lenstronomy.Sampling.Likelihoods package

Submodules

lenstronomy.Sampling.Likelihoods.flux_ratio_likelihood module

class FluxRatioLikelihood(lens_model_class, flux_ratios, flux_ratio_errors, flux_ratio_measurement_bool=None, num_point_sources=1, source_type='INF', window_size=0.1, grid_number=100, polar_grid=False, aspect_ratio=0.5, point_source_redshift_list=None)[source]

Bases: object

Likelihood class for magnification of multiply lensed images.

__init__(lens_model_class, flux_ratios, flux_ratio_errors, flux_ratio_measurement_bool=None, num_point_sources=1, source_type='INF', window_size=0.1, grid_number=100, polar_grid=False, aspect_ratio=0.5, point_source_redshift_list=None)[source]
Parameters:

  • lens_model_class – LensModel class instance

  • flux_ratios – ratio of fluxes of the multiple images (relative to the first appearing in same order as the images)

  • flux_ratio_errors – errors in the flux ratios (relative to the first appearing). Alternatively a log-normal covariance matrix. Note: in the case of a covariance matrix, the errors are assumed to be log-normal, i.e. the logarithms of the flux ratios, ln(F[i]/F[0]) are assumed to have a multivariate Gaussian distribution, with the given covariance matrix.

  • flux_ratio_measurement_bool – list of bools of length of the point source model, indicating for which point sources there is a flux ratio measurement

  • num_point_sources (int>=0) – number of point sources in the point source model output

  • source_type – string, type of source, ‘INF’ specifies a point source, while ‘GAUSSIAN’ specifies a finite-size source modeled as a Gaussian

  • window_size – size of window to compute the finite flux

  • grid_number – number of grid cells per axis in the window to numerically compute the flux

  • point_source_redshift_list – list of redshifts to the different sources

logL(ra_image_list, dec_image_list, kwargs_lens, kwargs_special)[source]
Parameters:

  • kwargs_lens – lens model keyword argument list

  • kwargs_special – dictionary of ‘special’ keyword parameters

Returns:

log likelihood of the measured flux ratios given a model

property num_data
Returns:

integer, number of data points associated with the flux ratios

lenstronomy.Sampling.Likelihoods.image_likelihood module

class ImageLikelihood(multi_band_list, multi_band_type, kwargs_model, bands_compute=None, image_likelihood_mask_list=None, source_marg=False, linear_prior=None, check_positive_flux=False, kwargs_pixelbased=None, linear_solver=True)[source]

Bases: object

Manages imaging data likelihoods.

__init__(multi_band_list, multi_band_type, kwargs_model, bands_compute=None, image_likelihood_mask_list=None, source_marg=False, linear_prior=None, check_positive_flux=False, kwargs_pixelbased=None, linear_solver=True)[source]
Parameters:

  • bands_compute – list of bools with same length as data objects, indicates which “band” to include in the fitting

  • image_likelihood_mask_list – list of boolean 2d arrays of size of images marking the pixels to be evaluated in the likelihood

  • source_marg – marginalization addition on the imaging likelihood based on the covariance of the inferred linear coefficients

  • linear_prior – float or list of floats (when multi-linear setting is chosen) indicating the range of linear amplitude priors when computing the marginalization term.

  • check_positive_flux – bool, option to punish models that do not have all positive linear amplitude parameters

  • kwargs_pixelbased – keyword arguments with various settings related to the pixel-based solver (see SLITronomy documentation)

  • linear_solver – bool, if True (default) fixes the linear amplitude parameters ‘amp’ (avoid sampling) such that they get overwritten by the linear solver solution.

logL(kwargs_lens=None, kwargs_source=None, kwargs_lens_light=None, kwargs_ps=None, kwargs_special=None, kwargs_extinction=None, **kwargs)[source]
Parameters:

  • kwargs_lens – lens model keyword argument list according to LensModel module

  • kwargs_source – source light keyword argument list according to LightModel module

