def __init__(self, psrs, bayesephem=True, gamma_common=4.33,
orf='hd', wideband=False, select=None):
# initialize standard model with fixed white noise and
# and powerlaw red and gw signal
self.pta = models.model_2a(psrs, psd='powerlaw', bayesephem=bayesephem,
gamma_common=gamma_common, wideband=wideband,
select=select)
# get frequencies here
self.freqs = self._get_freqs()
# get F-matrices and set up cache
self.Fmats = self.get_Fmats()
self._set_cache_parameters()
# pulsar locations
self.psrlocs = [p.pos for p in psrs]
# overlap reduction function
if orf == 'hd':
self.orf = utils.hd_orf
elif orf == 'dipole':
self.orf = utils.dipole_orf
elif orf == 'monopole':
self.orf = utils.monopole_orf
else:
raise ValueError('Unknown ORF!')
def test_model2a_pshift(dmx_psrs, caplog):
caplog.set_level(logging.CRITICAL)
m2a = models.model_2a(dmx_psrs,
noisedict=noise_dict,
pshift=True,
pseed=42)
assert hasattr(m2a, 'get_lnlikelihood')
def test_model2a_broken_plaw(dmx_psrs, caplog):
caplog.set_level(logging.CRITICAL)
m2a = models.model_2a(dmx_psrs,
psd='broken_powerlaw',
delta_common=0,
noisedict=noise_dict)
assert hasattr(m2a, 'get_lnlikelihood')
def test_hypermodel(dmx_psrs, caplog):
m2a = models.model_2a(dmx_psrs, noisedict=noise_dict)
m3a = models.model_3a(dmx_psrs, noisedict=noise_dict)
ptas = {0: m2a, 1: m3a}
hm = hypermodel.HyperModel(ptas)
assert hasattr(hm, 'get_lnlikelihood')
assert 'gw_log10_A' in hm.param_names
assert 'nmodel' in hm.param_names
def test_setup_sampler(dmx_psrs, caplog):
m2a = models.model_2a(dmx_psrs, noisedict=noise_dict)
samp = sampler.setup_sampler(m2a, outdir=outdir, human='tester')
assert hasattr(samp, "sample")
paramfile = os.path.join(outdir, "pars.txt")
assert os.path.isfile(paramfile)
with open(paramfile, "r") as f:
params = [line.rstrip('\n') for line in f]
for ptapar, filepar in zip(m2a.param_names, params):
assert ptapar == filepar
def test_jumpproposal(dmx_psrs,caplog):
m2a=models.model_2a(dmx_psrs,noisedict=noise_dict)
jp=sampler.JumpProposal(m2a)
assert jp.draw_from_prior.__name__ == 'draw_from_prior'
assert jp.draw_from_signal_prior.__name__ == 'draw_from_signal_prior'
assert (jp.draw_from_par_prior('J1713+0747').__name__ ==
'draw_from_J1713+0747_prior')
assert (jp.draw_from_par_log_uniform({'gw':(-20,-10)}).__name__ ==
'draw_from_gw_log_uniform')
assert (jp.draw_from_signal('red noise').__name__ ==
'draw_from_red noise_signal')
def __init__(self,
psrs,
bayesephem=True,
gamma_common=4.33,
orf='hd',
wideband=False,
select=None,
noisedict=None,
pta=None):
# initialize standard model with fixed white noise and
# and powerlaw red and gw signal
if pta is None:
self.pta = models.model_2a(psrs,
psd='powerlaw',
bayesephem=bayesephem,
gamma_common=gamma_common,
is_wideband=wideband,
select='backend',
noisedict=noisedict)
else:
self.pta = pta
self.gamma_common = gamma_common
# get frequencies here
self.freqs = self._get_freqs(psrs)
# set up cache
self._set_cache_parameters()
# pulsar locations
self.psrlocs = [p.pos for p in psrs]
# overlap reduction function
if orf == 'hd':
self.orf = model_orfs.hd_orf
elif orf == 'dipole':
self.orf = model_orfs.dipole_orf
elif orf == 'monopole':
self.orf = model_orfs.monopole_orf
elif orf == 'gw_monopole':
self.orf = model_orfs.gw_monopole_orf
elif orf == 'gw_dipole':
self.orf = model_orfs.gw_dipole_orf
elif orf == 'st':
self.orf = model_orfs.st_orf
else:
raise ValueError('Unknown ORF!')
