def modify_model(self, changes=None, removings=None, verbose=False, warnings=True, describe=False):
"""
Apply changes on the loaded model.
Parameters
----------
changes : [dict or None]
Dictionary containing the desired changes for the desired parameters.
As an example, changes={"zred"={"init"=2.}}) will change the initial value of the 'zred' parameter, without changing anything else.
Default is None.
removings : [string or list(string) or None]
List of every parameters you want to remove from the model.
Default is None.
Options
-------
verbose : [bool]
If True, print the SED model before and after modifications.
Default is False.
warnings : [bool]
If True, allow warnings to be printed.
Default is True.
describe : [bool]
If True, print the description for any used not native prior.
Default is False.
Returns
-------
Void
"""
from prospect.models.sedmodel import SedModel
_model = self.model.config_dict
if verbose:
print(tools.get_box_title(title="Previous model", box="\n#=#\n# {} #\n#=#\n"))
pprint(_model)
#Changes
_describe_priors = []
if changes is not None:
_model, _describe_priors = self._model_changes_(_model, changes, warnings)
if describe and len(_describe_priors) > 0:
self.describe_priors(_describe_priors)
#Removing
if removings is not None:
_model = self._model_removings_(_model, removings, warnings)
if verbose:
print(tools.get_box_title(title="New model", box="\n\n\n#=#\n# {} #\n#=#\n"))
pprint(_model)
self._model = SedModel(_model)
def load_model(zred=None, **extras):
# In principle (and we've done it) you could have the model depend on
# command line arguments (or anything in run_params) by making changes to
# `model_params` here before instantiation the SedModel object. Up to you.
# Here we are going to set the intial value (and the only value, since it
# is not a free parameter) of the redshift parameter to whatever was used
# to generate the mock, listed in the run_params dictionary.
pn = [p['name'] for p in model_params]
zind = pn.index('zred')
model_params[zind]['init'] = zred
return SedModel(model_params)
def build_model(binNum=0,
fit_metallicity=1,
fit_redshift=1,
infile=None,
add_neb=0,
**extras):
from astropy.table import Table
from prospect.models.priors import LogUniform, TopHat, Normal
from prospect.models.sedmodel import SedModel
from prospect.models.templates import TemplateLibrary
from prospect.sources.constants import cosmo
table = Table.read(infile)
zobs = table['z'][binNum]
max_age = cosmo.age(zobs).value
model_params = TemplateLibrary['parametric_sfh'] # delay-tau model
model_params['zred']['init'] = zobs
model_params['zred']['isfree'] = bool(fit_redshift)
model_params['zred']['prior'] = TopHat(mini=zobs - 0.01, maxi=zobs + 0.01)
model_params['mass']['init'] = 1e7
model_params['mass']['prior'] = LogUniform(mini=1e5, maxi=1e9)
model_params['logzsol']['init'] = -0.15
model_params['logzsol']['isfree'] = bool(fit_metallicity)
model_params['logzsol']['prior'] = Normal(mean=-0.15, sigma=0.22)
model_params['dust2']['init'] = 0.3
model_params['dust2']['prior'] = TopHat(mini=0.0, maxi=2.0)
model_params['tage']['init'] = 6.24
model_params['tage']['prior'] = TopHat(mini=0.001, maxi=max_age)
model_params['tau']['init'] = 0.29
model_params['tau']['prior'] = LogUniform(mini=0.1, maxi=10)
model_params['imf_type']['init'] = 1 # Chabrier (2003) IMF
model_params['dust_type']['init'] = 1 # Cardelli+ (1989) MW extinction
if add_neb:
model_params.update(TemplateLibrary['nebular'])
model = SedModel(model_params)
return model
def build_model(binNum=0, fit_metallicity=1, fit_redshift=1, infile=None,
**extras) :
import numpy as np
from astropy.table import Table
from prospect.models.priors import StudentT, TopHat
from prospect.models.sedmodel import SedModel
from prospect.models.templates import TemplateLibrary
table = Table.read(infile)
zobs = table['z'][binNum]
nbins = 10
model_params = TemplateLibrary['continuity_sfh'] # non-parametric SFH
model_params['zred']['init'] = zobs
model_params['zred']['isfree'] = bool(fit_redshift)
model_params['zred']['prior'] = TopHat(mini=zobs-0.01, maxi=zobs+0.01)
model_params['logmass']['init'] = 6
model_params['logmass']['prior'] = TopHat(mini=3, maxi=16)
model_params['mass']['N'] = nbins
model_params['logzsol']['init'] = -0.5
model_params['logzsol']['isfree'] = bool(fit_metallicity)
model_params['logzsol']['prior'] = TopHat(mini=-2, maxi=0.2)
model_params['dust2']['init'] = 0.5
model_params['dust2']['prior'] = TopHat(mini=0.0, maxi=4.0)
model_params['agebins']['N'] = nbins
model_params['agebins']['init'] = [[0.0, 7.47712125], [7.47712125, 8.0],
[8.0, 8.69897], [8.69897, 9.0], [9.0, 9.5261747], [9.5261747, 9.75720267],
[9.75720267, 9.90720604], [9.90720604, 10.01848862],
[10.01848862, 10.10699395], [10.10699395, 10.12927034]]
model_params['logsfr_ratios']['N'] = nbins - 1
model_params['logsfr_ratios']['init'] = np.full(nbins-1, 0.0)
model_params['logsfr_ratios']['prior'] = StudentT(mean=np.full(nbins-1,0.0),
scale=np.full(nbins-1,0.3),
df=np.full(nbins-1, 2))
model_params['imf_type']['init'] = 1 # Chabrier (2003) IMF
model_params['dust_type']['init'] = 1 # Cardelli+ (1989) MW extinction
model = SedModel(model_params)
return model
def load_model(zred=0.0,
add_neb=False,
complex_atten=False,
atten_bump=False,
**extras):
"""Instantiate and return a ProspectorParams model subclass.
:param zred: (optional, default: 0.1)
The redshift of the model
:param add_neb: (optional, default: False)
If True, turn on nebular emission and add relevant parameters to the
model.
