0.25] * len(lf_dat),
color='r',
label='Model SNR in sims')
ax.set_xscale('log')
_lgnd = ax.legend(loc='best')
fig.suptitle('Model SNR from LF')
fig_filename = 'snr_model.png'
fig_pathname = os.path.join(QC_dirname, fig_filename)
fig.savefig(fig_pathname)
## pixel SNR
fig, ax = plt.subplots()
bins = np.logspace(-2, 4, 60)
_n, _bins, _patches = ax.hist(input_data[cuts]['pixel_SNratio'],
bins=bins,
histtype='step',
color='k',
label='Pixel SNR in data')
_n, _bins, _patches = ax.hist(
lf_dat[:, lf_fieldnames.index('pixel SNratio')],
bins=bins,
histtype='step',
weights=[0.25] * len(lf_dat),
color='r',
label='Pixel SNR in sims')
ax.set_xscale('log')
_lgnd = ax.legend(loc='best')
fig.suptitle('Pixel SNR from LF')
fig_filename = 'snr_pixel.png'
fig_pathname = os.path.join(QC_dirname, fig_filename)
fig.savefig(fig_pathname)
plt.close('all')