def plot(self, filename=None,
axes=None,
fignum=None, figsize=(4,4),
data_kwargs=dict()):
""" Plot the SED using matpotlib. """
if axes is None:
fig = P.figure(fignum,figsize)
axes = fig.add_axes((0.22,0.15,0.75,0.8))
self.set_xlim(axes,
float(self.lower_energy[0]/self.energy_units),
float(self.upper_energy[-1]/self.energy_units))
self.axes = axes
ce = lambda x: units.tonumpy(x, self.energy_units)
cf = lambda y: units.tonumpy(
y.multiply_elementwise(self.energy).multiply_elementwise(self.energy),
self.flux_units/units.cm**2/units.s)
SED._plot_points(
x=ce(self.energy),
xlo=ce(self.lower_energy),
xhi=ce(self.upper_energy),
y=cf(self.dnde),
y_err=cf(self.dnde_err),
y_ul=cf(self.dnde_ul),
significant=self.significant,
energy_units=self.energy_units_str,
flux_units=self.flux_units_str,
axes=axes, **data_kwargs)
if filename is not None: P.savefig(filename)
return axes
def plot(self,
filename=None,
axes=None,
fignum=None,
figsize=(4, 4),
data_kwargs=dict()):
""" Plot the SED using matpotlib. """
if axes is None:
fig = P.figure(fignum, figsize)
axes = fig.add_axes((0.22, 0.15, 0.75, 0.8))
self.set_xlim(axes,
float(self.lower_energy[0] / self.energy_units),
float(self.upper_energy[-1] / self.energy_units))
self.axes = axes
ce = lambda x: units.tonumpy(x, self.energy_units)
cf = lambda y: units.tonumpy(
y.multiply_elementwise(self.energy).multiply_elementwise(
self.energy), self.flux_units / units.cm**2 / units.s)
SED._plot_points(x=ce(self.energy),
xlo=ce(self.lower_energy),
xhi=ce(self.upper_energy),
y=cf(self.dnde),
y_err=cf(self.dnde_err),
y_ul=cf(self.dnde_ul),
significant=self.significant,
energy_units=self.energy_units_str,
flux_units=self.flux_units_str,
axes=axes,
**data_kwargs)
if filename is not None: P.savefig(filename)
return axes
def __init__(self, like, name, *args, **kwargs):
keyword_options.process(self, kwargs)
self.like = like
self.name = name
if self.ul_algorithm not in BaseGtlikeSED.ul_choices:
raise Exception("Upper Limit Algorithm %s not in %s" % (self.ul_algorithm,str(BaseGtlikeSED.ul_choices)))
if self.bin_edges is not None:
if not BaseGtlikeSED.good_binning(like, self.bin_edges):
raise Exception("bin_edges is not commensurate with the underlying energy binning of pyLikelihood.")
self.bin_edges = np.asarray(self.bin_edges)
energy = np.sqrt(self.bin_edges[1:]*self.bin_edges[:-1])
else:
# These energies are always in MeV
self.bin_edges = like.energies
energy = like.e_vals
self.lower=self.bin_edges[:-1]
self.upper=self.bin_edges[1:]
source=self.like.logLike.getSource(name)
self.init_spectrum=source.spectrum()
self.init_model=build_pointlike_model(self.init_spectrum)
self.results = dict(
Name=name,
spectrum=name_to_spectral_dict(like,name, errors=True, covariance_matrix=True),
)
self._calculate(like)
super(GtlikeSED,self).__init__(self.results, **keyword_options.defaults_to_kwargs(self, SED))
def __init__(self, like, name, bin_edges, **kwargs):
""" Parameters:
* like - pyLikelihood object
* name - source to make an SED for
* bin_edges - if specified, calculate the SED in these bins.
"""
keyword_options.process(self, kwargs)
self.like = like
self.name = name
if not BaseGtlikeSED.good_binning(self.like, bin_edges):
raise Exception("bin_edges is not commensurate with the underlying energy binning of pyLikelihood.")
source=self.like.logLike.getSource(name)
self.init_spectrum=source.spectrum()
self.init_model=build_pointlike_model(self.init_spectrum)
self.init_energes = self.like.energies[[0,-1]]
bin_edges = np.asarray(bin_edges)
self.lower_energy=bin_edges[:-1]
self.upper_energy=bin_edges[1:]
self.middle_energy=np.sqrt(self.lower_energy*self.upper_energy)
if self.ul_algorithm not in self.ul_choices:
raise Exception("Upper Limit Algorithm %s not in %s" % (self.ul_algorithm,str(self.ul_choices)))
empty = lambda: np.empty_like(self.middle_energy)
self._calculate()
def gtlike_analysis(roi, hypothesis, upper_limit=False, cutoff=False):
print 'Performing Gtlike crosscheck for %s' % hypothesis
gtlike=Gtlike(roi)
like=gtlike.like
like.fit(covar=True)
r=sourcedict(like, name)
if upper_limit:
r['upper_limit'] = powerlaw_upper_limit(like, name, emin=emin, emax=emax, cl=.95)
if cutoff:
r['test_cutoff']=test_cutoff(like,name)
for kind, kwargs in [['4bpd',dict(bin_edges=np.logspace(2,5,13))],
['1bpd',dict(bin_edges=np.logspace(2,5,4))]]:
print 'Making %s SED' % kind
sed = SED(like, name, **kwargs)
sed.plot('sed_gtlike_%s_%s_%s.png' % (kind,hypothesis,name))
sed.verbosity=True
sed.save('sed_gtlike_%s_%s_%s.dat' % (kind,hypothesis,name))
return r
def gtlike_analysis(roi, hypothesis, upper_limit=False, cutoff=False):
print "Performing Gtlike crosscheck for %s" % hypothesis
gtlike = Gtlike(roi)
like = gtlike.like
like.fit(covar=True)
r = sourcedict(like, name)
if upper_limit:
r["upper_limit"] = powerlaw_upper_limit(like, name, emin=emin, emax=emax, cl=0.95)
if cutoff:
r["test_cutoff"] = test_cutoff(like, name)
for kind, kwargs in [
["4bpd", dict(bin_edges=np.logspace(2, 5, 13))],
["1bpd", dict(bin_edges=np.logspace(2, 5, 4))],
]:
print "Making %s SED" % kind
sed = SED(like, name, **kwargs)
sed.plot("sed_gtlike_%s_%s_%s.png" % (kind, hypothesis, name))
sed.verbosity = True
sed.save("sed_gtlike_%s_%s_%s.dat" % (kind, hypothesis, name))
return r
def initialize(self):
if (self.R is None): self.R = self.getRad(self.logg, self.m)
if (self.T is None): self.T = self.getTeff(self.L, self.R)
if (self.logg is None):
self.logg = self.getlogg(self.m, self.L, self.T)
#one option for getting the extinction
if (self.AV is None):
count = 0
while (self.AV is None and count < 100):
count += 1
self.AV = extinction.get_AV_infinity(self.RA,
self.Dec,
frame='icrs')
if (self.AV is None):
print("WARNING: No AV found", self.RA, self.Dec, self.AV,
count)
time.sleep(30)
#initialize the SED
self.SED = SED()
self.SED.filters = self.filters
self.SED.filterFilesRoot = self.filterFilesRoot
self.SED.T = self.T * units.K
self.SED.R = self.R * units.solRad
self.SED.L = self.L * units.solLum
self.SED.logg = self.logg
self.SED.M_H = self.M_H
self.SED.EBV = self.AV / self.RV #could use this to account for reddening in SED
self.SED.initialize()
#one option for getting the extinction
ext = F04(Rv=self.RV)
Lconst = self.SED.getLconst()
for f in self.filters:
self.Ared[f] = ext(self.wavelength[f] * units.nm) * self.AV
self.Fv[f] = self.SED.getFvAB(self.dist * units.kpc,
f,
Lconst=Lconst)
self.appMagMean[f] = -2.5 * np.log10(
self.Fv[f]) + self.Ared[f] #AB magnitude
def __init__(self, like, name, *args, **kwargs):
keyword_options.process(self, kwargs)
self.like = like
self.name = name
if self.ul_algorithm not in BaseGtlikeSED.ul_choices:
raise Exception("Upper Limit Algorithm %s not in %s" %
(self.ul_algorithm, str(BaseGtlikeSED.ul_choices)))
if self.bin_edges is not None:
if not BaseGtlikeSED.good_binning(like, self.bin_edges):
raise Exception(
"bin_edges is not commensurate with the underlying energy binning of pyLikelihood."