"""
from prospect.models.sedmodel import SedModel
from prospect.models.templates import TemplateLibrary
model_params = TemplateLibrary["alpha"]
model_params["fagn"]["isfree"] = False
model_params["agn_tau"]["isfree"] = False
model_params["sfh"]["init"] = 0
if add_neb == True:
model_params['add_neb_emission']['init'] = True
if complex_atten == True:
# --- Complexify dust attenuation ---
# Switch to Kriek and Conroy 2013
model_params["dust_type"] = {
'N': 1,
'isfree': False,
'init': 4,
'prior': None
}
if atten_bump == True:
# Slope of the attenuation curve, expressed as the index of the power-law
# that modifies the base Kriek & Conroy/Calzetti shape.
# I.e. a value of zero is basically calzetti with a 2175AA bump
model_params["dust_index"] = {
'N': 1,
'isfree': False,
'init': 0.0,
'prior': None
}
return SedModel(model_params)
def build_model(object_redshift=None,ldist=None,fixed_metallicity=None,add_duste=None,
add_nebular=True,add_burst=True,**extras):
model_params = TemplateLibrary["parametric_sfh"]
model_params["tau"]["init"] = 3.0
model_params["dust2"]["init"] = 0.1
model_params["logzsol"]["init"] = -0.8
model_params["tage"]["init"] = 6.5
model_params["mass"]["init"] = 1e8
model_params["tage"]["prior"] = priors.TopHat(mini=0.1, maxi=11.)
model_params["dust2"]["prior"] = priors.TopHat(mini=0.0, maxi=2.0)
model_params["tau"]["prior"] = priors.LogUniform(mini=1., maxi=7.)
model_params["mass"]["prior"] = priors.LogUniform(mini=1e7, maxi=1e12)
model_params["zred"]["prior"] = priors.TopHat(mini=0.005, maxi=4.0)
model_params["mass"]["disp_floor"] = 1e5
model_params["tau"]["disp_floor"] = 1.0
model_params["tage"]["disp_floor"] = 1.0
if fixed_metallicity is not None:
model_params["logzsol"]["isfree"] = False
model_params["logzsol"]['init'] = fixed_metallicity
if object_redshift is not None:
model_params["zred"]['isfree'] = False
model_params["zred"]['init'] = object_redshift
if object_redshift is None:
model_params["zred"]['isfree'] = True
model_params["zred"]['init'] = 0.1
if add_duste:
model_params.update(TemplateLibrary["dust_emission"])
if add_nebular:
model_params.update(TemplateLibrary["nebular"])
if add_burst:
model_params.update(TemplateLibrary["burst_sfh"])
model_params["mass"]["isfree"] = True
model_params["logzsol"]["isfree"] = True
model_params["dust2"]["isfree"] = True
model_params["tage"]["isfree"] = True
model_params["tau"]["isfree"] = True
model = SedModel(model_params)
return model
def build_model(object_redshift=None,
fixed_metallicity=None,
add_duste=False,
**extras):
"""Build a prospect.models.SedModel object
:param object_redshift: (optional, default: None)
If given, produce spectra and observed frame photometry appropriate
for this redshift. Otherwise, the redshift will be zero.
:param fixed_metallicity: (optional, default: None)
If given, fix the model metallicity (:math:`log(Z/Z_sun)`) to the given value.
:param add_duste: (optional, default: False)
If `True`, add dust emission and associated (fixed) parameters to the model.
:returns model:
An instance of prospect.models.SedModel
"""
# Get (a copy of) one of the prepackaged model set dictionaries.
# This is, somewhat confusingly, a dictionary of dictionaries, keyed by parameter name
model_params = TemplateLibrary["alpha"]
#model_params = TemplateLibrary["parametric_sfh"]
# Set agebins manually
#age_at_z = astropy_cosmo.age(object_redshift).value
nbins_sfh = 8
model_params['agebins']['N'] = nbins_sfh
model_params['mass']['N'] = nbins_sfh
# This will give the stellar mass as the surviving
# mass which is what we want. Otherwise by default
# it gives total mass formed.
#model_params["mass"]["units"] = 'mstar'
# Set dust type. # Fixed for parametric sfh
#model_params["dust_type"] = 4
# Let's make some changes to initial values appropriate for our objects and data
#model_params["logmass"]["init"] = 10
#model_params["logzsol"]["init"] = -0.5
#model_params["dust2"]["init"] = 0.05
#model_params["mass"]["init"] = 1e10
#model_params["tage"]["init"] = 4.0
#model_params["tau"]["init"] = 1.0
# Choose priors
# Priors not specified in here are left at default values
#model_params["mass"]["prior"] = priors.LogUniform(mini=1e8, maxi=1e12)
#model_params["tau"]["prior"] = priors.LogUniform(mini=1e-1, maxi=1e2)
# If we are going to be using emcee, it is useful to provide a
# minimum scale for the cloud of walkers (the default is 0.1)
#model_params["mass"]["disp_floor"] = 1e8
#model_params["tau"]["disp_floor"] = 1.0
#model_params["tage"]["disp_floor"] = 1.0
# Change the model parameter specifications based on some keyword arguments
if fixed_metallicity is not None:
# make it a fixed parameter
model_params["logzsol"]["isfree"] = False
#And use value supplied by fixed_metallicity keyword
model_params["logzsol"]['init'] = fixed_metallicity
if object_redshift is not None:
# make sure zred is fixed
model_params["zred"]['isfree'] = False
# And set the value to the object_redshift keyword
model_params["zred"]['init'] = object_redshift
if add_duste:
# Add dust emission (with fixed dust SED parameters)
# Since `model_params` is a dictionary of parameter specifications,
# and `TemplateLibrary` returns dictionaries of parameter specifications,
# we can just update `model_params` with the parameters described in the
# pre-packaged `dust_emission` parameter set.
model_params.update(TemplateLibrary["dust_emission"])
# Now instantiate the model object using this dictionary of parameter specifications
model = SedModel(model_params)
return model
def load_model(zred=None, add_neb=False, complex_atten=False, atten_bump=False, **extras):
"""Instantiate and return a ProspectorParams model subclass.
