def convolve_healpix(input_map, func=None, sigma=None, thetamax=10 ):
"""
Convolve a HEALPix map with a function, or Gaussian
input_map : array of float
a HEALPix array, RING indexing, nside a power of 2
func : The function of an integer el | None
returns the amplitude for spherical harmonic el
example: for a Gaussian with sigma in radians:
lambda el : np.exp(-0.5 * (el * (el + 1)) * sigma**2)
sigma : None | float (deg)
If not None, use gaussian for func
Returns: the convolved map
"""
import healpy
nside = int(np.sqrt(len(input_map)/12))
assert 12*nside**2 == len(input_map),'Bad length: expect power of 2'
if func is None:
assert sigma is not None, 'If no func, must specify sigma'
func= lambda el : np.exp(-0.5 * (el * (el + 1)) * np.radians(sigma)**2)
else:
assert func(thetamax)/func(0) <1e-3
alm = healpy.map2alm(input_map);
lmax = healpy.Alm.getlmax(len(alm))
if lmax < 0:
raise TypeError('Wrong alm size for the given '
'mmax (len(alms[%d]) = %d).'%(ialm, len(alm)))
ell = np.arange(lmax + 1.)
fact = np.array([func(x) for x in ell])
healpy.almxfl(alm, fact, inplace=True)
return healpy.alm2map(alm, nside=nside, verbose=False)
def run_MCMC_new(which_par,niter,save_title, renorm_var):
"""
Functions that runs the MCMC for a given set of parameters
Keyword Arguments:
which_par -- a list of indices, corresponding to the order defined above, exemple [0,2] means ombh2,tau if order is [ombh2,omch2,tau,As,ns,H0]
niter -- number of iterations in MCMC
renorm_var -- factor multipling the variance, to play around for better acceptance rate.
"""
cov_new_temp = cov_new[which_par,:][:,which_par] * renorm_var
string_temp = strings[which_par]
titles_temp = titles[which_par]
x_mean_temp = x_mean[which_par]
priors_central_temp = priors_central[which_par]
priors_invvar_temp = priors_invvar[which_par]
print titles_temp
# generate first guess parameters
guess_param = PS2P.prop_dist_form_params(x_mean_temp,cov_new_temp)
print "initial guess = ", guess_param
# generate first fluctuation map
dd2 = cb.update_dic(dd,guess_param,string_temp)
cl = cb.generate_spectrum(dd2)[:,1]
cl[:2] = 1.e-35
renorm = CG.renorm_term(cl,bl,nl)
fluc = hp.almxfl(CG.generate_w1term(cl[:lmax+1],bl[:lmax+1],nl[:lmax+1]) + CG.generate_w0term(cl[:lmax+1]),renorm)
mf = hp.almxfl(CG.generate_mfterm(dlm,cl[:lmax+1],bl[:lmax+1],nl[:lmax+1]),renorm)
# the core of the MCMC
testss = np.array(MH.MCMC_log_Jeff_new(guess_param, JJi.target_new,PS2P.prop_dist_form_params, PS2P.prop_func_form_params,niter,PS2P.Gaussian_priors_func,[[dlm,string_temp,dd,nl[:lmax+1],bl[:lmax+1]],[cl[:lmax+1],fluc,mf]],[x_mean_temp*0,np.matrix(cov_new_temp)],[priors_central_temp,priors_invvar_temp]))
np.save("chain_%s_%s_%d_%d.npy"%(save_title,str(which_par).replace(',','').replace('[','').replace(']','').replace(' ',''),np.random.randint(0,100000),niter),testss)
return testss
def _icov_diag(self, alm, lens_power=False):
'''
Return (in-place) inverse covariance weighted version if input.
Assuming that covariance is diagonal in multipole.
Parameters
----------
alm : (npol, nelem) array
Healpix-ordered alm array to be inverse weighted. Order: T, E.
lens_power : bool, optional
Include lensing power in invserse covariance.
Returns
-------
c_inv_a : (npol, nelem) complex array
Inverse covariance weighted input array.
Raises
------
ValueError
If shape input alm does not match npol or exceeds lmax.
'''
if alm.ndim == 1:
input_1d = True
alm = np.atleast_2d(alm)
else:
input_1d = False
if alm.shape[0] != self.npol:
raise ValueError(
'Shape alm: {} does not match with length of pol: {}.'
.format(alm.shape, self.npol))
lmax_alm = hp.Alm.getlmax(alm.shape[-1])
if lmax_alm > self.lmax:
raise ValueError('lmax alm exceeds lmax icov ({} > {})'.
format(lmax_alm, self.lmax))
if lens_power:
icov = self.icov_ell_lensed
else:
icov = self.icov_ell_nonlensed
if self.pol == ('T', 'E'):
# Would be nice get rid of this copy with C or cython.
tmp = alm.copy()
hp.almxfl(tmp[0], icov[2,:lmax_alm+1], inplace=True)
hp.almxfl(tmp[1], icov[2,:lmax_alm+1], inplace=True)
hp.almxfl(alm[0], icov[0,:lmax_alm+1], inplace=True)
hp.almxfl(alm[1], icov[1,:lmax_alm+1], inplace=True)
alm[1] += tmp[0]
alm[0] += tmp[1]
else:
hp.almxfl(alm[0], icov[0,:lmax_alm+1], inplace=True)
if input_1d:
return alm[0]
else:
return alm
def _get_alms(data, beams=None, lmax=None, weights=None, iter=3):
alms = []
for f, fdata in enumerate(data):
if weights is None:
alms.append(hp.map2alm(fdata, lmax=lmax, iter=iter))
else:
alms.append(
hp.map2alm(hp.ma(fdata) * weights, lmax=lmax, iter=iter))
logging.info(f"{f+1} of {len(data)} complete")
alms = np.array(alms)
if beams is not None:
logging.info('Correcting alms for the beams')
for fwhm, alm in zip(beams, alms):
bl = hp.gauss_beam(np.radians(fwhm / 60.0),
lmax,
pol=(alm.ndim == 2))
if alm.ndim == 1:
alm = [alm]
bl = [bl]
for i_alm, i_bl in zip(alm, bl.T):
hp.almxfl(i_alm, 1.0 / i_bl, inplace=True)
return alms
def calc_prep(m, s_cls, n_inv_filt):
"""Missing doc."""
kmap = np.copy(m)
n_inv_filt.apply_map(kmap)
alm = map2alm(kmap, lmax=len(n_inv_filt.b_transf) - 1, iter=0)
hp.almxfl(alm, n_inv_filt.b_transf * (len(m) / (4. * np.pi)), inplace=True)
return alm
def get_gpmap(self, idx, spin, k=None, xfilt=None):
"""
\sum_{lm} (Elm +- iBlm) sqrt(l+2 (l-1)) _1 Ylm(n).
sqrt(l-2 (l+3)) _3 Ylm(n).
Output is list with real and imaginary part of the spin 1 or 3 transforms.
