def get_foreground_model(fg_params, fg_model,
frequencies, ell,
requested_cls=["tt", "te", "ee"]):
normalisation = fg_model["normalisation"]
nu_0 = normalisation["nu_0"]
ell_0 = normalisation["ell_0"]
from fgspectra import cross as fgc
from fgspectra import power as fgp
from fgspectra import frequency as fgf
# We don't seem to be using this
# cirrus = fgc.FactorizedCrossSpectrum(fgf.PowerLaw(), fgp.PowerLaw())
ksz = fgc.FactorizedCrossSpectrum(fgf.ConstantSED(), fgp.kSZ_bat())
cibp = fgc.FactorizedCrossSpectrum(fgf.ModifiedBlackBody(), fgp.PowerLaw())
radio = fgc.FactorizedCrossSpectrum(fgf.PowerLaw(), fgp.PowerLaw())
tsz = fgc.FactorizedCrossSpectrum(fgf.ThermalSZ(), fgp.tSZ_150_bat())
cibc = fgc.FactorizedCrossSpectrum(fgf.CIB(), fgp.PowerLaw())
# Make sure to pass a numpy array to fgspectra
if not isinstance(frequencies, np.ndarray):
frequencies = np.array(frequencies)
model = {}
model["tt", "kSZ"] = fg_params["a_kSZ"] * ksz(
{"nu": frequencies},
{"ell": ell, "ell_0": ell_0})
model["tt", "cibp"] = fg_params["a_p"] * cibp(
{"nu": frequencies, "nu_0": nu_0,
"temp": fg_params["T_d"], "beta": fg_params["beta_p"]},
{"ell": ell, "ell_0": ell_0, "alpha": 2})
model["tt", "radio"] = fg_params["a_s"] * radio(
{"nu": frequencies, "nu_0": nu_0, "beta": -0.5 - 2},
{"ell": ell, "ell_0": ell_0, "alpha": 2})
model["tt", "tSZ"] = fg_params["a_tSZ"] * tsz(
{"nu": frequencies, "nu_0": nu_0},
{"ell": ell, "ell_0": ell_0})
model["tt", "cibc"] = fg_params["a_c"] * cibc(
{"nu": frequencies, "nu_0": nu_0,
"temp": fg_params["T_d"], "beta": fg_params["beta_c"]},
{"ell": ell, "ell_0": ell_0, "alpha": 2 - fg_params["n_CIBC"]})
components = fg_model["components"]
component_list = {s: components[s] for s in requested_cls}
fg_dict = {}
for c1, f1 in enumerate(frequencies):
for c2, f2 in enumerate(frequencies):
for s in requested_cls:
fg_dict[s, "all", f1, f2] = np.zeros(len(ell))
for comp in component_list[s]:
fg_dict[s, comp, f1, f2] = model[s, comp][c1, c2]
fg_dict[s, "all", f1, f2] += fg_dict[s, comp, f1, f2]
return fg_dict
def _init_foreground_model(self):
from fgspectra import cross as fgc
from fgspectra import frequency as fgf
from fgspectra import power as fgp
template_path = os.path.join(
os.path.dirname(os.path.abspath(fgp.__file__)), "data")
cibc_file = os.path.join(template_path, "cl_cib_Choi2020.dat")
# set pivot freq and multipole
self.fg_nu_0 = self.foregrounds["normalisation"]["nu_0"]
self.fg_ell_0 = self.foregrounds["normalisation"]["ell_0"]
# We don't seem to be using this
# cirrus = fgc.FactorizedCrossSpectrum(fgf.PowerLaw(), fgp.PowerLaw())
self.ksz = fgc.FactorizedCrossSpectrum(fgf.ConstantSED(),
fgp.kSZ_bat())
self.cibp = fgc.FactorizedCrossSpectrum(fgf.ModifiedBlackBody(),
fgp.PowerLaw())
self.radio = fgc.FactorizedCrossSpectrum(fgf.PowerLaw(),
fgp.PowerLaw())
self.tsz = fgc.FactorizedCrossSpectrum(fgf.ThermalSZ(),
fgp.tSZ_150_bat())
self.cibc = fgc.FactorizedCrossSpectrum(
fgf.CIB(), fgp.PowerSpectrumFromFile(cibc_file))
self.dust = fgc.FactorizedCrossSpectrum(fgf.ModifiedBlackBody(),
fgp.PowerLaw())
