def load_binned_ty(mb,
dtype='dr2_smica',
fltr='cinv',
cmask='Lmask',
mtype=0,
ytype='nilc',
bhe=['lens']):
# load filenames
d = prjlib.data_directory()
p = prjlib.init_analysis(dtype=dtype, fltr=fltr, wtype=cmask)
cy = init_compy(p.ids, masktype=mtype, ytype=ytype)
qobj = quad_func.quad(stag=p.stag,
root=d['root'],
ids=p.ids,
qtype='tau',
bhe=bhe,
rlmin=100,
rlmax=2048)
ftxy = init_cross(qobj, cy, p.ids, p.stag)
# optimal binning
yy = savgol_filter((np.loadtxt(cy.fclyy)).T[1], 51, 3)
al = (np.loadtxt(qobj.f['TT'].al)).T[1]
vl = np.sqrt(al * yy) / np.sqrt(qobj.l + 1e-30)
# results
return bn.binned_spec(mb, ftxy.xl, cn=1, opt=True, vl=vl)
def filename(self, ids):
d = prjlib.data_directory()
#//// public data ////#
# original mask file for ymap
self.fmask_org = plf.subd['pr2'][
'ysz'] + 'COM_CompMap_Compton-SZMap-masks_2048_R2.01.fits'
# ymap
self.fymap = plf.subd['pr2']['ysz'] + self.ytype + '_ymaps.fits'
#//// reduced data ////#
# reduced mask (multiplied ptsr mask and costheta mask)
self.fmask = d['win'] + 'ymask.fits'
self.famask = d['win'] + 'ymask_a' + str(self.ascale) + 'deg.fits'
if self.ascale == 0.0: self.famask = self.fmask
# yalm
ttag = ''
if self.tausig: ttag = '_tausig'
self.fyalm = [
d['ysz'] + 'alm/' + self.ytype + '_' + self.ymask + ttag + '_' +
x + '.pkl' for x in ids
]
# real clyy
self.fclyy = d['ysz'] + 'aps/' + self.ytype + '_' + self.ymask + '.dat'
def fname(self, qtag, mlist, etag, doreal):
#set directory
d = prjlib.data_directory()
ids = prjlib.rlz_index(doreal=doreal)
# delensing internal tag
ltag = 'le' + str(self.elmin) + '-' + str(self.elmax) + '_lk' + str(
self.klmin) + '-' + str(self.klmax)
ttag = ltag + '_' + self.kfltr + '_' + qtag + '_' + '-'.join(
mlist.keys()) + '_' + etag
#alm of lensing template B-modes
self.falm, self.fwlm, self.cl = {}, {}, {}
for k in self.klist:
self.falm[k] = [
d['del'] + 'alm/alm_' + k + '_' + ttag + '_' + x + '.pkl'
for x in ids
]
self.cl[k] = [
d['del'] + 'aps/rlz/cl_' + k + '_' + ttag + '_' + x + '.dat'
for x in ids
]
self.gtag = '_ideal'
# correlation coeff of templates
self.frho = d['del'] + 'aps/rho_' + '-'.join(self.klist) + '_' + ttag
def init_compy(ids, **kwargs):
d = prjlib.data_directory()
cy = compy(**kwargs)
compy.filename(cy, ids)
return cy
def __init__(self, qtag, ytag, ids):
d = prjlib.data_directory()
xaps = d['xcor'] + '/aps/'
xtag = qtag + '_' + ytag
self.xl = [xaps + '/rlz/xl_' + xtag + '_' + x + '.dat' for x in ids]
self.xcov = xaps + '/xcov_' + xtag + '.dat'
def init_qobj(stag, doreal, **kwargs):
# setup parameters for lensing reconstruction (see cmblensplus/utils/quad_func.py)
d = prjlib.data_directory()
ids = prjlib.rlz_index(doreal=doreal)
qobj = quad_func.reconstruction(d['root'],
ids,
stag=stag,
run=[],
**kwargs)
return qobj
def __init__( self, glob, qobj, lmin=8, lmax=2007, add_cmb=['TT'], add_gal=np.arange(6), add_cib=True ):
