Chains utilities
The chains_utils
module contains several functions
that can be used to analyze the MCMC chains produced by PTArcade. On this page,
we just highlight some of its functionalities. A more detailed discussion of
this module can be found in its reference page.
import_chains
-
This function can be used to load chains and model parameters of a PTArcade run. Just type
The
import_chains
function is particularly useful when you have multiple chains for the same model that you want to merge. In this case,import_chains
will do that for you by merging all the chains that are located inside the path that you pass to it. By default,import_chains
will also remove the first 25% of each chain before merging. If you want to change the amount of burn-in, you can do so via theburn_frac
argument.Finally, notice that by default,
import_chains
will only load the part of the chains corresponding to user-specified parameters, the likelihood, the posterior, and the hypermodel index. If you also want to load red noise and eventual DM parameters, you can do that by settingquick_import = False
. compute_bf
-
This function can be used to compute Bayes factors from runs where
mod_sel = True
in the configuration file. You can do this as follows:import ptarcade.chains_utils as utils params, chain = utils.import_chains('path_to_chains_folder') bf, bf_err = utils.compute_bf(chain, params)
This will give an estimate for the Bayes factor for the comparison of the user-specified signal against the SMBHB signal and the associated error. By default, the Bayes factor is calculated by dividing the number of points in the chain that fall in the hypermodel bin of the user-specified signal by the number of points falling in the bin of the reference SMBHB model. For a more precise estimate of the Bayes factor and associated error, you can set
bootstrap=True
. In this case, the Bayes factor and its standard deviation will be derived by using bootstrapping methods.