def read_maps(self, fname):
f = FITS(fname, 'r')
p = f[1].read() # noqa
m = f[2].read()
m = np.array([m[n] for n in f[2].get_colnames()])
f.close()
return m
def mcalcat_read(path):
cat = []
fits = FITS(path, 'r')
for c in fits[1:]:
cat.append(
DataFrame.from_records(
c.read().byteswap().newbyteorder()).set_index(
'coadd_objects_id'))
fits.close()
return cat
def mcalcat_addzbin(d, zbin):
fits = FITS(zbin)
zbin = fits[1].read(columns=[
'coadd_objects_id', 'zbin_mcal', 'zbin_mcal_1p', 'zbin_mcal_1m',
'zbin_mcal_2p', 'zbin_mcal_2m'
])
fits.close()
zbin = zbin.byteswap().newbyteorder()
zbin = DataFrame.from_records(zbin).set_index('coadd_objects_id')
zbin = zbin.rename(columns={"zbin_mcal": "zbin"})
cat = concat([d, zbin], axis=1)
return cat[cat['zbin'] > -1].dropna()
def get_filename_from_fits(self, ffile, ftype):
fname = None
try:
f = FITS(ffile)
h = f[0].read_header()
expnum = h['EXPNUM']
fname = 'DECam_%08i' % expnum
fname += '.' + ftype
f.close()
except Exception as e:
self.logger.info('ERROR ' + str(e))
finally:
return fname
def mcalcat_load(cat):
fields = [
'coadd_objects_id', 'ra', 'dec', 'psf_e1', 'psf_e2', 'e1', 'e2', 'R11',
'R22', 'flags_select'
]
fits = FITS(cat)
cat = fits[1].read(columns=fields)
fits.close()
cat = cat.byteswap().newbyteorder()
cat = DataFrame.from_records(cat).set_index('coadd_objects_id')
# Select usable objects
cat = cat[cat.flags_select == 0]
# Weights
cat['w'] = ones(cat.shape[0])
return cat.drop(['flags_select'], axis=1)
def write_fit(data, filename, count=1, clobber=False):
import fitsio
from fitsio import FITS, FITSHDR
import os.path
from astropy.io import fits
#If the first time it enters the file exist does not write
#anything write only if clobber is set to True
if count == 1:
#If does not exist, create it and write data
if not os.path.isfile(filename):
hdu = fits.PrimaryHDU()
hdul = fits.HDUList([hdu])
corrhdu = fits.BinTableHDU(data, name='')
hdul.insert(1, corrhdu)
hdul.writeto(filename, overwrite=clobber)
print("Creating file: ", filename)
#If exist open it and add data
else:
if clobber:
hdu = fits.PrimaryHDU()
hdul = fits.HDUList([hdu])
corrhdu = fits.BinTableHDU(data, name='')
hdul.insert(1, corrhdu)
hdul.writeto(filename, overwrite=clobber)
print("Clobering file: ", filename)
else:
raise Exception(
'You tried to overwrite an existing file, without setting clobber'
)
else:
fits = FITS(filename, 'rw')
fits[-1].append(data)
print("Apending File: ", filename)
def write_fit(data, file_name):
import fitsio
from fitsio import FITS, FITSHDR
import os.path
if not os.path.isfile(file_name):
fitsio.write(file_name, data, clobber=False)
else:
fits = FITS(file_name, 'rw')
fits[-1].append(data)
def load(cls, fname):
fits_inventory = DESFITSInventory(fname)
hdus_present = sorted(fits_inventory.raws)
images = cls()
with FITS(fname) as fits:
images.images = [
DESDataImage.load_from_open(fits, ext) for ext in hdus_present
]
return images
def read_table(filename, hdu=1, subset=None):
"""
Read the first HDU table from a fits file
"""
file = FITS(filename, upper=True)
if subset is not None:
column, start, end = subset
return numpy.array(file[hdu][column][start:end], copy=True)
else:
return numpy.array(file[hdu].read(), copy=True)
def write_fit(data, file_name):
import fitsio
from fitsio import FITS, FITSHDR
import os.path
import pandas
if not os.path.isfile(file_name):
fitsio.write(file_name, data, clobber=False)
else:
fits = FITS(file_name, 'rw')
#appending data to the las hdu
fits[-1].append(data)
def read_image(filename, hdu=0):
"""
Read the zeroth image HDU from a fits file
"""
file = FITS(filename, upper=True)
