def test_fits():
try:
import fitsio
except ImportError:
print('Skipping FITS tests, since fitsio is not installed')
return
get_from_wiki('Aardvark.fit')
file_name = os.path.join('data','Aardvark.fit')
config = treecorr.read_config('Aardvark.yaml')
config['verbose'] = 1
# Just test a few random particular values
cat1 = treecorr.Catalog(file_name, config)
np.testing.assert_equal(len(cat1.ra), 390935)
np.testing.assert_equal(cat1.nobj, 390935)
np.testing.assert_almost_equal(cat1.ra[0], 56.4195 * (pi/180.))
np.testing.assert_almost_equal(cat1.ra[390934], 78.4782 * (pi/180.))
np.testing.assert_almost_equal(cat1.dec[290333], 83.1579 * (pi/180.))
np.testing.assert_almost_equal(cat1.g1[46392], 0.0005066675)
np.testing.assert_almost_equal(cat1.g2[46392], -0.0001006742)
np.testing.assert_almost_equal(cat1.k[46392], -0.0008628797)
# The catalog doesn't have x, y, or w, but test that functionality as well.
del config['ra_col']
del config['dec_col']
config['x_col'] = 'RA'
config['y_col'] = 'DEC'
config['w_col'] = 'MU'
config['flag_col'] = 'INDEX'
config['ignore_flag'] = 64
cat2 = treecorr.Catalog(file_name, config)
np.testing.assert_almost_equal(cat2.x[390934], 78.4782, decimal=4)
np.testing.assert_almost_equal(cat2.y[290333], 83.1579, decimal=4)
np.testing.assert_almost_equal(cat2.w[46392], 0.) # index = 1200379
np.testing.assert_almost_equal(cat2.w[46393], 0.9995946) # index = 1200386
# Test using a limited set of rows
config['first_row'] = 101
config['last_row'] = 50000
cat3 = treecorr.Catalog(file_name, config)
np.testing.assert_equal(len(cat3.x), 49900)
np.testing.assert_equal(cat3.ntot, 49900)
np.testing.assert_equal(cat3.nobj, sum(cat3.w != 0))
np.testing.assert_equal(cat3.sumw, sum(cat3.w))
np.testing.assert_equal(cat3.sumw, sum(cat2.w[100:50000]))
np.testing.assert_almost_equal(cat3.g1[46292], 0.0005066675)
np.testing.assert_almost_equal(cat3.g2[46292], -0.0001006742)
np.testing.assert_almost_equal(cat3.k[46292], -0.0008628797)
cat4 = treecorr.read_catalogs(config, key='file_name', is_rand=True)[0]
np.testing.assert_equal(len(cat4.x), 49900)
np.testing.assert_equal(cat4.ntot, 49900)
np.testing.assert_equal(cat4.nobj, sum(cat4.w != 0))
np.testing.assert_equal(cat4.sumw, sum(cat4.w))
np.testing.assert_equal(cat4.sumw, sum(cat2.w[100:50000]))
assert cat4.g1 is None
assert cat4.g2 is None
assert cat4.k is None
def test_direct_perp():
# This is the same as the above test, but using the perpendicular distance metric
get_from_wiki('nn_perp_data.dat')
get_from_wiki('nn_perp_rand.dat')
ngal = 100
s = 10.
numpy.random.seed(8675309)
x1 = numpy.random.normal(312, s, (ngal,) )
y1 = numpy.random.normal(728, s, (ngal,) )
z1 = numpy.random.normal(-932, s, (ngal,) )
r1 = numpy.sqrt( x1*x1 + y1*y1 + z1*z1 )
dec1 = numpy.arcsin(z1/r1)
ra1 = numpy.arctan2(y1,x1)
cat1 = treecorr.Catalog(ra=ra1, dec=dec1, r=r1, ra_units='rad', dec_units='rad')
x2 = numpy.random.normal(312, s, (ngal,) )
y2 = numpy.random.normal(728, s, (ngal,) )
z2 = numpy.random.normal(-932, s, (ngal,) )
r2 = numpy.sqrt( x2*x2 + y2*y2 + z2*z2 )
dec2 = numpy.arcsin(z2/r2)
ra2 = numpy.arctan2(y2,x2)
cat2 = treecorr.Catalog(ra=ra2, dec=dec2, r=r2, ra_units='rad', dec_units='rad')
min_sep = 1.
max_sep = 50.
nbins = 50
dd = treecorr.NNCorrelation(min_sep=min_sep, max_sep=max_sep, nbins=nbins, bin_slop=0.)
dd.process(cat1, cat2, metric='Rperp')
print('dd.npairs = ',dd.npairs)
log_min_sep = numpy.log(min_sep)
log_max_sep = numpy.log(max_sep)
true_npairs = numpy.zeros(nbins)
bin_size = (log_max_sep - log_min_sep) / nbins
for i in range(ngal):
for j in range(ngal):
rsq = (x1[i]-x2[j])**2 + (y1[i]-y2[j])**2 + (z1[i]-z2[j])**2
rsq -= (r1[i] - r2[j])**2
logr = 0.5 * numpy.log(rsq)
k = int(numpy.floor( (logr-log_min_sep) / bin_size ))
if k < 0: continue
if k >= nbins: continue
true_npairs[k] += 1
print('true_npairs = ',true_npairs)
print('diff = ',dd.npairs - true_npairs)
numpy.testing.assert_array_equal(dd.npairs, true_npairs)
# Can also specify coords directly as x,y,z
cat1 = treecorr.Catalog(x=x1, y=y1, z=z1)
cat2 = treecorr.Catalog(x=x2, y=y2, z=z2)
dd.process(cat1, cat2, metric='Rperp')
numpy.testing.assert_array_equal(dd.npairs, true_npairs)
def test_direct_perp():
# This is the same as the above test, but using the perpendicular distance metric
get_from_wiki('nn_perp_data.dat')
get_from_wiki('nn_perp_rand.dat')
ngal = 100
s = 10.
numpy.random.seed(8675309)
x1 = numpy.random.normal(312, s, (ngal,) )
y1 = numpy.random.normal(728, s, (ngal,) )
z1 = numpy.random.normal(-932, s, (ngal,) )
r1 = numpy.sqrt( x1*x1 + y1*y1 + z1*z1 )
dec1 = numpy.arcsin(z1/r1)
ra1 = numpy.arctan2(y1,x1)
cat1 = treecorr.Catalog(ra=ra1, dec=dec1, r=r1, ra_units='rad', dec_units='rad')
x2 = numpy.random.normal(312, s, (ngal,) )
y2 = numpy.random.normal(728, s, (ngal,) )
z2 = numpy.random.normal(-932, s, (ngal,) )
r2 = numpy.sqrt( x2*x2 + y2*y2 + z2*z2 )
dec2 = numpy.arcsin(z2/r2)
ra2 = numpy.arctan2(y2,x2)
cat2 = treecorr.Catalog(ra=ra2, dec=dec2, r=r2, ra_units='rad', dec_units='rad')
min_sep = 1.