:param zred: (optional, default: 0.1)
The redshift of the model
:param add_neb: (optional, default: False)
If True, turn on nebular emission and add relevant parameters to the
model.
"""
from prospect.models.sedmodel import SedModel
from prospect.models.templates import TemplateLibrary
model_params = TemplateLibrary["parametric_sfh"]
if complex_atten == True:
# --- Complexify dust attenuation ---
# Switch to Kriek and Conroy 2013
model_params["dust_type"] = {'N': 1, 'isfree': False,
'init': 4, 'prior': None}
if atten_bump == True:
# Slope of the attenuation curve, expressed as the index of the power-law
# that modifies the base Kriek & Conroy/Calzetti shape.
# I.e. a value of zero is basically calzetti with a 2175AA bump
model_params["dust_index"] = {'N': 1, 'isfree': False,
'init': 0.0, 'prior': None}
# --- Distance ---
model_params.update({'zred': {'N': 1,
'isfree': False,
'init': zred,
'units': '',
'prior': priors.TopHat(mini=0.0, maxi=5.8)}})
# --- SFH --------
# FSPS parameter
model_params.update({'sfh': {'N': 1,
'isfree': False,
'init': 4, # This is delay-tau
'units': 'type',
'prior': None}})
model_params.update({'mass':{'N': 1, 'isfree': True,
'init': 1e10,
'init_disp': 1e9,
'units': r'M_\odot',
'prior': priors.LogUniform(mini=1e6, maxi=1e12)}})
model_params.update({'logzsol': {'N': 1,
'isfree': True,
'init': -0.3,
'init_disp': 0.3,
'units': r'$\log (Z/Z_\odot)$',
'prior': priors.TopHat(mini=-2.0, maxi=0.19)}})
# If zcontinuous > 1, use 3-pt smoothing
model_params.update({'pmetals': {'N': 1,
'isfree': False,
'init': -99,
'prior': None}})
# FSPS parameter
model_params.update({'tau': {'N': 1,
'isfree': True,
'init': 1.0,
'init_disp': 0.5,
'units': 'Gyr',
'prior':priors.LogUniform(mini=0.101, maxi=100)}})
# FSPS parameter
model_params.update({'tage':{'N': 1,
'isfree': True,
'init': 5.0,
'init_disp': 3.0,
'units': 'Gyr',
'prior': priors.TopHat(mini=0.101, maxi=13.6)}})
model_params.update({'fage_burst': {'N': 1,
'isfree': False,
'init': 0.0,
'units': 'time at wich burst happens, as a fraction of `tage`',
'prior': priors.TopHat(mini=0.9, maxi=1.0)}})
# If we are going to be using emcee, it is useful to provide a
# minimum scale for the cloud of walkers (the default is 0.1)
model_params["mass"]["disp_floor"] = 1e7
model_params["mass"]["init_disp"] = 1e8
model_params["tau"]["disp_floor"] = 1.0
model_params["tage"]["disp_floor"] = 1.0
# FSPS parameter
model_params.update({'tburst': {'N': 1,
'isfree': False,
'init': 0.0,
'units': '',
'prior': None,}})
#'depends_on': tburst_fage}}) # uncomment if using bursts.
# FSPS parameter
model_params.update({'fburst': {'N': 1,
'isfree': False,
'init': 0.0,
'units': '',
'prior': priors.TopHat(mini=0.0, maxi=0.5)}})
# ------ IMF ---------
model_params.update({'imf_type': {'N': 1,
'isfree': False,
'init': 1, #1 = chabrier
'units': None,
'prior': None}})
# --- Dust ---------
# FSPS parameter
model_params.update({'dust_type': {'N': 1,
'isfree': False,
'init': 0, # power-laws
'units': 'index',
'prior': None}})
# FSPS parameter
model_params.update({'dust2': {'N': 1,
'isfree': True,
'init': 0.35,
'reinit': True,
'init_disp': 0.3,
'units': 'Diffuse dust optical depth towards all stars at 5500AA',
'prior': priors.TopHat(mini=0.0, maxi=2.0)}})
# FSPS parameter
model_params.update({'dust1': {'N': 1,
'isfree': False,
'init': 0.0,
'units': 'Extra optical depth towards young stars at 5500AA',
'prior': None}})
# FSPS parameter
model_params.update({'dust_tesc': {'N': 1,
'isfree': False,
'init': 7.0,
'units': 'definition of young stars for the purposes of the CF00 dust model, log(Gyr)',
'prior': None}})
# FSPS parameter
model_params.update({'dust_index': {'N': 1,
'isfree': False,
'init': -0.7,
'units': 'power law slope of the attenuation curve for diffuse dust',
'prior': priors.TopHat(mini=-1.0, maxi=0.4)}})
# FSPS parameter
model_params.update({'dust1_index': {'N': 1,
'isfree': False,
'init': -1.0,
'units': 'power law slope of the attenuation curve for young-star dust',
'prior': None}})
# ---- Dust Emission ---
# FSPS parameter
model_params.update({'add_dust_emission': {'N': 1,
'isfree': False,
'init': False,
'units': 'index',
'prior': None}})
# An example of the parameters controlling the dust emission SED. There are others!
model_params.update({'duste_gamma': {'N': 1,
'isfree': False,
'init': 0.01,
'init_disp': 0.2,
'disp_floor': 0.15,
'units': None,
'prior': priors.TopHat(mini=0.0, maxi=1.0)}})
model_params.update({'duste_umin': {'N': 1,
'isfree': False,
'init': 1.0,
'init_disp': 5.0,
'disp_floor': 4.5,
'units': None,
'prior': priors.TopHat(mini=0.1, maxi=25.0)}})
model_params.update({'duste_qpah': {'N': 1,
'isfree': False,
'init': 2.0,
'init_disp': 3.0,
'disp_floor': 3.0,
'units': 'percent',
'prior': priors.TopHat(mini=0.0, maxi=7.0)}})
