h0rton.h0_inference.h0_utils

Module Contents

Classes

CosmoConverter

Convert the time-delay distance to H0 and vice versa

Functions

gaussian_ll_pdf(x, mu, sigma)	Evaluates the (unnormalized) log of the normal PDF at point x
reorder_to_tdlmc(img_array, increasing_dec_i, abcd_ordering_i)	Apply the permutation scheme for reordering the list of ra, dec, and time delays to conform to the order in the TDLMC challenge
pred_to_natural_gaussian(pred_mu, pred_cov_mat, shift, scale)	Convert the BNN-predicted multivariate Gaussian parameters into the natural space counterparts by reverse transformation
get_lognormal_stats(all_samples)	Compute lognormal stats robustly, using median stats, assuming the samples are drawn from a lognormal distribution
get_lognormal_stats_naive(all_samples, all_weights=None)	Compute lognormal stats assuming the samples are drawn from a lognormal distribution
get_normal_stats(all_samples)
get_normal_stats_naive(all_samples, all_weights)
remove_outliers_from_lognormal(data, level=3)	Remove extreme outliers corresponding to level-STD away from the mean
combine_lenses(likelihood_type, z_lens, z_src, true_Om0, samples_save_path=None, corner_save_path=None, n_run=100, n_burn=400, n_walkers=10, **posterior_parameters)	Combine lenses in the D_dt space

h0rton.h0_inference.h0_utils.gaussian_ll_pdf(x, mu, sigma)

Evaluates the (unnormalized) log of the normal PDF at point x

x : float or array-like

point at which to evaluate the log pdf

mu : float or array-like

mean of the normal on a linear scale

sigma : float or array-like

standard deviation of the normal on a linear scale

h0rton.h0_inference.h0_utils.reorder_to_tdlmc(img_array, increasing_dec_i, abcd_ordering_i)

Apply the permutation scheme for reordering the list of ra, dec, and time delays to conform to the order in the TDLMC challenge

img_array : array-like

array of properties corresponding to the AGN images

array-like

img_array reordered to the TDLMC order

h0rton.h0_inference.h0_utils.pred_to_natural_gaussian(pred_mu, pred_cov_mat, shift, scale)

Convert the BNN-predicted multivariate Gaussian parameters into the natural space counterparts by reverse transformation

pred_mu : np.array of shape [Y_dim,] pred_cov_mat : np.array of shape [Y_dim, Y_dim] scale : np.array of shape [Y_dim,]

vector by which the features were scaled, e.g. the training-set feature standard deviations

shift : np.array of shape [Y_dim,]

vector by which the features were shifted, e.g. the training-set feature means

Derive it or go here: https://math.stackexchange.com/questions/332441/affine-transformation-applied-to-a-multivariate-gaussian-random-variable-what

mu : np.array of shape [Y_dim,]

mu in natural space

cov_mat : np.array of shape [Y_dim, Y_dim]

covariance matrix in natural space

class h0rton.h0_inference.h0_utils.CosmoConverter(z_lens, z_src, H0=70.0, Om0=0.3)

Convert the time-delay distance to H0 and vice versa

This was modified from lenstronomy.Cosmo.cosmo_solver to handle array types.

get_H0(self, D_dt)

get_D_dt(self, H0)

h0rton.h0_inference.h0_utils.get_lognormal_stats(all_samples)

Compute lognormal stats robustly, using median stats, assuming the samples are drawn from a lognormal distribution

h0rton.h0_inference.h0_utils.get_lognormal_stats_naive(all_samples, all_weights=None)

Compute lognormal stats assuming the samples are drawn from a lognormal distribution

h0rton.h0_inference.h0_utils.get_normal_stats(all_samples)

h0rton.h0_inference.h0_utils.get_normal_stats_naive(all_samples, all_weights)

h0rton.h0_inference.h0_utils.remove_outliers_from_lognormal(data, level=3)

Remove extreme outliers corresponding to level-STD away from the mean

data : np.array

data expected to follow a lognormal distribution

h0rton.h0_inference.h0_utils.combine_lenses(likelihood_type, z_lens, z_src, true_Om0, samples_save_path=None, corner_save_path=None, n_run=100, n_burn=400, n_walkers=10, **posterior_parameters)

Combine lenses in the D_dt space

true_Om0 : float

true Om0, not inferred

likelihood_type : str

‘DdtGaussian’, ‘DdtLogNorm’, ‘DdtHistKDE’ supported. ‘DdtGaussian’ must have ‘ddt_mean’, ‘ddt_sigma’. ‘DdtLogNorm’ must have ‘ddt_mu’ and ‘ddt_sigma’. ‘DdtHistKDE’ must have ‘lens_ids’ and ‘samples_dir’.