max_sep = 50.
nbins = 50
dd = treecorr.NNCorrelation(min_sep=min_sep, max_sep=max_sep, nbins=nbins, bin_slop=0.)
dd.process(cat1, cat2, metric='Rperp')
print('dd.npairs = ',dd.npairs)
log_min_sep = numpy.log(min_sep)
log_max_sep = numpy.log(max_sep)
true_npairs = numpy.zeros(nbins)
bin_size = (log_max_sep - log_min_sep) / nbins
for i in range(ngal):
for j in range(ngal):
rsq = (x1[i]-x2[j])**2 + (y1[i]-y2[j])**2 + (z1[i]-z2[j])**2
rsq -= (r1[i] - r2[j])**2
logr = 0.5 * numpy.log(rsq)
k = int(numpy.floor( (logr-log_min_sep) / bin_size ))
if k < 0: continue
if k >= nbins: continue
true_npairs[k] += 1
print('true_npairs = ',true_npairs)
print('diff = ',dd.npairs - true_npairs)
numpy.testing.assert_array_equal(dd.npairs, true_npairs)
# Can also specify coords directly as x,y,z
cat1 = treecorr.Catalog(x=x1, y=y1, z=z1)
cat2 = treecorr.Catalog(x=x2, y=y2, z=z2)
dd.process(cat1, cat2, metric='Rperp')
numpy.testing.assert_array_equal(dd.npairs, true_npairs)
def test_fits():
get_from_wiki('Aardvark.fit')
file_name = os.path.join('data','Aardvark.fit')
config = treecorr.read_config('Aardvark.yaml')
config['verbose'] = 1
# Just test a few random particular values
cat1 = treecorr.Catalog(file_name, config)
numpy.testing.assert_equal(len(cat1.ra), 390935)
numpy.testing.assert_equal(cat1.nobj, 390935)
numpy.testing.assert_almost_equal(cat1.ra[0], 56.4195 * (pi/180.))
numpy.testing.assert_almost_equal(cat1.ra[390934], 78.4782 * (pi/180.))
numpy.testing.assert_almost_equal(cat1.dec[290333], 83.1579 * (pi/180.))
numpy.testing.assert_almost_equal(cat1.g1[46392], 0.0005066675)
numpy.testing.assert_almost_equal(cat1.g2[46392], -0.0001006742)
numpy.testing.assert_almost_equal(cat1.k[46392], -0.0008628797)
# The catalog doesn't have x, y, or w, but test that functionality as well.
del config['ra_col']
del config['dec_col']
config['x_col'] = 'RA'
config['y_col'] = 'DEC'
config['w_col'] = 'MU'
config['flag_col'] = 'INDEX'
config['ignore_flag'] = 64
cat2 = treecorr.Catalog(file_name, config)
numpy.testing.assert_almost_equal(cat2.x[390934], 78.4782, decimal=4)
numpy.testing.assert_almost_equal(cat2.y[290333], 83.1579, decimal=4)
numpy.testing.assert_almost_equal(cat2.w[46392], 0.) # index = 1200379
numpy.testing.assert_almost_equal(cat2.w[46393], 0.9995946) # index = 1200386
# Test using a limited set of rows
config['first_row'] = 101
config['last_row'] = 50000
cat3 = treecorr.Catalog(file_name, config)
numpy.testing.assert_equal(len(cat3.x), 49900)
numpy.testing.assert_equal(cat3.ntot, 49900)
numpy.testing.assert_equal(cat3.nobj, sum(cat3.w != 0))
numpy.testing.assert_equal(cat3.sumw, sum(cat3.w))
numpy.testing.assert_equal(cat3.sumw, sum(cat2.w[100:50000]))
numpy.testing.assert_almost_equal(cat3.g1[46292], 0.0005066675)
numpy.testing.assert_almost_equal(cat3.g2[46292], -0.0001006742)
numpy.testing.assert_almost_equal(cat3.k[46292], -0.0008628797)
def test_fits():
get_from_wiki('Aardvark.fit')
file_name = os.path.join('data', 'Aardvark.fit')
config = treecorr.read_config('Aardvark.params')
# Just test a few random particular values
cat1 = treecorr.Catalog(file_name, config)
numpy.testing.assert_equal(len(cat1.ra), 390935)
numpy.testing.assert_equal(cat1.nobj, 390935)
numpy.testing.assert_almost_equal(cat1.ra[0], 56.4195 * (pi / 180.))
numpy.testing.assert_almost_equal(cat1.ra[390934], 78.4782 * (pi / 180.))
numpy.testing.assert_almost_equal(cat1.dec[290333], 83.1579 * (pi / 180.))
numpy.testing.assert_almost_equal(cat1.g1[46392], 0.0005066675)
numpy.testing.assert_almost_equal(cat1.g2[46392], -0.0001006742)
numpy.testing.assert_almost_equal(cat1.k[46392], -0.0008628797)
# The catalog doesn't have x, y, or w, but test that functionality as well.
del config['ra_col']
del config['dec_col']
config['x_col'] = 'RA'
config['y_col'] = 'DEC'
config['w_col'] = 'MU'
config['flag_col'] = 'INDEX'
config['ignore_flag'] = 64
cat2 = treecorr.Catalog(file_name, config)
numpy.testing.assert_almost_equal(cat2.x[390934], 78.4782, decimal=4)
numpy.testing.assert_almost_equal(cat2.y[290333], 83.1579, decimal=4)
numpy.testing.assert_almost_equal(cat2.w[46392], 0.) # index = 1200379
numpy.testing.assert_almost_equal(cat2.w[46393],
0.9995946) # index = 1200386
# Test using a limited set of rows
config['first_row'] = 101
config['last_row'] = 50000
cat3 = treecorr.Catalog(file_name, config)
numpy.testing.assert_equal(len(cat3.x), 49900)
numpy.testing.assert_equal(cat3.ntot, 49900)
numpy.testing.assert_equal(cat3.nobj, sum(cat3.w != 0))
numpy.testing.assert_equal(cat3.sumw, sum(cat3.w))
numpy.testing.assert_equal(cat3.sumw, sum(cat2.w[100:50000]))
numpy.testing.assert_almost_equal(cat3.g1[46292], 0.0005066675)
numpy.testing.assert_almost_equal(cat3.g2[46292], -0.0001006742)
numpy.testing.assert_almost_equal(cat3.k[46292], -0.0008628797)
def setup():
from test_helper import get_from_wiki
file_name = os.path.join('data', 'Aardvark.fit')
patch_file = os.path.join('data', 'mpi_patches.fits')
# Make sure we have Aardvark.fit
get_from_wiki('Aardvark.fit')
