def get_cross_pairwise_ksz(df1, df2, params):
'''
Wrapper for calling pairwise_cross_ksz using a cross_dataset
object.
params: params object from paramTools.py
'''
dT1 = df1.dT.values
dT2 = df2.dT.values
z = df1.z.values # this requires Dc and z values to be the same
Dc = df1.Dc.values
if params.GET_TZAV_FAST:
tzav1 = get_tzav_fast(dT1, z, params.SIGMA_Z)
tzav2 = get_tzav_fast(dT2, z, params.SIGMA_Z)
else:
tzav1 = get_tzav(dT1, z, params.SIGMA_Z)
tzav2 = get_tzav(dT2, z, params.SIGMA_Z)
# more common values
ra_deg = df1.ra.values
dec_deg = df1.dec.values
assert params.UNEVEN_BINS # only uneven bins supported for this
bin_edges = params.BIN_EDGES
if params.DO_VARIANCE_WEIGHTED:
assert False # not implemented
else:
r_sep, p_uk = cross_pairwise_ksz_uneven_bins(Dc, ra_deg, dec_deg,
tzav1, tzav2, dT1, dT2,
bin_edges)
return r_sep, p_uk
def test_get_pairwise_ksz():
testPath = '/'.join((catalogTools.__file__).split('/')[:-2]) + '/tests/'
testParamFileFullPath = os.path.join(testPath, 'data_toTestAPI/params.ini')
params = paramTools.params(testParamFileFullPath)
df = produceFakeCatalog()
rsep, p_uk = pairwiser.get_pairwise_ksz(df, params, multithreading=False)
tzav = pairwiser.get_tzav(df.dT.values, df.z.values, params.SIGMA_Z)
if not params.UNEVEN_BINS:
rsep0, p_uk0 = pairwiser.pairwise_ksz(df.Dc.values,
df.ra.values,
df.dec.values,
tzav,
df.dT.values,
params.BIN_SIZE_MPC,
params.N_BINS,
multithreading=False)
else:
rsep0, p_uk0 = pairwiser.pairwise_ksz_uneven_bins(df.Dc.values,
df.ra.values,
df.dec.values,
tzav,
df.dT.values,
params.BIN_EDGES,
multithreading=False)
rsep_diff_sq = np.sum((rsep - rsep0)**2)
p_uk_diff_sq = np.sum((p_uk - p_uk0)**2)
assert rsep_diff_sq < 1e-10
assert p_uk_diff_sq < 1e-10
def test_get_tzav_fast():
N_gals = 15000
sigma_z = 0.01
z = np.random.uniform(size=N_gals)
dT = np.random.normal(size=N_gals)
tzav = pairwiser.get_tzav(dT, z, sigma_z)
tzav_fast = pairwiser.get_tzav_fast(dT, z, sigma_z)
chisq = np.sum((tzav - tzav_fast)**2)
assert chisq < 1e-8
def test_get_tzav():
size = 1000
dTs = np.random.uniform(low=0, high=100, size=size)
zs = np.random.uniform(low=0, high=100, size=size)
dTs = dTs + zs
sigma_z = 10
numer = np.empty_like(dTs)
denom = np.empty_like(dTs)
for j in range(size):
numer[j] = np.sum(dTs * np.exp(-(zs[j] - zs)**2 / (2 * sigma_z**2)))
denom[j] = np.sum(np.exp(-(zs[j] - zs)**2 / (2 * sigma_z**2)))
Tz_numpy = numer / denom
Tz_numba = pairwiser.get_tzav(dTs, zs, sigma_z)
diff_sq = (Tz_numpy - Tz_numba)**2
error = diff_sq.sum()
assert error < 1e-10
def test_varianceWeighted():
'''Tests variance_weighted_pairwise_ksz and
variance_weighted_pairwise_one_row'''
testPath = '/'.join((catalogTools.__file__).split('/')[:-2]) + '/tests/'
testParamFileFullPath = os.path.join(testPath, 'data_toTestAPI/params.ini')
params = paramTools.params(testParamFileFullPath)
df = produceFakeCatalog()
rsep, p_uk = pairwiser.get_pairwise_ksz(df, params, multithreading=False)
tzav = pairwiser.get_tzav(df.dT.values, df.z.values, params.SIGMA_Z)
div = np.ones(len(tzav))
rsep0, p_uk0 = pairwiser.variance_weighted_pairwise_ksz(
df.Dc.values,
df.ra.values, # noqa
df.dec.values,
tzav,
df.dT.values,
div,
params.BIN_SIZE_MPC,
params.N_BINS,
multithreading=False)
chisq = np.sum((p_uk - p_uk0)**2)
assert chisq < 1.0e10
''' reads a preprocessed catalog and computes tzav for it.
Writes a file with the numpy array
Written by: P. Gallardo.
'''
import numpy as np # noqa
from iskay import catalogTools # noqa
from iskay import pairwiser # noqa
#look in parent directory
df = catalogTools.preProcessedCat(directory='../ApPhotoResults').df
sigma_z = 0.01
dT = df.dT.values
z = df.z.values
tzav = pairwiser.get_tzav(dT, z, sigma_z)
np.savez('tzav_allCat', tzav)