# --- Stellar Pops ------------
# One could imagine changing these, though doing so *during* the fitting will
# be dramatically slower.
# FSPS parameter
model_params.update({'tpagb_norm_type': {'N': 1,
'isfree': False,
'init': 2,
'units': 'index',
'prior': None}})
# FSPS parameter
model_params.update({'add_agb_dust_model': {'N': 1,
'isfree': False,
'init': True,
'units': 'index',
'prior': None}})
# FSPS parameter
model_params.update({'agb_dust': {'N': 1,
'isfree': False,
'init': 1,
'units': 'index',
'prior': None}})
# --- Nebular Emission ------
# For speed we turn off nebular emission in the demo
model_params.update({'add_neb_emission': {'N': 1,
'isfree': False,
'init': False,
'prior': None}})
# FSPS parameter
model_params.update({'gas_logz': {'N': 1,
'isfree': False,
'init': 0.0,
'units': r'log Z/Z_\odot',
# 'depends_on': stellar_logzsol,
'prior': priors.TopHat(mini=-2.0, maxi=0.5)}})
# FSPS parameter
model_params.update({'gas_logu': {'N': 1,
'isfree': False,
'init': -2.0,
'units': '',
'prior': priors.TopHat(mini=-4, maxi=-1)}})
# --- Calibration ---------
# Only important if using a NoiseModel
model_params.update({'phot_jitter': {'N': 1,
'isfree': False,
'init': 0.0,
'units': 'mags',
'prior': priors.TopHat(mini=0.0, maxi=0.2)}})
return SedModel(model_params)
def build_model(binNum=0,
fit_metallicity=1,
fit_redshift=1,
infile=None,
add_neb=0,
metalfile=None,
cluster=None,
ID=0,
**extras):
import numpy as np
from astropy.table import Table
from prospect.models.priors import ClippedNormal, StudentT, TopHat
from prospect.models.sedmodel import SedModel
from prospect.models.templates import TemplateLibrary
from prospect.sources.constants import cosmo
table = Table.read(infile)
zobs = table['z'][binNum]
max_age, nbins = cosmo.age(zobs).value, 10
# sma, smb = table['sma'][binNum], table['smb'][binNum]
# R_e, width = table['R_e'][binNum], table['width'][binNum]
# radius = (sma - width)*np.sqrt(smb/sma)/R_e
metallicities = Table.read(metalfile)
idx = np.where((metallicities['cluster'] == cluster)
& (metallicities['ID'] == ID))[0][0]
central_metallicity = metallicities['centrallogZ'][idx]
sigma_metallicity = metallicities['sigma'][idx]
init_metallicity = central_metallicity # - 0.1*radius
model_params = TemplateLibrary['continuity_sfh'] # non-parametric model
model_params['zred']['init'] = zobs
model_params['zred']['isfree'] = bool(fit_redshift)
model_params['zred']['prior'] = TopHat(mini=zobs - 0.01, maxi=zobs + 0.01)
model_params['logmass']['init'] = 7
model_params['logmass']['prior'] = TopHat(mini=5, maxi=10)
model_params['mass']['N'] = nbins
model_params['logzsol']['init'] = init_metallicity
model_params['logzsol']['isfree'] = bool(fit_metallicity)
model_params['logzsol']['prior'] = ClippedNormal(mean=init_metallicity,
sigma=sigma_metallicity,
mini=-2,
maxi=0.2)
model_params['dust2']['init'] = 0.3
model_params['dust2']['prior'] = TopHat(mini=0.0, maxi=2.0)
model_params['agebins']['N'] = nbins
agelims = np.concatenate([
np.log10([1e-9, 0.03, 0.1, 0.5]),
np.linspace(0, np.log10(0.95 * max_age), 6),
np.log10([max_age])
]) + 9
model_params['agebins']['init'] = np.array([agelims[:-1], agelims[1:]]).T
model_params['logsfr_ratios']['N'] = nbins - 1
model_params['logsfr_ratios']['init'] = np.zeros(nbins - 1)
model_params['logsfr_ratios']['prior'] = StudentT(
mean=np.zeros(nbins - 1),
scale=0.3 * np.ones(nbins - 1),
df=2 * np.ones(nbins - 1))
model_params['imf_type']['init'] = 1 # Chabrier (2003) IMF
model_params['dust_type']['init'] = 1 # Cardelli+ (1989) MW extinction
if add_neb:
model_params.update(TemplateLibrary['nebular'])
model = SedModel(model_params)
return model
def build_model(gal_ebv, object_redshift=None, init_theta=None):
"""Build a prospect.models.SedModel object
:param object_redshift: (optional, default: None)
If given, produce spectra and observed frame photometry appropriate
for this redshift. Otherwise, the redshift will be zero.
:param init_theta: (optional, default: [1e10, -1, 10, 1])
The initial guess on the parameters for mcmc.
:returns model:
An instance of prospect.models.SedModel
"""