)
self.bin_edges = np.asarray(self.bin_edges)
energy = np.sqrt(self.bin_edges[1:] * self.bin_edges[:-1])
else:
# These energies are always in MeV
self.bin_edges = like.energies
energy = like.e_vals
self.lower = self.bin_edges[:-1]
self.upper = self.bin_edges[1:]
source = self.like.logLike.getSource(name)
self.init_spectrum = source.spectrum()
self.init_model = build_pointlike_model(self.init_spectrum)
self.results = dict(
Name=name,
spectrum=name_to_spectral_dict(like,
name,
errors=True,
covariance_matrix=True),
)
self._calculate(like)
super(GtlikeSED,
self).__init__(self.results,
**keyword_options.defaults_to_kwargs(self, SED))
def define_model(self):
self.user_indices = tf.placeholder(tf.int32,
shape=[None],
name='user_indices')
self.user_cdcf_indices = tf.placeholder(tf.int32,
shape=[None],
name='user_cdcf_indices')
self.item_indices = tf.placeholder(tf.int32,
shape=[None],
name='item_indices')
self.true_rating = tf.placeholder(tf.float32,
shape=[None],
name='true_ratings')
self.keep_prob = tf.placeholder(tf.float32, name='keep_prob')
self.keep_prob_layer = tf.placeholder(tf.float32,
name='keep_prob_layer')
self.domain_indices = tf.placeholder(tf.int32,
shape=[None],
name='domain_indices')
self.valid_clip = tf.placeholder(tf.float32, name='valid_clip')
self.mu_embeddings = tf.Variable(tf.constant(
[self.mu_target, self.mu_source], dtype=tf.float32),
dtype=tf.float32,
name='mu_embedding',
trainable=False)
self.user_bias_embeddings = tf.Variable(
self.initializer(shape=[self.num_users]),
dtype=tf.float32,
name='user_bias_embedding')
self.user_bias_cdcf_embeddings = tf.Variable(
tf.zeros(shape=[2 * self.num_users]),
dtype=tf.float32,
name='user_bias_embedding')
self.item_bias_embeddings = tf.Variable(
self.initializer(shape=[self.num_items]),
dtype=tf.float32,
name='item_bias_embedding')
self.ubias_embeds = tf.nn.embedding_lookup(self.user_bias_embeddings,
self.user_indices)
self.ubias_cdcf_embeds = tf.nn.embedding_lookup(
self.user_bias_cdcf_embeddings, self.user_cdcf_indices)
self.ibias_embeds = tf.nn.embedding_lookup(self.item_bias_embeddings,
self.item_indices)
self.mu_embeds = tf.nn.embedding_lookup(self.mu_embeddings,
self.domain_indices)
# create gmf object and define gmf model
if self.method == 'gcmf':
self.gcmf_model = GCMF(self.params)
self.gcmf_model.define_model(self.user_indices,
self.user_cdcf_indices,
self.item_indices, self.true_rating,
self.keep_prob, self.keep_prob_layer,
self.domain_indices)
self.pred_rating_model = self.gcmf_model.pred_rating
elif self.method == 'mlp':
self.mlp_model = MLP(self.params)
self.mlp_model.define_model(self.user_indices,
self.user_cdcf_indices,
self.item_indices, self.true_rating,
self.keep_prob, self.keep_prob_layer,
self.domain_indices)
self.pred_rating_model = self.mlp_model.pred_rating
elif self.method == 'sed':
self.sed_model = SED(self.params)
self.sed_model.define_model(self.user_indices,
self.user_cdcf_indices,
self.item_indices, self.true_rating,
self.keep_prob, self.keep_prob_layer,
self.domain_indices)
self.pred_rating_model = self.sed_model.pred_rating
elif self.method == 'gcmf_mlp':
self.gcmf_mlp_model = GCMF_MLP(self.params)
self.gcmf_mlp_model.define_model(self.user_indices,
self.user_cdcf_indices,
self.item_indices,
self.true_rating, self.keep_prob,
self.keep_prob_layer,
self.domain_indices)
self.pred_rating_model = self.gcmf_mlp_model.pred_rating
#elif self.method == 'gcmf_sed':
elif self.method == 'gcmf_sed':
self.gcmf_sed_model = GCMF_SED(self.params)
self.gcmf_sed_model.define_model(self.user_indices,
self.user_cdcf_indices,
self.item_indices,
self.true_rating, self.keep_prob,
self.keep_prob_layer,
self.domain_indices)
self.pred_rating_model = self.gcmf_sed_model.pred_rating
elif self.method == 'gcmf_mlp_sed':
self.gcmf_mlp_sed_model = GCMF_MLP_SED(self.params)
self.gcmf_mlp_sed_model.define_model(
self.user_indices, self.user_cdcf_indices, self.item_indices,
self.true_rating, self.keep_prob, self.keep_prob_layer,
self.domain_indices)
self.pred_rating_model = self.gcmf_mlp_sed_model.pred_rating
#====
'''
self.pred_rating = (self.pred_rating_model +
self.ubias_embeds + self.ubias_cdcf_embeds +
self.ibias_embeds +self.mu_embeds)
'''
self.pred_rating = (self.pred_rating_model)
self.pred_rating = (
(1 - self.valid_clip) * self.pred_rating +
self.valid_clip * tf.clip_by_value(
self.pred_rating, self.rating_min_val, self.rating_max_val))
print('PRED_RATING : with bias')
class NeuCDCF(object):
def __init__(self, params):
self.num_users = params.num_users
self.num_items = params.num_items
self.method = params.method
self.reg_bias = params.reg_bias
self.params = params
self.mu = params.mu
self.mu_source = params.mu_source
self.mu_target = params.mu_target
self.initializer = params.initializer
#self.valid_clip = params.valid_clip
self.rating_min_val = params.rating_min_val
self.rating_max_val = params.rating_max_val
def define_model(self):
self.user_indices = tf.placeholder(tf.int32,
shape=[None],
name='user_indices')
self.user_cdcf_indices = tf.placeholder(tf.int32,
shape=[None],
name='user_cdcf_indices')
self.item_indices = tf.placeholder(tf.int32,
shape=[None],
name='item_indices')
self.true_rating = tf.placeholder(tf.float32,
shape=[None],
name='true_ratings')
self.keep_prob = tf.placeholder(tf.float32, name='keep_prob')
self.keep_prob_layer = tf.placeholder(tf.float32,
name='keep_prob_layer')
self.domain_indices = tf.placeholder(tf.int32,
shape=[None],
name='domain_indices')
self.valid_clip = tf.placeholder(tf.float32, name='valid_clip')
self.mu_embeddings = tf.Variable(tf.constant(
[self.mu_target, self.mu_source], dtype=tf.float32),
dtype=tf.float32,
name='mu_embedding',
trainable=False)
self.user_bias_embeddings = tf.Variable(
self.initializer(shape=[self.num_users]),
dtype=tf.float32,
name='user_bias_embedding')
self.user_bias_cdcf_embeddings = tf.Variable(
tf.zeros(shape=[2 * self.num_users]),
dtype=tf.float32,
name='user_bias_embedding')
self.item_bias_embeddings = tf.Variable(
self.initializer(shape=[self.num_items]),
dtype=tf.float32,
name='item_bias_embedding')
self.ubias_embeds = tf.nn.embedding_lookup(self.user_bias_embeddings,
self.user_indices)
self.ubias_cdcf_embeds = tf.nn.embedding_lookup(
self.user_bias_cdcf_embeddings, self.user_cdcf_indices)
self.ibias_embeds = tf.nn.embedding_lookup(self.item_bias_embeddings,
self.item_indices)