# A copy of data is made before the file object
# is dereferenced. This is to ensure no back references
# to the fits object and file gets closed in a timely
# fashion.
return numpy.array(file[hdu].read(), copy=True)
def mcalcat_write(cat, path):
from os import remove
if exists(path):
remove(path)
fits = FITS(path, 'rw')
for c in cat:
fits.write(c.reset_index().to_records(index=False))
fits.close()
def write_fit(data, file_name):
import fitsio
from fitsio import FITS,FITSHDR
import os.path
if not os.path.isfile(file_name):
fitsio.write(file_name, data, clobber=False)
else:
try:
fits = FITS(file_name,'rw')
fits[-1].append(data)
except OSError as e:
print("Ignore OSError from writing:", data['expnum'])
print(e)
pass
def write_fit(data, names, filename):
import fitsio
from fitsio import FITS, FITSHDR
import os.path
if not os.path.isfile(filename):
fits = FITS(filename, 'rw')
fits.write(data, names=names, clobber=False)
print("Writing file: ", filename)
else:
fits = FITS(filename, 'rw')
fits[-1].append(data)
print("Apending File: ", filename)
def random_preprocess():
randoms = fitsio.read('flagship_randoms_v2.fits')
ra, dec, jk_label = randoms["RA"], randoms["DEC"], randoms["JK_LABEL"]
centers = np.loadtxt("flagship_jk_centers_v2.txt")
for jk in range(len(centers)):
jk_mask = jk_label != jk
random_jk = {"RA": ra[jk_mask], "DEC": dec[jk_mask]}
fits = FITS('data/random_jk_' + str(jk) + '.fits', 'rw')
fits.write(random_jk, names=["RA", "DEC"])
fits.close()
return None
def read_maps(self, fname):
f = FITS(fname, 'r')
p = f[0].read() # noqa
m = f[1].read()
f.close()
return m
def write_maps(self, maps, fname):
f = FITS(fname, 'rw', clobber=True)
f.write(self.goodpix)
f.write(maps)
f.close()
def write_catalog(fname,
x,
y,
z,
w,
type="DATA",
format="fits",
coord="PSEUDO_EQUATORIAL"):
if format == "txt":
np.savetxt(fname, np.transpose([x, y, z, w]))
elif format == "fits":
# current DM supported
DM_VERSION = "2.5.0"
# Extracted from AsciiToFits.py written by Daniele Tavagnacco
if coord == "PSEUDO_EQUATORIAL":
columns = [['SOURCE_ID', -1, 'f8'], ['RA', x, 'f8'],
['DEC', y, 'f8'], ['REDSHIFT', z, 'f8'],
['WEIGHT', -1, 'f8'], ['DENSITY', w, 'f8']]
elif coord == "CARTESIAN":
columns = [['SOURCE_ID', -1, 'f8'], ['COMOV_X', x, 'f8'],
['COMOV_Y', y, 'f8'], ['COMOV_Z', z, 'f8'],
['WEIGHT', -1, 'f8'], ['DENSITY', w, 'f8']]
header_keywords = {
"TELESCOP": "EUCLID",
"INSTRUME": "LE3GC-MOCKS",
"FILENAME": fname,
"CAT_ID": "MOCK",
"COORD": coord,
"ANGLE": "DEGREES"
}
# fname_nodir = fname[[pos for pos, char in enumerate(fname) if char == '/'][-1]+1:]
# zshell = fname[fname.find('zshell'):-5]
# header_keywords = {'TELESCOP' : 'EUCLID ',
# 'INSTRUME' : 'LE3IDSELID',
# 'FILENAME' : fname_nodir,
# 'CAT_TYPE' : type,
# 'CAT_NAME' : 'MOCK-LE3GC',
# 'CAT_NOBJ' : x.size,
# 'COORD ' : 'EQUATORIAL',
# 'ANGLE ' : 'DEG ',
# 'COMPLETE' : 1.0,
# 'PURITY ' : 1.0,
# 'SELECT ' : zshell}
extension = "CATALOG"
xmlKeys = {
"pf": "PK_LE3_GC_WindowMultipoles",
"instr": "LE3_GC_MOCKS",
"id": "MOCK",
"coord": coord
}
print("Preparing FITS structure")
types = []
# keep just wanted columns ...bad but..
for c in columns:
types.append((c[0], c[2]))
hdr = FITSHDR()
print("+ Add keywords")
for k in header_keywords:
hdr[k] = header_keywords[k]
keep_table = {}
for c in columns:
# if required but not existing (-1) fill with ones
if (c[1] is -1):
# add tmp column with correct name and position, bu only ones
print(str("+ -Col '%s' filled" % c[0]))
keep_table[c[0]] = np.ones_like(columns[1][1],
dtype=np.float64)
else:
# keep column with requested position in the input file
print(str("+ -Col '%s' from '%s'" % (c[0], c[1])))
keep_table[c[0]] = c[1].astype(np.float64)
fullsize = len(keep_table) * len(c[1]) * 8 / 1024 / 1024.