from prospect.models.sedmodel import SedModel
from prospect.models.templates import TemplateLibrary
from prospect.models import priors
# Get (a copy of) one of the prepackaged model set dictionaries.
model_params = TemplateLibrary["parametric_sfh"]
model_params["sfh"]["init"] = 4
Av_init = gal_ebv * 3.1
# print(init_theta)
# Changing the initial values appropriate for our objects and data
model_params["mass"]["init"] = init_theta[0]
model_params["logzsol"]["init"] = init_theta[1]
model_params["dust2"]["init"] = Av_init
model_params["tage"]["init"] = init_theta[2]
model_params["tau"]["init"] = init_theta[3]
# Setting the priors forms and limits
model_params["mass"]["prior"] = priors.LogUniform(mini=1e6, maxi=1e12)
model_params["logzsol"]["prior"] = priors.Uniform(mini=-1.8, maxi=0.3)
model_params["dust2"]["prior"] = priors.ClippedNormal(mean=Av_init, sigma=0.05, mini=Av_init,
maxi=Av_init + 0.1)
# model_params["dust2"]["prior"] = priors.Uniform(mini=0.0, maxi=0.5)
model_params["tage"]["prior"] = priors.Uniform(mini=0.1, maxi=13.8)
model_params["tau"]["prior"] = priors.Uniform(mini=0.001, maxi=30)
# Setting the spread of the walkers for the mcmc sampling
model_params["mass"]["disp_floor"] = 1e8
model_params["logzsol"]['disp_floor'] = 0.5
model_params["dust2"]["disp_floor"] = 0.01
model_params["tage"]["disp_floor"] = 5.0
model_params["tau"]["disp_floor"] = 3.0
model_params["mass"]["init_disp"] = 1e8
model_params["logzsol"]['init_disp'] = 0.5
model_params["dust2"]["init_disp"] = 0.01
model_params["tage"]["init_disp"] = 5.0
model_params["tau"]["init_disp"] = 3.0
# Fixing and defining the object redshift
model_params["zred"]['isfree'] = False
model_params["zred"]['init'] = object_redshift
# ldist = cosmo.luminosity_distance(object_redshift).value
# model_params["lumdist"] = {"N": 1, "isfree": False, "init": ldist, "units": "Mpc"}
# Now instantiate the model object using this dictionary of parameter specifications
model = SedModel(model_params)
return model
def load_model(obs, template_library='delayed-tau', verbose=False):
"""
http://dfm.io/python-fsps/current/stellarpop_api/#api-reference
https://github.com/moustakas/siena-astrophysics/blob/master/research/redmapper/redmapper-stellar-mass.py#L125-L197
"""
from prospect.models import priors
from prospect.models.sedmodel import SedModel
from prospect.models.templates import TemplateLibrary
from prospect.models.transforms import dustratio_to_dust1
def base_delayed_tau():
model_params = TemplateLibrary['parametric_sfh']
# Initialize with sensible numbers.
model_params['tau']['init'] = 10.0
model_params['tage']['init'] = 1.0
model_params['logzsol']['init'] = 0.2
# optimize log-stellar mass, not linear stellar mass
model_params['logmass'] = {
'N': 1,
'isfree': True,
'init': 11.0,
'prior': priors.TopHat(mini=10.0, maxi=12.0),
'units': '$M_{\odot}$'
}
model_params['mass']['isfree'] = False
model_params['mass']['init'] = 10**model_params['logmass']['init']
model_params['mass']['prior'] = None
model_params['mass']['depends_on'] = logmass2mass
# Adjust the prior ranges.
model_params['tau']['prior'] = priors.LogUniform(mini=0.01, maxi=30.0)
model_params['tage']['prior'] = priors.LogUniform(mini=0.1, maxi=13.0)
model_params['logzsol']['prior'] = priors.TopHat(mini=-0.5, maxi=0.3)
#print('HACK!!!!!!!!!!!!!')
#model_params['tau']['isfree'] = False
#model_params['tage']['isfree'] = False
#model_params['logzsol']['isfree'] = False
#model_params['dust2']['isfree'] = False
return model_params
if template_library == 'delayed-tau':
# Underlying delayed tau model.
model_params = base_delayed_tau()
if template_library == 'bursty':
# Underlying delayed tau model.
model_params = base_delayed_tau()
# Add bursts
model_params.update(TemplateLibrary['burst_sfh'])
model_params['fburst']['isfree'] = True
model_params['fburst']['init'] = 0.1
model_params['fburst']['prior'] = priors.TopHat(mini=0.0, maxi=1.0)
model_params['fage_burst']['isfree'] = True
model_params['fage_burst']['init'] = 0.9
model_params['fage_burst']['prior'] = priors.TopHat(mini=0.5, maxi=1.0)
# Add dust emission (with fixed dust SED parameters).
model_params.update(TemplateLibrary['dust_emission'])
model_params['duste_gamma']['isfree'] = True
model_params['dust2']['init'] = 1.0 # diffuse dust
model_params['dust2']['prior'] = priors.TopHat(mini=0.0, maxi=4.0)
# Add more dust flexibility.
if True:
model_params['dust_type'] = {
'N': 1,
'isfree': False,
'init': 0,
'units': 'dust model'
}
model_params['dust_index'] = {
'N': 1,
'isfree': False,
'init': -0.7,
'units': 'power-law index',
'prior': None
}
#model_params['dust1'] = {'N': 1, 'isfree': False, 'init': 0.0, 'prior': None,
# 'units': 'optical depth towards young stars',
# 'depends_on': dustratio_to_dust1}
#model_params['dust_ratio'] = {'N': 1, 'isfree': True, 'init': 1.0,
# 'prior': priors.TopHat(mini=1.0, maxi=10.0),
# 'units': 'dust1/dust2 ratio (optical depth to young stars vs diffuse)'}
## Add nebular emission.
#model_params.update(TemplateLibrary['nebular'])
##model_params['add_neb_continuum']['init'] = False
#model_params['gas_logu']['init'] = -1.0 # harder radiation field [default is -2.0]
# Fixed redshift.
model_params['zred']['init'] = obs['redshift']
model_params['zred']['isfree'] = False
# Change the IMF from Kroupa to Salpeter.
model_params['imf_type']['init'] = 0
# Now instantiate the model using this new dictionary of parameter specifications
model = SedModel(model_params)
if verbose:
print(model)
return model
def build_model(object_redshift=None,
ldist=10.0,
fixed_metallicity=None,
add_duste=False,
**extras):
"""Build a prospect.models.SedModel object
:param object_redshift: (optional, default: None)
If given, produce spectra and observed frame photometry appropriate
for this redshift. Otherwise, the redshift will be zero.
:param ldist: (optional, default: 10)
The luminosity distance (in Mpc) for the model. Spectra and observed
frame (apparent) photometry will be appropriate for this luminosity distance.
:param fixed_metallicity: (optional, default: None)
If given, fix the model metallicity (:math:`log(Z/Z_sun)`) to the given value.
:param add_duste: (optional, default: False)
If `True`, add dust emission and associated (fixed) parameters to the model.