"""
assert spin in [1, 3]
assert xfilt is None, 'not implemented'
Glm = self.ivfs.get_sim_emliklm(idx)
Clm = self.ivfs.get_sim_bmliklm(idx)
assert Glm.size == Clm.size, (Clm.size, Clm.size)
lmax = hp.Alm.getlmax(Glm.size)
if spin == 1:
fl = np.arange(2, lmax + 3, dtype=float) * (np.arange(-1, lmax))
elif spin == 3:
fl = np.arange(-2, lmax - 1, dtype=float) * (np.arange(
3, lmax + 4))
else:
assert 0
fl[:spin] *= 0.
fl = np.sqrt(fl)
hp.almxfl(Glm, fl, inplace=True)
hp.almxfl(Clm, fl, inplace=True)
return hp.alm2map_spin([Glm, Clm], self.nside, spin, lmax)
def test_t():
lmax = 200
nside = 256
cls_unl = np.ones(lmax + 1, dtype=float)
tunl = hp.synalm(cls_unl, new=True)
dlm = np.zeros_like(tunl)
hp.almxfl(dlm,
np.sqrt(
np.arange(lmax + 1) * np.arange(1, lmax + 2, dtype=float)),
inplace=True)
T = hp.alm2map(tunl, nside)
T1 = lensing.alm2lenmap(tunl, [dlm, None],
nside,
verbose=True,
nband=8,
facres=-2)
assert np.max(np.abs(T - T1)) / np.std(T) < 1e-5
d1Re, d1Im = hp.alm2map_spin([dlm, np.zeros_like(dlm)], nside, 1, lmax)
T2 = lensing.alm2lenmap(tunl, [d1Re, d1Im],
nside,
verbose=True,
nband=8,
facres=-2)
assert np.max(np.abs(T - T2)) / np.std(T) < 1e-5
T3 = lensing.alm2lenmap(tunl, [dlm, dlm.copy()],
nside,
verbose=True,
nband=8,
facres=-2)
assert np.all(T2 == T3)
def draw_realization(self, synalm_lmax=None, seeds=None):
if seeds is None:
seeds = (None, None)
if synalm_lmax is None:
synalm_lmax = min(16384, 3 * self.nside - 1)
np.random.seed(seeds[0])
alm_small_scale = hp.synalm(
list(self.small_scale_cl.value) +
[np.zeros_like(self.small_scale_cl[0])] * 3,
lmax=synalm_lmax,
new=True,
)
alm_small_scale = [
hp.almxfl(each, np.ones(min(synalm_lmax, 3 * self.nside - 1)))
for each in alm_small_scale
]
map_small_scale = hp.alm2map(alm_small_scale, nside=self.nside)
# need later for beta
modulate_map_I = hp.alm2map(self.modulate_alm[0].value, self.nside)
map_small_scale[0] *= modulate_map_I
map_small_scale[1:] *= hp.alm2map(self.modulate_alm[1].value,
self.nside)
map_small_scale += hp.alm2map(
self.template_largescale_alm.value,
nside=self.nside,
)
output_IQU = (utils.log_pol_tens_to_map(map_small_scale) *
self.template_largescale_alm.unit)
np.random.seed(seeds[1])
output_unit = np.sqrt(1 * self.small_scale_cl_pl_index.unit).unit
alm_small_scale = hp.synalm(
self.small_scale_cl_pl_index.value,
lmax=synalm_lmax,
new=True,
)
alm_small_scale = hp.almxfl(
alm_small_scale, np.ones(min(3 * self.nside - 1, synalm_lmax + 1)))
pl_index = hp.alm2map(alm_small_scale, nside=self.nside) * output_unit
pl_index *= modulate_map_I
pl_index += (hp.alm2map(
self.largescale_alm_pl_index.value,
nside=self.nside,
) * output_unit)
pl_index -= 3.1 * u.dimensionless_unscaled
# Fixed values for comparison with s4
# pl_index = -3.1 * u.dimensionless_unscaled
return (output_IQU[0], output_IQU[1], output_IQU[2], pl_index)
def Step_MC(guess, *arg):
"""
Keyword Arguments:
x -- the vector of params
*args are:
arg[0] -- a target class object
arg[1] -- Cl_old, fluc_lm_old
"""
##print guess, arg
dlm,strings,params,nl,bl = arg[0]
Cl_old, fluc_lm_GS,mf_old = arg[1]
dd = cb.update_dic(params,guess,strings)
#generate new spectrum from new params
lmax = len(Cl_old)
Cl_new = cb.generate_spectrum(dd)[:lmax,1]
# avoid dividing by 0
Cl_new[:2] = 1.e-35
#print "new = ",Cl_new[50]
#print "old = ",Cl_old[50]
# renormalization, i.e. lhs part of eq (24) and (25)
renorm = CG.renorm_term(Cl_new,bl,nl)
# generate mean field map using new PS
mf_lm_new = hp.almxfl(CG.generate_mfterm(dlm,Cl_new,bl,nl),renorm)
# get deterministic fluctuation map (new rescaling with noise)
fluc_lm_determ = hp.almxfl(fluc_lm_GS,np.sqrt((1./Cl_old))/np.sqrt((1./Cl_new)))
# get GS fluctuation map "for next iteration"
fluc_lm_GS_next = hp.almxfl(CG.generate_w1term(Cl_new,bl,nl)+CG.generate_w0term(Cl_new),renorm)
# Chi2 part of the likelihood
return fluc_lm_determ,mf_lm_new, fluc_lm_GS_next, Cl_new
def test_fluctuation_term(guess, *arg):
"""
Keyword Arguments:
x -- the vector of params
*args are:
arg[0] -- a target class object
arg[1] -- Cl_old, fluc_lm_old
"""
##print guess, arg
dlm,strings,params,nl,bl = arg[0]
Cl_old, fluc_lm_old = arg[1]
dd = cb.update_dic(params,guess,strings)
#generate new spectrum from new params
lmax = len(Cl_old)
Cl_new = cb.generate_spectrum(dd)[:lmax,1]
# avoid dividing by 0
Cl_new[:2] = 1.e-35
#print "new = ",Cl_new[50]
#print "old = ",Cl_old[50]
# renormalization, i.e. lhs part of eq (24) and (25)
renorm = CG.renorm_term(Cl_new,bl,nl)
# generate fluctuation map using new PS, this is actually the 'step 2', with acceptance 1, won't be used anymore in this function
fluc_lm_type2 = hp.almxfl(CG.generate_w1term(Cl_new,bl,nl)+CG.generate_w0term(Cl_new),renorm)
#print "new = ",fluc_lm_type2[50]
#print "old = ",fluc_lm_old[50]
# get deterministic fluctuation map (new rescaling with noise)
fluc_lm_determ = hp.almxfl(fluc_lm_old,np.sqrt((1./Cl_old+bl**2/nl))/np.sqrt((1./Cl_new+bl**2/nl)))
# "fluctuation part" of the likelihood
tt3 = -1/2. *np.real(np.vdot((fluc_lm_determ).T,hp.almxfl((fluc_lm_determ),1/nl*bl**2)))