print(str("+ ~%.2f MB in memory" % fullsize))
# now write some data
print(str("+ Write FITS: %s" % fname))
fits = FITS(fname, 'rw', clobber=True)
fits.write_table(data=keep_table, header=hdr, extname=extension)
fits.close()
print("+ Preparing XML product")
with open(fname.replace(input.cat4le3_format, "xml"), "w+") as f:
f.write(
'''<?xml version="1.0" encoding="UTF-8" standalone="no" ?>\n'''
)
if (type == "RANDOM"):
f.write(
'''<p1:DpdLE3gcInputRandCat xmlns:p1="http://euclid.esa.org/schema/dpd/le3/gc/inp/catrandin">\n'''
)
else:
f.write(
'''<p1:DpdLE3gcInputDataCat xmlns:p1="http://euclid.esa.org/schema/dpd/le3/gc/inp/catdatain">\n'''
)
f.write(''' <Header>\n''')
f.write(
str(''' <ProductId>%s</ProductId>\n''' %
header_keywords["FILENAME"].split('.')[0]))
f.write(
''' <ProductType>dpdLE3gcInputRandCat</ProductType>\n''')
f.write(''' <SoftwareName>LE3_GC_test</SoftwareName>\n''')
f.write(''' <SoftwareRelease>1.0</SoftwareRelease>\n''')
f.write(
''' <ManualValidationStatus>UNKNOWN</ManualValidationStatus>\n'''
)
f.write(''' <PipelineRun>LE3_GC_Test_Inputs</PipelineRun>\n''')
f.write(''' <ExitStatusCode>OK</ExitStatusCode>\n''')
f.write(
str(''' <DataModelVersion>%s</DataModelVersion>\n''' %
DM_VERSION))
f.write(
str(''' <MinDataModelVersion>%s</MinDataModelVersion>\n''' %
DM_VERSION))
f.write(''' <ScientificCustodian>LE3</ScientificCustodian>\n''')
f.write(''' <AccessRights>\n''')
f.write(
''' <EuclidConsortiumRead>true</EuclidConsortiumRead>\n'''
)
f.write(
''' <EuclidConsortiumWrite>true</EuclidConsortiumWrite>\n'''
)
f.write(
''' <ScientificGroupRead>true</ScientificGroupRead>\n''')
f.write(
''' <ScientificGroupWrite>true</ScientificGroupWrite>\n'''
)
f.write(''' </AccessRights>\n''')
f.write(''' <Curator>\n''')
f.write(''' <Name>SDC-IT</Name>\n''')
f.write(''' </Curator>\n''')
f.write(str(''' <Creator>%s</Creator>\n''' % xmlKeys["pf"]))
f.write(
''' <CreationDate>2019-10-31T12:12:12Z</CreationDate>\n''')
f.write(''' </Header>\n''')
f.write(''' <Data>\n''')
f.write(
str(''' <Instrument>%s</Instrument>\n''' % xmlKeys["instr"]))
f.write(str(''' <Catalog_ID>%s</Catalog_ID>\n''' % xmlKeys["id"]))
f.write(str(''' <CoordType>%s</CoordType>\n''' %
xmlKeys["coord"]))
f.write(''' <Catalog format="le3.gc.cat.test" version="0.2">\n''')
f.write(''' <DataContainer filestatus="PROPOSED">\n''')
f.write(str(''' <FileName>%s.fits</FileName>\n''' % fname))
f.write(''' </DataContainer>\n''')
f.write(''' </Catalog>\n''')
f.write(''' </Data>\n''')
if (type == "RANDOM"):
f.write('''</p1:DpdLE3gcInputRandCat>\n''')
else:
f.write('''</p1:DpdLE3gcInputDataCat>\n''')
f.close()
print("files %s and %s written" %
(fname, fname.replace(input.cat4le3_format, "xml")))
else:
print("ERROR: unrecognized format in write_catalog")
sys.exit(-1)
['cl_rand', cl_rand, 'f8']]
header_keywords = {}
types = []
# keep just wanted columns ...bad but..
for c in columns:
types.append((c[0], c[2]))
hdr = FITSHDR()
print("+ Add keywords")
for k in header_keywords:
hdr[k] = header_keywords[k]
keep_table = {}
for c in columns:
# keep column with requested position in the input file
print(str("+ -Col '%s'" % (c[0])))
keep_table[c[0]] = c[1] #.astype(np.float64)
fullsize = len(keep_table) * len(c[1]) * 8 / 1024 / 1024.
print(str("+ ~%.2f MB in memory" % fullsize))
fname = input.cls_fname(zmin, zmax, run=input.pinocchio_first_run)
print("## Writing FITS: %s" % fname)
fitsfile = FITS(fname, 'rw')
fitsfile.write_table(data=keep_table, header=hdr)
fitsfile.close()
print("# DONE!")
def read_metadata(filename, hdu=0):
"""
Read the metadata of a HDU table
"""
file = FITS(filename, upper=True)
return dict(file[hdu].read_header())
def size_table(filename, hdu=1):
"""
Read the first HDU table from a fits file
"""
file = FITS(filename, upper=True)
return file[hdu].get_nrows()
def execute_cmd(x_script: str, out_file: str, src: str, run_type: str, timeout: float) \
-> Tuple[Union[FITS, str], str, bool, list, list]:
"""
This function is called for the local compute option. It will run the supplied XSPEC script, then check
parse the output for errors and check that the expected output file has been created.
:param str x_script: The path to an XSPEC script to be run.
:param str out_file: The expected path for the output file of that XSPEC script.
:param str src: A string representation of the source object that this fit is associated with.
:param str run_type: A flag that tells this function what type of run this is; e.g. fit or conv_factors.
:param float timeout: The length of time (in seconds) which the XSPEC script is allowed to run for before being
killed.
:return: FITS object of the results, string repr of the source associated with this fit, boolean variable
describing if this fit can be used, list of any errors found, list of any warnings found.
:rtype: Tuple[Union[FITS, str], str, bool, list, list]
"""
if XSPEC_VERSION is None:
raise XSPECNotFoundError(
"There is no XSPEC installation detectable on this machine.")
# We assume the output will be usable to start with
usable = True
cmd = "xspec - {}".format(x_script)
# I add exec to the beginning to make sure that the command inherits the same process ID as the shell, which
# allows the timeout to kill the XSPEC run rather than the shell process. Entirely thanks to slayton on
# https://stackoverflow.com/questions/4789837/how-to-terminate-a-python-subprocess-launched-with-shell-true
xspec_proc = Popen("exec " + cmd, shell=True, stdout=PIPE, stderr=PIPE)
# This makes sure the process is killed if it does timeout
try:
out, err = xspec_proc.communicate(timeout=timeout)
except TimeoutExpired:
xspec_proc.kill()
out, err = xspec_proc.communicate()
# Need to infer the name of the source to supply it in the warning
source_name = x_script.split('/')[-1].split("_")[0]
warnings.warn("An XSPEC fit for {} has timed out".format(source_name))
usable = False
out = out.decode("UTF-8").split("\n")
err = err.decode("UTF-8").split("\n")
err_out_lines = [
line.split("***Error: ")[-1] for line in out if "***Error" in line
]
warn_out_lines = [
line.split("***Warning: ")[-1] for line in out if "***Warning" in line
]
err_err_lines = [
line.split("***Error: ")[-1] for line in err if "***Error" in line
]
warn_err_lines = [
line.split("***Warning: ")[-1] for line in err if "***Warning" in line
]
if usable and len(err_out_lines) == 0 and len(err_err_lines) == 0:
usable = True
else:
usable = False
error = err_out_lines + err_err_lines
warn = warn_out_lines + warn_err_lines
if os.path.exists(out_file + "_info.csv") and run_type == "fit":