def test_hyper_sampler(dmx_psrs, caplog):
m2a = models.model_2a(dmx_psrs, noisedict=noise_dict)
m3a = models.model_3a(dmx_psrs, noisedict=noise_dict)
ptas = {0: m2a, 1: m3a}
hm = hypermodel.HyperModel(ptas)
samp = hm.setup_sampler(outdir=outdir, human='tester')
assert hasattr(samp, "sample")
paramfile = os.path.join(outdir, "pars.txt")
assert os.path.isfile(paramfile)
with open(paramfile, "r") as f:
params = [line.rstrip('\n') for line in f]
for ptapar, filepar in zip(hm.param_names, params):
assert ptapar == filepar
x0 = hm.initial_sample()
assert len(x0) == len(hm.param_names)
def test_extend_emp_dists_1d_kde(dmx_psrs, caplog):
with open(datadir+'/emp_dist_samples.pkl', 'rb') as fin:
tmp_data = pickle.load(fin)
m2a = models.model_2a(dmx_psrs, noisedict=noise_dict)
new_dist = make_empirical_distributions_KDE(m2a, tmp_data['names'], tmp_data['names'],
tmp_data['samples'], save_dists=False)
new_dist = sampler.extend_emp_dists(m2a, new_dist)
for ii in range(len(tmp_data['names'])):
m2a.params[ii].prior._defaults['pmin'] -= 0.1
new_dist = sampler.extend_emp_dists(m2a, new_dist)
assert len(new_dist) == 6
for i in range(6):
assert new_dist[i].minval <= m2a.params[i].prior._defaults['pmin']
assert new_dist[i].maxval >= m2a.params[i].prior._defaults['pmax']
def test_extend_emp_dists_1d(dmx_psrs, caplog):
with open(datadir+'/emp_dist_samples.pkl', 'rb') as fin:
tmp_data = pickle.load(fin)
m2a = models.model_2a(dmx_psrs, noisedict=noise_dict)
new_dist = make_empirical_distributions(m2a, tmp_data['names'], tmp_data['names'],
tmp_data['samples'], save_dists=False)
# run extend when edges match priors
new_dist = sampler.extend_emp_dists(m2a, new_dist)
# change priors so they don't match edges of
# empirical distribution
for ii in range(len(tmp_data['names'])):
m2a.params[ii].prior._defaults['pmin'] -= 0.1
new_dist = sampler.extend_emp_dists(m2a, new_dist)
assert len(new_dist) == 6
for i in range(6):
assert new_dist[i]._edges[0] <= m2a.params[i].prior._defaults['pmin']
assert new_dist[i]._edges[-1] >= m2a.params[i].prior._defaults['pmax']
def test_extend_emp_dists_2d(dmx_psrs, caplog):
with open(datadir+'/emp_dist_samples.pkl', 'rb') as fin:
tmp_data = pickle.load(fin)
m2a = models.model_2a(dmx_psrs, noisedict=noise_dict)
parnames = [[tmp_data['names'][0], tmp_data['names'][1]],
[tmp_data['names'][2], tmp_data['names'][3]],
[tmp_data['names'][4], tmp_data['names'][5]]]
new_dist = make_empirical_distributions(m2a, parnames, tmp_data['names'],
tmp_data['samples'], save_dists=False)
# case 1, edges match priors
new_dist = sampler.extend_emp_dists(m2a, new_dist)
# case 2, edges don't match priors (set priors to be different)
for ii in range(len(tmp_data['names'])):
m2a.params[ii].prior._defaults['pmin'] -= 0.1
m2a.params[ii].prior._defaults['pmax'] += 0.1
new_dist = sampler.extend_emp_dists(m2a, new_dist)
assert len(new_dist) == 3
for i in range(3):
k = 2 * i