#print tt3
# we return Cl_new and fluc_lm_type2 for next iteration.
return tt3,Cl_new,fluc_lm_type2,fluc_lm_determ
def _get_sim_alm(self, idx, idf):
# FIXME : triangularise this
ret = hp.almxfl(self.lib_pha.get_sim(idx, idf=0), self.rmat[:, idf, 0])
for _i in range(1, len(self.fields)):
ret += hp.almxfl(self.lib_pha.get_sim(idx, idf=_i),
self.rmat[:, idf, _i])
return ret
def diff_fixed_like(fluc_lm_determ,fluc_lm_GS,mf_lm_new,mf_lm_old,Cl_new,Cl_old, *arg):
"""
Keyword Arguments:
x -- the vector of params
*args are:
arg[0] -- a target class object
arg[1] -- Cl_old, fluc_lm_old
"""
##print guess, arg
dlm,strings,params,nl,bl = arg[0]
tt1_new = -1/2.*np.real(np.vdot(CG.complex2real_alm(dlm-hp.almxfl(mf_lm_new,bl).T),CG.complex2real_alm(hp.almxfl((dlm-hp.almxfl(mf_lm_new,bl)),1/nl))))
tt1_old = -1/2.*np.real(np.vdot(CG.complex2real_alm(dlm-hp.almxfl(mf_lm_old,bl).T),CG.complex2real_alm(hp.almxfl((dlm-hp.almxfl(mf_lm_old,bl)),1/nl))))
#print "chi**2 = ",tt1_new-tt1_old
# "mean field part" of the likelihood
tt2_new = -1/2. *np.real(np.vdot(CG.complex2real_alm(mf_lm_new.T),CG.complex2real_alm(hp.almxfl((mf_lm_new),1./Cl_new))))
tt2_old = -1/2. *np.real(np.vdot(CG.complex2real_alm(mf_lm_old.T),CG.complex2real_alm(hp.almxfl((mf_lm_old),1./Cl_old))))
#print "'mf like' = ",tt2_new-tt2_old
# "fluctuation part" of the likelihood
tt3_new = -1/2. *np.real(np.vdot(CG.complex2real_alm(fluc_lm_determ.T),CG.complex2real_alm(hp.almxfl((fluc_lm_determ),1/nl*bl**2))))
tt3_old = -1/2. *np.real(np.vdot(CG.complex2real_alm(fluc_lm_GS.T),CG.complex2real_alm(hp.almxfl((fluc_lm_GS),1/nl*bl**2))))
#print "'fluc like' = ",tt3_new-tt3_old
tt_new = tt1_new + tt2_new + tt3_new
tt_old = tt1_old + tt2_old + tt3_old
return tt_new,tt_old
def analworker(i):
print "This is analysis worker starting for map", i[1]+1, "/", nmaps
alms = hp.map2alm(i[0],lmax=ellmax-1)
hp.almxfl(alms,pixrecip,inplace=True) #Correcting for pixwin
#TESTING
'''cls = hp.alm2cl(alms)
map_bandlim = hp.alm2map(alms,nside=256,pixwin=True)
hp.write_map('/Users/keir/Documents/s2let_ilc_planck/planck2015_2_cmb_map_1_bandlim300.fits',map_bandlim)'''
wav_maps, scal_maps = ps.analysis_lm2wav_manualtiling(alms,ellmax,ndir,spin,scal_tiles,wav_tiles.T.ravel(),scal_bandlims,wav_bandlims)
del alms
np.save(scal_outfits[i[1]],scal_maps)
del scal_maps
#Splitting up output wavelet maps
offset = 0
for j in xrange(jmax+1):
for n in xrange(ndir):
bandlim = wav_bandlims[j]
nelem = bandlim*(2.*bandlim-1.)
wav_outfits = wav_outfits_root[i[1]] + '_' + wavparam_code + str(j) + '_n' + str(n+1) + '.npy'
np.save(wav_outfits,wav_maps[offset:offset+nelem])
offset += nelem
del wav_maps
return 0 #wav_maps,scal_maps #,map_bandlim,alms,cls
def _cache_eblm(self, idx):
elm = self.unlcmbs.get_sim_elm(idx)
blm = None if 'b' not in self.fields else self.unlcmbs.get_sim_blm(idx)
dlm = self.get_sim_plm(idx)
assert 'o' not in self.fields, 'not implemented'
lmaxd = hp.Alm.getlmax(dlm.size)
hp.almxfl(dlm,
np.sqrt(
np.arange(lmaxd + 1, dtype=float) *
np.arange(1, lmaxd + 2)),
inplace=True)
Qlen, Ulen = self.lens_module.alm2lenmap_spin([elm, blm], [dlm, None],
self.nside_lens,
2,
nband=self.nbands,
facres=self.facres,
verbose=self.verbose)
elm, blm = hp.map2alm_spin([Qlen, Ulen], 2, lmax=self.lmax)
del Qlen, Ulen
hp.write_alm(os.path.join(self.lib_dir, 'sim_%04d_elm.fits' % idx),
elm)
del elm
hp.write_alm(os.path.join(self.lib_dir, 'sim_%04d_blm.fits' % idx),
blm)
def apply_ivf(self, det, tmap, pmap):
assert(self.lmax == 1000)
mask_t = qcinv.util.load_map(self.mask_t)
mask_p = qcinv.util.load_map(self.mask_p)
bl = dmc.get_bl(self.sim_lib.year, det)[0:self.lmax+1]
pxw = hp.pixwin( self.sim_lib.nside )[0:self.lmax+1]
# qcinv filtering for temperature
dcf = self.lib_dir + "/dense_cache_det_" + det + ".pk"
# id preconditioners lmax nside im em tr cache
chain_descr = [ [ 2, ["split(dense("+dcf+"), 64, diag_cl)"], 256, 128, 3, 0.0, qcinv.cd_solve.tr_cg, qcinv.cd_solve.cache_mem()],
[ 1, ["split(stage(2), 256, diag_cl)"], 512, 256, 3, 0.0, qcinv.cd_solve.tr_cg, qcinv.cd_solve.cache_mem()],
[ 0, ["split(stage(1), 512, diag_cl)"], 1000, 512, np.inf, 1.0e-6, qcinv.cd_solve.tr_cg, qcinv.cd_solve.cache_mem()] ]
ninv = ( hp.read_map( dmc.get_fname_iqumap(self.sim_lib.year, det, self.sim_lib.forered), hdu=1, field=3 ) /
dmc.sigma0[(self.sim_lib.year, self.sim_lib.forered, 'T')][det]**2 / 1e6 * mask_t )
n_inv_filt = qcinv.opfilt_tt.alm_filter_ninv( ninv, bl*pxw, marge_monopole=True, marge_dipole=True, marge_maps=[] )
chain = qcinv.multigrid.multigrid_chain( qcinv.opfilt_tt, chain_descr, self.cl, n_inv_filt )
tlm = np.zeros( qcinv.util_alm.lmax2nlm(self.lmax), dtype=np.complex )
chain.solve( tlm, tmap )