  • kwargs_lens_light – deflector light (not lensed) keyword argument list according to LightModel module

  • kwargs_ps – point source keyword argument list according to PointSource module

  • kwargs_special – special keyword argument list as part of the Param module

  • kwargs_extinction – extinction parameter keyword argument list according to LightModel module

Returns:

log likelihood of the data given the model, linear parameter inversion list

property num_data
Returns:

number of image data points

num_param_linear(kwargs_lens=None, kwargs_source=None, kwargs_lens_light=None, kwargs_ps=None, kwargs_special=None, kwargs_extinction=None, kwargs_tracer_source=None)[source]
Returns:

number of linear parameters solved for during the image reconstruction process

reset_point_source_cache(cache=True)[source]
Parameters:

cache – boolean

Returns:

None

lenstronomy.Sampling.Likelihoods.kinematic_2D_likelihood module

class KinLikelihood(kinematic_bin_2d_class, lens_model_class, lens_light_model_class, kwargs_data, idx_lens=0, idx_lens_light=0)[source]

Bases: object

Class to compute the likelihood associated to binned 2D kinematic maps.

__init__(kinematic_bin_2d_class, lens_model_class, lens_light_model_class, kwargs_data, idx_lens=0, idx_lens_light=0)[source]
Parameters:

  • kinematic_data_2d_class – KinBin class instance

  • lens_model_class – LensModel class instance

  • lens_light_model_class – LightModel class instance

  • kwargs_data – kwargs describing image rotation

  • idx_lens – int, index of the LensModel mass profile to consider for kinematics

  • idx_lens_light – int, index of the lens LightModel profile to consider for kinematics

calc_vrms(kwargs_lens, kwargs_lens_light, kwargs_special, verbose=False)[source]

Calculates binned vrms using SKiNN.

Parameters:

  • kwargs_lens – lens model kwargs list

  • kwargs_lens_light – lens light kwargs list

  • kwargs_special – cosmology and other kwargs

  • verbose – default False; if True print statements when out of bounds return binned vrms [km/s]; if SKiNN not installed return nan

logL(kwargs_lens, kwargs_lens_light, kwargs_special, verbose=False)[source]

Calculates Log likelihood from 2D kinematic likelihood.

Parameters:

  • kwargs_lens – lens model kwargs list

  • kwargs_lens_light – lens light kwargs list

  • kwargs_special – cosmology and other kwargs

  • verbose – default False; if True print statements when out of bounds return kinematics log likelihood

convert_to_nn_params(kwargs_lens, kwargs_lens_light, kwargs_special)[source]

Converts lenstronomy kwargs into input vector for SKiNN, also returns whether or not mass and light are aligned.

Parameters:

  • kwargs_lens – lens model kwargs list

  • kwargs_lens_light – lens light kwargs list

  • kwargs_special – cosmology and other kwargs

Returns:

parameters in GLEE convention to be input into NN

rescale_distance(image, kwargs_special)[source]

Rescales velocity map according to distance, requires lens redshift.

Parameters:

  • image – vrms image [km/s]

  • kwargs_special – kwargs with cosmological distances for rescaling return rescaled vrms image [km/s]

kwargs_data2image_input(kwargs_data)[source]

Creates the kwargs of the image needed for 2D kinematic likelihood.

Parameters:

kwargs_data – kwargs giving image and describing imaging data coordinate transformation

Return kwargs:

coordinate transformation kwargs as input for KinNNImageAlign class

update_image_input(kwargs_lens)[source]

Updates the image_input for rotation with the new values of orientation and center of the lens model.

Parameters:

kwargs_lens – lens kwargs with center and PA positions to be used as input

auto_binning(rotated_map, light_map)[source]

Function to convolve and bin the NN rotated output.

Parameters:

  • rotated_map – model vrms map [km/s] in data pixel coordinates

  • light_map – model light map in data pixel coordinates for weighting

Returns:

binned vrms [km/s] for comparison with data

lenstronomy.Sampling.Likelihoods.kinematic_NN_call module

class KinematicNN[source]

Bases: object

Class to call the NN to emulate JAM kinematics.