assert new_dist[i]._edges[0][0] <= m2a.params[k].prior._defaults['pmin']
assert new_dist[i]._edges[0][-1] <= m2a.params[k].prior._defaults['pmax']
assert new_dist[i]._edges[1][0] <= m2a.params[k + 1].prior._defaults['pmin']
assert new_dist[i]._edges[1][-1] <= m2a.params[k + 1].prior._defaults['pmax']
del psrs[idx]
Outdir = args.outdir + '{0}/'.format(args.nyears)
with open(args.noisepath, 'r') as fin:
noise_dict = json.load(fin)
if args.vary_gamma:
gamma_gw = None
else:
gamma_gw = args.gamma_gw
if args.model == 'model_2a':
pta = models.model_2a(psrs,
psd=args.psd,
noisedict=noise_dict,
n_gwbfreqs=args.nfreqs,
gamma_common=gamma_gw,
upper_limit=args.gwb_ul,
bayesephem=args.bayes_ephem,
wideband=args.wideband,
pshift=args.pshift,
select='backend')
if args.model == 'model_3a':
pta = models.model_3a(psrs,
psd=args.psd,
noisedict=noise_dict,
n_gwbfreqs=args.nfreqs,
gamma_common=gamma_gw,
upper_limit=args.gwb_ul,
bayesephem=args.bayes_ephem,
wideband=args.wideband,
pshift=args.pshift,
select='backend')
def test_model2a_5gwb(dmx_psrs, caplog):
caplog.set_level(logging.CRITICAL)
m2a = models.model_2a(dmx_psrs, n_gwbfreqs=5, noisedict=noise_dict)
assert hasattr(m2a, 'get_lnlikelihood')
with open(args.noisepath, 'r') as fin:
noise = json.load(fin)
if args.rn_psrs[0] == 'all':
rn_psrs = 'all'
else:
rn_psrs = args.rn_psrs
pta_crn = models.model_2a(psrs,
psd=args.psd,
noisedict=noise,
n_gwbfreqs=args.nfreqs,
gamma_common=args.gamma_gw,
delta_common=0.,
upper_limit=False,
bayesephem=args.bayes_ephem,
be_type='setIII',
wideband=False,
rn_psrs=rn_psrs,
select='backend',
pshift=False)
####Sky Scramble script
if args.sky_scramble is None:
pass
else:
scr_npz = np.load(args.sky_scramble)
thetas = scr_npz['thetas'][args.process, :]
phis = scr_npz['phis'][args.process, :]
for p, theta, phi in zip(psrs, thetas, phis):
def pta_model2a(dmx_psrs, caplog):
m2a = models.model_2a(dmx_psrs, noisedict=noise_dict)
return m2a
def pta_model2a(dmx_psrs, caplog):
m2a = models.model_2a(dmx_psrs, noisedict=noise_dict, tnequad=True)
return m2a
upper_limit=False,
bayesephem=args.bayes_ephem,
be_type='setIII',
wideband=False,
rn_psrs=rn_psrs,
pshift=args.pshift,
pseed=args.process,
psr_models=False)
pta_crn = models.model_2a(psrs,
psd='powerlaw',
noisedict=noise,
n_gwbfreqs=args.nfreqs,
gamma_common=13 / 3.,
upper_limit=False,
bayesephem=args.bayes_ephem,
be_type='setIII',
wideband=False,
rn_psrs=rn_psrs,
pshift=args.pshift,
pseed=args.process,
select='backend',
psr_models=False)
print(pta_gw.get_lnlikelihood(np.hstack(p.sample() for p in pta_gw.params)))
print(pta_crn.get_lnlikelihood(np.hstack(p.sample() for p in pta_crn.params)))
#
# #Load in T matrix from previous run to check same random phase shifts
# if os.path.exists(args.outdir+'Tmats.npy'):
# Tmats = np.load(args.outdir+'Tmats.npy')
# print('Loading Tmat list from earlier incarnation.')
# else: # Save random phase shifted T matrices