# simple filtering for polarization.
elm, blm = hp.map2alm_spin( (pmap.real * mask_p, pmap.imag * mask_p), 2, lmax=self.lmax )
ftl, fel, fbl = self.get_ftebl(det)
hp.almxfl( elm, fel / bl / pxw, inplace=True )
hp.almxfl( blm, fbl / bl / pxw, inplace=True )
return tlm, elm, blm
def vmaps2vmap_P(pix_vmaps, weights_e, weights_b, nside):
"""From individual Q and U freq pixel variance maps and weights create expected pixel variance map
Args:
pix_vmaps: list of pixel variance maps
weights_e: weights for E-mode freq. weighting (as applied onto the noise maps)
weights_b: weights for B-mode freq. weighting (as applied onto the noise maps)
nside: desired output map resolution
Note:
the pix_vmaps in pol in this routine are expected to be ~ 1/2 (s2_Q + s2_U)
See Planck 2018 gravitational lensing paper Eqs 16-17
"""
assert len(pix_vmaps) == len(weights_e), (len(pix_vmaps), len(weights_e))
assert len(pix_vmaps) == len(weights_b), (len(pix_vmaps), len(weights_b))
nf, lmaxp1_e = weights_e.shape
nf, lmaxp1_b = weights_b.shape
lmax_out = min(2 * max(lmaxp1_e, lmaxp1_b) - 2, 3 * nside - 1)
ret_lm = np.zeros(hp.Alm.getsize(lmax_out), dtype=complex)
for i, (pix_vmap, wle, wlb) in enumerate_progress(
list(zip(pix_vmaps, weights_e, weights_b))):
m = read_map(pix_vmap)
vpix = hp.nside2pixarea(hp.npix2nside(m.size), degrees=False)
this_s2lm = hp.map2alm(m, iter=0, lmax=lmax_out)
wl2 = 0.25 * vpix * _w2wsq(wle + wlb, 2, 2, lmax_out)
wl2 += 0.25 * vpix * _w2wsq(wle - wlb, 2, -2, lmax_out)
hp.almxfl(this_s2lm, wl2, inplace=True)
ret_lm += this_s2lm
return hp.alm2map(ret_lm, nside, verbose=False)
def convolve_healpix(input_map, func, thetamax, quiet=True, lmax_limit=None ):
"""
Convolve a HEALPix map with a function
input_map : array of float
a HEALPix array, RING indexing, nside a power of 2
func : function of theta to convolve with
thetamax : float
estimate of largest theta for function
Returns: the convolved map
"""
nside = int(np.sqrt(len(input_map)/12))
assert 12*nside**2 == len(input_map),'Bad length'
alm = healpy.map2alm(input_map);
lmax = healpy.Alm.getlmax(len(alm))
if lmax < 0:
raise TypeError('Wrong alm size for the given '
'mmax (len(alms[%d]) = %d).'%(ialm, len(alm)))
if lmax_limit is not None and lmax>lmax_limit:
lmax=lmax_limit
ell = np.arange(lmax + 1.)
if not quiet:
print 'Creating spherical harmonic content, lmax={}'.format(lmax)
fact = convolution.SphericalHarmonicContent(func,lmax, thetamax, quiet=quiet)(ell)
healpy.almxfl(alm, fact, inplace=True)
return healpy.alm2map(alm, nside=nside, verbose=False)
def almworker(i):
print "This is (map2alm & alm2map) worker starting for another map"
alms_worker = hp.map2alm(i[0], lmax=ellmax - 1)
hp.almxfl(alms_worker, i[1],
inplace=True) #Correcting for pixwin & smoothing
map_worker = hp.alm2map(alms_worker, nside_out, pixwin=True)
del alms_worker
return map_worker
def calc(self, alm):
tmat = self.slinv
relm = hp.almxfl(alm.elm, tmat[:, 0, 0], inplace=False) + hp.almxfl(
alm.blm, tmat[:, 0, 1], inplace=False)
rblm = hp.almxfl(alm.elm, tmat[:, 1, 0], inplace=False) + hp.almxfl(
alm.blm, tmat[:, 1, 1], inplace=False)
return eblm([relm, rblm])
def A_matrix_func(data):
"""
cf eq 25 of eriksen 2004
"""
Chalf = np.sqrt(data.cl_th[: data.lmax + 1]) * data.beam[: data.lmax + 1]
map2 = hp.alm2map(hp.almxfl(real2complex_alm(data.alm), Chalf), data.nside) * data.invvar
alm2 = hp.almxfl(hp.map2alm(map2, data.lmax, use_weights=False) * hp.nside2npix(data.nside) / 4.0 / np.pi, Chalf)
return data.alm + complex2real_alm(alm2)
def get_sim_qumap(self,idx):
elm = self.sims_cmb_len.get_sim_elm(idx)
hp.almxfl(elm,self.cl_ptransf,inplace=True)
blm = self.sims_cmb_len.get_sim_blm(idx)
hp.almxfl(blm, self.cl_ptransf, inplace=True)
Q,U = hp.alm2map_spin([elm,blm], self.nside, 2,hp.Alm.getlmax(elm.size))
del elm,blm
return [Q + self.get_sim_qnoise(idx),U + self.get_sim_unoise(idx)]
def get_deriv(mp) :
ns=hp.npix2nside(len(mp))
l=np.arange(3*ns)
alpha1i=np.sqrt(l*(l+1.))
alpha2i=np.sqrt((l-1.)*l*(l+1.)*(l+2.))
mpd1=hp.alm2map(hp.almxfl(hp.map2alm(mp),alpha1i),nside=ns,verbose=False)
mpd2=hp.alm2map(hp.almxfl(hp.map2alm(mp),alpha2i),nside=ns,verbose=False)
return mpd1,mpd2
def apply_alm(self, alm):
"""Missing doc. """
npix = len(self.n_inv)
hp.almxfl(alm, self.b_transf, inplace=True)
kmap = alm2map(alm, hp.npix2nside(npix), verbose=False)
self.apply_map(kmap)
alm[:] = map2alm(kmap, lmax=hp.Alm.getlmax(alm.size), iter=0)
hp.almxfl(alm, self.b_transf * (npix / (4. * np.pi)), inplace=True)
def get_irestmap(self, idx, xfilt=None):
reslm = self.ivfs.get_sim_tlm(idx)
if xfilt is not None:
assert isinstance(xfilt, dict) and 't' in xfilt.keys()
hp.almxfl(reslm, xfilt['t'], inplace=True)
return hp.alm2map(reslm,
self.nside,
lmax=hp.Alm.getlmax(reslm.size),
verbose=False)
def get_sim_pmap(self, idx):
elm = hp.almxfl(cmb_len_ffp10.get_sim_elm(idx), self.transf)
blm = hp.almxfl(cmb_len_ffp10.get_sim_blm(idx), self.transf)
Q, U = hp.alm2map_spin((elm, blm), self.nside, 2, hp.Alm.getlmax(elm.size))
del elm, blm
nlevp_pix = self.nlevp / np.sqrt(hp.nside2pixarea(self.nside, degrees=True)) / 60.
