#
# If using `pblum_mode='dataset-scaled'` while optimizing, it is generally a good idea to disable this (set to 'component-coupled' or 'dataset-coupled' and sample over `pblum` to account for any correlations between the luminosities and your other sampled parameters. For more details, see the [pblum tutorial](./pblum.ipynb).
#
# If your observational uncertainties are not reliable, you may also want to sample over `sigmas_lnf` (the [emcee fitting a line tutorial](https://emcee.readthedocs.io/en/stable/tutorials/line/) has a nice overview on the mathematics, and the [inverse paper](http://phoebe-project.org/publications/2020Conroy+) describes the implementation within PHOEBE).
#
# See the [distributions tutorial](./distributions.ipynb) for more details on adding distributions.
# In[12]:
b.add_distribution({'sma@binary': phoebe.gaussian_around(0.1),
'incl@binary': phoebe.gaussian_around(5),
't0_supconj': phoebe.gaussian_around(0.001),
'requiv@primary': phoebe.gaussian_around(0.2),
'pblum@primary': phoebe.gaussian_around(0.2),
'sigmas_lnf@lc01': phoebe.uniform(-1e9, -1e4),
}, distribution='ball_around_guess')
# It is useful to make sure that the model-parameter space represented by this initializing distribution covers the observations themselves.
# In[13]:
b.run_compute(compute='fastcompute', sample_from='ball_around_guess',
sample_num=20, model='init_from_model')
# In[14]:
# b.add_distribution({'sma@binary': pb.gaussian_around(0.1),
# 'incl@binary': pb.gaussian_around(5),
# 't0_supconj': pb.gaussian_around(0.001),
# 'requiv@primary': pb.gaussian_around(0.2),
# 'pblum@primary': pb.gaussian_around(0.2),
# 'sigmas_lnf@lc01': pb.uniform(-1e9, -1e4),
# }, distribution='ball_around_guess')
b.add_distribution(
{
't0_supconj': pb.gaussian_around(0.001),
'incl@binary': pb.gaussian_around(5),
'q': pb.gaussian_around(1),
'ecc': pb.gaussian_around(0.005),
'per0': pb.gaussian_around(0.2),
'sigmas_lnf@lc01': pb.uniform(-1e9, -1e4),
},
distribution='ball_around_guess')
b.run_compute(model='EMCEE_Fit',
compute='fastcompute',
sample_from='ball_around_guess',
sample_num=20)
b.set_value('init_from', 'ball_around_guess')
b.set_value(
'nwalkers', solver='emcee_solver', value=12
) #Define number of walkers. Must be twice number of parameters
b.set_value('niters', solver='emcee_solver',
value=250) #Define number of iterations
# * [rename_distribution](../api/phoebe.frontend.bundle.Bundle.rename_distribution.md)
# * [remove_distribution](../api/phoebe.frontend.bundle.Bundle.remove_distribution.md)
# In[7]:
print(b.get_distribution(distribution='mydist'))
# Now let's add another distribution, with the same `distribution` tag, to the inclination of the binary.
# In[8]:
b.add_distribution(qualifier='incl', component='binary',
value=phoebe.uniform(80,90),
distribution='mydist')
# In[9]:
print(b.get_distribution(distribution='mydist'))
# Accessing & Plotting Distributions
# --------------------
# The parameters we've created and attached are [DistributionParameters](../api/phoebe.parameters.DistributionParameter.md) and live in `context='distribution'`, with all other tags matching the parameter they're referencing. For example, let's filter and look at the distributions we've added.
# In[10]:
# In[4]:
b = phoebe.default_binary()
# In[5]:
b.set_value('latex_repr', component='binary', value='orb')
b.set_value('latex_repr', component='primary', value='1')
b.set_value('latex_repr', component='secondary', value='2')
# In[6]:
b.add_distribution(
{
'sma@binary': phoebe.uniform(5, 8),
'incl@binary': phoebe.gaussian(75, 10)
},
distribution='mydist')
# # Plotting Distributions
# In[7]:
dist = b.get_parameter('sma', component='binary',
context='component').get_distribution('mydist')
plt.clf()
figure = plt.figure(figsize=(4, 4))
_ = dist.plot(plot_uncertainties=False)
_ = plt.tight_layout()
# * [b.calculate_residuals](../api/phoebe.parameters.ParameterSet.calculate_residuals.md)
# * [b.calculate_chi2](../api/phoebe.parameters.ParameterSet.calculate_chi2.md)
# * [b.calculate_lnlikelihood](../api/phoebe.parameters.ParameterSet.calculate_lnlikelihood.md)
# * [b.calculate_lnp](../api/phoebe.frontend.bundle.Bundle.calculate_lnp.md)
#
# The log-probability used as the merit function within optimizers and samplers is defined as [calculate_lnp](../api/phoebe.frontend.bundle.Bundle.calculate_lnp.md)`(priors, combine=priors_combine)` + [calculate_lnlikelihood](../api/phoebe.parameters.ParameterSet.calculate_lnlikelihood).
#
# To see the affect of `priors_combine`, we can pass the `solver` tag directly to [b.get_distribution_collection](../api/phoebe.frontend.bundle.Bundle.get_distribution_collection.md), [b.plot_distribution_collection](../api/phoebe.frontend.bundle.Bundle.plot_distribution_collection.md), or [b.calculate_lnp](../api/phoebe.frontend.bundle.Bundle.calculate_lnp.md).
# In[14]:
b.add_distribution('teff@primary', phoebe.gaussian(6000,100), distribution='mydist01')
b.add_distribution('teff@secondary', phoebe.gaussian(5500,600), distribution='mydist01')
b.add_distribution('teff@primary', phoebe.uniform(5800,6200), distribution='mydist02')
# In[15]:
b.add_solver('sampler.emcee', priors=['mydist01', 'mydist02'], solver='myemceesolver')
# In[16]:
print(b.filter(qualifier='prior*'))
# Now we'll look at the affect of `priors_combine` on the resulting priors distributions that would be sent to the merit function.