# The original version of the xga_output.tcl script output everything as one nice neat fits file
# but life is full of extraordinary inconveniences and for some reason it didn't work if called from
# a Jupyter Notebook. So now I'm going to smoosh all the csv outputs into one fits.
results = pd.read_csv(out_file + "_results.csv", header="infer")
# This is the csv with the fit results in, creates new fits file and adds in
fitsio.write(out_file + ".fits",
results.to_records(index=False),
extname="results",
clobber=True)
del results
# The information about individual spectra, exposure times, luminosities etc.
spec_info = pd.read_csv(out_file + "_info.csv", header="infer")
# Gets added into the existing file
fitsio.write(out_file + ".fits",
spec_info.to_records(index=False),
extname="spec_info")
del spec_info
# This finds all of the matching spectrum plot csvs were generated
rel_path = "/".join(out_file.split('/')[0:-1])
# This is mostly just used to find how many files there are
spec_tabs = [
rel_path + "/" + sp for sp in os.listdir(rel_path)
if "{}_spec".format(out_file) in rel_path + "/" + sp
]
for spec_i in range(1, len(spec_tabs) + 1):
# Loop through and redefine names like this to ensure they're in the right order
spec_plot = pd.read_csv(out_file + "_spec{}.csv".format(spec_i),
header="infer")
# Adds all the plot tables into the existing fits file in the right order
fitsio.write(out_file + ".fits",
spec_plot.to_records(index=False),
extname="plot{}".format(spec_i))
del spec_plot
# This reads in the fits we just made
with FITS(out_file + ".fits") as res_tables:
tab_names = [tab.get_extname() for tab in res_tables]
if "results" not in tab_names or "spec_info" not in tab_names:
usable = False
# I'm going to try returning the file path as that should be pickleable
res_tables = out_file + ".fits"
elif os.path.exists(out_file) and run_type == "conv_factors":
res_tables = out_file
usable = True
else:
res_tables = None
usable = False
return res_tables, src, usable, error, warn
def force_phot (filenames, ra_deg, dec_deg, fits_out, nsigma=5,
keys2add=None, keys2add_dtypes=None, tel='ML1'):
# initialize rows to be converted to fits table
rows = []
# convert keys2add and keys2add_dtypes to lists
keys2add_list = keys2add.upper().split(',')
dtypes_list = keys2add_dtypes.split(',')
# loop through [filenames]
for nfile, filename in enumerate(filenames):
# start row
row = [filename]
# read in header and data
data, header = read_hdulist (filename, get_header=True)
# add header keywords to output table
for key in keys2add_list:
if key in header:
row.append (header[key])
else:
row.append (None)
log.error ('key {} not in header of {}'
.format(key, filename))
# convert input [ra_deg] and [dec_deg] to pixel coordinates
xcoord, ycoord = WCS(header).all_world2pix(ra_deg, dec_deg, 1)
# skip if coordinates not on the image
ysize, xsize = data.shape
dpix_edge = 10
if (xcoord < dpix_edge or xcoord > xsize-dpix_edge or
ycoord < dpix_edge or ycoord > ysize-dpix_edge):
log.warning ('pixel coordinates (x,y)=({},{}) not on image for {}'
'; skipping it'.format(int(xcoord), int(ycoord),
filename))
continue
# determine optimal flux
# ----------------------
# read PSFEx binary fits table
base = filename.split('.fits')[0]
psfex_bintable = '{}_psf.fits'.format(base)
# background STD
fits_bkg_std = '{}_bkg_std.fits'.format(base)
if os.path.exists(fits_bkg_std):
data_bkg_std = read_hdulist (fits_bkg_std, dtype='float32')
else:
# if it does not exist, create it from the background mesh
fits_bkg_std_mini = '{}_bkg_std_mini.fits'.format(base)
data_bkg_std_mini, header_mini = read_hdulist (
fits_bkg_std_mini, get_header=True, dtype='float32')
if 'BKG-SIZE' in header_mini:
bkg_size = header_mini['BKG-SIZE']
else:
bkg_size = get_par(set_zogy.bkg_boxsize,tel)
if True:
# determine scalar bkg_std value from mini image at
# xcoord, ycoord
x_index_mini = int(int(xcoord-0.5)/bkg_size)
y_index_mini = int(int(ycoord-0.5)/bkg_size)