Q += self.pix_libphas.get_sim(idx, idf=1) * nlevp_pix
U += self.pix_libphas.get_sim(idx, idf=2) * nlevp_pix
return Q, U
def get_irespmap(self, idx, xfilt=None):
reselm = self.ivfs.get_sim_elm(idx)
resblm = self.ivfs.get_sim_blm(idx)
assert hp.Alm.getlmax(reselm.size) == hp.Alm.getlmax(resblm.size)
if xfilt is not None:
assert isinstance(xfilt, dict) and 'e' in xfilt.keys() and 'b' in xfilt.keys()
hp.almxfl(reselm, xfilt['e'], inplace=True)
hp.almxfl(resblm, xfilt['b'], inplace=True)
fac = 0.5
return hp.alm2map_spin([reselm * fac, resblm * fac], self.nside, 2, hp.Alm.getlmax(reselm.size))
def calc_prep(map, s_cls, n_inv_filt):
tmap = np.copy(map)
n_inv_filt.apply_map(tmap)
lmax = len(n_inv_filt.b_transf) - 1
npix = len(map)
alm = hp.map2alm(tmap, lmax=lmax, iter=3)
alm *= npix / (4. * np.pi)
hp.almxfl(alm, n_inv_filt.b_transf, inplace=True)
return alm
def _apply_ivf_p(self, pmap, soltn=None):
assert len(pmap[0]) == hp.nside2npix(self.nside) and len(
pmap[0]) == len(pmap[1])
elm, blm = hp.map2alm_spin([m for m in pmap], 2, lmax=self.lmax_fl)
elm = hp.almxfl(
elm,
self.get_fel() * utils.cli(self.transf[:len(self.fel)]))
blm = hp.almxfl(
blm,
self.get_fbl() * utils.cli(self.transf[:len(self.fbl)]))
return elm, blm
def test_forward_backward_spin(self):
""" compare map2alm_spin outputs to alm2map_spin inputs. tolerances are very loose. """
for spin in [1,2,3]:
tcl = np.ones(self.lmax+1); tcl[0:spin] = 0.
almg = hp.almxfl(self.almg, tcl, inplace=False)
almc = hp.almxfl(self.almc, tcl, inplace=False)
rmap, imap = hp.alm2map_spin( [almg, almc], self.nside, spin, self.lmax )
tglm, tclm = hp.map2alm_spin( [rmap, imap], spin, self.lmax )
np.testing.assert_allclose( almg, tglm, rtol=1.e-2, atol=1.e-2)
np.testing.assert_allclose( almc, tclm, rtol=1.e-2, atol=1.e-2)
def simulate_tebp_correlated(cl_tebp_arr,nside,lmax) : alms=healpy.synalm(cl_tebp_arr,lmax=lmax,new=True) aphi=alms[-1] acmb=alms[0:-1] #Set to zero above map resolution to avoid aliasing beam_cut=np.ones(3*nside) for ac in acmb : healpy.almxfl(ac,beam_cut,inplace=True) cmb=np.array(healpy.alm2map(acmb,nside,pol=True)) return cmb,aphi
def calc_prep(maps, s_cls, n_inv_filt):
qmap, umap = np.copy(maps[0]), np.copy(maps[1])
assert len(qmap) == len(umap)
lmax = len(n_inv_filt.b_transf) - 1
npix = len(qmap)
n_inv_filt.apply_map([qmap, umap])
elm, blm = map2alm_spin([qmap, umap], 2, lmax=lmax)
hp.almxfl(elm, n_inv_filt.b_transf * npix / (4. * np.pi), inplace=True)
hp.almxfl(blm, n_inv_filt.b_transf * npix / (4. * np.pi), inplace=True)
return eblm([elm, blm])
def apply_ivf(self, det, tmap, pmap):
mask_t = qcinv.util.load_map(self.mask_t)
bl = self.bl
pxw = hp.pixwin(self.nside)[0:self.lmax+1]
tlm = hp.map2alm( tmap * mask_t, lmax=self.lmax, iter=0, regression=False )
hp.almxfl( tlm, self.ftl / bl / pxw, inplace=True )
return tlm
def test_forward_backward_spin(self):
""" compare map2alm_spin outputs to alm2map_spin inputs. tolerances are very loose. """
for spin in [1,2,3]:
tcl = np.ones(self.lmax+1); tcl[0:spin] = 0.
almg = hp.almxfl(self.almg, tcl, inplace=False)
almc = hp.almxfl(self.almc, tcl, inplace=False)
rmap, imap = hp.alm2map_spin( [almg, almc], self.nside, spin, self.lmax )
tglm, tclm = hp.map2alm_spin( [rmap, imap], spin, self.lmax )
np.testing.assert_allclose( almg, tglm, rtol=1.e-2, atol=1.e-2)
np.testing.assert_allclose( almc, tclm, rtol=1.e-2, atol=1.e-2)
def calc_prep(map, s_cls, n_inv_filt):
tmap = np.copy(map)
n_inv_filt.apply_map(tmap)
lmax = len(n_inv_filt.b_transf) - 1
npix = len(map)
alm = hp.map2alm(tmap, lmax=lmax, iter=0, regression=False)
alm *= npix / (4.*np.pi)
hp.almxfl( alm, n_inv_filt.b_transf, inplace=True )
return alm
def simulate_tebp_correlated(cl_tebp_arr,nside,lmax,seed):
np.random.seed(seed)
alms=hp.synalm(cl_tebp_arr,lmax=lmax,new=True)
aphi=alms[-1]
acmb=alms[0:-1]
#Set to zero above map resolution to avoid aliasing
beam_cut=np.ones(3*nside)
for ac in acmb :
hp.almxfl(ac,beam_cut,inplace=True)
cmb=np.array(hp.alm2map(acmb,nside,pol=True,verbose=False))
return cmb,aphi
def apply_alm(self, alm):
# applies Y^T N^{-1} Y
npix = len(self.n_inv)
hp.almxfl(alm, self.b_transf, inplace=True)
tmap = hp.alm2map(alm, hp.npix2nside(npix))
self.apply_map(tmap)
alm[:] = hp.map2alm(tmap, lmax=util_alm.nlm2lmax(len(alm)), iter=0)
alm[:] *= (npix / (4. * np.pi))
hp.almxfl(alm, self.b_transf, inplace=True)
def simulate(self, det, idx):
assert( det in (dmc.wmap_das + dmc.wmap_bands) )
tlm = self.sim_tlm_cmb.get_sim_tlm(idx)
hp.almxfl(tlm, self.get_beam(det), inplace=True)
tmap = hp.alm2map(tlm, self.nside)
tmap_nse = self.simulate_nse(det)
tmap += tmap_nse; del tmap_nse
return tmap
def apply_ivf(self, det, tmap):
mask_t = qcinv.util.load_map(self.mask_t)
tlm = hp.map2alm( tmap * mask_t, lmax=self.lmax, iter=0, regression=False )
bl = dmc.get_bl(self.sim_lib.year, det)[0:self.lmax+1]
pxw = hp.pixwin( self.sim_lib.nside )[0:self.lmax+1]
ftl = self.get_ftl(det)
hp.almxfl( tlm, ftl / bl / pxw, inplace=True )
return tlm
def smoothworker(
i): #(Rflat[i],smoothing_lmax,spin,gausssmooth,scale_lmax,n,i,j)
print "Smoothing another independent covariance element"
alms = ps.map2alm_mw(
i[0], i[1],
i[2]) #No pixwin correct. with MW sampling - calc alms to smooth
#del i[0] #Everything gets moved down one index
hp.almxfl(alms, i[3], inplace=True) #Multiply by gaussian beam
'''if i[5] != -1: #If n != -1 (i.e. the maps are directional)
print "Picking out directional component of covariance map"
#Testing convolving gaussian-smoothed covariance maps with directional wavelet
jmin_min = 1
#Need to truncate alm's to scale_lmax (Is this necessary?)