self.mu_embeds = tf.nn.embedding_lookup(self.mu_embeddings,
self.domain_indices)
# create gmf object and define gmf model
if self.method == 'gcmf':
self.gcmf_model = GCMF(self.params)
self.gcmf_model.define_model(self.user_indices,
self.user_cdcf_indices,
self.item_indices, self.true_rating,
self.keep_prob, self.keep_prob_layer,
self.domain_indices)
self.pred_rating_model = self.gcmf_model.pred_rating
elif self.method == 'mlp':
self.mlp_model = MLP(self.params)
self.mlp_model.define_model(self.user_indices,
self.user_cdcf_indices,
self.item_indices, self.true_rating,
self.keep_prob, self.keep_prob_layer,
self.domain_indices)
self.pred_rating_model = self.mlp_model.pred_rating
elif self.method == 'sed':
self.sed_model = SED(self.params)
self.sed_model.define_model(self.user_indices,
self.user_cdcf_indices,
self.item_indices, self.true_rating,
self.keep_prob, self.keep_prob_layer,
self.domain_indices)
self.pred_rating_model = self.sed_model.pred_rating
elif self.method == 'gcmf_mlp':
self.gcmf_mlp_model = GCMF_MLP(self.params)
self.gcmf_mlp_model.define_model(self.user_indices,
self.user_cdcf_indices,
self.item_indices,
self.true_rating, self.keep_prob,
self.keep_prob_layer,
self.domain_indices)
self.pred_rating_model = self.gcmf_mlp_model.pred_rating
#elif self.method == 'gcmf_sed':
elif self.method == 'gcmf_sed':
self.gcmf_sed_model = GCMF_SED(self.params)
self.gcmf_sed_model.define_model(self.user_indices,
self.user_cdcf_indices,
self.item_indices,
self.true_rating, self.keep_prob,
self.keep_prob_layer,
self.domain_indices)
self.pred_rating_model = self.gcmf_sed_model.pred_rating
elif self.method == 'gcmf_mlp_sed':
self.gcmf_mlp_sed_model = GCMF_MLP_SED(self.params)
self.gcmf_mlp_sed_model.define_model(
self.user_indices, self.user_cdcf_indices, self.item_indices,
self.true_rating, self.keep_prob, self.keep_prob_layer,
self.domain_indices)
self.pred_rating_model = self.gcmf_mlp_sed_model.pred_rating
#====
'''
self.pred_rating = (self.pred_rating_model +
self.ubias_embeds + self.ubias_cdcf_embeds +
self.ibias_embeds +self.mu_embeds)
'''
self.pred_rating = (self.pred_rating_model)
self.pred_rating = (
(1 - self.valid_clip) * self.pred_rating +
self.valid_clip * tf.clip_by_value(
self.pred_rating, self.rating_min_val, self.rating_max_val))
print('PRED_RATING : with bias')
def define_loss(self, loss_type='all'): # no_reg ##
self.recon_error = tf.constant(0.0, dtype=tf.float32)
if self.method == 'gcmf':
self.gcmf_model.define_loss(loss_type=loss_type)
self.regularization_loss = self.gcmf_model.regularization_loss
elif self.method == 'mlp':
self.mlp_model.define_loss(loss_type=loss_type)
self.regularization_loss = self.mlp_model.regularization_loss
elif self.method == 'sed':
self.sed_model.define_loss(loss_type=loss_type)
self.regularization_loss = self.sed_model.regularization_loss
self.recon_error += self.sed_model.recon_error
elif self.method == 'gcmf_mlp':
self.gcmf_mlp_model.define_loss(loss_type=loss_type)
self.regularization_loss = self.gcmf_mlp_model.regularization_loss
elif self.method == 'gcmf_sed':
self.gcmf_sed_model.define_loss(loss_type=loss_type)
self.regularization_loss = self.gcmf_sed_model.regularization_loss
self.recon_error += self.gcmf_sed_model.recon_error
elif self.method == 'gcmf_mlp_sed':
self.gcmf_mlp_sed_model.define_loss(loss_type=loss_type)
self.regularization_loss = self.gcmf_mlp_sed_model.regularization_loss
self.recon_error += self.gcmf_mlp_sed_model.recon_error
self.mse_loss = tf.losses.mean_squared_error(self.true_rating,
self.pred_rating)
self.mae_loss = tf.losses.absolute_difference(self.true_rating,
self.pred_rating)
self.se_loss = tf.reduce_sum(
tf.squared_difference(self.true_rating, self.pred_rating))
self.regularization_loss = (
self.regularization_loss + self.reg_bias *
(tf.nn.l2_loss(self.ubias_embeds) + tf.nn.l2_loss(
self.ubias_cdcf_embeds) + tf.nn.l2_loss(self.ibias_embeds)))
#if loss_type == 'all':
self.loss = self.mse_loss + self.regularization_loss + self.recon_error
def getrMagSingle(L1, T1, g1, r1, M_H, dist, AV, extVal):
SED1 = SED()
SED1.filters = ['r_']
SED1.filterFilesRoot = filterFilesRoot
SED1.T = T1 * units.K
SED1.R = r1 * units.solRad
SED1.L = L1 * units.solLum
SED1.logg = g1
SED1.M_H = M_H
SED1.EBV = AV / RV #could use this to account for reddening in SED
SED1.initialize()
Lconst1 = SED1.getLconst()
Ared = extVal * AV
Fv = SED1.getFvAB(dist * units.kpc, 'r_', Lconst=Lconst1)
return -2.5 * np.log10(Fv) + Ared #AB magnitude
def getrMagBinary(L1, T1, g1, r1, L2, T2, g2, r2, M_H, dist, AV, extVal):
SED1 = SED()
SED1.filters = ['r_']
SED1.filterFilesRoot = filterFilesRoot
SED1.T = T1 * units.K
SED1.R = r1 * units.solRad
SED1.L = L1 * units.solLum
SED1.logg = g1
SED1.M_H = M_H
SED1.EBV = AV / RV #could use this to account for reddening in SED
SED1.initialize()
SED2 = SED()
SED2.filters = ['r_']
SED2.filterFilesRoot = filterFilesRoot
SED2.T = T2 * units.K
SED2.R = r2 * units.solRad
SED2.L = L2 * units.solLum
SED2.logg = g2
SED2.M_H = M_H
SED2.EBV = AV / RV #could use this to account for reddening in SED
SED2.initialize()
Lconst1 = SED1.getLconst()
Lconst2 = SED2.getLconst()
Ared = extVal * AV
Fv1 = SED1.getFvAB(dist * units.kpc, 'r_', Lconst=Lconst1)
Fv2 = SED2.getFvAB(dist * units.kpc, 'r_', Lconst=Lconst2)
Fv = Fv1 + Fv2
return -2.5 * np.log10(Fv) + Ared #AB magnitude
flux_densities = False
redshifts = [3.00,2.794,2.541,2.43,1.759,1.411,2.59,1.552,0.667,2.086,1.996,5.000,2.497,0.769,1.721,1.314]
c = const.c.value
wavelength = 1.3e-3
freq = c/wavelength
if flux_densities:
N = len(chain[0][int(burn_in*len(chain[0])):])
Temps, C_vals = [],[]
for i in range(16):
Temps.append(chain[2*i][int(burn_in*len(chain[2*i])):])
C_vals.append(chain[(2*i)+1][int(burn_in*(len(chain[(2*i)+1]))):])
SED_chain = [[] for i in range(16)]
for i in range(int(N)):
for j in range(16):
SED_chain[j].append(SED(freq,Temps[j][i],1.5,redshifts[j],C_vals[j][i],2.0, nu_cut(Temps[j][i],1.5,redshifts[j],C_vals[j][i],2.0)))
for i in SED_chain:
print "Maximum Likelihood Value and 68% Confidence Interval: ", statistics_hist(i,0.68,"stats")
#plt.figure()
#plt.hist(SED_chain[0],np.linspace(min(SED_chain[0]),max(SED_chain[0]),50))
#plt.xlabel("Flux Density (Jy)")
#plt.show()
#************************************************************************
"""
Star Formation Rate
- testing not adequate at the moment: wait for longer chain.