data_bkg_std = data_bkg_std_mini[y_index_mini, x_index_mini]
else:
# determine full bkg_std image from mini image
# determine whether interpolation is allowed across different
# channels in [mini2back] using function get_Xchan_bool
chancorr = get_par(set_zogy.MLBG_chancorr,tel)
interp_Xchan_std = get_Xchan_bool (tel, chancorr, 'new',
std=True)
data_bkg_std = mini2back (data_bkg_std_mini, data.shape,
order_interp=1, bkg_boxsize=bkg_size,
interp_Xchan=interp_Xchan_std,
timing=get_par(set_zogy.timing,tel))
# data mask
fits_mask = filename.replace('_red.fits', '_mask.fits')
data_mask = read_hdulist (fits_mask, dtype='uint8')
# object mask - needs to be created with source extractor; for
# now, assume entire image is masked, leading to local
# background not being used
objmask = np.ones (data.shape, dtype=bool)
# determine optimal fluxes at pixel coordinates
flux_opt, fluxerr_opt = get_psfoptflux (
psfex_bintable, data, data_bkg_std**2, data_mask,
np.array([xcoord]), np.array([ycoord]),
imtype='new', fwhm=header['PSF-FWHM'],
D_objmask=objmask, set_zogy=set_zogy, tel=tel)
# flux_opt and fluxerr_opt are 1-element arrays
s2n = flux_opt[0] / fluxerr_opt[0]
# convert fluxes to magnitudes by applying the zeropoint
keys = ['EXPTIME', 'FILTER', 'DATE-OBS']
exptime, filt, obsdate = [header[key] for key in keys]
# get zeropoint from [header]
if 'PC-ZP' in header:
zp = header['PC-ZP']
else:
log.warning ('keyword PC-ZP not in header of {}; skipping it'
.format(filename))
continue
# determine object airmass, unless input image is a combined
# image
if 'R-V' in header or 'R-COMB-M' in header:
airmass = 1.0
else:
lat = get_par(set_zogy.obs_lat,tel)
lon = get_par(set_zogy.obs_lon,tel)
height = get_par(set_zogy.obs_height,tel)
airmass = get_airmass(ra_deg, dec_deg, obsdate, lat, lon, height)
log.info ('airmass: {}'.format(airmass))
ext_coeff = get_par(set_zogy.ext_coeff,tel)[filt]
mag_opt, magerr_opt = apply_zp (flux_opt, zp, airmass, exptime, filt,
ext_coeff, fluxerr=fluxerr_opt)
# read limiting magnitude at pixel coordinates
filename_limmag = filename.replace('_red.fits', '_red_limmag.fits')
if os.path.isfile(filename_limmag):
x_index = int(xcoord-0.5)
y_index = int(ycoord-0.5)
t0 = time.time()
try:
data_limmag = FITS(filename_limmag)[-1]
limmag = data_limmag[y_index, x_index][0][0]
except:
data_limmag = read_hdulist(filename_limmag)
limmag = data_limmag[y_index, x_index]
# convert limmag from 5-sigma to nsigma
limmag += -2.5*np.log10(nsigma/5)
row += [mag_opt[0], magerr_opt[0], s2n, limmag]
# append row to rows
rows.append(row)
# initialize output table
names = ['FILENAME']
dtypes = ['U100']
for ikey, key in enumerate(keys2add_list):
names.append(key)
dtype_str = dtypes_list[ikey]
if dtype_str in ['float', 'int', 'bool']:
dtype = eval(dtype_str)
else:
dtype = dtype_str
dtypes.append(dtype)
names += ['MAG_OPT', 'MAGERR_OPT', 'S2N', 'LIMMAG_{}SIGMA'.format(nsigma)]
dtypes += [float, float, float, float]
if len(rows) == 0:
# rows without entries: create empty table
table = Table(names=names, dtype=dtypes)
else:
table = Table(rows=rows, names=names, dtype=dtypes)
# save output fits table
table.write(fits_out, overwrite=True)
def __init__(self, fname):
self.hdr = []
with FITS(fname) as fits:
for hdu in fits:
self.hdr.append(hdu.read_header())
def write_maps(self, maps, fname):
f = FITS(fname, 'rw', clobber=True)
f.write([self.goodpix], names=['pixels'])
names = ['map_%d' % i for i in range(len(maps))]
f.write(list(maps), names=names)
f.close()
def summary_result(aim, mock_n, reff, mock_g1, mock_g2, dx, dy, lens_g1,
lens_g2, init_flux, KSB, ID):
if aim == 'mock_image' or aim == 'HST_decon':
gain = 1.0
zero_point = 1.0
expt = 1.0
read_noise = 1.0
elif aim == 'Eu_conv' or aim == 'Euclidiz':
gain = config.Eu_gain
zero_point = config.Eu_zp
expt = config.Eu_expt
read_noise = config.Eu_rn
elif aim == 'HST_conv':
gain = config.HST_gain
zero_point = config.HST_zp
expt = config.HST_expt
read_noise = config.HST_rn