self.tSZ_and_CIB = fgc.SZxCIB_Choi2020()
components = self.foregrounds["components"]
self.fg_component_list = {s: components[s] for s in self.requested_cls}
def _get_foreground_model(self, fg_params):
# Might change given different lmax
lmin, lmax = 2, self.lmax
l = np.arange(lmin, lmax)
foregrounds = self.foregrounds
normalisation = foregrounds["normalisation"]
nu_0 = normalisation["nu_0"]
ell_0 = normalisation["ell_0"]
T_CMB = normalisation["T_CMB"]
all_freqs = np.concatenate(
[v for exp in self.experiments for k, v in exp.items()])
from fgspectra import cross as fgc
from fgspectra import power as fgp
from fgspectra import frequency as fgf
cirrus = fgc.FactorizedCrossSpectrum(fgf.PowerLaw(), fgp.PowerLaw())
ksz = fgc.FactorizedCrossSpectrum(fgf.ConstantSED(), fgp.kSZ_bat())
cibp = fgc.FactorizedCrossSpectrum(fgf.ModifiedBlackBody(),
fgp.PowerLaw())
radio = fgc.FactorizedCrossSpectrum(fgf.PowerLaw(), fgp.PowerLaw())
tsz = fgc.FactorizedCrossSpectrum(fgf.ThermalSZ(), fgp.tSZ_150_bat())
cibc = fgc.FactorizedCrossSpectrum(fgf.CIB(), fgp.PowerLaw())
model = {}
model["tt", "kSZ"] = fg_params["a_kSZ"] * ksz({"nu": all_freqs}, {
"ell": l,
"ell_0": ell_0
})
model["tt", "cibp"] = fg_params["a_p"] * cibp(
{
"nu": all_freqs,
"nu_0": nu_0,
"temp": fg_params["T_d"],
"beta": fg_params["beta_p"]
}, {
"ell": l,
"ell_0": ell_0,
"alpha": 2
})
model["tt", "radio"] = fg_params["a_s"] * radio(
{
"nu": all_freqs,
"nu_0": nu_0,
"beta": -0.5 - 2
}, {
"ell": l,
"ell_0": ell_0,
"alpha": 2
})
model["tt",
"tSZ"] = fg_params["a_tSZ"] * tsz({
"nu": all_freqs,
"nu_0": nu_0
}, {
"ell": l,
"ell_0": ell_0
})
model["tt", "cibc"] = fg_params["a_c"] * cibc(
{
"nu": all_freqs,
"nu_0": nu_0,
"temp": fg_params["T_d"],
"beta": fg_params["beta_c"]
}, {
"ell": l,
"ell_0": ell_0,
"alpha": 2 - fg_params["n_CIBC"]
})
components = foregrounds["components"]
component_list = {s: components[s] for s in self.requested_cls}
fg_model = {}
for c1, f1 in enumerate(all_freqs):
for c2, f2 in enumerate(all_freqs):
for s in self.requested_cls:
fg_model[s, "all", f1, f2] = np.zeros(len(l))
for comp in component_list[s]:
fg_model[s, comp, f1, f2] = model[s, comp][c1, c2]
fg_model[s, "all", f1, f2] += fg_model[s, comp, f1, f2]
return fg_model
def test_ACT_models():
# define the models from fgspectra
ksz = fgc.FactorizedCrossSpectrum(fgf.ConstantSED(), fgp.kSZ_bat())
cibp = fgc.FactorizedCrossSpectrum(fgf.ModifiedBlackBody(), fgp.PowerLaw())
radio = fgc.FactorizedCrossSpectrum(fgf.PowerLaw(), fgp.PowerLaw())
cirrus = fgc.FactorizedCrossSpectrum(fgf.PowerLaw(), fgp.PowerLaw())
# if there are correlations between components,
# have to define them in a joined spectrum
tSZ_and_CIB = fgc.CorrelatedFactorizedCrossSpectrum(
fgf.Join(fgf.ThermalSZ(), fgf.CIB()), fgp.SZxCIB_Addison2012())
# for testing purposes we'll also compute the tSZ and clustered CIB alone
tsz = fgc.FactorizedCrossSpectrum(fgf.ThermalSZ(), fgp.tSZ_150_bat())
cibc = fgc.FactorizedCrossSpectrum(fgf.CIB(), fgp.PowerLaw())
ells = np.array([2000])
par = {
'nu_0': 150.0,
'ell_0': 3000,
'T_CMB': 2.725,
'T_d': 9.7,