alms_fname = 'alms_dirwav_' + str(i[7]) + '_' + str(i[5]) + '_' + str(i[6]) + '.fits'
hp.write_alm(alms_fname,alms,lmax=i[4]-1,mmax=i[4]-1)
del alms
alms_truncate = hp.read_alm(alms_fname)
print "Analysing covariance map" #Could increase wavparam!?
wav,scal = ps.analysis_lm2wav(alms_truncate,wavparam,i[4],jmin_min,ndir,spin,upsample)
del alms_truncate
#Delete wrong directions by zero-ing them
print "Deleting wrong directions"
jmax_min = ps.pys2let_j_max(wavparam,i[4],jmin_min)
for j in xrange(jmin_min,jmax_min+1):
for n in xrange(0,ndir):
if n != i[5]:
offset,new_scale_lmax,nelem,nelem_wav = ps.wav_ind(j,n,wavparam,i[4],ndir,jmin_min,upsample)
wav[offset:offset+nelem] = 0.
print "Synthesising directional covariance map"
alms = ps.synthesis_wav2lm(wav,scal,wavparam,i[4],jmin_min,ndir,spin,upsample)
del wav,scal
#Expand alm's with zero-padding
print "Zero-padding the alm's"
nzeros = i[1] - i[4] #No. zeros to pad
new_alms_temp = np.concatenate((alms[:i[4]],np.zeros(nzeros)))
for em in xrange(1,i[4]):
startindex = em*i[4] - .5*em*(em-1)
new_alms_temp = np.concatenate((new_alms_temp,alms[startindex:(startindex+i[4]-em)],np.zeros(nzeros)))
del alms
print "Temporary length of alm's =", len(new_alms_temp)
nfinalzeros = hp.Alm.getsize(i[1]-1) - len(new_alms_temp)
alms = np.concatenate((new_alms_temp,np.zeros(nfinalzeros)))
del new_alms_temp
print "Final length of alm's =", len(alms)'''
#hp.almxfl(alms,i[3],inplace=True) #Multiply by gaussian beam
print "Synthesising smoothed covariance map"
Rsmoothflat = ps.alm2map_mw(
alms, i[1], i[2]) #Smooth covariance in MW - calc final map to scale
del alms
return Rsmoothflat
def calc(self, alm):
tmat = self.slinv
rtlm = hp.almxfl(alm.tlm, tmat[:, 0, 0]) + hp.almxfl(
alm.elm, tmat[:, 0, 1]) + hp.almxfl(alm.blm, tmat[:, 0, 2])
relm = hp.almxfl(alm.tlm, tmat[:, 1, 0]) + hp.almxfl(
alm.elm, tmat[:, 1, 1]) + hp.almxfl(alm.blm, tmat[:, 1, 2])
rblm = hp.almxfl(alm.tlm, tmat[:, 2, 0]) + hp.almxfl(
alm.elm, tmat[:, 2, 1]) + hp.almxfl(alm.blm, tmat[:, 2, 2])
return teblm([rtlm, relm, rblm])
def target_distrib(guess, *arg):
"""
Keyword Arguments:
x -- the vector of params
*args are:
arg[0] -- a target class object
arg[1] -- Cl_old, fluc_lm_old
"""
#print guess, arg
dlm,strings,params,nl,bl = arg[0]
Cl_old, fluc_lm_old = arg[1]
dd = cb.update_dic(params,guess,strings)
Cl_new = cb.generate_spectrum(dd)[:,1]
Cl_new[:2] = 1.e-35
#print "new = ",Cl_new[50]
#print "old = ",Cl_old[50]
renorm = CG.renorm_term(Cl_new,bl,nl)
mf_lm_new = hp.almxfl(CG.generate_mfterm(dlm,Cl_new,bl,nl),renorm)
fluc_lm_type2 = hp.almxfl(CG.generate_w1term(Cl_new,bl,nl)+CG.generate_w0term(Cl_new),renorm)
#print "new = ",fluc_lm_type2[50]
#print "old = ",fluc_lm_old[50]
fluc_lm_determ = hp.almxfl(fluc_lm_old,np.sqrt(Cl_new/Cl_old))
tt1 = -1/2.*np.real(np.vdot((dlm-hp.almxfl(mf_lm_new,bl)).T,hp.almxfl((dlm-hp.almxfl(mf_lm_new,bl)),1/nl)))
#print tt1
tt2 = -1/2. *np.real(np.vdot((mf_lm_new).T,hp.almxfl((mf_lm_new),1./Cl_new)))
#print tt2
tt3 = -1/2. *np.real(np.vdot((fluc_lm_determ).T,hp.almxfl((fluc_lm_determ),1/nl*bl**2)))
#print tt3
#tt4 = - 1./2 *(np.arange(1,np.size(Cl_new)+1)*np.log(Cl_new)).sum()
##print tt4
return [tt1,tt2,tt3],Cl_new,fluc_lm_type2
def apply_alm(self, alm):
# applies Y^T N^{-1} Y
npix = len(self.n_inv)
hp.almxfl(alm, self.b_transf, inplace=True)
tmap = hp.alm2map(alm, hp.npix2nside(npix))
self.apply_map(tmap)
alm[:] = hp.map2alm(tmap, lmax=util_alm.nlm2lmax(len(alm)), iter=0, regression=False)
alm[:] *= (npix / (4.*np.pi))
hp.almxfl(alm, self.b_transf, inplace=True)
def get_sim_tmap(self, idx):
"""Returns temperature healpy map for a simulation
Args:
idx: simulation index
Returns:
healpy map
"""
tmap = self.sims_cmb_len.get_sim_tlm(idx)
hp.almxfl(tmap, self.cl_transf, inplace=True)
