MCMCResults
- class dysmalpy.fitting.MCMCResults(model=None, sampler_results=None, linked_posterior_names=None, blob_name=None, nPostBins=50)[source]
Bases:
BayesianFitResults
,FitResults
Class to hold results of MCMC fitting to DYSMALPY models.
Notes:
emcee
sampler_results object is ported to a dictionary inmcmcResults.sampler_results
The name of the free parameters in the chain are accessed through
mcmcResults.chain_param_names
, or more generally (separate model + parameter names) throughmcmcResults.free_param_names
Optional Attribute:
linked_posterior_names
Indicates if best-fit parameters should be measured in multi-dimensional histogram space. It takes a list of linked parameter sets, where each set consists of len-2 tuples/lists of the component + parameter names.
Structure Explanation:
To analyze component + param 1 and 2 together, and then 3 and 4 together:
linked_posterior_names = [joint_param_bundle1, joint_param_bundle2]
withjoint_param_bundle1 = [[cmp1, par1], [cmp2, par2]]
andjoint_param_bundle2 = [[cmp3, par3], [cmp4, par4]]
, for a full array of:linked_posterior_names = [[[cmp1, par1], [cmp2, par2]],[[cmp3, par3], [cmp4, par4]]]
.- To analyze component + param 1 and 2 together:
linked_posterior_names = [joint_param_bundle1]
withjoint_param_bundle1 = [[cmp1, par1], [cmp2, par2]]
, for a full array oflinked_posterior_names = [[[cmp1, par1], [cmp2, par2]]]
. - Example: Look at halo: mvirial and disk+bulge: total_mass together
linked_posterior_names = [[['halo', 'mvirial'], ['disk+bulge', 'total_mass']]]
- To analyze component + param 1 and 2 together:
Methods Summary
reload_sampler_results
([filename])Reload the MCMC sampler_results saved earlier
Methods Documentation