'a_tSZ': 4.66,
'a_kSZ': 1.60,
'a_p': 6.87,
'beta_p': 2.08,
'a_c': 6.10,
'beta_c': 2.08,
'n_CIBC': 1.20,
'xi': 0.09,
'a_s': 3.50,
'a_g': 0.88,
'f0_sz': 146.9,
'f0_synch': 147.6,
'f0_dust': 149.7
}
fsz = np.array([par['f0_sz']])
fsynch = np.array([par['f0_synch']])
fdust = np.array([par['f0_dust']])
result = (par['a_tSZ'] * tsz({
'nu': fsz,
'nu_0': par['nu_0']
}, {
'ell': ells,
'ell_0': par['ell_0']
}), par['a_kSZ'] * ksz({'nu': fsz}, {
'ell': ells,
'ell_0': par['ell_0']
}), par['a_p'] * cibp(
{
'nu': fdust,
'nu_0': par['nu_0'],
'temp': par['T_d'],
'beta': par['beta_p']
}, {
'ell': ells,
'ell_0': par['ell_0'],
'alpha': 2
}), par['a_c'] * cibc(
{
'nu': fdust,
'nu_0': par['nu_0'],
'temp': par['T_d'],
'beta': par['beta_c']
}, {
'ell': ells,
'ell_0': par['ell_0'],
'alpha': 2 - par['n_CIBC']
}),
tSZ_and_CIB(
{
'kwseq': ({
'nu': fsz,
'nu_0': par['nu_0']
}, {
'nu': fdust,
'nu_0': par['nu_0'],
'temp': par['T_d'],
'beta': par['beta_c']
})
}, {
'kwseq': ({
'ell': ells,
'ell_0': par['ell_0'],
'amp': par['a_tSZ']
}, {
'ell': ells,
'ell_0': par['ell_0'],
'alpha': 2 - par['n_CIBC'],
'amp': par['a_c']
}, {
'ell':
ells,
'ell_0':
par['ell_0'],
'amp':
-par['xi'] * np.sqrt(par['a_tSZ'] * par['a_c'])
})
}), par['a_s'] *
radio({
'nu': fsynch,
'nu_0': par['nu_0'],
'beta': -0.5 - 2
}, {
'ell': ells,
'ell_0': par['ell_0'],
'alpha': 2
}), par['a_g'] *
cirrus({
'nu': fdust,
'nu_0': par['nu_0'],
'beta': 3.8 - 2
}, {
'ell': ells,
'ell_0': par['ell_0'],
'alpha': -0.7
}))
"""
The following array `dunk13` was generated by inserting this code block to
line 188 of `ACT_equa_likelihood.f90` in the Fortran ACT multifrequency
likelihood.
```fortran
if(il==2000) then
write(*,*) (f1*f1)/(f0*f0)*amp_tsz*cl_tsz(il)
write(*,*) amp_ksz*cl_ksz(il)
write(*,*) amp_d*cl_p(il)*(f1_dust/feff)**(2.d0*beta_d)*(planckratiod1*fluxtempd1)**2.d0
write(*,*) amp_c*cl_c(il)*(f1_dust/feff)**(2.d0*beta_c)*(planckratiod1*fluxtempd1)**2.d0
write(*,*) -2.d0*sqrt(amp_c*amp_tsz)*xi*cl_szcib(il)*((f1_dust**beta_c*f1*planckratiod1*fluxtempd1)/(feff**beta_c*f0))
write(*,*) amp_s*cl_p(il)*(f1_synch/feff)**(2.d0*alpha_s)*fluxtemps1**2.d0
write(*,*) amp_ge*cl_cir(il)*(f1_dust**2.d0/feff**2.d0)**beta_g*fluxtempd1**2.d0
endif
```
"""
dunk13 = np.array([
4.46395567509345, # tsz
1.37903889069153, # ksz
3.02039291213271, # cibp
4.36478789541715, # cibc
-0.611930900267790, # cibxtsz
1.63099887443075, # sync
1.15557760337592
]) # cirrus
# for testing purposes, subtract tsz and cibc alone
# from the combined tsz,cibc,tszxcibc.
result = np.array([r[0][0][0] for r in result])
result[-3] -= result[0]
result[-3] -= result[3]
tsz_fg, ksz_fg, cibp_fg, cibc_fg, cibxtxz_fg, sync_fg, cirrus_fg = result
assert (np.abs(tsz_fg - dunk13[0]) / dunk13[0] < 1e-3)
assert (np.abs(ksz_fg - dunk13[1]) / dunk13[1] < 1e-3)
assert (np.abs(cibp_fg - dunk13[2]) / dunk13[2] < 1e-3)
assert (np.abs(cibc_fg - dunk13[3]) / dunk13[3] < 1e-3)
assert (np.abs(cibxtxz_fg - dunk13[4]) / dunk13[4] < 1e-3)
assert (np.abs(sync_fg - dunk13[5]) / dunk13[5] < 1e-3)
assert (np.abs(cirrus_fg - dunk13[6]) / dunk13[6] < 1e-3)