# And all the tests will use these patches. Make them once and save them.
# For a real-life example, this might be made once and saved.
# Or it might be made from a smaller version of the catalog:
# either with the every_nth option, or maybe on a redmagic catalog or similar,
# which would be smaller than the full source catalog, etc.
if not os.path.exists(patch_file):
part_cat = treecorr.Catalog(file_name,
ra_col='RA',
dec_col='DEC',
ra_units='deg',
dec_units='deg',
npatch=8)
part_cat.write_patch_centers(patch_file)
del part_cat
def download():
# The download can be a bit slow, and the files need to be merged to get something
# that includes both ra/dec and e1,e2. So we did this block once and uploaded
# the results to the wiki. Normal test running can just get the result from the wiki.
# Download the public DES SV files (if not already downloaded)
host = 'http://desdr-server.ncsa.illinois.edu/despublic/sva1_files/'
lens_file = 'redmagic_sva1_public_v6.3_faint.fits.gz'
get_from_wiki(lens_file, host=host)
lens_file = os.path.join('data', lens_file)
ngmix_file = 'sva1_gold_r1.0_ngmix.fits.gz'
get_from_wiki(ngmix_file, host=host)
ngmix_file = os.path.join('data', ngmix_file)
info_file = 'sva1_gold_r1.0_wlinfo.fits.gz'
get_from_wiki(info_file, host=host)
info_file = os.path.join('data', info_file)
source_file = os.path.join('data', 'sva1_gold_r1.0_merged.fits')
if not os.path.exists(source_file):
print('Reading ngmix_data')
ngmix_data = fitsio.read(ngmix_file)
print('Reading info_data')
info_data = fitsio.read(info_file)
col_names = ['RA', 'DEC']
cols = [info_data[n] for n in col_names] # These come from wlinfo
col_names += ['E_1', 'E_2', 'W']
cols += [ngmix_data[n] for n in col_names[2:]] # These are in ngmix
# combine the two sensitivity estimates
col_names += ['SENS']
cols += [(ngmix_data['SENS_1'] + ngmix_data['SENS_2']) / 2.]
# Save time by cutting to only flag != 0 objects here.
use = info_data['NGMIX_FLAG'] == 0
print('total number of galaxies = ', len(use))
print('number to use = ', np.sum(use))
cols = [col[use] for col in cols]
print('writing merged file: ', source_file)
treecorr.util.gen_write(source_file, col_names, cols, file_type='FITS')
return source_file, lens_file
def test_fits():
try:
import fitsio
except ImportError:
print('Skipping FITS tests, since fitsio is not installed')
return
get_from_wiki('Aardvark.fit')
file_name = os.path.join('data','Aardvark.fit')
config = treecorr.read_config('Aardvark.yaml')
config['verbose'] = 1
config['kk_file_name'] = 'kk.fits'
config['gg_file_name'] = 'gg.fits'
# Just test a few random particular values
cat1 = treecorr.Catalog(file_name, config)
np.testing.assert_equal(len(cat1.ra), 390935)
np.testing.assert_equal(cat1.nobj, 390935)
np.testing.assert_almost_equal(cat1.ra[0], 56.4195 * (pi/180.))
np.testing.assert_almost_equal(cat1.ra[390934], 78.4782 * (pi/180.))
np.testing.assert_almost_equal(cat1.dec[290333], 83.1579 * (pi/180.))
np.testing.assert_almost_equal(cat1.g1[46392], 0.0005066675)
np.testing.assert_almost_equal(cat1.g2[46392], -0.0001006742)
np.testing.assert_almost_equal(cat1.k[46392], -0.0008628797)
assert_raises(ValueError, treecorr.Catalog, file_name, config, ra_col='invalid')
assert_raises(ValueError, treecorr.Catalog, file_name, config, dec_col='invalid')
assert_raises(ValueError, treecorr.Catalog, file_name, config, r_col='invalid')
assert_raises(ValueError, treecorr.Catalog, file_name, config, w_col='invalid')
assert_raises(ValueError, treecorr.Catalog, file_name, config, wpos_col='invalid')
assert_raises(ValueError, treecorr.Catalog, file_name, config, flag_col='invalid')
assert_raises(ValueError, treecorr.Catalog, file_name, config, g1_col='invalid')
assert_raises(ValueError, treecorr.Catalog, file_name, config, g2_col='invalid')
assert_raises(ValueError, treecorr.Catalog, file_name, config, k_col='invalid')
assert_raises(ValueError, treecorr.Catalog, file_name, config, ra_col='0')
assert_raises(ValueError, treecorr.Catalog, file_name, config, dec_col='0')
assert_raises(ValueError, treecorr.Catalog, file_name, config, x_col='x')
assert_raises(ValueError, treecorr.Catalog, file_name, config, y_col='y')
assert_raises(ValueError, treecorr.Catalog, file_name, config, z_col='z')
assert_raises(ValueError, treecorr.Catalog, file_name, config, ra_col='0', dec_col='0')
assert_raises(ValueError, treecorr.Catalog, file_name, config, g1_col='0')
assert_raises(ValueError, treecorr.Catalog, file_name, config, g2_col='0')
assert_raises(ValueError, treecorr.Catalog, file_name, config, k_col='0')
assert_raises(TypeError, treecorr.Catalog, file_name, config, x_units='arcmin')
assert_raises(TypeError, treecorr.Catalog, file_name, config, y_units='arcmin')
del config['ra_units']
assert_raises(TypeError, treecorr.Catalog, file_name, config)
del config['dec_units']
assert_raises(TypeError, treecorr.Catalog, file_name, config, ra_units='deg')