:returns model:
An instance of prospect.models.SedModel
"""
# Get (a copy of) one of the prepackaged model set dictionaries.
# This is, somewhat confusingly, a dictionary of dictionaries, keyed by parameter name
model_params = TemplateLibrary["parametric_sfh"]
# Now add the lumdist parameter by hand as another entry in the dictionary.
# This will control the distance since we are setting the redshift to zero.
# In `build_obs` above we used a distance of 10Mpc to convert from absolute to apparent magnitudes,
# so we use that here too, since the `maggies` are appropriate for that distance.
model_params["lumdist"] = {
"N": 1,
"isfree": False,
"init": ldist,
"units": "Mpc"
}
# Let's make some changes to initial values appropriate for our objects and data
model_params["zred"]["init"] = 0.0
model_params["dust2"]["init"] = 0.05
model_params["logzsol"]["init"] = -0.5
model_params["tage"]["init"] = 13.
model_params["mass"]["init"] = 1e8
# These are dwarf galaxies, so lets also adjust the metallicity prior,
# the tau parameter upward, and the mass prior downward
model_params["dust2"]["prior"] = priors.TopHat(mini=0.0, maxi=2.0)
model_params["tau"]["prior"] = priors.LogUniform(mini=1e-1, maxi=1e2)
model_params["mass"]["prior"] = priors.LogUniform(mini=1e6, maxi=1e10)
# If we are going to be using emcee, it is useful to provide a
# minimum scale for the cloud of walkers (the default is 0.1)
model_params["mass"]["disp_floor"] = 1e6
model_params["tau"]["disp_floor"] = 1.0
model_params["tage"]["disp_floor"] = 1.0
# Change the model parameter specifications based on some keyword arguments
if fixed_metallicity is not None:
# make it a fixed parameter
model_params["logzsol"]["isfree"] = False
#And use value supplied by fixed_metallicity keyword
model_params["logzsol"]['init'] = fixed_metallicity
if object_redshift is not None:
# make sure zred is fixed
model_params["zred"]['isfree'] = False
# And set the value to the object_redshift keyword
model_params["zred"]['init'] = object_redshift
if add_duste:
# Add dust emission (with fixed dust SED parameters)
# Since `model_params` is a dictionary of parameter specifications,
# and `TemplateLibrary` returns dictionaries of parameter specifications,
# we can just update `model_params` with the parameters described in the
# pre-packaged `dust_emission` parameter set.
model_params.update(TemplateLibrary["dust_emission"])
# Now instantiate the model object using this dictionary of parameter specifications
model = SedModel(model_params)
return model
def build_model(self, model=None, templates=None, verbose=False, describe=False):
"""
Build the model parameters to fit on measurements.
Parameters
----------
templates : [string or list(string) or None]
Prospector prepackaged parameter set(s) to load.
Can be one template, or a list.
Default is None.
Options
-------
model : [prospect.models.sedmodel.SedModel or dict or list or None]
Can load an already existing prospector compatible 'model' SedModel.
Can update a SedModel compatible 'model' dictionary before to create the SedModel on it.
Default is None.
verbose : [bool]
If True, print the built model.
Default is False.
describe : [bool]
If True, print the description for any used SED model template and not native prior.
Default is False.
Returns
-------
Void
"""
from prospect.models.sedmodel import SedModel
from prospect.models.templates import TemplateLibrary
if type(model) is SedModel:
self._model = model
return
if describe:
self.describe_templates(templates=templates)
_model = {} if model is None else model
if isinstance(_model, list):
_model = {_p["name"]:_p for _p in _model}
for _t in np.atleast_1d(templates):
if _t in TemplateLibrary._descriptions.keys():
_model.update(TemplateLibrary[_t])
_describe_priors = []
for _p, _pv in _model.items():
if "prior" in _pv.keys() and type(_pv["prior"]) == dict:
_describe_priors.append(_pv["prior"])
_pv["prior"] = self.build_prior(_pv["prior"])
if describe and len(_describe_priors) > 0:
self.describe_priors(_describe_priors)
if self.has_z():
_model["zred"].update({"init":self.obs["zspec"], "isfree":False})
else:
_model["zred"].update({"isfree":True})
if verbose:
print(tools.get_box_title(title="Built model", box="\n#=#\n# {} #\n#=#\n"))
pprint(_model)
self._model = SedModel(_model)
def build_model_un(object_redshift=None,
fixed_metallicity=None,
add_duste=False,
**extras):
from prospect.models.sedmodel import SedModel
from prospect.models.templates import TemplateLibrary
from prospect.models import priors
from prospect.models import transforms
from prospect.models.templates import adjust_continuity_agebins
from astropy.cosmology import WMAP9 as cosmos
import prospect
# Get (a copy of) one of the prepackaged model set dictionaries.
# Get the 2018 prospector-alpha model manually
model_params = (TemplateLibrary["continuity_sfh"])
model_params.update(TemplateLibrary["dust_emission"])
model_params.update(TemplateLibrary["nebular"])
model_params.update(TemplateLibrary["agn"])
# Set the dust and agn emission free
model_params["fagn"]["isfree"] = True
model_params["agn_tau"]["isfree"] = True
# Complexify the dust attenuation
model_params["dust_type"] = {
"N": 1,
"isfree": False,
"init": 4,
"units": "FSPS index"
}
model_params["dust2"]["prior"] = priors.TopHat(mini=0.0, maxi=4.0)
model_params["dust1"] = {
"N": 1,
"isfree": False,
'depends_on': transforms.dustratio_to_dust1,
"init": 0.0,
"units": "optical depth towards young stars"
}
model_params["dust_ratio"] = {
"N": 1,
"isfree": True,
"init": 1.0,
"units": "ratio of birth-cloud to diffuse dust",
"prior": priors.ClippedNormal(mini=0.0, maxi=2.0, mean=1.0, sigma=0.3)
}
model_params["dust_index"] = {
"N": 1,
"isfree": True,
"init": 0.0,
"units": "power-law multiplication of Calzetti",
"prior": priors.TopHat(mini=-2.0, maxi=0.5)
}
# in Gyr
tuniv = cosmos.age(object_redshift).value
model_params = adjust_continuity_agebins(model_params, tuniv)
model_params["duste_qpah"]["isfree"] = False
model_params["duste_umin"]["isfree"] = False
model_params["duste_gamma"]["isfree"] = False
model_params["duste_qpah"]["init"] = 2.0
model_params["duste_umin"]["init"] = 1.0
model_params["duste_gamma"]["init"] = 0.01
model_params["duste_qpah"]["prior"] = priors.TopHat(mini=0.0, maxi=7.0)
model_params["duste_umin"]["prior"] = priors.TopHat(mini=0.1, maxi=25.0)
model_params["duste_gamma"]["prior"] = priors.TopHat(mini=0.0, maxi=1.0)
model_params["duste_qpah"]["disp_floor"] = 3.0
model_params["duste_umin"]["disp_floor"] = 4.5
model_params["duste_gamma"]["disp_floor"] = 0.15
model_params["duste_qpah"]["init_disp"] = 3.0
model_params["duste_umin"]["init_disp"] = 5.0
model_params["duste_gamma"]["init_disp"] = 0.2
model_params['gas_logz']["isfree"] = True
model_params['gas_logz']["init"] = 0.0
model_params['gas_logz']["prior"] = priors.TopHat(mini=-2.0, maxi=0.5)
model_params['gas_logu']["isfree"] = False
model_params['gas_logu']["init"] = -1.0
model_params['gas_logu']["prior"] = priors.TopHat(mini=-4.0, maxi=-1.0)