#Dictionary to store a row after each run of the code
dict = np.zeros(1,
dtype=[('aim', 'S32'), ('truth_pxscale', 'f8'),
('truth_size', 'f8'), ('mock_idx', 'f8'),
('mock_flux_mag', 'f8'), ('mock_SNR', 'f8'),
('mock_mu', 'f8'), ('mock_n', 'f8'),
('mock_reff', 'f8'), ('mock_g1', 'f8'),
('mock_g2', 'f8'), ('mock_flux', 'f8'),
('dx', 'f8'), ('dy', 'f8'), ('lens_g1', 'f8'),
('lens_g2', 'f8'), ('Eu_rn', 'f8'),
('Eu_gain', 'f8'), ('Eu_expt', 'f8'),
('HST_rn', 'f8'), ('HST_gain', 'f8'),
('HST_expt', 'f8'), ('init_n', 'f8'),
('init_reff', 'f8'), ('init_flux', 'f8'),
('init_flux_mag', 'f8'), ('g1_obs', 'f8'),
('g2_obs', 'f8'),
('g1_cor', 'f8'), ('g2_cor', 'f8'),
('shape_cor_err', 'f8'), ('m_sigma', 'f8'),
('m_amp', 'f8'), ('m_rho4', 'f8'), ('m_n', 'f8'),
('KSB_status', 'f8')])
dict['aim'] = aim
dict['truth_pxscale'] = config.truth_pxscale
dict['truth_size'] = config.truth_size
dict['mock_idx'] = inp.idx
dict['mock_flux_mag'] = inp.mock_mag
dict['mock_SNR'] = inp.Eu_snr
dict['mock_mu'] = config.mock_mu
dict['mock_n'] = '%.2f' % mock_n
dict['mock_reff'] = '%.2f' % reff
dict['mock_g1'] = '%.4f' % mock_g1
dict['mock_g2'] = '%.4f' % mock_g2
dict['mock_flux'] = '%.1f' % config.mock_flux
dict['dx'] = dx
dict['dy'] = dy
dict['lens_g1'] = '%.5f' % lens_g1
dict['lens_g2'] = '%.5f' % lens_g2
dict['Eu_rn'] = config.Eu_rn
dict['Eu_gain'] = config.Eu_gain
dict['Eu_expt'] = config.Eu_expt
dict['HST_rn'] = config.HST_rn
dict['HST_gain'] = config.HST_gain
dict['HST_expt'] = config.HST_expt
dict['init_n'] = config.init_n
dict['init_reff'] = config.init_reff
dict[
'init_flux'] = init_flux / 1000 # 10000 is the same value you can find in the model fitting to avoid the fitting failure
dict['init_flux_mag'] = '%.2f' % (zero_point - 2.5 * np.log10(
(init_flux / 1000) * gain / expt))
#dict['bestfit_flux']='%.5f'%(fit_mod.flux[0]/1000) # 10000 is the same value you can find in the model fitting to avoid the fitting failure
#dict['bestfit_flux_mag']='%.2f'%(zero_point - 2.5*np.log10((fit_mod.flux[0]/1000)*gain/expt))
#dict['bestfit_reff_pix']='%.5f'%fit_mod.r_eff[0]
#dict['bestfit_reff_arcsec']='%.5f'%(config.truth_pxscale*fit_mod.r_eff[0])
#dict['bestfit_n']='%.5f'%fit_mod.n[0]
#dict['bestfit_x0']='%.5f'%fit_mod.x_0[0]
#dict['bestfit_y0']='%.5f'%fit_mod.y_0[0]
#dict['bestfit_g1']='%.5f'%fit_mod.g1[0]
#dict['bestfit_g2']='%.5f'%fit_mod.g2[0]
dict['g1_obs'] = '%.5f' % KSB.observed_shape.g1
dict['g2_obs'] = '%.5f' % KSB.observed_shape.g2
dict['g1_cor'] = '%.5f' % KSB.corrected_g1
dict['g2_cor'] = '%.5f' % KSB.corrected_g2
dict['shape_cor_err'] = '%.5f' % KSB.corrected_shape_err
dict['m_sigma'] = '%.5f' % KSB.moments_sigma
dict['m_amp'] = '%.5f' % KSB.moments_amp
dict['m_rho4'] = '%.5f' % KSB.moments_rho4
dict['m_n'] = float(KSB.moments_n_iter)
#dict['chi2']=chi2
#dict['red_chi2']=red_chi2
#dict['SNR']=SNR
#dict['nfev']='%.5f'%fitter.fit_info['nfev']
#dict['fvec']=fitter.fit_info['fvec']
#dict['ierr']='%.5f'%fitter.fit_info['ierr']
dict['KSB_status'] = float(KSB.moments_status)
#Creation of fits table containing one lines of results
table = make_table_file()
table.write(config.result_filename + '_' + str(aim) + '_' + 'ID' +
str(ID) + config.fits)
f = FITS(
config.result_filename + '_' + str(aim) + '_' + 'ID' + str(ID) +
config.fits, 'rw')
f[-1].append(dict)
f.close()
return
def incomplete_shear_preprocess(zmin, zmax):
shear = fitsio.read("source_v2.fits",
columns=[
"ra_gal", "dec_gal", "observed_redshift_gal",
"gamma1", "gamma2"
])
ra, dec, z = shear["ra_gal"], shear["dec_gal"], shear[
"observed_redshift_gal"]
g1, g2 = shear["gamma1"], shear["gamma2"]
dz = 0.05 * (1 + z) #redshift uncertainties
z = np.random.normal(z, dz) #perturbed redshifts
mask = (z > zmax + 0.1)
z = z[mask]
ra, dec = ra[mask], dec[mask]
g1, g2 = g1[mask], g2[mask]
print("length of the catalog after applying the cut", len(ra))
coord = np.vstack([ra, dec]).T
centers = np.loadtxt("flagship_jk_centers_v2.txt")
NJK = centers.shape[0]
print("Segmentation begins!")