# The catalog doesn't have x, y, or w, but test that functionality as well.
del config['ra_col']
del config['dec_col']
config['x_col'] = 'RA'
config['y_col'] = 'DEC'
config['w_col'] = 'MU'
config['flag_col'] = 'INDEX'
config['ignore_flag'] = 64
cat2 = treecorr.Catalog(file_name, config)
np.testing.assert_almost_equal(cat2.x[390934], 78.4782, decimal=4)
np.testing.assert_almost_equal(cat2.y[290333], 83.1579, decimal=4)
np.testing.assert_almost_equal(cat2.w[46392], 0.) # index = 1200379
np.testing.assert_almost_equal(cat2.w[46393], 0.9995946) # index = 1200386
assert_raises(ValueError, treecorr.Catalog, file_name, config, x_col='invalid')
assert_raises(ValueError, treecorr.Catalog, file_name, config, y_col='invalid')
assert_raises(ValueError, treecorr.Catalog, file_name, config, z_col='invalid')
assert_raises(ValueError, treecorr.Catalog, file_name, config, ra_col='invalid')
assert_raises(ValueError, treecorr.Catalog, file_name, config, dec_col='invalid')
assert_raises(ValueError, treecorr.Catalog, file_name, config, r_col='invalid')
assert_raises(ValueError, treecorr.Catalog, file_name, config, w_col='invalid')
assert_raises(ValueError, treecorr.Catalog, file_name, config, wpos_col='invalid')
assert_raises(ValueError, treecorr.Catalog, file_name, config, flag_col='invalid')
assert_raises(ValueError, treecorr.Catalog, file_name, config, g1_col='invalid')
assert_raises(ValueError, treecorr.Catalog, file_name, config, g2_col='invalid')
assert_raises(ValueError, treecorr.Catalog, file_name, config, k_col='invalid')
# Test using a limited set of rows
config['first_row'] = 101
config['last_row'] = 50000
cat3 = treecorr.Catalog(file_name, config)
np.testing.assert_equal(len(cat3.x), 49900)
np.testing.assert_equal(cat3.ntot, 49900)
np.testing.assert_equal(cat3.nobj, sum(cat3.w != 0))
np.testing.assert_equal(cat3.sumw, sum(cat3.w))
np.testing.assert_equal(cat3.sumw, sum(cat2.w[100:50000]))
np.testing.assert_almost_equal(cat3.g1[46292], 0.0005066675)
np.testing.assert_almost_equal(cat3.g2[46292], -0.0001006742)
np.testing.assert_almost_equal(cat3.k[46292], -0.0008628797)
cat4 = treecorr.read_catalogs(config, key='file_name', is_rand=True)[0]
np.testing.assert_equal(len(cat4.x), 49900)
np.testing.assert_equal(cat4.ntot, 49900)
np.testing.assert_equal(cat4.nobj, sum(cat4.w != 0))
np.testing.assert_equal(cat4.sumw, sum(cat4.w))
np.testing.assert_equal(cat4.sumw, sum(cat2.w[100:50000]))
assert cat4.g1 is None
assert cat4.g2 is None
assert cat4.k is None
do_pickle(cat1)
do_pickle(cat2)
do_pickle(cat3)
do_pickle(cat4)
assert_raises(ValueError, treecorr.Catalog, file_name, config, first_row=-10)
assert_raises(ValueError, treecorr.Catalog, file_name, config, first_row=0)
assert_raises(ValueError, treecorr.Catalog, file_name, config, first_row=60000)
assert_raises(ValueError, treecorr.Catalog, file_name, config, first_row=50001)
assert_raises(TypeError, treecorr.read_catalogs, config)
assert_raises(TypeError, treecorr.read_catalogs, config, key='file_name', list_key='file_name')
# If gg output not given, it is still invalid to only have one or the other of g1,g2.
del config['gg_file_name']
assert_raises(ValueError, treecorr.Catalog, file_name, config, g1_col='0')
assert_raises(ValueError, treecorr.Catalog, file_name, config, g2_col='0')
def test_dessv():
try:
import fitsio
except ImportError:
print('Skipping dessv test, since fitsio is not installed')
return
#treecorr.set_omp_threads(1);
get_from_wiki('des_sv.fits')
file_name = os.path.join('data', 'des_sv.fits')
cat = treecorr.Catalog(file_name,
ra_col='ra',
dec_col='dec',
ra_units='deg',
dec_units='deg')
# Use an odd number to make sure we force some of the shuffle bits in InitializeCenters
# to happen.
npatch = 43
field = cat.getNField()
t0 = time.time()
patches = field.run_kmeans(npatch)
t1 = time.time()
print('patches = ', np.unique(patches))
assert len(patches) == cat.ntot
assert min(patches) == 0
assert max(patches) == npatch - 1
# KMeans minimizes the total inertia.
# Check this value and the rms size, which should also be quite small.
xyz = np.array([cat.x, cat.y, cat.z]).T
cen = np.array([xyz[patches == i].mean(axis=0) for i in range(npatch)])
inertia = np.array(
[np.sum((xyz[patches == i] - cen[i])**2) for i in range(npatch)])
sizes = np.array(
[np.mean((xyz[patches == i] - cen[i])**2) for i in range(npatch)])**0.5
sizes *= 180. / np.pi * 60. # convert to arcmin
counts = np.array([np.sum(patches == i) for i in range(npatch)])
print('With standard algorithm:')
print('time = ', t1 - t0)
print('total inertia = ', np.sum(inertia))
print('mean inertia = ', np.mean(inertia))
print('rms inertia = ', np.std(inertia))
print('mean size = ', np.mean(sizes))
print('rms size = ', np.std(sizes))
assert np.sum(
inertia
) < 200. # This is specific to this particular field and npatch.
assert np.std(inertia) < 0.3 * np.mean(
inertia) # rms is usually < 0.2 * mean
assert np.std(sizes) < 0.1 * np.mean(
sizes) # sizes have even less spread usually.
# Should all have similar number of points. Nothing is required here though.
print('mean counts = ', np.mean(counts))
print('min counts = ', np.min(counts))
print('max counts = ', np.max(counts))
# Check the alternate algorithm. rms inertia should be lower.
t0 = time.time()
patches = field.run_kmeans(npatch, alt=True)
t1 = time.time()
assert len(patches) == cat.ntot
assert min(patches) == 0
assert max(patches) == npatch - 1
cen = np.array([xyz[patches == i].mean(axis=0) for i in range(npatch)])
inertia = np.array(
[np.sum((xyz[patches == i] - cen[i])**2) for i in range(npatch)])
sizes = np.array(
[np.mean((xyz[patches == i] - cen[i])**2) for i in range(npatch)])**0.5
sizes *= 180. / np.pi * 60. # convert to arcmin
counts = np.array([np.sum(patches == i) for i in range(npatch)])
print('With alternate algorithm:')
print('time = ', t1 - t0)
print('total inertia = ', np.sum(inertia))
print('mean inertia = ', np.mean(inertia))
print('rms inertia = ', np.std(inertia))
print('mean size = ', np.mean(sizes))
print('rms size = ', np.std(sizes))
assert np.sum(
inertia
) < 200. # Total shouldn't increase much. (And often decreases.)
assert np.std(inertia) < 0.1 * np.mean(
inertia) # rms should be even smaller here.
assert np.std(sizes) < 0.1 * np.mean(
sizes) # This is only a little bit smaller.