__init__()[source]
generate_map(input_p, verbose=False)[source]

Generate velocity map given input parameters.

Parameters:

input_p – vector of input parameters

Returns:

the velocity map associated with input_p

plot_map(ax, input_p)[source]
check_bounds(input_p, same_orientation=True, verbose=False)[source]

Checks to see if input parameters lie in bounds used for the training set.

Parameters:

  • input_p – input parameters to NN

  • same_orientation – default True; confirms that mass and light have same position angles

  • verbose – default False; if True prints statements when out of bounds

lenstronomy.Sampling.Likelihoods.position_likelihood module

class PositionLikelihood(point_source_class, image_position_uncertainty=0.005, astrometric_likelihood=False, image_position_likelihood=False, ra_image_list=None, dec_image_list=None, source_position_likelihood=None, source_position_tolerance=None, source_position_sigma=0.001, force_no_add_image=False, restrict_image_number=False, max_num_images=None)[source]

Bases: object

Likelihood of positions of multiply imaged point sources.

__init__(point_source_class, image_position_uncertainty=0.005, astrometric_likelihood=False, image_position_likelihood=False, ra_image_list=None, dec_image_list=None, source_position_likelihood=None, source_position_tolerance=None, source_position_sigma=0.001, force_no_add_image=False, restrict_image_number=False, max_num_images=None)[source]
Parameters:

  • point_source_class – Instance of PointSource() class

  • image_position_uncertainty – uncertainty in image position uncertainty (1-sigma Gaussian radially), this is applicable for astrometric uncertainties as well as if image positions are provided as data

  • astrometric_likelihood – bool, if True, evaluates the astrometric uncertainty of the predicted and modeled image positions with an offset ‘delta_x_image’ and ‘delta_y_image’

  • image_position_likelihood – bool, if True, evaluates the likelihood of the model predicted image position given the data/measured image positions

  • ra_image_list – list or RA image positions per model component

  • dec_image_list – list or DEC image positions per model component

  • source_position_likelihood – bool, if True, ray-traces image positions back to source plane and evaluates relative errors in respect ot the position_uncertainties in the image plane (image_position_uncertainty)

  • source_position_tolerance (None or float) – tolerance level (in arc seconds in the source plane) of the different images. If set =! None, then the backwards ray tracing is performed on the images and demand on the same position of the source is meant to match the requirements, otherwise a punishing likelihood term is introduced

  • source_position_sigma – r.m.s. value corresponding to a 1-sigma Gaussian likelihood accepted by the model precision in matching the source position transformed from the image plane. Used for source_position_likelihood

  • force_no_add_image – bool, if True, will punish additional images appearing in the frame of the modelled image(first calculate them)

  • restrict_image_number – bool, if True, searches for all appearing images in the frame of the data and compares with max_num_images

  • max_num_images – integer, maximum number of appearing images. Default is the number of images given in the Param() class

logL(kwargs_lens, kwargs_ps, kwargs_special, verbose=False)[source]
Parameters:

  • kwargs_lens – lens model parameter keyword argument list

  • kwargs_ps – point source model parameter keyword argument list

  • kwargs_special – special keyword arguments

  • verbose – bool

Returns:

log likelihood of the optional likelihoods being computed

check_additional_images(kwargs_ps, kwargs_lens)[source]

Checks whether additional images have been found and placed in kwargs_ps.

Parameters:

  • kwargs_ps – point source kwargs

  • kwargs_lens – lens model keyword arguments

Returns:

bool, True if more image positions are found than originally been assigned

static astrometric_likelihood(kwargs_ps, kwargs_special, sigma)[source]

Evaluates the astrometric uncertainty of the model plotted point sources (only available for ‘LENSED_POSITION’ point source model) and predicted image position by the lens model including an astrometric correction term.