gal_labels_jk = kmeans_radec.find_nearest(coord, centers)
print("Done with assigning jacknife labels to galaxies")
gals = {
"RA": ra,
"DEC": dec,
"gamma1": g1,
"gamma2": g2,
"redshift": z,
"JK_LABEL": gal_labels_jk
}
fits = FITS('data/incomplete_shear_zmax_' + str(round(zmax, 1)) + '.fits',
'rw')
fits.write(gals,
names=["RA", "DEC", "gamma1", "gamma2", "redshift", "JK_LABEL"])
fits.close()
for jk in range(len(centers)):
gal_jk_mask = gals["JK_LABEL"] != jk
gal_jk = {
"RA": ra[gal_jk_mask],
"DEC": dec[gal_jk_mask],
"gamma1": g1[gal_jk_mask],
"gamma2": g2[gal_jk_mask],
"redshift": z[gal_jk_mask]
}
fits = FITS(
'data/incomplete_shear_zmax_' + str(round(zmax, 1)) + '_jk_' +
str(jk) + '.fits', 'rw')
fits.write(gal_jk, names=["RA", "DEC", "gamma1", "gamma2", "redshift"])
fits.close()
return None
def wrapper(*args, **kwargs):
# The first argument of all of these XSPEC functions will be the source object (or a list of),
# so rather than return them from the XSPEC model function I'll just access them like this.
if isinstance(args[0], BaseSource):
sources = [args[0]]
elif isinstance(args[0], (list, BaseSample)):
sources = args[0]
else:
raise TypeError(
"Please pass a source object, or a list of source objects.")
# This is the output from whatever function this is a decorator for
# First return is a list of paths of XSPEC scripts to execute, second is the expected output paths,
# and 3rd is the number of cores to use.
# run_type describes the type of XSPEC script being run, for instance a fit or a fakeit run to measure
# countrate to luminosity conversion constants
script_list, paths, cores, run_type, src_inds, radii, timeout = xspec_func(
*args, **kwargs)
src_lookup = {
repr(src): src_ind
for src_ind, src in enumerate(sources)
}
rel_src_repr = [repr(sources[src_ind]) for src_ind in src_inds]
# Make sure the timeout is converted to seconds, then just stored as a float
timeout = timeout.to('second').value
# This is what the returned information from the execute command gets stored in before being parceled out
# to source and spectrum objects
results = {s: [] for s in src_lookup}
if run_type == "fit":
desc = "Running XSPEC Fits"
elif run_type == "conv_factors":
desc = "Running XSPEC Simulations"
if len(script_list) > 0:
# This mode runs the XSPEC locally in a multiprocessing pool.
with tqdm(total=len(script_list),
desc=desc) as fit, Pool(cores) as pool:
def callback(results_in):
"""
Callback function for the apply_async pool method, gets called when a task finishes
and something is returned.
"""
nonlocal fit # The progress bar will need updating
nonlocal results # The dictionary the command call results are added to
if results_in[0] is None:
fit.update(1)
return
else:
res_fits, rel_src, successful, err_list, warn_list = results_in
results[rel_src].append(
[res_fits, successful, err_list, warn_list])
fit.update(1)
for s_ind, s in enumerate(script_list):
pth = paths[s_ind]
src = rel_src_repr[s_ind]
pool.apply_async(execute_cmd,
args=(s, pth, src, run_type, timeout),
callback=callback)
pool.close() # No more tasks can be added to the pool
pool.join(
) # Joins the pool, the code will only move on once the pool is empty.
elif len(script_list) == 0:
warnings.warn("All XSPEC operations had already been run.")
# Now we assign the fit results to source objects
for src_repr in results:
# Made this lookup list earlier, using string representations of source objects.
# Finds the ind of the list of sources that we should add these results to
ind = src_lookup[src_repr]
s = sources[ind]
# This flag tells this method if the current set of fits are part of an annular spectra or not
ann_fit = False
ann_results = {}
ann_lums = {}
ann_obs_order = {}
for res_set in results[src_repr]:
if len(res_set) != 0 and res_set[1] and run_type == "fit":
with FITS(res_set[0]) as res_table:
global_results = res_table["RESULTS"][0]
model = global_results["MODEL"].strip(" ")
# Just define this to check if this is an annular fit or not
first_key = res_table["SPEC_INFO"][0][
"SPEC_PATH"].strip(" ").split("/")[-1].split(
'ra')[-1]
first_key = first_key.split('_spec.fits')[0]
if "_ident" in first_key:
ann_fit = True
inst_lums = {}
obs_order = []
for line_ind, line in enumerate(
res_table["SPEC_INFO"]):
sp_info = line["SPEC_PATH"].strip(" ").split(
"/")[-1].split("_")
# Want to derive the spectra storage key from the file name, this strips off some
# unnecessary info
sp_key = line["SPEC_PATH"].strip(" ").split(
"/")[-1].split('ra')[-1].split('_spec.fits')[0]
# If its not an AnnularSpectra fit then we can just fetch the spectrum from the source
# the normal way
if not ann_fit:
# This adds ra back on, and removes any ident information if it is there
sp_key = 'ra' + sp_key
# Finds the appropriate matching spectrum object for the current table line
spec = s.get_products("spectrum",
sp_info[0],
sp_info[1],
extra_key=sp_key)[0]
else:
obs_order.append([sp_info[0], sp_info[1]])
ann_id = int(
sp_key.split("_ident")[-1].split("_")[1])
sp_key = 'ra' + sp_key.split('_ident')[0]
first_part = sp_key.split('ri')[0]
second_part = "_" + "_".join(
sp_key.split('ro')[-1].split("_")[1:])
ann_sp_key = first_part + "ar" + "_".join(
radii[ind].value.astype(str)) + second_part
ann_specs = s.get_products(
"combined_spectrum", extra_key=ann_sp_key)
if len(ann_specs) > 1:
raise MultipleMatchError(
"I have found multiple matches for that AnnularSpectra, "
"this is the developers fault, not yours."