# This doesn't keep the counts as equal as the standard algorithm.
print('mean counts = ', np.mean(counts))
print('min counts = ', np.min(counts))
print('max counts = ', np.max(counts))
# Finally, use a field with lots of top level cells to check the other branch in
# InitializeCenters.
field = cat.getNField(min_top=10)
t0 = time.time()
patches = field.run_kmeans(npatch)
t1 = time.time()
assert len(patches) == cat.ntot
assert min(patches) == 0
assert max(patches) == npatch - 1
cen = np.array([xyz[patches == i].mean(axis=0) for i in range(npatch)])
inertia = np.array(
[np.sum((xyz[patches == i] - cen[i])**2) for i in range(npatch)])
sizes = np.array(
[np.mean((xyz[patches == i] - cen[i])**2) for i in range(npatch)])**0.5
sizes *= 180. / np.pi * 60. # convert to arcmin
counts = np.array([np.sum(patches == i) for i in range(npatch)])
# This doesn't give as good an initialization, so these are a bit worse usually.
print('With min_top=10:')
print('time = ', t1 - t0)
print('total inertia = ', np.sum(inertia))
print('mean inertia = ', np.mean(inertia))
print('rms inertia = ', np.std(inertia))
print('mean size = ', np.mean(sizes))
print('rms size = ', np.std(sizes))
assert np.sum(inertia) < 210.
assert np.std(inertia) < 0.4 * np.mean(
inertia) # I've seen over 0.3 x mean here.
assert np.std(sizes) < 0.15 * np.mean(sizes)
print('mean counts = ', np.mean(counts))
print('min counts = ', np.min(counts))
print('max counts = ', np.max(counts))
def test_perp_minmax():
"""This test is based on a bug report from Erika Wagoner where the lowest bins were
getting spuriously high w(rp) values. It stemmed from a subtlety about how large
rp can be compared to minsep. The maximum rp is more than just rp + s1 + s2.
So this test checks that when the min and max are expanded a bit, the number of pairs
doesn't change much in the bins that used to be the min and max.
"""
# Just use Erika's files for data and rand.
config = {
'ra_col' : 1,
'dec_col' : 2,
'ra_units' : 'deg',
'dec_units' : 'deg',
'r_col' : 3,
'min_sep' : 20,
'bin_size' : 0.036652,
'nbins' : 50,
'verbose' : 1
}
# Speed up for nosetests runs
if __name__ != "__main__":
config['nbins'] = 5
config['bin_size'] = 0.1
get_from_wiki('nn_perp_data.dat')
dcat = treecorr.Catalog('data/nn_perp_data.dat', config)
dd1 = treecorr.NNCorrelation(config)
dd1.process(dcat, metric='Rperp')
lower_min_sep = config['min_sep'] * numpy.exp(-2.*config['bin_size'])
more_nbins = config['nbins'] + 4
dd2 = treecorr.NNCorrelation(config, min_sep=lower_min_sep, nbins=more_nbins)
dd2.process(dcat, metric='Rperp')
print('dd1 npairs = ',dd1.npairs)
print('dd2 npairs = ',dd2.npairs[2:-2])
# First a basic sanity check. The values not near the edge should be identical.
numpy.testing.assert_equal(dd1.npairs[2:-2], dd2.npairs[4:-4])
# The edge bins may differ slightly from the binning approximations (bin_slop and such),
# but the differences should be very small. (When Erika reported the problem, the differences
# were a few percent, which ended up making a bit difference in the correlation function.)
numpy.testing.assert_almost_equal( dd1.npairs / dd2.npairs[2:-2], 1., decimal=4)
if __name__ == '__main__':
# If we're running from the command line, go ahead and finish the calculation
# This catalog has 10^6 objects, which takes quite a while. I should really investigate
# how to speed up the Rperp distance calculation. Probably by having a faster over-
# and under-estimate first, and then only do the full calculation when it seems like we
# will actually need it.
# Anyway, until then, let's not take forever by using last_row=200000
get_from_wiki('nn_perp_rand.dat')
rcat = treecorr.Catalog('data/nn_perp_rand.dat', config, last_row=200000)
rr1 = treecorr.NNCorrelation(config)
rr1.process(rcat, metric='Rperp')
rr2 = treecorr.NNCorrelation(config, min_sep=lower_min_sep, nbins=more_nbins)
rr2.process(rcat, metric='Rperp')
print('rr1 npairs = ',rr1.npairs)
print('rr2 npairs = ',rr2.npairs[2:-2])
numpy.testing.assert_almost_equal( rr1.npairs / rr2.npairs[2:-2], 1., decimal=4)
dr1 = treecorr.NNCorrelation(config)
dr1.process(dcat, rcat, metric='Rperp')
dr2 = treecorr.NNCorrelation(config, min_sep=lower_min_sep, nbins=more_nbins)
dr2.process(dcat, rcat, metric='Rperp')
print('dr1 npairs = ',dr1.npairs)
print('dr2 npairs = ',dr2.npairs[2:-2])
numpy.testing.assert_almost_equal( dr1.npairs / dr2.npairs[2:-2], 1., decimal=4)
xi1, varxi1 = dd1.calculateXi(rr1, dr1)
xi2, varxi2 = dd2.calculateXi(rr2, dr2)
print('xi1 = ',xi1)
print('xi2 = ',xi2[2:-2])
numpy.testing.assert_almost_equal( xi1 / xi2[2:-2], 1., decimal=2)
# Check that we get the same result with the corr2 executable.
import subprocess
corr2_exe = get_script_name('corr2')
p = subprocess.Popen( [corr2_exe,"nn_rperp.yaml"] )
p.communicate()
corr2_output = numpy.genfromtxt(os.path.join('output','nn_rperp.out'),names=True,skip_header=1)
print('xi = ',xi1)
print('from corr2 output = ',corr2_output['xi'])
print('ratio = ',corr2_output['xi']/xi1)
print('diff = ',corr2_output['xi']-xi1)
numpy.testing.assert_almost_equal(corr2_output['xi']/xi1, 1., decimal=3)
def test_aardvark():