Parameters:

  • kwargs_ps – point source model kwargs list

  • kwargs_special – kwargs list, should include the astrometric corrections ‘delta_x’, ‘delta_y’

  • sigma – 1-sigma Gaussian uncertainty in the astrometry

Returns:

log likelihood of the astrometirc correction between predicted image positions and model placement of the point sources

image_position_likelihood(kwargs_ps, kwargs_lens, sigma)[source]

Computes the likelihood of the model predicted image position relative to measured image positions with an astrometric error. This routine requires the ‘ra_image_list’ and ‘dec_image_list’ being declared in the initiation of the class.

Parameters:

  • kwargs_ps – point source keyword argument list

  • kwargs_lens – lens model keyword argument list

  • sigma – 1-sigma uncertainty in the measured position of the images

Returns:

log likelihood of the model predicted image positions given the data/measured image positions.

source_position_likelihood(kwargs_lens, kwargs_ps, sigma, hard_bound_rms=None, verbose=False)[source]

Computes a likelihood/punishing factor of how well the source positions of multiple images match given the image position and a lens model. The likelihood level is computed in respect of a displacement in the image plane and transposed through the Hessian into the source plane.

Parameters:

  • kwargs_lens – lens model keyword argument list

  • kwargs_ps – point source keyword argument list

  • sigma – 1-sigma Gaussian uncertainty in the image plane

  • hard_bound_rms – hard bound deviation between the mapping of the images back to the source plane (in source frame)

  • verbose – bool, if True provides print statements with useful information.

Returns:

log likelihood of the model reproducing the correct image positions given an image position uncertainty

property num_data
Returns:

integer, number of data points associated with the class instance

lenstronomy.Sampling.Likelihoods.prior_likelihood module

class PriorLikelihood(prior_lens=None, prior_source=None, prior_lens_light=None, prior_ps=None, prior_special=None, prior_extinction=None, prior_lens_kde=None, prior_source_kde=None, prior_lens_light_kde=None, prior_ps_kde=None, prior_special_kde=None, prior_extinction_kde=None, prior_lens_lognormal=None, prior_source_lognormal=None, prior_lens_light_lognormal=None, prior_ps_lognormal=None, prior_special_lognormal=None, prior_extinction_lognormal=None)[source]

Bases: object

Class containing additional Gaussian priors to be folded into the likelihood.

__init__(prior_lens=None, prior_source=None, prior_lens_light=None, prior_ps=None, prior_special=None, prior_extinction=None, prior_lens_kde=None, prior_source_kde=None, prior_lens_light_kde=None, prior_ps_kde=None, prior_special_kde=None, prior_extinction_kde=None, prior_lens_lognormal=None, prior_source_lognormal=None, prior_lens_light_lognormal=None, prior_ps_lognormal=None, prior_special_lognormal=None, prior_extinction_lognormal=None)[source]
Parameters:

  • prior_lens – list of [index_model, param_name, mean, 1-sigma priors]

  • prior_source – list of [index_model, param_name, mean, 1-sigma priors]

  • prior_lens_light – list of [index_model, param_name, mean, 1-sigma priors]

  • prior_ps – list of [index_model, param_name, mean, 1-sigma priors]

  • prior_special – list of [param_name, mean, 1-sigma priors]

  • prior_extinction – list of [index_model, param_name, mean, 1-sigma priors]

  • prior_lens_kde – list of [index_model, param_name, samples]

  • prior_source_kde – list of [index_model, param_name, samples]

  • prior_lens_light_kde – list of [index_model, param_name, samples]

  • prior_ps_kde – list of [index_model, param_name, samples]

  • prior_special_kde – list of [param_name, samples]

  • prior_extinction_kde – list of [index_model, param_name, samples]

  • prior_lens_lognormal – list of [index_model, param_name, mean, 1-sigma priors]

  • prior_source_lognormal – list of [index_model, param_name, mean, 1-sigma priors]

  • prior_lens_light_lognormal – list of [index_model, param_name, mean, 1-sigma priors]