"""
from scipy.integrate import simps
like.fit(covar=True)
r['gtlike']=sourcedict(like,name,'_at_pulsar')
# calculate gtlike upper limits
# calculate TScutoff
roi.save('roi_%s.dat' % name)
open('results_%s.yaml' % name,'w').write(
yaml.dump(
tolist(results)
)
)
# save stuff out
roi.plot_tsmap(filename='residual_tsmap_%s.pdf' % name, size=8)
roi.zero_source(which=name)
roi.plot_tsmap(filename='source_tsmap_%s.pdf' % name, size=8)
roi.unzero_source(which=name)
roi.plot_sources(filename='sources_%s.pdf' % name, size=8, label_psf=False)
sed = SED(like,name)
sed.save('sed_gtlike_%s.dat' % name)
sed.plot('sed_gtlike_%s.png' % name)
class SingleStar(object):
def __init__(self, *args, **kwargs):
self.SED = None
self.m = None #*units.solMass
self.R = None #*units.solRad
self.L = None #*units.solLum
self.M_H = 0. #metallicity
self.AV = None
self.RA = None
self.Dec = None
self.dist = None #*units.kpc
#these will be calculated
self.T = None #Kelvin
self.logg = None
self.Fv = dict()
self.Ared = dict()
self.BC = dict()
self.appMagMean = dict()
#don't touch these
self.filters = ['u_', 'g_', 'r_', 'i_', 'z_', 'y_']
self.filterFilesRoot = '../input/filters/'
self.RV = 3.1
#from https://www.lsst.org/scientists/keynumbers
#in nm
self.wavelength = {
'u_': (324. + 395.) / 2.,
'g_': (405. + 552.) / 2.,
'r_': (552. + 691.) / 2.,
'i_': (691. + 818.) / 2.,
'z_': (818. + 921.) / 2.,
'y_': (922. + 997.) / 2.
}
#also in EclipsingBinary
def getTeff(self, L, R):
#use stellar radius and stellar luminosity to get the star's effective temperature
logTeff = 3.762 + 0.25 * np.log10(L) - 0.5 * np.log10(R)
return 10.**logTeff
#Some approximate function for deriving stellar parameters
def getRad(self, logg, m):
#g = GM/r**2
g = 10.**logg * units.cm / units.s**2.
r = ((constants.G * m * units.Msun / g)**0.5).decompose().to(
units.Rsun).value
return r
def getlogg(self, m, L, T):
#use stellar mass, luminosity, and effective temperature to get log(gravity)
return np.log10(m) + 4. * np.log10(T) - np.log10(L) - 10.6071
def initialize(self):
if (self.R is None): self.R = self.getRad(self.logg, self.m)
if (self.T is None): self.T = self.getTeff(self.L, self.R)
if (self.logg is None):
self.logg = self.getlogg(self.m, self.L, self.T)
#one option for getting the extinction
if (self.AV is None):
count = 0
while (self.AV is None and count < 100):
count += 1
self.AV = extinction.get_AV_infinity(self.RA,
self.Dec,
frame='icrs')
if (self.AV is None):
print("WARNING: No AV found", self.RA, self.Dec, self.AV,
count)
time.sleep(30)
#initialize the SED
self.SED = SED()
self.SED.filters = self.filters
self.SED.filterFilesRoot = self.filterFilesRoot
self.SED.T = self.T * units.K
self.SED.R = self.R * units.solRad
self.SED.L = self.L * units.solLum
self.SED.logg = self.logg
self.SED.M_H = self.M_H
self.SED.EBV = self.AV / self.RV #could use this to account for reddening in SED
self.SED.initialize()
#one option for getting the extinction
ext = F04(Rv=self.RV)
Lconst = self.SED.getLconst()
for f in self.filters:
self.Ared[f] = ext(self.wavelength[f] * units.nm) * self.AV
self.Fv[f] = self.SED.getFvAB(self.dist * units.kpc,
f,
Lconst=Lconst)
self.appMagMean[f] = -2.5 * np.log10(
self.Fv[f]) + self.Ared[f] #AB magnitude
def initialize(self):
#should I initialize the seed here?
#No I am initializing it in LSSTEBworker.py
#self.initializeSeed()
self.q = self.m2 / self.m1
self.a = self.getafromP(self.m1 * units.solMass,
self.m2 * units.solMass,
self.period * units.day).to(units.solRad).value
self.R_1 = self.r1 / self.a
self.R_2 = self.r2 / self.a
self.R_1e = self.r1 / self.Eggleton_RL(
self.m1 / self.m2, self.a * (1. - self.eccentricity))
self.R_2e = self.r2 / self.Eggleton_RL(
self.m2 / self.m1, self.a * (1. - self.eccentricity))
self.f_c = np.sqrt(self.eccentricity) * np.cos(
self.omega * np.pi / 180.)
self.f_s = np.sqrt(self.eccentricity) * np.sin(
self.omega * np.pi / 180.)
if (self.RA is None):
coord = SkyCoord(x=self.xGx,
y=self.yGx,
z=self.zGx,
unit='pc',
representation='cartesian',
frame='galactocentric')
self.RA = coord.icrs.ra.to(units.deg).value
self.Dec = coord.icrs.dec.to(units.deg).value
for f in self.filters:
self.appMagMean[f] = -999.
self.deltaMag[f] = None
self.LSS[f] = -999.
self.appMagObs[f] = [None]
self.appMagObsErr[f] = [None]
self.deltaMag[f] = 0.
self.obsDates[f] = [None]
self.eclipseDepthFrac[f] = 0.
self.appMagMean[
'r_'] = self.TRILEGALrmag #for binaries that don't pass the preCheck
self.maxDeltaMag = 0.
self.preCheckIfObservable()
if (self.observable):
if (self.T1 is None): self.T1 = self.getTeff(self.L1, self.r1)
if (self.T2 is None): self.T2 = self.getTeff(self.L2, self.r2)
if (self.g1 is None):
self.g1 = self.getlogg(self.m1, self.L1, self.T1)
if (self.g2 is None):
self.g2 = self.getlogg(self.m2, self.L2, self.T2)
self.sbratio = (self.L2 / self.r2**2.) / (self.L1 / self.r1**2.)
#one option for getting the extinction
if (self.AV is None):
count = 0
while (self.AV is None and count < 100):
count += 1
self.AV = extinction.get_AV_infinity(self.RA,
self.Dec,
frame='icrs')
if (self.AV is None):
print("WARNING: No AV found", self.RA, self.Dec,
self.AV, count)
time.sleep(30)
self.SED1 = SED()
self.SED1.filters = self.filters
self.SED1.filterFilesRoot = self.filterFilesRoot
self.SED1.T = self.T1 * units.K
self.SED1.R = self.r1 * units.solRad
self.SED1.L = self.L1 * units.solLum
self.SED1.logg = self.g1
self.SED1.M_H = self.M_H
self.SED1.EBV = self.AV / self.RV #could use this to account for reddening in SED
self.SED1.initialize()
self.SED2 = SED()
self.SED2.filters = self.filters
self.SED2.filterFilesRoot = self.filterFilesRoot
self.SED2.T = self.T2 * units.K
self.SED2.R = self.r2 * units.solRad
self.SED2.L = self.L2 * units.solLum
self.SED2.logg = self.g2
self.SED2.M_H = self.M_H
self.SED2.EBV = self.AV / self.RV #could use this to account for reddening in SED
self.SED2.initialize()
#estimate a combined Teff value, as I do in the N-body codes (but where does this comes from?)
logLb = np.log10(self.L1 + self.L2)
logRb = 0.5 * np.log10(self.r1**2. + self.r2**2.)
self.T12 = 10.**(3.762 + 0.25 * logLb - 0.5 * logRb)
#print(self.L1, self.L2, self.T1, self.T2, self.T12)
#account for reddening and the different filter throughput functions (currently related to a blackbody)
self.appMagMeanAll = 0.