# Eric Suchyta did a brute force calculation of the Aardvark catalog, so it is useful to
# compare the output from my code with that.
get_from_wiki('Aardvark.fit')
file_name = os.path.join('data', 'Aardvark.fit')
config = treecorr.read_config('Aardvark.params')
cat1 = treecorr.Catalog(file_name, config)
gg = treecorr.GGCorrelation(config)
gg.process(cat1)
direct_file_name = os.path.join('data', 'Aardvark.direct')
direct_data = numpy.genfromtxt(direct_file_name)
direct_xip = direct_data[:, 3]
direct_xim = direct_data[:, 4]
#print('gg.xip = ',gg.xip)
#print('direct.xip = ',direct_xip)
xip_err = gg.xip - direct_xip
print('xip_err = ', xip_err)
print('max = ', max(abs(xip_err)))
assert max(abs(xip_err)) < 2.e-7
print('xip_im = ', gg.xip_im)
print('max = ', max(abs(gg.xip_im)))
assert max(abs(gg.xip_im)) < 3.e-7
xim_err = gg.xim - direct_xim
print('xim_err = ', xim_err)
print('max = ', max(abs(xim_err)))
assert max(abs(xim_err)) < 1.e-7
print('xim_im = ', gg.xim_im)
print('max = ', max(abs(gg.xim_im)))
assert max(abs(gg.xim_im)) < 1.e-7
# However, after some back and forth about the calculation, we concluded that Eric hadn't
# done the spherical trig correctly to get the shears relative to the great circle joining
# the two positions. So let's compare with my own brute force calculation (i.e. using
# bin_slop = 0):
# This also has the advantage that the radial bins are done the same way -- uniformly
# spaced in log of the chord distance, rather than the great circle distance.
bs0_file_name = os.path.join('data', 'Aardvark.bs0')
bs0_data = numpy.genfromtxt(bs0_file_name)
bs0_xip = bs0_data[:, 2]
bs0_xim = bs0_data[:, 3]
#print('gg.xip = ',gg.xip)
#print('bs0.xip = ',bs0_xip)
xip_err = gg.xip - bs0_xip
print('xip_err = ', xip_err)
print('max = ', max(abs(xip_err)))
assert max(abs(xip_err)) < 1.e-7
xim_err = gg.xim - bs0_xim
print('xim_err = ', xim_err)
print('max = ', max(abs(xim_err)))
assert max(abs(xim_err)) < 5.e-8
# Check that we get the same result using the corr2 executable:
# Note: This is the only test of the corr2 executable that we do with nosetests.
# The other similar tests are blocked out with: if __name__ == '__main__':
import subprocess
corr2_exe = get_script_name('corr2')
p = subprocess.Popen([corr2_exe, "Aardvark.params"])
p.communicate()
corr2_output = numpy.genfromtxt(os.path.join('output', 'Aardvark.out'),
names=True)
print('gg.xip = ', gg.xip)
print('from corr2 output = ', corr2_output['xip'])
print('ratio = ', corr2_output['xip'] / gg.xip)
print('diff = ', corr2_output['xip'] - gg.xip)
numpy.testing.assert_almost_equal(corr2_output['xip'] / gg.xip,
1.,
decimal=3)
print('gg.xim = ', gg.xim)
print('from corr2 output = ', corr2_output['xim'])
print('ratio = ', corr2_output['xim'] / gg.xim)
print('diff = ', corr2_output['xim'] - gg.xim)
numpy.testing.assert_almost_equal(corr2_output['xim'] / gg.xim,
1.,
decimal=3)
print('xip_im from corr2 output = ', corr2_output['xip_im'])
print('max err = ', max(abs(corr2_output['xip_im'])))
assert max(abs(corr2_output['xip_im'])) < 3.e-7
print('xim_im from corr2 output = ', corr2_output['xim_im'])
print('max err = ', max(abs(corr2_output['xim_im'])))
assert max(abs(corr2_output['xim_im'])) < 1.e-7
# As bin_slop decreases, the agreement should get even better.
# This test is slow, so only do it if running test_gg.py directly.
if __name__ == '__main__':
config['bin_slop'] = 0.2
gg = treecorr.GGCorrelation(config)
gg.process(cat1)
#print('gg.xip = ',gg.xip)
#print('bs0.xip = ',bs0_xip)
xip_err = gg.xip - bs0_xip
print('xip_err = ', xip_err)
print('max = ', max(abs(xip_err)))
assert max(abs(xip_err)) < 1.e-8
xim_err = gg.xim - bs0_xim
print('xim_err = ', xim_err)
print('max = ', max(abs(xim_err)))
assert max(abs(xim_err)) < 1.e-8
def test_aardvark():
# Eric Suchyta did a brute force calculation of the Aardvark catalog, so it is useful to
# compare the output from my code with that.
get_from_wiki('Aardvark.fit')
file_name = os.path.join('data','Aardvark.fit')
config = treecorr.read_config('Aardvark.yaml')
cat1 = treecorr.Catalog(file_name, config)
gg = treecorr.GGCorrelation(config)
gg.process(cat1)
direct_file_name = os.path.join('data','Aardvark.direct')
direct_data = numpy.genfromtxt(direct_file_name)
direct_xip = direct_data[:,3]
direct_xim = direct_data[:,4]
#print('gg.xip = ',gg.xip)
#print('direct.xip = ',direct_xip)
xip_err = gg.xip - direct_xip
print('xip_err = ',xip_err)
print('max = ',max(abs(xip_err)))
assert max(abs(xip_err)) < 2.e-7
print('xip_im = ',gg.xip_im)
print('max = ',max(abs(gg.xip_im)))
assert max(abs(gg.xip_im)) < 3.e-7
xim_err = gg.xim - direct_xim
print('xim_err = ',xim_err)
print('max = ',max(abs(xim_err)))
assert max(abs(xim_err)) < 1.e-7
print('xim_im = ',gg.xim_im)
print('max = ',max(abs(gg.xim_im)))
assert max(abs(gg.xim_im)) < 1.e-7
# However, after some back and forth about the calculation, we concluded that Eric hadn't
# done the spherical trig correctly to get the shears relative to the great circle joining
# the two positions. So let's compare with my own brute force calculation (i.e. using
# bin_slop = 0):
# This also has the advantage that the radial bins are done the same way -- uniformly
# spaced in log of the chord distance, rather than the great circle distance.
bs0_file_name = os.path.join('data','Aardvark.bs0')
bs0_data = numpy.genfromtxt(bs0_file_name)
bs0_xip = bs0_data[:,2]
bs0_xim = bs0_data[:,3]
#print('gg.xip = ',gg.xip)
#print('bs0.xip = ',bs0_xip)
xip_err = gg.xip - bs0_xip
print('xip_err = ',xip_err)
print('max = ',max(abs(xip_err)))
assert max(abs(xip_err)) < 1.e-7
xim_err = gg.xim - bs0_xim
print('xim_err = ',xim_err)
print('max = ',max(abs(xim_err)))