  • prior_ps_lognormal – list of [index_model, param_name, mean, 1-sigma priors]

  • prior_special_lognormal – list of [param_name, mean, 1-sigma priors]

  • prior_extinction_lognormal – list of [index_model, param_name, mean, 1-sigma priors]

logL(kwargs_lens=None, kwargs_source=None, kwargs_lens_light=None, kwargs_ps=None, kwargs_special=None, kwargs_extinction=None, kwargs_tracer_source=None)[source]
Parameters:

kwargs_lens – lens model parameter list

Returns:

log likelihood of lens center

lenstronomy.Sampling.Likelihoods.time_delay_likelihood module

class TimeDelayLikelihood(time_delays_measured, time_delays_uncertainties, lens_model_class, point_source_class, time_delay_measurement_bool_list=None, bimodal_measurement=False)[source]

Bases: object

Class to compute the likelihood of a model given a measurement of time delays.

__init__(time_delays_measured, time_delays_uncertainties, lens_model_class, point_source_class, time_delay_measurement_bool_list=None, bimodal_measurement=False)[source]
Parameters:

  • time_delays_measured – relative time delays (in days) in respect to the first image of the point source

  • time_delays_uncertainties – time-delay uncertainties in same order as time_delay_measured. Alternatively a full covariance matrix that describes the likelihood.

  • lens_model_class – instance of the LensModel() class

  • point_source_class – instance of the PointSource() class, note: the first point source type is the one the time delays are imposed on

  • time_delay_measurement_bool_list – list of list of bool to indicate for which point source model a measurement is available. This list must have the same length as time_delays_measured and time_delays_uncertainties. Example for two point sources, imaged 4 times each: [[True, False, True], [True, True, True]]

  • bimodal_measurement (bool) – if True, two sets of delays are required. Only allowed for one set of point sources

logL(kwargs_lens, kwargs_ps, kwargs_cosmo, lambda_mst=1)[source]

Routine to compute the log likelihood of the time-delay distance.

Parameters:

  • kwargs_lens – lens model kwargs list

  • kwargs_ps – point source kwargs list

  • kwargs_cosmo – cosmology and other kwargs

  • lambda_mst (float or int) – mass-sheet transform of the input lens model that is not accounted for in the lens model parameters

Returns:

log likelihood of the model given the time delay data.

property num_data
Returns:

number of time delay measurements

lenstronomy.Sampling.Likelihoods.tracer_likelihood module

class TracerLikelihood(tracer_data, kwargs_model, kwargs_imaging, tracer_likelihood_mask=None, linear_solver=True)[source]

Bases: object

Class to evaluate the tracer map.

__init__(tracer_data, kwargs_model, kwargs_imaging, tracer_likelihood_mask=None, linear_solver=True)[source]
Parameters:

  • tracer_bands_compute – list of bools with same length as data objects, indicates which “band” to include in the fitting

  • tracer_likelihood_mask – boolean 2d arrays of size of images marking the pixels to be

  • linear_solver – boolean, if True, linear solver for surface brigthness amplitude is activated evaluated in the likelihood

logL(kwargs_tracer_source, kwargs_lens=None, kwargs_source=None, kwargs_lens_light=None, kwargs_ps=None, kwargs_special=None, kwargs_extinction=None, param=None)[source]
Parameters:

  • kwargs_lens – lens model keyword argument list according to LensModel module

  • kwargs_source – source light keyword argument list according to LightModel module

  • kwargs_lens_light – deflector light (not lensed) keyword argument list according to LightModel module

  • kwargs_ps – point source keyword argument list according to PointSource module

  • kwargs_special – special keyword argument list as part of the Param module

  • kwargs_extinction – extinction parameter keyword argument list according to LightModel module

Returns:

log likelihood of the data given the model

property num_data
Returns:

number of image data points

reset_point_source_cache(cache=True)[source]
Parameters:

cache – boolean

Returns:

None

Module contents