#one option for getting the extinction
#ext = F04(Rv=self.RV)
ext = F04(Rv=self.RV)
#a check
# self.Ltest = self.SED.getL(self.T1*units.K, self.r1*units.solRad)
#definitely necessary for Kurucz because these models are not normalized
#for the bb I'm getting a factor of 2 difference for some reason
Lconst1 = self.SED1.getLconst()
Lconst2 = self.SED2.getLconst()
#print(np.log10(Lconst1), np.log10(Lconst2))
for f in self.filters:
#print(extinction.fitzpatrick99(np.array([self.wavelength[f]*10.]), self.AV, self.RV, unit='aa')[0] , ext(self.wavelength[f]*units.nm)*self.AV)
#self.Ared[f] = extinction.fitzpatrick99(np.array([self.wavelength[f]*10.]), self.AV, self.RV, unit='aa')[0] #or ccm89
self.Ared[f] = ext(self.wavelength[f] * units.nm) * self.AV
self.Fv1[f] = self.SED1.getFvAB(self.dist * units.kpc,
f,
Lconst=Lconst1)
self.Fv2[f] = self.SED2.getFvAB(self.dist * units.kpc,
f,
Lconst=Lconst2)
Fv = self.Fv1[f] + self.Fv2[f]
self.appMagMean[f] = -2.5 * np.log10(Fv) + self.Ared[
f] #AB magnitude
#print(self.wavelength[f], self.appMagMean[f], self.Ared[f], self.T1)
self.appMagMeanAll += self.appMagMean[f]
self.appMagMeanAll /= len(self.filters)
#check if we can observe this based on the magnitude
self.magCheckIfObservable()
#if we're using OpSim, then get the field ID
#get the field ID number from OpSim where this binary would be observed
if (self.useOpSimDates and self.observable
and self.OpSim.fieldID[0] is None):
self.OpSim.setFieldID(self.RA, self.Dec)
#this now covers the galaxy and the cluster, if available
if (self.observable and self.crowding is not None):
self.crowding.input_rFlux = self.Fv1['r_'] + self.Fv2[
'r_'] #NOTE; if I don't have a r_ band, this will crash
self.crowding.getCrowding()
for f in self.filters:
self.light_3[f] = self.crowding.backgroundFlux[f] / (
self.Fv1[f] + self.Fv2[f])
class EclipsingBinary(object):
def __init__(self, *args, **kwargs):
self.MbolSun = 4.73 #as "recommended" for Flower's bolometric correction in http://iopscience.iop.org/article/10.1088/0004-6256/140/5/1158/pdf
#these is defined by the user
self.m1 = None #*units.solMass
self.m2 = None #*units.solMass
self.r1 = None #*units.solRad
self.r2 = None #*units.solRad
self.L1 = None #*units.solLum
self.L2 = None #*units.solLum
self.k1 = None #BSE stellar type
self.k2 = None #BSE stellar type
self.period = None #*units.day
self.eccentricity = None
self.OMEGA = None
self.omega = None
self.inclination = None
self.t_zero = None
self.dist = None #*units.kpc
self.xGx = None #*units.parsec
self.yGx = None #*units.parsec
self.zGx = None #*units.parsec
self.verbose = False
self.RV = 3.1
self.M_H = 0. #metallicity
self.TRILEGALrmag = -999. #r magnitude from TRILEGAL
#for SED
self.filterFilesRoot = '../input/filters/'
self.SED1 = None
self.SED2 = None
#for Galaxy
self.Galaxy = None
self.SEDsingle = None
#for star cluster crowding model
self.crowding = None
#from https://www.lsst.org/scientists/keynumbers
#in nm
self.wavelength = {
'u_': (324. + 395.) / 2.,
'g_': (405. + 552.) / 2.,
'r_': (552. + 691.) / 2.,
'i_': (691. + 818.) / 2.,
'z_': (818. + 921.) / 2.,
'y_': (922. + 997.) / 2.
}
#these will be calculated after calling self.initialize()
self.RL1 = None
self.RL2 = None
self.T1 = None
self.T2 = None
self.T12 = None
self.L1 = None
self.L2 = None
self.g1 = None
self.g2 = None
self.a = None
self.q = None
self.f_c = None
self.f_s = None
self.R_1 = None
self.R_2 = None
self.sbratio = None
self.RA = None
self.Dec = None
self.Mbol = None
self.AV = None
self.appMagMean = dict()
self.Fv1 = dict()
self.Fv2 = dict()
self.appMagMeanAll = None
self.Ared = dict()
self.BC = dict()
#for light curves
self.SED = None
self.OpSim = None
self.filters = ['u_', 'g_', 'r_', 'i_', 'z_', 'y_']
self.ld_1 = 'claret'
self.ld_2 = 'claret'
self.grid_1 = 'default'
self.grid_2 = 'default'
self.shape_1 = 'sphere'
self.shape_2 = 'sphere'
self.sigma_sys = 0.005 #systematic photometric error
self.obsDates = dict()
self.m_5 = dict()
self.appMag = dict()
self.appMagObs = dict()
self.appMagObsErr = dict()
self.deltaMag = dict()
self.eclipseDepthFrac = dict()
self.maxDeltaMag = 0.
self.useOpSimDates = True
self.observable = True
self.appmag_failed = 0
self.eclipseDepth_failed = 0
self.incl_failed = 0
self.period_failed = 0
self.radius_failed = 0
self.OpSimi = 0
self.years = 10.
self.totaltime = 365. * self.years
self.cadence = 3.
self.Nfilters = 6.
self.nobs = 0
self.light_3 = {}
for f in self.filters:
self.light_3[f] = 0.
#this is for the magnitude uncertainties
#https://arxiv.org/pdf/0805.2366.pdf
self.sigmaDict = {
'u_': {
'gamma': 0.037,
'seeing': 0.77,
'm_sky': 22.9,
'C_m': 22.92,
'k_m': 0.451
},
'g_': {
'gamma': 0.038,
'seeing': 0.73,
'm_sky': 22.3,
'C_m': 24.29,
'k_m': 0.163
},
'r_': {
'gamma': 0.039,
'seeing': 0.70,
'm_sky': 21.2,
'C_m': 24.33,
'k_m': 0.087
},
'i_': {
'gamma': 0.039,
'seeing': 0.67,
'm_sky': 20.5,
'C_m': 24.20,
'k_m': 0.065
},
'z_': {
'gamma': 0.040,
'seeing': 0.65,
'm_sky': 19.6,
'C_m': 24.07,
'k_m': 0.043
},
'y_': {
#from Ivezic et al 2008 - https://arxiv.org/pdf/0805.2366.pdf - Table 2 (p26)
'gamma': 0.0039,
'seeing':
0.65, #not sure where this is from - not in Ivezic; still the z value
'm_sky': 18.61,
'C_m': 23.73,
'k_m': 0.170
}
}
self.magLims = np.array(
[15.8, 25.]
) #lower and upper limits on the magnitude detection assumed for LSST: 15.8 = rband saturation from Science Book page 57, before Section 3.3; 24.5 is the desired detection limit
self.eclipseDepthLim = 3. #depth / error
#set within the "driver" code, for gatspy
self.LSS = dict()
self.LSSmodel = dict()
self.LSM = -999.
self.LSMmodel = None
self.seed = None
def getFlowerBCV(self, Teff):
#from http://iopscience.iop.org/article/10.1088/0004-6256/140/5/1158/pdf
#which updates/corrects from Flower, P. J. 1996, ApJ, 469, 355
lt = np.log10(Teff)
a = 0.
b = 0.
c = 0.
d = 0.
e = 0.
f = 0.
if (lt < 3.7):
a = -0.190537291496456 * 10.**5.
b = 0.155144866764412 * 10.**5.
c = -0.421278819301717 * 10.**4.
d = 0.381476328422343 * 10.**3.
if (lt >= 3.7 and lt < 3.9):
a = -0.370510203809015 * 10.**5.
b = 0.385672629965804 * 10.**5.
c = -0.150651486316025 * 10.**5.
d = 0.261724637119416 * 10.**4.
e = -0.170623810323864 * 10.**3.
if (lt >= 3.9):
a = -0.118115450538963 * 10.**6.
b = 0.137145973583929 * 10.**6.
c = -0.636233812100225 * 10.**5.
d = 0.147412923562646 * 10.**5.
e = -0.170587278406872 * 10.**4.
f = 0.788731721804990 * 10.**2.