assert max(abs(xim_err)) < 5.e-8
# Check that we get the same result using the corr2 executable:
# Note: This is the only test of the corr2 executable that we do with nosetests.
# The other similar tests are blocked out with: if __name__ == '__main__':
import subprocess
corr2_exe = get_script_name('corr2')
p = subprocess.Popen( [corr2_exe,"Aardvark.yaml"] )
p.communicate()
corr2_output = numpy.genfromtxt(os.path.join('output','Aardvark.out'), names=True)
print('gg.xip = ',gg.xip)
print('from corr2 output = ',corr2_output['xip'])
print('ratio = ',corr2_output['xip']/gg.xip)
print('diff = ',corr2_output['xip']-gg.xip)
numpy.testing.assert_almost_equal(corr2_output['xip']/gg.xip, 1., decimal=3)
print('gg.xim = ',gg.xim)
print('from corr2 output = ',corr2_output['xim'])
print('ratio = ',corr2_output['xim']/gg.xim)
print('diff = ',corr2_output['xim']-gg.xim)
numpy.testing.assert_almost_equal(corr2_output['xim']/gg.xim, 1., decimal=3)
print('xip_im from corr2 output = ',corr2_output['xip_im'])
print('max err = ',max(abs(corr2_output['xip_im'])))
assert max(abs(corr2_output['xip_im'])) < 3.e-7
print('xim_im from corr2 output = ',corr2_output['xim_im'])
print('max err = ',max(abs(corr2_output['xim_im'])))
assert max(abs(corr2_output['xim_im'])) < 1.e-7
# As bin_slop decreases, the agreement should get even better.
# This test is slow, so only do it if running test_gg.py directly.
if __name__ == '__main__':
config['bin_slop'] = 0.2
gg = treecorr.GGCorrelation(config)
gg.process(cat1)
#print('gg.xip = ',gg.xip)
#print('bs0.xip = ',bs0_xip)
xip_err = gg.xip - bs0_xip
print('xip_err = ',xip_err)
print('max = ',max(abs(xip_err)))
assert max(abs(xip_err)) < 1.e-8
xim_err = gg.xim - bs0_xim
print('xim_err = ',xim_err)
print('max = ',max(abs(xim_err)))
assert max(abs(xim_err)) < 1.e-8
def test_fits_reader():
try:
import fitsio
except ImportError:
print('Skipping FitsReader tests, since fitsio not installed.')
return
get_from_wiki('Aardvark.fit')
r = FitsReader(os.path.join('data', 'Aardvark.fit'))
# Check things not allowed if not in context
with assert_raises(RuntimeError):
r.read(['RA'], slice(0, 10, 2), 1)
with assert_raises(RuntimeError):
r.read('RA')
with assert_raises(RuntimeError):
r.row_count('DEC', 1)
with assert_raises(RuntimeError):
r.row_count()
with assert_raises(RuntimeError):
r.names(1)
with assert_raises(RuntimeError):
r.names()
with assert_raises(RuntimeError):
1 in r
with r:
assert_raises(ValueError, r.check_valid_ext, 'invalid')
assert_raises(ValueError, r.check_valid_ext, 0)
r.check_valid_ext('AARDWOLF')
r.check_valid_ext(1)
# Default ext is 1
assert r.default_ext == 1
# Default ext is "in" reader
assert 1 in r
# Probably can slice, but depends on installed fitsio version
assert r.can_slice == (fitsio.__version__ > '1.0.6')
s = slice(0, 10, 2)
for ext in [1, 'AARDWOLF']:
data = r.read(['RA'], s, ext)
dec = r.read('DEC', s, ext)
assert data['RA'].size == 5
assert dec.size == 5
assert r.row_count('RA', ext) == 390935
assert r.row_count('GAMMA1', ext) == 390935
assert set(r.names(ext)) == set(
"INDEX RA DEC Z EPSILON GAMMA1 GAMMA2 KAPPA MU".split())
assert set(r.names(ext)) == set(r.names())
# Can read without slice or ext to use defaults
assert r.row_count() == 390935
g2 = r.read('GAMMA2')
assert len(g2) == 390935
d = r.read(['KAPPA', 'MU'])
assert len(d['KAPPA']) == 390935
assert len(d['MU']) == 390935
# check we can also index by integer, not just number
d = r.read(['DEC'], np.arange(10), 'AARDWOLF')
assert d.size == 10
if sys.version_info < (3, ): return # mock only available on python 3
from unittest import mock
# Again check things not allowed if not in context
with assert_raises(RuntimeError):
r.read(['RA'], slice(0, 10, 2), 1)
with assert_raises(RuntimeError):
r.read('RA')
with assert_raises(RuntimeError):
r.row_count('DEC', 1)
with assert_raises(RuntimeError):
r.row_count()
with assert_raises(RuntimeError):
r.names(1)
with assert_raises(RuntimeError):
r.names()
with assert_raises(RuntimeError):
1 in r
# Regardless of the system's fitsio version, check the two cases in code.
with mock.patch('fitsio.__version__', '1.0.6'):
with FitsReader(os.path.join('data', 'Aardvark.fit')) as r:
assert not r.can_slice
with mock.patch('fitsio.__version__', '1.1.0'):
with FitsReader(os.path.join('data', 'Aardvark.fit')) as r:
assert r.can_slice
def test_hdf_reader():
try:
import h5py
except ImportError:
print('Skipping HdfReader tests, since h5py not installed.')