)
elif len(ann_specs) == 0:
raise NoMatchFoundError(
"Somehow I haven't found the AnnularSpectra that you "
"fitted, this is the developers fault, not yours"
)
else:
ann_spec = ann_specs[0]
spec = ann_spec.get_spectra(
ann_id, sp_info[0], sp_info[1])
# Adds information from this fit to the spectrum object.
spec.add_fit_data(
str(model), line,
res_table["PLOT" + str(line_ind + 1)])
# The add_fit_data method formats the luminosities nicely, so we grab them back out
# to help grab the luminosity needed to pass to the source object 'add_fit_data' method
processed_lums = spec.get_luminosities(model)
if spec.instrument not in inst_lums:
inst_lums[spec.instrument] = processed_lums
# Ideally the luminosity reported in the source object will be a PN lum, but its not impossible
# that a PN value won't be available. - it shouldn't matter much, lums across the cameras are
# consistent
if "pn" in inst_lums:
chosen_lums = inst_lums["pn"]
# mos2 generally better than mos1, as mos1 has CCD damage after a certain point in its life
elif "mos2" in inst_lums:
chosen_lums = inst_lums["mos2"]
else:
chosen_lums = inst_lums["mos1"]
if ann_fit:
ann_results[spec.annulus_ident] = global_results
ann_lums[spec.annulus_ident] = chosen_lums
ann_obs_order[spec.annulus_ident] = obs_order
elif not ann_fit:
# Push global fit results, luminosities etc. into the corresponding source object.
s.add_fit_data(model, global_results, chosen_lums,
sp_key)
elif len(res_set
) != 0 and res_set[1] and run_type == "conv_factors":
res_table = pd.read_csv(res_set[0],
dtype={
"lo_en": str,
"hi_en": str
})
# Gets the model name from the file name of the output results table
model = res_set[0].split("_")[-3]
# We can infer the storage key from the name of the results table, just makes it easier to
# grab the correct spectra
storage_key = res_set[0].split('/')[-1].split(
s.name)[-1][1:].split(model)[0][:-1]
# Grabs the ObsID+instrument combinations from the headers of the csv. Makes sure they are unique
# by going to a set (because there will be two columns for each ObsID+Instrument, rate and Lx)
# First two columns are skipped because they are energy limits
combos = list(
set([c.split("_")[1] for c in res_table.columns[2:]]))
# Getting the spectra for each column, then assigning rates and lums
for comb in combos:
spec = s.get_products("spectrum",
comb[:10],
comb[10:],
extra_key=storage_key)[0]
spec.add_conv_factors(
res_table["lo_en"].values,
res_table["hi_en"].values,
res_table["rate_{}".format(comb)].values,
res_table["Lx_{}".format(comb)].values, model)
elif len(res_set) != 0 and not res_set[1]:
for err in res_set[2]:
raise XSPECFitError(err)
if ann_fit:
# We fetch the annular spectra object that we just fitted, searching by using the set ID of
# the last spectra that was opened in the loop
ann_spec = s.get_annular_spectra(set_id=spec.set_ident)
try:
ann_spec.add_fit_data(model, ann_results, ann_lums,
ann_obs_order)
# The most likely reason for running XSPEC fits to a profile is to create a temp. profile
# so we check whether constant*tbabs*apec (single_temp_apec function)has been run and if so
# generate a Tx profile automatically
if model == "constant*tbabs*apec":
temp_prof = ann_spec.generate_profile(
model, 'kT', 'keV')
s.update_products(temp_prof)
# Normalisation profiles can be useful for many things, so we generate them too
norm_prof = ann_spec.generate_profile(
model, 'norm', 'cm^-5')
s.update_products(norm_prof)
if 'Abundanc' in ann_spec.get_results(
0, 'constant*tbabs*apec'):
met_prof = ann_spec.generate_profile(
model, 'Abundanc', '')
s.update_products(met_prof)
else:
raise NotImplementedError(
"How have you even managed to fit this model to a profile?! Its not"
" supported yet.")
except ValueError:
warnings.warn(
"{src} annular spectra profile fit was not successful".
format(src=ann_spec.src_name))
# If only one source was passed, turn it back into a source object rather than a source
# object in a list.
if len(sources) == 1:
sources = sources[0]
return sources
def gal_preprocess(zmin, zmax):
lens = fitsio.read("lens.fits",
columns=["ra_gal", "dec_gal", "observed_redshift_gal"])
ra, dec, z = lens["ra_gal"], lens["dec_gal"], lens["observed_redshift_gal"]
mask = (z > zmin) & (z < zmax)
ra, dec, z = ra[mask], dec[mask], z[mask]
print("length of the catalog after applying the cut", len(ra))
coord = np.vstack([ra, dec]).T
centers = np.loadtxt("flagship_jk_centers_v2.txt")
NJK = centers.shape[0]
print("Segmentation begins!")
gal_labels_jk = kmeans_radec.find_nearest(coord, centers)
print("Done with assigning jacknife labels to galaxies")
gals = {"RA": ra, "DEC": dec, "redshift": z, "JK_LABEL": gal_labels_jk}
fits = FITS('data/gal_zmax_' + str(zmax) + '.fits', 'rw')
fits.write(gals, names=["RA", "DEC", "redshift", "JK_LABEL"])
fits.close()
for jk in range(len(centers)):
gal_jk_mask = gals["JK_LABEL"] != jk
gal_jk = {
"RA": ra[gal_jk_mask],
"DEC": dec[gal_jk_mask],
"redshift": z[gal_jk_mask]
}
fits = FITS('data/gal_zmax_' + str(zmax) + '_jk_' + str(jk) + '.fits',
'rw')
fits.write(gal_jk, names=["RA", "DEC", "redshift"])
fits.close()
return None