BCV = a + b * lt + c * lt**2. + d * lt**3. + e * lt**4 + f * lt**5.
return BCV
#Some approximate function for deriving stellar parameters
def getRad(self, logg, m):
#g = GM/r**2
g = 10.**logg * units.cm / units.s**2.
r = ((constants.G * m * units.Msun / g)**0.5).decompose().to(
units.Rsun).value
return r
def getRadOld(self, m):
#(not needed with Katie's model, but included here in case needed later)
#use stellar mass to get stellar radius (not necessary, read out by Katie's work)
if (m > 1): #*units.solMass
eta = 0.57
else:
eta = 0.8
return (m)**eta #* units.solRad (units.solMass)
def getTeff(self, L, R):
#use stellar radius and stellar luminosity to get the star's effective temperature
logTeff = 3.762 + 0.25 * np.log10(L) - 0.5 * np.log10(R)
return 10.**logTeff
def getlogg(self, m, L, T):
#use stellar mass, luminosity, and effective temperature to get log(gravity)
return np.log10(m) + 4. * np.log10(T) - np.log10(L) - 10.6071
def getLum(self, m):
#https://en.wikipedia.org/wiki/Mass%E2%80%93luminosity_relation
#(not needed with Katie's model, but included here in case needed later)
#use stellar mass to return stellar luminosity (not necessary, read out by Katie's work)
if (m < 0.43):
cons = 0.23
coeff = 2.3
if (m >= 0.43 and m < 2.0):
cons = 1
coeff = 4
if (m >= 2.0 and m < 55.0):
cons = 1.4
coeff = 3.5
if (m >= 55):
cons = 32000
coeff = 1
return cons * (m**coeff)
def getafromP(self, m1, m2, P):
#returns the semimajor axis from the period and stellar masses
return (((P**2.) * constants.G * (m1 + m2) /
(4 * np.pi**2.))**(1. / 3.)).decompose().to(units.AU)
def Eggleton_RL(self, q, a):
#Eggleton (1983) Roche Lobe
#assuming synchronous rotation
#but taking the separation at pericenter
return a * 0.49 * q**(2. / 3.) / (0.6 * q**(2. / 3.) +
np.log(1. + q**(1. / 3.)))
def setLightCurve(self,
filt,
t_vis=30.,
X=1.,
useDates=[None],
useT0=None,
light_3=None):
def getSig2Rand(filt, magnitude, m_5=[None]):
#returns 2 sigma random error based on the pass band (y-values may be wonky - need to check for seeing and
# against others)
#X = 1. #function of distance??
#t_vis = 30. #seconds
if (m_5[0] is None):
m_5 = self.sigmaDict[filt]['C_m'] + (
0.50 * (self.sigmaDict[filt]['m_sky'] - 21.)
) + (2.50 * np.log10(0.7 / self.sigmaDict[filt]['seeing'])) + (
1.25 *
np.log10(t_vis / 30.)) - (self.sigmaDict[filt]['k_m'] *
(X - 1.))
return (0.04 - self.sigmaDict[filt]['gamma']) * (10**(
0.4 * (magnitude - m_5))) + self.sigmaDict[filt]['gamma'] * (
(10**(0.4 * (magnitude - m_5)))**2) * (magnitude**2)
# Function to get y-band LDCs for any Teff, logg, M_H
# written by Andrew Bowen, Northwestern undergraduate, funded by LSSTC grant (summer 2018)
def get_y_LDC(Teff, logg, M_H):
# All filters/wavelength arrays
SDSSfilters = ['u_', 'g_', 'r_', 'i_', 'z_', "J", 'H',
"K"] #Only 2MASS/SDSS filters (8 in total)
SDSSwavelength = np.array(
[354, 464, 621.5, 754.5, 870, 1220, 1630, 2190])
y_wavelength = np.array(1004)
# Getting coefficients from ELLC and appending them to specific coeff arrays
a1_array = np.array([])
a2_array = np.array([])
a3_array = np.array([])
a4_array = np.array([])
# Gets LDCs for all filters
for w, f in zip(SDSSwavelength, SDSSfilters):
ldy_filt = ellc.ldy.LimbGravityDarkeningCoeffs(f)
a1, a2, a3, a4, y = ldy_filt(Teff, logg, M_H)
a1_array = np.append(a1_array, a1)
a2_array = np.append(a2_array, a2)
a3_array = np.append(a3_array, a3)
a4_array = np.append(a4_array, a4)
# Sets up interpolation for y-band for each coeff
find_y_a1 = np.interp(y_wavelength, SDSSwavelength, a1_array)
find_y_a2 = np.interp(y_wavelength, SDSSwavelength, a2_array)
find_y_a3 = np.interp(y_wavelength, SDSSwavelength, a3_array)
find_y_a4 = np.interp(y_wavelength, SDSSwavelength, a4_array)
return find_y_a1, find_y_a2, find_y_a3, find_y_a4
#in case the user did not initialize
if (self.T1 is None):
self.initialize()
#limb darkenning
# T1 = np.clip(self.T1, 3500., 50000.)
# T2 = np.clip(self.T2, 3500., 50000.)
# g1 = np.clip(self.g1, 0., 5.)
# g2 = np.clip(self.g2, 0., 5.)
# MH = np.clip(self.M_H, -2.5, 0.5)
#there is a complicated exclusion region in the limb darkening. See testing/limbDarkening/checkClaret.ipynb .
#I could possibly account for that, but for now I will simply not use limb darkening in those cases.
T1 = np.clip(self.T1, 3500., 40000.)
T2 = np.clip(self.T2, 3500., 40000.)
g1 = np.clip(self.g1, 0., 5.)
g2 = np.clip(self.g2, 0., 5.)
MH = np.clip(self.M_H, -5, 1.)
# print(T1, T2, g1, g2, self.g1, self.g2, self.M_H)
if (filt == 'y_'):
a1_1, a2_1, a3_1, a4_1 = get_y_LDC(T1, g1, MH)
a1_2, a2_2, a3_2, a4_2 = get_y_LDC(T2, g2, MH)
else:
ldy_filt = ellc.ldy.LimbGravityDarkeningCoeffs(filt)
a1_1, a2_1, a3_1, a4_1, y = ldy_filt(T1, g1, MH)
a1_2, a2_2, a3_2, a4_2, y = ldy_filt(T2, g2, MH)
ldc_1 = [a1_1, a2_1, a3_1, a4_1]
ldc_2 = [a1_2, a2_2, a3_2, a4_2]
# print(ldc_1, ldc_2)
#light curve
# self.period = 5
# self.inclination = 90
# self.R_1 = 0.05
# self.R_2 = 0.05
# self.sbratio = 1.
# self.q = 1.
# print(self.t_zero, self.period, self.a, self.q,
# self.R_1, self.R_2, self.inclination, self.sbratio)
#This is in arbitrary units... H ow do we get this into real units??
