def fitbias(ph, spectra, binit=[1, 0, 0, 0, 1], k=None, kmax=None):
if k is not None and kmax is not None:
ik = np.where(k > kmax)[0][0]
else: ik = len(ph)
tomin = lambda b: sum( (ph - tools.getmodel(spectra, [1] + list(b[:-1])) - b[-1]) [:ik]**2 ) #added shot noise
rep = minimize(tomin, binit, method='Nelder-Mead', options={'maxfev':10000})
#rep = minimize(tomin, binit, options={'maxfev':10000})
return rep
spectra = np.loadtxt(
'./output/%s/spectra-%04d-%04d-%04d-R%d.txt' %
(subf, aa * 10000, bs, nc, Rsm)).T
k, spectra = spectra[0], spectra[1:]
header = header.split(',')
iv = len(header)
fig, axar = plt.subplots(1, 2, figsize=(8, 4))
for ik, kmax in enumerate([0.1, 0.3, 0.5, 0.8, 1.0]):
binit = [1., 0, 0, 0, sn]
rep = fitbias(ph, spectra, k=k, kmax=kmax, binit=binit)
print(rep)
bvec = [1] + list(rep.x[:-1])
model = tools.getmodel(spectra, bvec) + rep.x[-1]
axis = axar[0]
if ik == 0: axis.plot(k, ph, 'k', label='Halo', lw=2)
axis.plot(k, model, 'C%d--' % ik, lw=2, alpha=0.7)
axis.set_xlabel('k (h/Mpc)', fontsize=12)
axis.set_ylabel('$P$', fontsize=12)
axis.legend(fontsize=12)
axis.loglog()
axis.grid()
axis = axar[1]
axis.plot(k,
model / ph,
'C%d-' % ik,
label='k=%0.2f' % kmax,
header = '1, b1, b2, bg, bk'
if zadisp:
np.savetxt('./output/%s/spectraza-%04d-%04d-%04d-R%d.txt' %
(subf, aa * 10000, bs, nc, Rsm),
np.vstack([k, spectra]).T.real,
header='k / ' + header,
fmt='%0.4e')
else:
np.savetxt('./output/%s/spectra-%04d-%04d-%04d-R%d.txt' %
(subf, aa * 10000, bs, nc, Rsm),
np.vstack([k, spectra]).T.real,
header='k / ' + header,
fmt='%0.4e')
header = header.split(',')
bvec = [1, 1, 1, 1, 1]
model = tools.getmodel(spectra, bvec)
iv = len(header)
fig, ax = plt.subplots(1, iv, figsize=(15, 4), sharex=True)
counter = 0
for i in range(iv):
for j in range(i, iv):
ax[i].plot(k,
spectra[counter],
'-C%d' % j,
label=header[j])
ax[i].plot(k, -spectra[counter], '--C%d' % j)
counter += 1
ax[i].set_title(header[i])