# Now add the lumdist parameter by hand as another entry in the dictionary.
# This will control the distance since we are setting the redshift to zero.
# In `build_obs` above we used a distance of 10pc to convert from absolute to apparent magnitudes,
# so we use that here too, since the `maggies` are appropriate for that distance.
# Let's make some changes to values appropriate for our objects and data
model_params["logzsol"]["prior"] = priors.TopHat(mini=-2.0, maxi=0.2)
model_params["dust_index"]["prior"] = priors.TopHat(mini=-2.2, maxi=0.4)
model_params["agn_tau"]["prior"] = priors.LogUniform(mini=5, maxi=1.5e2)
model_params["dust2"]["prior"] = priors.TopHat(mini=1e-6, maxi=3.0)
model_params['dust_type']['init'] = 0
model_params['fagn']['init'] = 0.5
model_params['dust2']['init'] = 0.1
# If we are going to be using emcee, it is useful to provide a
# minimum scale for the cloud of walkers (the default is 0.1)
model_params["agn_tau"]["disp_floor"] = 1
model_params["dust2"]["disp_floor"] = 1e-2
model_params["logzsol"]["disp_floor"] = 1e-3
model_params['fagn']['disp_floor'] = 1e-3
model_params['dust_index']['disp_floor'] = 1e-3
# Change the model parameter specifications based on some keyword arguments
if object_redshift is not None:
# make sure zred is fixed
model_params["zred"]['isfree'] = False
# And set the value to the object_redshift keyword
model_params["zred"]['init'] = object_redshift
# Change fit orders
tparams = {}
parnames = [m for m in model_params]
fit_order = [
'logmass', 'dust_index', 'dust2', 'logzsol', 'fagn', 'dust_ratio'
]
for param in fit_order:
tparams[param] = model_params[param]
for param in model_params:
if param not in fit_order:
tparams[param] = model_params[param]
model_params = tparams
########
model_params['add_neb_emission']['init'] = False
########
model_params['sfh']['init'] = 0
# Now instantiate the model object using this dictionary of parameter specifications
model = SedModel(model_params)
#print(model_params['agebins'])
return model
def build_model(redshift,
fixed_metallicity=None,
dust=False,
agn_mass=None,
agn_eb_v=None,
agn_torus_mass=None,
igm_absorbtion=True,
inclination=True,
**extras):
"""Build a prospect.models.SedModel object
:param object_redshift: (optional, default: None)
If given, produce spectra and observed frame photometry appropriate
for this redshift. Otherwise, the redshift will be zero.
:param ldist: (optional, default: 10)
The luminosity distance (in Mpc) for the model. Spectra and observed
frame (apparent) photometry will be appropriate for this luminosity distance.
:param fixed_metallicity: (optional, default: None)
If given, fix the model metallicity (:math:`log(Z/Z_sun)`) to the given value.
:param dust: (optional, default: False)
If `True`, add dust emission and associated (fixed) parameters to the model.
:returns model:
An instance of prospect.models.SedModel
"""
logging.debug(redshift)
logging.debug(fixed_metallicity)
logging.debug(dust)
logging.debug(agn_mass)
logging.debug(agn_eb_v)
logging.debug(agn_torus_mass)
logging.debug(igm_absorbtion)
logging.debug(extras)
# TODO increase galaxy extinction to 0.6
# Get (a copy of) one of the prepackaged model set dictionaries.
# This is, somewhat confusingly, a dictionary of dictionaries, keyed by parameter name
model_params = TemplateLibrary["parametric_sfh"] # add sfh and tau
# dust2 init -.6, uniform [0, 2] uniform prior
# delay-tau model with tau [0.1, 30] log-uniform prior
# tage (burst start?) init 1, uniform (tophat) [0.001, 13.8]
model_params['dust_type'] = {
"N": 1,
"isfree": False,
"init": 2
} # Calzetti, as opposed to 0 for power law
# fixed: Kroupa IMF
# mass: lower 10^9, upper 10^12
# TODO set to logzsol=0 fixed, make this assumption
# dust (optical depth at 5500A): init=0.6, prior [0, 2], hopefully this is okay?
# sfh: 'delay-tau' is 4
# _parametric_["tau"] = {"N": 1, "isfree": True,
# "init": 1, "units": "Gyr^{-1}",
# "prior": priors.LogUniform(mini=0.1, maxi=30)}
# my convention: if None, don't model. if value, model as fixed to value. If True, model as free.
# Change the model parameter specifications based on some keyword arguments
if fixed_metallicity is None:
logging.info('Using free metallicity')
model_params['logzsol']['isfree'] = True
else:
# make it a fixed parameter
logging.info(f'Using fixed metallicity of {fixed_metallicity}')
model_params["logzsol"]["isfree"] = False
#And use value supplied by fixed_metallicity keyword
model_params["logzsol"]['init'] = fixed_metallicity
if dust:
# Add dust emission (with fixed dust SED parameters)
logging.info('Including dust emission fixed parameters')
model_params.update(TemplateLibrary["dust_emission"])
else:
logging.warning('Not using dust emission')
if igm_absorbtion:
logging.info('Using fixed IGM absorption (0.1 by default)')
# Add (fixed) IGM absorption parameters: madau attenuation (fixed to 1.)
model_params.update(TemplateLibrary["igm"])
else:
logging.warning('Not using IGM absorbtion')
if redshift is None:
raise ValueError('Redshift set to None - must be included in model!')