tmap = hp.alm2map(tmap, self.nside)
return tmap + self.get_sim_tnoise(idx)
def simulate_tebp_correlated(cl_tebp_arr, nside, lmax, seed):
"""This generates correlated T,E,B and Phi maps
"""
np.random.seed(seed)
alms = hp.synalm(cl_tebp_arr, lmax=lmax, new=True)
aphi = alms[-1]
acmb = alms[0:-1]
# Set to zero above map resolution to avoid aliasing
beam_cut = np.ones(3 * nside)
for ac in acmb:
hp.almxfl(ac, beam_cut, inplace=True)
cmb = np.array(hp.alm2map(acmb, nside, pol=True, verbose=False))
return cmb, aphi
def deconvolve_alm(alms,
lmax=None,
fwhm_in=0.0,
fwhm_out=0.0,
f_sky=1.0,
pol=False,
wiener=True,
sky_prior=None):
fwhm = np.sqrt(fwhm_in**2 - fwhm_out**2)
if fwhm == 0.0:
return alms
factor = (2.0 * np.sqrt(2.0 * np.log(2.0)))
sigma = fwhm / factor
retalm = []
if lmax is None:
lmax = hp.Alm.getlmax(len(alms[0] if pol else alms), None)
if wiener:
wiener_filter = wiener_filter_for_alm(alms[0] if pol else alms,
lmax,
f_sky=f_sky,
sky_prior=sky_prior,
fwhm=fwhm_in)
wiener_smooth = fl.filter_butter(wiener_filter, lmax, 100)
#wiener_smooth[:1500] = 1.0
if pol:
for ialm, alm in enumerate(alms):
ell = np.arange(lmax + 1)
if ialm >= 1:
s = 2
else:
s = 0
fact = np.exp(0.5 * (ell * (ell + 1) - s**2) * sigma**2)
if wiener:
fact /= wiener_smooth
res = hp.almxfl(alm, fact, inplace=False)
retalm.append(res)
else:
lmax = hp.Alm.getlmax(len(alms), None)
ell = np.arange(lmax + 1)
fact = np.exp(0.5 * (ell * (ell + 1)) * sigma**2)
if wiener:
fact /= wiener_smooth
retalm = hp.almxfl(alms, fact, inplace=False)
return retalm
def generate_gaus_map(self, readGmap=-1):
if (readGmap<0):
self.gausalm0=hp.synalm(self.inputCls)
else:
self.gausalm0=hp.read_alm(self.mapsdir+"gmap_"+str(readGmap)+".fits")
self.gausalm1=np.copy(self.gausalm0); self.gausalm1[0]=0.0
if (self.nodipole):
ndxfl=np.ones(len(self.inputCls))
ndxfl[1]=0.0
hp.almxfl(self.gausalm1, ndxfl, inplace=True)
hp.almxfl(self.gausalm0, ndxfl, inplace=True)
self.gausmap0=hp.alm2map(self.gausalm0, nside=self.NSIDE) # includes the monopole bit
self.gausmap1=hp.alm2map(self.gausalm1, nside=self.NSIDE) # does not include the monopole
def solve_CG(params_i,dd):
dd = cb.update_dic(dd,params_i,strings)
cl = cb.generate_spectrum(dd)[:,1]
# define the first guess, the b from eq 25 since the code was design with this at first (might not be best idea ever)
b_mf = CG.generate_mfterm(dlm_filt,cl[:lmax+1],bl[:lmax+1],nl[:lmax+1])
b_w1 = CG.generate_w1term(cl[:lmax+1],bl[:lmax+1],nl[:lmax+1])
b_w0 = CG.generate_w0term(cl[:lmax+1])
b = b_mf + b_w1 + b_w0
###### left hand side factor
renorm= np.sqrt(cl[:lmax+1])/(1+cl[:lmax+1]*bl[:lmax+1]**2/(nl[:lmax+1]))
out = hp.almxfl(b,renorm)
out_mf = hp.almxfl(b_mf,renorm)
out_w1 = hp.almxfl(b_w1,renorm)
out_w0 = hp.almxfl(b_w0,renorm)
return out,out_mf,out_w0,out_w1
def calc_prep(maps, s_cls, n_inv_filt):
qmap, umap = np.copy(maps[0]), np.copy(maps[1])
assert(len(qmap) == len(umap))
npix = len(qmap)
n_inv_filt.apply_map([qmap, umap])
lmax = len(n_inv_filt.b_transf) - 1
tlm, elm, blm = hp.map2alm( [qmap, qmap, umap], lmax=lmax, iter=0, regression=False, use_weights=False, pol=True )
del tlm
elm *= npix / (4.*np.pi); blm *= npix / (4.*np.pi)
hp.almxfl( elm, n_inv_filt.b_transf, inplace=True )
hp.almxfl( blm, n_inv_filt.b_transf, inplace=True )
return eblm([elm, blm])
def fisher_single(par,v) :
# v -> seen pixels
nb=len(par.bins)-1
npix=hp.nside2npix(par.nside)
npix_seen=len(par.ip_seen)
lmax=3*par.nside-1
larr=np.arange(lmax+1)
fisher=np.zeros([nb,nb])
pixsize=4*np.pi/hp.nside2npix(par.nside)
v_map=np.zeros(npix); v_map[par.ip_seen]=v
vcm1=invert_covar(par,v_map)
v_lm=hp.map2alm(v_map,iter=0)
vcm1_lm=hp.map2alm(vcm1,iter=0)
for iba in np.arange(nb) :
# print " Row %d"%iba
transfer=np.zeros(lmax+1); transfer[par.bins[iba]:par.bins[iba+1]]=1.