# print("calling ellc", filt, self.obsDates[filt], ldc_1, ldc_2, self.t_zero, self.period, self.a, self.q,
# self.f_c, self.f_s, self.ld_1, self.ld_2, self.R_1, self.R_2, self.inclination, self.sbratio,
# self.shape_1, self.shape_2, self.grid_1,self.grid_2)
t_zero = self.t_zero
if (useT0 is not None):
t_zero = useT0
dates = self.obsDates[filt]
if (useDates[0] is not None):
dates = useDates
if (light_3 is None):
light_3 = self.light_3[filt]
#print('using light_3', filt, light_3)
if (np.isfinite(ldc_1[0]) and np.isfinite(ldc_2[0])):
lc = ellc.lc(dates,
ldc_1=ldc_1,
ldc_2=ldc_2,
t_zero=t_zero,
period=self.period,
a=self.a,
q=self.q,
f_c=self.f_c,
f_s=self.f_s,
ld_1=self.ld_1,
ld_2=self.ld_2,
radius_1=self.R_1,
radius_2=self.R_2,
incl=self.inclination,
sbratio=self.sbratio,
shape_1=self.shape_1,
shape_2=self.shape_2,
grid_1=self.grid_1,
grid_2=self.grid_2,
light_3=light_3)
else:
print(
f"WARNING: nan's in ldc filter={filt}, ldc_1={ldc_1}, T1={T1}, logg1={g1}, ldc_2={ldc_2}, T2={T2}, logg2={g2}, [M/H]={MH}"
)
lc = ellc.lc(dates,
t_zero=t_zero,
period=self.period,
a=self.a,
q=self.q,
f_c=self.f_c,
f_s=self.f_s,
radius_1=self.R_1,
radius_2=self.R_2,
incl=self.inclination,
sbratio=self.sbratio,
shape_1=self.shape_1,
shape_2=self.shape_2,
grid_1=self.grid_1,
grid_2=self.grid_2,
light_3=light_3)
lc = lc / np.max(lc) #maybe there's a better normalization?
if (min(lc) > 0):
#this is mathematically the same as below
# #let's redefine these here, but with the lc accounted for
# absMag = self.MbolSun - 2.5*np.log10( (self.L1f[filt] + self.L2f[filt])*lc) #This may not be strictly correct? Should I be using the Sun's magnitude in the given filter? But maybe this is OK because, L1f and L2f are in units of LSun, which is related to the bolometric luminosity?
# self.appMag[filt] = absMag + 5.*np.log10(self.dist*100.) + self.Ared[filt] #multiplying by 1000 to get to parsec units
Fv = self.Fv1[filt] + self.Fv2[filt]
self.appMag[filt] = -2.5 * np.log10(
Fv * lc) + self.Ared[filt] #AB magnitude
# plt.plot((self.obsDates[filt] % self.period), lc,'.')
# plt.ylim(min(lc), max(lc))
# plt.show()
# plt.plot((self.obsDates[filt] % self.period), self.appMag[filt],'.', color='red')
# plt.plot((self.obsDates[filt] % self.period), self.appMagMean[filt] - 2.5*np.log10(lc), '.', color='blue')
# plt.ylim(max(self.appMag[filt]), min(self.appMag[filt]))
# plt.show()
# print( (self.appMagMean[filt] - 2.5*np.log10(lc)) - self.appMag[filt])
# raise
m_5 = [None]
if (self.useOpSimDates):
if (self.m_5):
m_5 = self.m_5[filt]
else:
m_5 = np.full(len(useDates),
np.mean(self.OpSim.m_5[self.OpSimi][filt]))
#Ivezic 2008, https://arxiv.org/pdf/0805.2366.pdf , Table 2
sigma2_rand = getSig2Rand(filt, self.appMag[filt],
m_5=m_5) #random photometric error
self.appMagObsErr[filt] = ((self.sigma_sys**2.) +
(sigma2_rand))**(1. / 2.)
#now add the uncertainty onto the magnitude
self.appMagObs[filt] = np.array([
np.random.normal(loc=x, scale=sig)
for (x, sig) in zip(self.appMag[filt], self.appMagObsErr[filt])
])
#self.deltaMag[filt] = abs(min(self.appMagObs[filt]) - max(self.appMagObs[filt]))
#self.eclipseDepthFrac[filt] = abs(self.deltaMag[filt]/np.mean(self.appMagObsErr[filt]))
maxpos = np.argmax(self.appMagObs[filt])
base = -2.5 * np.log10(Fv) + self.Ared[filt] #AB magnitude
self.deltaMag[filt] = abs(self.appMagObs[filt][maxpos] - base)
self.eclipseDepthFrac[filt] = abs(self.deltaMag[filt] /
self.appMagObsErr[filt][maxpos])
def checkEclipse(self, r1, r2, rp, inclination):
ratio = (r1 + r2) / rp #rp = 2a for circular orbits
if (ratio <= 1):
theta = np.arcsin(ratio) * 180. / np.pi
min_incl = 90. - theta
max_incl = 90. + theta
if (inclination <= min_incl or inclination >= max_incl):
return False
return True
def preCheckIfObservable(self):
#check for eclipse (accounting for eccentricity)
#primary
tp = radvel.orbit.timetrans_to_timeperi(np.array([self.t_zero]),
self.period, self.eccentricity,
self.omega * np.pi / 180.)
ta = radvel.orbit.true_anomaly(np.array([self.t_zero]), tp,
self.period, self.eccentricity)
rp = self.a * (1. - self.eccentricity * self.eccentricity) / (
1. + self.eccentricity * np.cos(ta))
eclipse_pri = self.checkEclipse(self.r1, self.r2, rp, self.inclination)
#secondary
tp = radvel.orbit.timetrans_to_timeperi(np.array([self.t_zero]),
self.period,
self.eccentricity,
self.omega * np.pi / 180.,
secondary=True)
ta = radvel.orbit.true_anomaly(np.array([self.t_zero]), tp,
self.period, self.eccentricity)
rp = self.a * (1. - self.eccentricity * self.eccentricity) / (
1. + self.eccentricity * np.cos(ta))
eclipse_sec = self.checkEclipse(self.r1, self.r2, rp, self.inclination)
if (not eclipse_pri and not eclipse_sec):
self.incl_failed = 1
#check for overlap of radii at peri
ta = 0.0
rp1 = self.a * (1. - self.eccentricity * self.eccentricity) / (
1. + self.eccentricity * np.cos(ta))
rp2 = self.a * (1. - self.eccentricity * self.eccentricity) / (
1. + self.eccentricity * np.cos(ta + np.pi))
rp = rp1 + rp2
ratio = (self.r1 + self.r2) / rp
if (ratio > 1 or self.R_1 * (1. - self.eccentricity) <= 0
or self.R_1 * (1. - self.eccentricity) >= 1
or self.R_2 * (1. - self.eccentricity) <= 0
or self.R_2 * (1. - self.eccentricity) >= 1 or self.R_1e >= 1
or self.R_2e >= 1):
self.radius_failed = 1
if (self.useOpSimDates):
#redefine the totaltime based on the maximum OpSim date range over all filters
for filt in self.filters:
#print("check filt, totaltime, obs", filt, self.totaltime, self.OpSim.obsDates[self.OpSimi])
#print("check obs", filt, self.OpSim.obsDates[self.OpSimi][filt])
if (self.OpSim.obsDates[self.OpSimi][filt][0] is not None):
self.totaltime = max(
self.totaltime,
(max(self.OpSim.obsDates[self.OpSimi][filt]) -
min(self.OpSim.obsDates[self.OpSimi][filt])))
# if (self.period >= self.totaltime):
# self.period_failed = 1
if (self.radius_failed or self.period_failed or self.incl_failed):
self.observable = False
if (self.verbose):
print("precheck observable", self.observable, self.radius_failed,
self.period_failed, self.incl_failed)
def magCheckIfObservable(self):
if (
self.appMagMean['r_'] <= self.magLims[0]
or self.appMagMean['r_'] >= self.magLims[1]
): #15.8 = rband saturation from Science Book page 57, before Section 3.3; 24.5 is the desired detection limit
self.appmag_failed = 1
self.observable = False
#print("checking mag", self.appMagMean['r_'], self.observable)
if (self.verbose):
print("mag observable", self.observable)
def initializeSeed(self):
if (self.seed is None):
np.random.seed()
else:
np.random.seed(seed=self.seed)
def initialize(self):
#should I initialize the seed here?