#lmax = 2007 #For now, set by the lmax in Byenoghee's spectra
#lmin = 8 #For now, set by the lmin in Byenoghee's spectra
# construct list of mass tracers to be combined
self.klist_cmb = {}
self.klist_gal = {}
self.klist_cib = {}
kid = 0
for k in add_cmb:
self.klist_cmb[k] = kid
kid += 1
for z in add_gal:
self.klist_gal['g'+str(z)] = kid
kid += 1
if add_cib:
self.klist_cib['cib'] = kid
self.klist = { **self.klist_cmb, **self.klist_gal, **self.klist_cib }
self.klist_ext = { **self.klist_gal, **self.klist_cib }
print(self.klist)
self.lmin = lmin
self.lmax = lmax
self.nkap = len(self.klist)
self.nlkk = {}
for k, n in self.klist_cmb.items():
self.nlkk[n] = np.loadtxt( qobj.f[k].al, unpack=True )[1][:lmax+1]
#set directory
d = prjlib.data_directory()
ids = prjlib.rlz_index( doreal=glob.doreal )
# cls
self.fspec = '/global/project/projectdirs/sobs/delensing/multitracer_forBBgroup/spectra_of_tracers/'
# kappa alm of each mass tracer
self.fklm = {}
for k in self.klist:
self.fklm[k] = [ d['del'] + 'mass/' + k + '_' + str(i) + '.pkl' for i in ids ]
# kappa alm of combined mass tracer
qtag = glob.stag + qobj.ltag
self.fcklm = [ d['del'] + 'mass/comb_' + qtag + '_' + '-'.join(self.klist.keys()) + '_' + str(i) + '.pkl' for i in ids ]
def load_binned_tt(mb,
dtype='dr2_smica',
fltr='cinv',
cmask='Lmask',
bhe=['lens'],
snmax=100):
# filename
d = prjlib.data_directory()
p = prjlib.init_analysis(snmax=snmax, dtype=dtype, fltr=fltr, wtype=cmask)
qobj = quad_func.quad(stag=p.stag,
root=d['root'],
ids=p.ids,
qtype='tau',
bhe=bhe,
rlmin=100,
rlmax=2048)
# optimal filter
al = (np.loadtxt(qobj.f['TT'].al)).T[1]
vl = al / np.sqrt(qobj.l + 1e-30)
# binned spectra
mtt, __, stt, ott = bn.binned_spec(mb,
qobj.f['TT'].cl,
cn=1,
doreal=True,
opt=True,
vl=vl)
# noise bias
nb = bn.binning((np.loadtxt(qobj.f['TT'].n0bs)).T[1], mb, vl=vl)
rd = np.array([(np.loadtxt(qobj.f['TT'].rdn0[i])).T[1] for i in p.rlz])
rb = bn.binning(rd, mb, vl=vl)
# debias
ott = ott - rb[0] - nb / (qobj.mfsim)
stt = stt - rb[1:, :] - nb / (qobj.mfsim - 1)
# sim mean and std
mtt = mtt - np.mean(rb[1:, :], axis=0) - nb / (qobj.mfsim - 1)
vtt = np.std(stt, axis=0)
# subtract average of sim
ott = ott - mtt
return mtt, vtt, stt, ott
def init_quad(ids, stag, rlz=[], **kwargs):
d = prjlib.data_directory()
# setup parameters for lensing reconstruction (see cmblensplus/utils/quad_func.py)
qtau = quad_func.quad(rlz=rlz,
stag=stag,
root=d['root'],
ids=ids,
qtype='tau',
**kwargs)
qlen = quad_func.quad(rlz=rlz,
stag=stag,
root=d['root'],
ids=ids,
qtype='lens',
**kwargs)
qsrc = quad_func.quad(rlz=rlz,
stag=stag,
root=d['root'],
ids=ids,
qtype='src',
**kwargs)
qtbh = quad_func.quad(rlz=rlz,
stag=stag,
root=d['root'],
ids=ids,
qtype='tau',
bhe=['lens'],
**kwargs)
qtBH = quad_func.quad(rlz=rlz,
stag=stag,
root=d['root'],
ids=ids,
qtype='tau',
bhe=['lens', 'src'],
**kwargs)
#qtau.fname(d['root'],ids,stag)
#qlen.fname(d['root'],ids,stag)