v_map2=hp.alm2map(hp.almxfl(v_lm,transfer),par.nside,verbose=False)/pixsize #Q_a * v
v_map2cm1=invert_covar(par,v_map2) #C^-1 * Q_a * v
va_lm=hp.map2alm(v_map2cm1,iter=0)
cl_vcm1_va=(2*larr+1)*hp.alm2cl(vcm1_lm,alms2=va_lm)
for ibb in np.arange(nb-iba)+iba :
fisher[iba,ibb]=np.sum(cl_vcm1_va[par.bins[ibb]:par.bins[ibb+1]])/pixsize**2
if iba!=ibb :
fisher[ibb,iba]=fisher[iba,ibb]
return fisher
def generate_mfterm(dlm, Cl, bl, nl):
"""
"""
lmax = np.size(Cl) - 1
out = hp.almxfl(dlm, np.sqrt(Cl[: lmax + 1]) * bl[: lmax + 1] / nl[: lmax + 1])
return out
def functional_form_params_n(x,*arg):
"""
Keyword Arguments:
x -- the vector of params
*args are:
dlm -- input map
x_str -- the dictionary strings corresponding to x
params -- a camber dictionnary
noise -- a noise power spectrum
beam -- a beam power spectrum
"""
dlm = arg[0]
strings = arg[1]
params = arg[2].copy()
noise = arg[3]
beam = arg[4]
Cl_old = arg[5]
lmax = len(Cl_old)-1
#params["output_root"] = '../Codes/CG_git/MH_MCMC/camb_ini/test%d'%np.random.randint(100)
for i in range(np.size(x)):
##print strings[i]
if strings[i]=='scalar_amp(1)':
##print params[strings[i]]
params[strings[i]]=np.exp(x[i])*1e-10
##print params[strings[i]]
else:
params[strings[i]]=x[i]
Cl = cb.generate_spectrum(params)
#lmax = Cl.shape[0]-1
tt = -1./2 * np.real(np.vdot(CG.complex2real_alm(dlm.T),CG.complex2real_alm(hp.almxfl(dlm,1/(beam[:lmax+1]**2*Cl[:lmax+1,1]+noise[:lmax+1])))))
#determinant is the product of the diagonal element: in log:
tt2 = - 1./2 *((2*np.arange(2,lmax+1)+1)*np.log(noise[2:lmax+1]+Cl[2:lmax+1,1]*beam[2:lmax+1]**2)).sum()
return tt+tt2,Cl[:,1]
def direct(self, input, output):
if input.ndim == 1:
input = input[:, None]
output = output[:, None]
for i, o in zip(input.T, output.T):
ialm = hp.map2alm(i)
alm_smoothed = hp.almxfl(ialm, self.fl)
o[...] = hp.alm2map(alm_smoothed, hp.npix2nside(len(i)))# * \
def return_map(map_class):
"""
We solve for C^{-1/2}x, here is to recover x
"""
Shalf = np.sqrt(map_class.cl_th[: map_class.lmax + 1])
alm_out = hp.almxfl(real2complex_alm(map_class.alm), Shalf)
cl_out = hp.alm2cl(alm_out)
map_out = hp.alm2map(alm_out, map_class.nside)
return cl_out, map_out
def rs_w1_matrix_func(data, w1):
"""
cf eq 25 of eriksen 2004
"""
Chalf = np.sqrt(data.cl_th[: data.lmax]) * data.beam[: data.lmax]
map2 = w1 * np.sqrt(data.invvar)
# map2 = hp.alm2map(real2complex_alm(data.alm),data.nside)*np.sqrt(data.invvar)
alm2 = hp.almxfl(hp.map2alm(map2, data.lmax, use_weights=False) * hp.nside2npix(data.nside) / 4.0 / np.pi, Chalf)
return complex2real_alm(alm2)
def plot_tests():
spec_mf_b = hp.alm2cl(hp.almxfl(mf_lm_new,bl))
spec_mf = hp.alm2cl((mf_lm_new))
spec = hp.alm2cl(dlm)
diff = hp.alm2cl(dlm - (hp.almxfl(mf_lm_new,bl)))
fluc_l = hp.alm2cl(fluc)
nnn = hp.alm2cl(dlm-hp.almxfl(mf_lm_new,bl))
plt.figure()
plt.plot(spec_mf,label='mean field ($\hat{s}$)')
plt.plot(spec_mf_b,label='beamed mean field ($A\hat{s}$)')
plt.plot(spec,label='data (d)')
plt.plot(spec - spec_mf_b,label='($d_\ell - (A\hat{s})_\ell$)')
plt.plot(diff,label='($d - (A\hat{s})$)$_\ell$')
plt.plot(fluc_l,"-.",label='($\hat{f}$)$_\ell$')
plt.plot(nl,"-.",label="noise")
plt.plot(cl,"-.",label="trial spectrum")
plt.yscale("log")
plt.legend(loc = 'best')
def calc(self, talm):
if ( np.all(talm == 0) ): # do nothing if zero
return talm
alm = np.copy(talm)
self.n_inv_filt.apply_alm(alm)
alm += hp.almxfl(talm, self.cltt_inv)
return alm
def test_der1(self):
""" compare output of alm2map_der1 with the equivalent spin-1 transform using alm2map_spin """
m, dt_der1, dp_der1 = hp.alm2map_der1( self.almg, self.nside )
alm_spin = hp.almxfl( self.almg, np.array( [np.sqrt(l*(l+1.)) for l in xrange(0,self.lmax+1)] ), inplace=False )
dt_spin, dp_spin = hp.alm2map_spin( [alm_spin, alm_spin*0.], self.nside, 1, self.lmax )
np.testing.assert_array_almost_equal( dt_der1, dt_spin, decimal=8)
np.testing.assert_array_almost_equal( dp_der1, dp_spin, decimal=8)
def test_loglike(dlm,Cl,noise,beam):
"""
returns the log likelihood for a given Cl, beam, noise, and data.
"""
lmax = Cl.shape[0]
tt_exp = -1./2 * np.real(np.vdot(dlm.T,hp.almxfl(dlm,1/(beam[:lmax]**2*Cl[:,1]+noise[:lmax]))))
#plt.plot(Cl[:,1])
tt_det = - 1./2 *(np.arange(1,lmax+1)*np.log((noise[:lmax]+Cl[:,1]*beam[:lmax]**2))).sum()
tt_f = tt_exp + tt_det
return tt_exp,tt_det,tt_f#,Cl[:,1]
def CG_algo_precond_diag(Matrix, Precond_diag, b, data_start, i_max, eps):
"""
Matrix is the function to apply the matrix on a vector
data_start is a data_class class, with real alms
"""
i = 0
x = data_start.alm.copy()
cl_th = data_start.cl_th
beam = data_start.beam
sigma = data_start.sigma
lmax = data_start.lmax
nside = data_start.nside
invvar = data_start.invvar
out = data_class([0, cl_th, beam, sigma, invvar, lmax, nside])
r = b - Matrix(data_start)
d = data_class([0, cl_th, beam, sigma, invvar, lmax, nside])
d.alm = r.copy()
d.alm = complex2real_alm(hp.almxfl(real2complex_alm(r), Precond_diag))
delt_n = np.dot(r.T, d.alm)
delt_0 = delt_n.copy()
iter_out_map = []
iter_out_cl = []
while i < i_max and delt_n > (eps ** 2 * delt_0):
q = Matrix(d)
alph = np.float(delt_n) / np.dot(d.alm.T, q)
x = x + alph * d.alm
if i % 10 == 0:
dat_temp = data_class([0, cl_th, beam, sigma, invvar, lmax, nside])
dat_temp.alm = x
r = b - Matrix(dat_temp)
else:
r = r - alph * q
s = complex2real_alm(hp.almxfl(real2complex_alm(r), Precond_diag))
# s = hp.almxfl(r,Precond_diag)
delt_old = delt_n.copy()
delt_n = np.dot(r.T, s)
bet = delt_n / delt_old
d.alm = s + bet * d.alm
i += 1
out.alm = x
iter_out_map.append(return_map(out)[1])
iter_out_cl.append(return_map(out)[0])
return iter_out_map, iter_out_cl
def calc(self, talm):
if ( np.all(talm == 0) ): # do nothing if zero
return talm
ret = hp.almxfl(talm, self.cltt_inv)
for n_inv_filt in self.n_inv_filts:
alm = np.copy(talm)
n_inv_filt.apply_alm(alm)
ret += alm
return ret