return
get_from_wiki('Aardvark.hdf5')
r = HdfReader(os.path.join('data', 'Aardvark.hdf5'))
# Check things not allowed if not in context
with assert_raises(RuntimeError):
r.read(['RA'], slice(0, 10, 2), '/')
with assert_raises(RuntimeError):
r.read('RA')
with assert_raises(RuntimeError):
r.row_count('DEC', '/')
with assert_raises(RuntimeError):
r.row_count('DEC')
with assert_raises(RuntimeError):
r.names('/')
with assert_raises(RuntimeError):
r.names()
with assert_raises(RuntimeError):
'/' in r
with r:
# '/' is the only extension in this file.
# TODO: Add an hdf5 example with other valid choices for ext
assert_raises(ValueError, r.check_valid_ext, 'invalid')
r.check_valid_ext('/')
# Default ext is '/'
assert r.default_ext == '/'
# Default ext is "in" reader
assert '/' in r
# Can always slice
assert r.can_slice
s = slice(0, 10, 2)
data = r.read(['RA'], s)
dec = r.read('DEC', s)
assert data['RA'].size == 5
assert dec.size == 5
assert r.row_count('RA') == 390935
assert r.row_count('RA', '/') == 390935
assert r.row_count('GAMMA1') == 390935
# Unlike the other readers, this needs a column name.
assert_raises(TypeError, r.row_count)
assert set(r.names()) == set(
"INDEX RA DEC Z EPSILON GAMMA1 GAMMA2 KAPPA MU".split())
assert set(r.names('/')) == set(r.names())
# Again check things not allowed if not in context
with assert_raises(RuntimeError):
r.read(['RA'], slice(0, 10, 2), '/')
with assert_raises(RuntimeError):
r.read('RA')
with assert_raises(RuntimeError):
r.row_count('DEC', '/')
with assert_raises(RuntimeError):
r.row_count('DEC')
with assert_raises(RuntimeError):
r.names('/')
with assert_raises(RuntimeError):
r.names()
with assert_raises(RuntimeError):
'/' in r
def test_dessv():
try:
import fitsio
except ImportError:
print('Skipping dessv test, since fitsio is not installed')
return
#treecorr.set_omp_threads(1);
get_from_wiki('des_sv.fits')
file_name = os.path.join('data','des_sv.fits')
cat = treecorr.Catalog(file_name, ra_col='ra', dec_col='dec', ra_units='deg', dec_units='deg')
# Use an odd number to make sure we force some of the shuffle bits in InitializeCenters
# to happen.
npatch = 43
field = cat.getNField()
t0 = time.time()
patches = field.run_kmeans(npatch)
t1 = time.time()
print('patches = ',np.unique(patches))
assert len(patches) == cat.ntot
assert min(patches) == 0
assert max(patches) == npatch-1
# KMeans minimizes the total inertia.
# Check this value and the rms size, which should also be quite small.
xyz = np.array([cat.x, cat.y, cat.z]).T
cen = np.array([xyz[patches==i].mean(axis=0) for i in range(npatch)])
inertia = np.array([np.sum((xyz[patches==i] - cen[i])**2) for i in range(npatch)])
sizes = np.array([np.mean((xyz[patches==i] - cen[i])**2) for i in range(npatch)])**0.5
sizes *= 180. / np.pi * 60. # convert to arcmin
counts = np.array([np.sum(patches==i) for i in range(npatch)])
print('With standard algorithm:')
print('time = ',t1-t0)
print('total inertia = ',np.sum(inertia))
print('mean inertia = ',np.mean(inertia))
print('rms inertia = ',np.std(inertia))
print('mean size = ',np.mean(sizes))
print('rms size = ',np.std(sizes))
assert np.sum(inertia) < 200. # This is specific to this particular field and npatch.
assert np.std(inertia) < 0.2 * np.mean(inertia) # rms is usually small mean
assert np.std(sizes) < 0.1 * np.mean(sizes) # sizes have even less spread usually.
# Should all have similar number of points. Nothing is required here though.
print('mean counts = ',np.mean(counts))
print('min counts = ',np.min(counts))
print('max counts = ',np.max(counts))
# Check the alternate algorithm. rms inertia should be lower.
t0 = time.time()
patches = field.run_kmeans(npatch, alt=True)
t1 = time.time()
assert len(patches) == cat.ntot
assert min(patches) == 0
assert max(patches) == npatch-1
cen = np.array([xyz[patches==i].mean(axis=0) for i in range(npatch)])
inertia = np.array([np.sum((xyz[patches==i] - cen[i])**2) for i in range(npatch)])
sizes = np.array([np.mean((xyz[patches==i] - cen[i])**2) for i in range(npatch)])**0.5
sizes *= 180. / np.pi * 60. # convert to arcmin
counts = np.array([np.sum(patches==i) for i in range(npatch)])
print('With alternate algorithm:')
print('time = ',t1-t0)
print('total inertia = ',np.sum(inertia))
print('mean inertia = ',np.mean(inertia))
print('rms inertia = ',np.std(inertia))
print('mean size = ',np.mean(sizes))
print('rms size = ',np.std(sizes))
assert np.sum(inertia) < 200. # Total shouldn't increase much. (And often decreases.)
assert np.std(inertia) < 0.1 * np.mean(inertia) # rms should be even smaller here.
assert np.std(sizes) < 0.1 * np.mean(sizes) # This is only a little bit smaller.
# This doesn't keep the counts as equal as the standard algorithm.
print('mean counts = ',np.mean(counts))
print('min counts = ',np.min(counts))
print('max counts = ',np.max(counts))
# Finally, use a field with lots of top level cells to check the other branch in
# InitializeCenters.
field = cat.getNField(min_top=10)
t0 = time.time()
patches = field.run_kmeans(npatch)
t1 = time.time()
assert len(patches) == cat.ntot
assert min(patches) == 0
assert max(patches) == npatch-1
cen = np.array([xyz[patches==i].mean(axis=0) for i in range(npatch)])
inertia = np.array([np.sum((xyz[patches==i] - cen[i])**2) for i in range(npatch)])
sizes = np.array([np.mean((xyz[patches==i] - cen[i])**2) for i in range(npatch)])**0.5
sizes *= 180. / np.pi * 60. # convert to arcmin
counts = np.array([np.sum(patches==i) for i in range(npatch)])
# This doesn't give as good an initialization, so these are a bit worse usually.
print('With min_top=10:')
print('time = ',t1-t0)
print('total inertia = ',np.sum(inertia))
print('mean inertia = ',np.mean(inertia))
print('rms inertia = ',np.std(inertia))
print('mean size = ',np.mean(sizes))
print('rms size = ',np.std(sizes))
assert np.sum(inertia) < 210.
assert np.std(inertia) < 0.4 * np.mean(inertia) # I've seen over 0.3 x mean here.
assert np.std(sizes) < 0.1 * np.mean(sizes)
print('mean counts = ',np.mean(counts))
print('min counts = ',np.min(counts))
print('max counts = ',np.max(counts))