#No I am initializing it in LSSTEBworker.py
#self.initializeSeed()
self.q = self.m2 / self.m1
self.a = self.getafromP(self.m1 * units.solMass,
self.m2 * units.solMass,
self.period * units.day).to(units.solRad).value
self.R_1 = self.r1 / self.a
self.R_2 = self.r2 / self.a
self.R_1e = self.r1 / self.Eggleton_RL(
self.m1 / self.m2, self.a * (1. - self.eccentricity))
self.R_2e = self.r2 / self.Eggleton_RL(
self.m2 / self.m1, self.a * (1. - self.eccentricity))
self.f_c = np.sqrt(self.eccentricity) * np.cos(
self.omega * np.pi / 180.)
self.f_s = np.sqrt(self.eccentricity) * np.sin(
self.omega * np.pi / 180.)
if (self.RA is None):
coord = SkyCoord(x=self.xGx,
y=self.yGx,
z=self.zGx,
unit='pc',
representation='cartesian',
frame='galactocentric')
self.RA = coord.icrs.ra.to(units.deg).value
self.Dec = coord.icrs.dec.to(units.deg).value
for f in self.filters:
self.appMagMean[f] = -999.
self.deltaMag[f] = None
self.LSS[f] = -999.
self.appMagObs[f] = [None]
self.appMagObsErr[f] = [None]
self.deltaMag[f] = 0.
self.obsDates[f] = [None]
self.eclipseDepthFrac[f] = 0.
self.appMagMean[
'r_'] = self.TRILEGALrmag #for binaries that don't pass the preCheck
self.maxDeltaMag = 0.
self.preCheckIfObservable()
if (self.observable):
if (self.T1 is None): self.T1 = self.getTeff(self.L1, self.r1)
if (self.T2 is None): self.T2 = self.getTeff(self.L2, self.r2)
if (self.g1 is None):
self.g1 = self.getlogg(self.m1, self.L1, self.T1)
if (self.g2 is None):
self.g2 = self.getlogg(self.m2, self.L2, self.T2)
self.sbratio = (self.L2 / self.r2**2.) / (self.L1 / self.r1**2.)
#one option for getting the extinction
if (self.AV is None):
count = 0
while (self.AV is None and count < 100):
count += 1
self.AV = extinction.get_AV_infinity(self.RA,
self.Dec,
frame='icrs')
if (self.AV is None):
print("WARNING: No AV found", self.RA, self.Dec,
self.AV, count)
time.sleep(30)
self.SED1 = SED()
self.SED1.filters = self.filters
self.SED1.filterFilesRoot = self.filterFilesRoot
self.SED1.T = self.T1 * units.K
self.SED1.R = self.r1 * units.solRad
self.SED1.L = self.L1 * units.solLum
self.SED1.logg = self.g1
self.SED1.M_H = self.M_H
self.SED1.EBV = self.AV / self.RV #could use this to account for reddening in SED
self.SED1.initialize()
self.SED2 = SED()
self.SED2.filters = self.filters
self.SED2.filterFilesRoot = self.filterFilesRoot
self.SED2.T = self.T2 * units.K
self.SED2.R = self.r2 * units.solRad
self.SED2.L = self.L2 * units.solLum
self.SED2.logg = self.g2
self.SED2.M_H = self.M_H
self.SED2.EBV = self.AV / self.RV #could use this to account for reddening in SED
self.SED2.initialize()
#estimate a combined Teff value, as I do in the N-body codes (but where does this comes from?)
logLb = np.log10(self.L1 + self.L2)
logRb = 0.5 * np.log10(self.r1**2. + self.r2**2.)
self.T12 = 10.**(3.762 + 0.25 * logLb - 0.5 * logRb)
#print(self.L1, self.L2, self.T1, self.T2, self.T12)
#account for reddening and the different filter throughput functions (currently related to a blackbody)
self.appMagMeanAll = 0.
#one option for getting the extinction
#ext = F04(Rv=self.RV)
ext = F04(Rv=self.RV)
#a check
# self.Ltest = self.SED.getL(self.T1*units.K, self.r1*units.solRad)
#definitely necessary for Kurucz because these models are not normalized
#for the bb I'm getting a factor of 2 difference for some reason
Lconst1 = self.SED1.getLconst()
Lconst2 = self.SED2.getLconst()
#print(np.log10(Lconst1), np.log10(Lconst2))
for f in self.filters:
#print(extinction.fitzpatrick99(np.array([self.wavelength[f]*10.]), self.AV, self.RV, unit='aa')[0] , ext(self.wavelength[f]*units.nm)*self.AV)
#self.Ared[f] = extinction.fitzpatrick99(np.array([self.wavelength[f]*10.]), self.AV, self.RV, unit='aa')[0] #or ccm89
self.Ared[f] = ext(self.wavelength[f] * units.nm) * self.AV
self.Fv1[f] = self.SED1.getFvAB(self.dist * units.kpc,
f,
Lconst=Lconst1)
self.Fv2[f] = self.SED2.getFvAB(self.dist * units.kpc,
f,
Lconst=Lconst2)
Fv = self.Fv1[f] + self.Fv2[f]
self.appMagMean[f] = -2.5 * np.log10(Fv) + self.Ared[
f] #AB magnitude
#print(self.wavelength[f], self.appMagMean[f], self.Ared[f], self.T1)
self.appMagMeanAll += self.appMagMean[f]
self.appMagMeanAll /= len(self.filters)
#check if we can observe this based on the magnitude
self.magCheckIfObservable()
#if we're using OpSim, then get the field ID
#get the field ID number from OpSim where this binary would be observed
if (self.useOpSimDates and self.observable
and self.OpSim.fieldID[0] is None):
self.OpSim.setFieldID(self.RA, self.Dec)
#this now covers the galaxy and the cluster, if available
if (self.observable and self.crowding is not None):
self.crowding.input_rFlux = self.Fv1['r_'] + self.Fv2[
'r_'] #NOTE; if I don't have a r_ band, this will crash
self.crowding.getCrowding()
for f in self.filters:
self.light_3[f] = self.crowding.backgroundFlux[f] / (
self.Fv1[f] + self.Fv2[f])
def observe(self, filt, light_3=None):
#get the observation dates
if (self.useOpSimDates):
#print("using OpSimDates...")
#check if we already have the observing dates
if (self.OpSim is not None):
#print("have OpSim", self.OpSim.obsDates)
if (filt in self.OpSim.obsDates[self.OpSimi]):
self.obsDates[filt] = self.OpSim.obsDates[
self.OpSimi][filt]
self.m_5[filt] = self.OpSim.m_5[self.OpSimi][filt]
#otherwise get them
if (filt not in self.obsDates):
#print("getting dates")
self.obsDates[filt], self.m_5[filt] = self.OpSim.getDates(
self.OpSim.fieldID[self.OpSimi], filt)
#print("received dates", filt, self.obsDates[filt])
if (self.verbose):
print(
f'observing with OpSim in filter {filt}, have {len(self.obsDates[filt])} observations'
)
else:
#print("not using OpSimDates...")
nobs = int(round(self.totaltime / (self.cadence * self.Nfilters)))
self.obsDates[filt] = np.sort(self.totaltime *
np.random.random(size=nobs))
self.nobs += len(self.obsDates[filt])
#get the light curve, and related information
if (self.obsDates[filt][0] is not None):
#print("calling light curve", filt)
self.setLightCurve(filt, light_3=light_3)
if (filt == 'r_'):
if (self.eclipseDepthFrac[filt] < self.eclipseDepthLim):
self.eclipseDepth_failed = 1
self.observable = False
def outputLCtoFile(self, fname):
obs = dict()
for filt in self.filters:
obs[filt] = dict()
obs[filt]['OpSimDates'] = self.obsDates[filt]
obs[filt]['mag'] = self.appMag[filt]
obs[filt]['magObs'] = self.appMagObs[filt]
obs[filt]['e_magObs'] = self.appMagObsErr[filt]
file = open(fname, 'wb')
pickle.dump(obs, file)
file.close()