search

jiwoncpark / h0rton

Deep Modeling of Strong Gravitational Time Delay Lenses for Bayesian Inference of the Hubble Constant
MIT License
10 stars 1 forks source link

Generate cosmological summaries for individual and rung-level lenses #30

Closed jiwoncpark closed 3 years ago

jiwoncpark commented 4 years ago

Individual lenses To evaluate the TDLMC metrics, I need the individual H0 posteriors. For the simple MC default: k_ext-corrected H0 --> k_ext-corrected D_dt samples For the MCMC and hybrid: rescale D_dt samples by 1/(1-k_ext) --> k_ext-corrected D_dt samples --> cache the k_ext-corrected D_dt samples for individual lenses (correction: let's not) --> compute and log parameters of lognormal likelihood (D_dt_mu, D_dt_sigma) --> log the redshifts as well (for rung-level combination) --> sample from lognormal --> H0 samples --> compute and log parameters of Gaussian likelihood (H0_mean, H0_std)

Rung-level combination Start from D_dt_mu, D_dt_sigma of individual lenses --> format them for the hierarc MCMC sampler with likelihood_type = 'TDLogNorm' --> Obtain the rung-level H0 samples

jiwoncpark commented 3 years ago

A bit disorganized, but this branch will also contain the latest unit tests and miscellaneous updates.