#qsrc.fname(d['root'],ids,stag)
#qtbh.fname(d['root'],ids,stag)
#qtBH.fname(d['root'],ids,stag)
return qtau, qlen, qsrc, qtbh, qtBH
def __init__(self,doreal=False,telescope='la',ntype='base',freq='145',snmin=0,snmax=100,overwrite=False,verbose=True):
self.telescope = str.upper(telescope)
self.ntype = ntype
self.band = int(freq)
self.doreal = doreal
self.rlz = np.linspace(snmin,snmax,snmax-snmin+1,dtype=np.int)
self.overwrite = overwrite
self.verbose = verbose
if 'base' in ntype:
self.mode = 'baseline'
elif 'goal' in ntype:
self.mode = 'goal'
elif ntype == '':
self.mode = 'signal'
print('signal calculation')
else:
sys.exit('unknown noise level')
if 'roll' in ntype:
self.roll = int(ntype[ntype.find('roll')+4:])
else:
self.roll = 0
self.nside, self.npix = prjlib.mapres(telescope)
# set directory
d = prjlib.data_directory()
d_map = d['cmb'] + 'map/'
# map filename
ids = prjlib.rlz_index(doreal=doreal)
if ntype == '':
#cmb signal map
if telescope == 'id': # use LAT signal sim
self.fmap = [d_map+'/cmb_uKCMB_la145_nside'+str(self.nside)+'_'+x+'.fits' for x in ids]
else:
self.fmap = [d_map+'/cmb_uKCMB_'+telescope+freq+'_nside'+str(self.nside)+'_'+x+'.fits' for x in ids]
else:
#cmb noise map
self.fmap = [d_map+'/noise_uKCMB_'+telescope+freq+'_'+ntype+'_nside'+str(self.nside)+'_'+x+'.fits' for x in ids]
def interface(run=[], kwargs_ov={}, kwargs_cmb={}, kwargs_qrec={}, ep=1e-30):
if kwargs_cmb['t'] != 'la':
sys.exit('only la is supported')
# Define parameters, filenames for input CMB
p = prjlib.analysis_init(**kwargs_cmb)
# Load pre-computed w-factor which is used for correction normalization of spectrum
wn = prjlib.wfac(p.telescope)
# Compute filtering
if p.fltr == 'none': # for none-filtered alm
# Load "observed" aps containing signal, noise, and some residual.
# This aps will be used for normalization calculation
ocl = prjlib.loadocl(p.fcmb.scl['o'], lTmin=p.lTmin, lTmax=p.lTmax)
# CMB alm will be multiplied by 1/ifl before reconstruction process
ifl = ocl #p.lcl[0:3,:]
elif p.fltr == 'cinv': # for C^-1 wiener-filtered alm
pc = prjlib.analysis_init(t=p.telescope,
freq='com',
fltr='none',
ntype=p.ntype)
# Compute aps appropriate for C^-1 filtering case.
ocl, ifl = quad_filter(p.fcmb.scl,
pc.fcmb.scl,
p.lmax,
p.lcl,
lTmin=p.lTmin,
lTmax=p.lTmax)
ocl[ocl <= 0.] = 1e30
ifl[ifl <= 0.] = 1e30
wn[:] = wn[0]
else:
sys.exit('unknown filtering')
if 'iso' in p.ntype: #fullsky case
wn[:] = 1.
d = prjlib.data_directory()
ids = prjlib.rlz_index(doreal=p.doreal)
qobj = quad_func.reconstruction(d['root'],
ids,
rlz=p.rlz,
stag=p.stag,
run=run,
wn=wn,
lcl=p.lcl,
ocl=ocl,
ifl=ifl,
falm=p.fcmb.alms['o'],
**kwargs_ov,
**kwargs_qrec)
# Aps of reconstructed phi
if 'aps' in run:
aps(p.fltr, qobj, p.rlz, p.fpalm, wn)
# Cinv kappa
if 'kcinv' in run:
compute_kcninv(qobj.f,
p.rlz,
p.fltr,
p.kk[:qobj.olmax + 1],
p.fpalm,
Snmin=p.snmin,
Snmax=p.snmax,
qlist=qobj.qlist,
**kwargs_ov)