if isinstance(redshift, float):
logging.info('Using fixed redshift of {}'.format(redshift))
# Set redshift as fixed to value
model_params["zred"]['isfree'] = False
model_params["zred"]['init'] = redshift
else:
logging.info('Using free redshift')
# Set redshift as free
assert redshift == True # not just truthy, but exactly True/bool
model_params["zred"]['isfree'] = True
if agn_mass is None:
logging.warning('No AGN mass supplied - AGN not modelled')
# finish early
model = SedModel(model_params)
return model
elif isinstance(agn_mass, float):
logging.info('AGN mass will be fixed at {}'.format(agn_mass))
model_params['agn_mass'] = {
'N': 1,
'isfree': False,
'init': agn_mass
} # units?
else:
assert agn_mass == True
logging.info('AGN mass will be free parameter')
model_params['agn_mass'] = {
'N': 1,
'isfree': True,
'init': 1.,
'prior': priors.LogUniform(mini=1e-7, maxi=15)
}
if agn_eb_v is None:
logging.warning('AGN extinction not modelled')
else:
assert agn_mass
if isinstance(agn_eb_v, float):
logging.info('Using fixed agn_eb_v of {}'.format(agn_eb_v))
assert agn_mass
model_params['agn_eb_v'] = {
"N": 1,
"isfree": False,
"init": agn_eb_v,
"units": "",
'prior': priors.TopHat(mini=0., maxi=0.5)
}
elif agn_mass == True:
logging.info('Using free agn_eb_v')
model_params['agn_eb_v'] = {
"N": 1,
"isfree": True,
"init": 0.1,
"units": "",
'prior': priors.TopHat(mini=0., maxi=0.5)
}
else:
logging.warning('Not modelling AGN disk')
if agn_torus_mass is None:
logging.warning('Not modelling AGN torus')
elif isinstance(agn_torus_mass, float):
logging.info('Using fixed obscured torus of {}'.format(agn_torus_mass))
model_params['agn_torus_mass'] = {
"N": 1,
"isfree": False,
"init": agn_torus_mass,
"units": "",
'prior': priors.LogUniform(mini=1e-7, maxi=15)
}
else:
logging.info('Using free obscured torus')
model_params['agn_torus_mass'] = {
"N": 1,
"isfree": True,
"init": .1,
"units": "",
'prior': priors.LogUniform(mini=1e-7, maxi=15)
}
if inclination is None:
raise ValueError(
'No model inclination preference supplied - set float to fix, or set True to leave free, but set something!'
)
elif isinstance(inclination, float):
logging.info('Using fixed inclination of {}'.format(inclination))
model_params['inclination'] = {
"N": 1,
"isfree": False,
"init": inclination,
"units": "",
'prior': priors.TopHat(mini=0., maxi=90.)
}
else:
logging.info('Using free inclination')
model_params['inclination'] = {
"N": 1,
"isfree": True,
"init": 60.,
"units": "",
'prior': priors.TopHat(mini=0., maxi=90.)
}
# explicitly no FSPS dusty torus
# model_params['fagn'] = None
# model_params['agn_tau'] = None
# Now instantiate the model object using this dictionary of parameter specifications
model = SedModel(model_params)
return model
def build_model(binNum=0,
fit_metallicity=1,
fit_redshift=1,
infile=None,
add_neb=0,
metalfile=None,
cluster=None,
ID=0,
**extras):
import numpy as np
from astropy.table import Table
from prospect.models.priors import ClippedNormal, LogUniform, TopHat
from prospect.models.sedmodel import SedModel
from prospect.models.templates import TemplateLibrary
from prospect.sources.constants import cosmo
table = Table.read(infile)
zobs = table['z'][binNum]
max_age = cosmo.age(zobs).value
sma, smb = table['sma'][binNum], table['smb'][binNum]
R_e, width = table['R_e'][binNum], table['width'][binNum]
radius = (sma - width) * np.sqrt(smb / sma) / R_e
metallicities = Table.read(metalfile)
idx = np.where((metallicities['cluster'] == cluster)
& (metallicities['ID'] == ID))[0][0]
central_metallicity = metallicities['centrallogZ'][idx]
sigma_metallicity = metallicities['sigma'][idx]
init_metallicity = central_metallicity - 0.1 * radius
model_params = TemplateLibrary['parametric_sfh'] # delay-tau model
model_params['zred']['init'] = zobs
model_params['zred']['isfree'] = bool(fit_redshift)
model_params['zred']['prior'] = TopHat(mini=zobs - 0.01, maxi=zobs + 0.01)
model_params['mass']['init'] = 1e7
model_params['mass']['prior'] = LogUniform(mini=1e5, maxi=1e10)
model_params['logzsol']['init'] = init_metallicity
model_params['logzsol']['isfree'] = bool(fit_metallicity)
model_params['logzsol']['prior'] = ClippedNormal(mean=init_metallicity,
sigma=sigma_metallicity,
mini=-2,
maxi=0.2)
model_params['dust2']['init'] = 0.3
model_params['dust2']['prior'] = TopHat(mini=0.0, maxi=2.0)
model_params['tage']['init'] = 6.24
model_params['tage']['prior'] = TopHat(mini=0.001, maxi=max_age)
model_params['tau']['init'] = 0.29
model_params['tau']['prior'] = LogUniform(mini=0.1, maxi=10)
model_params['imf_type']['init'] = 1 # Chabrier (2003) IMF
model_params['dust_type']['init'] = 1 # Cardelli+ (1989) MW extinction
if add_neb:
model_params.update(TemplateLibrary['nebular'])
model = SedModel(model_params)
return model