autolens.twins.ObservationSetup#

class ObservationSetup[source]#

Bases: object

How a lens is observed: mask, PSF, observed noise map and sampling.

Parameters:

mask (Mask2D) – The mask of the observation; twins are simulated and stored on it.
psf (Optional[Convolver]) – The PSF as an al.Convolver (already normalised, as held by the dataset).
noise_map (Array2D) – The observed noise map (standard deviation per pixel) on mask. It is the default noise of every twin (ruling R2).
pixel_scales (Optional[Tuple[float, float]]) – Default to the mask’s values.
origin (Optional[Tuple[float, float]]) – Default to the mask’s values.
over_sample_size_lp (Union[int, Array2D]) – Over-sampling of the light-profile and pixelization grids (an int or an aa.Array2D on mask), as used by the fit.
over_sample_size_pixelization (Union[int, Array2D]) – Over-sampling of the light-profile and pixelization grids (an int or an aa.Array2D on mask), as used by the fit.
cosmology – The cosmology the twin is interpreted in; None means the tracer’s own.
exposure_time (Optional[float]) – Only used by the "poisson" noise route, which requires both.
background_sky_level (Optional[float]) – Only used by the "poisson" noise route, which requires both.

Methods

`dataset_from`	An `al.Imaging` of `data` observed with this setup: masked with the setup mask and over-sampled with its over-sample sizes, following the aggregator's reload order (build unmasked, `apply_mask` once, `apply_over_sampling`).
`dataset_from_fit_directory`	Rebuild the masked, over-sampled `Imaging` a model-fit used from its `image/dataset.fits` (HDUs: 0 mask, 1 data, 2 noise map, 3 PSF, 4 light-profile over-sampling, 5 pixelization over-sampling), as written by `AnalysisImaging.save_attributes` (`autolens/imaging/model/analysis.py`).
`from_dataset`	The setup of a masked `al.Imaging` dataset: its mask, PSF, noise map and over-sampling.
`from_fit_directory`	The setup of a model-fit output directory (the folder holding `image/` and `files/`), read from `image/dataset.fits` exactly as the aggregator does (see `dataset_from_fit_directory`).
`with_`	A copy of this setup with the given fields replaced, e.g. `setup.with_(exposure_time=2000.0, background_sky_level=1.0)`.

Attributes

`fields`	The constructor arguments of this setup, by name.
`grid`	The masked lens-plane grid, over-sampled with `over_sample_size_lp`.

property fields: dict#: The constructor arguments of this setup, by name.

classmethod from_dataset(dataset, cosmology=None, exposure_time=None, background_sky_level=None)[source]#: The setup of a masked al.Imaging dataset: its mask, PSF, noise map and over-sampling. The PSF and noise map are copied so later changes to the dataset do not leak into the setup.

static dataset_from_fit_directory(path)[source]#

Rebuild the masked, over-sampled Imaging a model-fit used from its image/dataset.fits (HDUs: 0 mask, 1 data, 2 noise map, 3 PSF, 4 light-profile over-sampling, 5 pixelization over-sampling), as written by AnalysisImaging.save_attributes (autolens/imaging/model/analysis.py).

This replicates PyAutoGalaxy’s aggregator reload _imaging_from (autogalaxy/aggregator/imaging/imaging.py lines 29-115) and agg_util.mask_header_from line for line, for a single (non-summed) analysis, without needing an af.Fit. If that recipe changes, change this.

classmethod from_fit_directory(path, cosmology=None, exposure_time=None, background_sky_level=None)[source]#: The setup of a model-fit output directory (the folder holding image/ and files/), read from image/dataset.fits exactly as the aggregator does (see dataset_from_fit_directory).

with_(**overrides)[source]#

A copy of this setup with the given fields replaced, e.g. setup.with_(exposure_time=2000.0, background_sky_level=1.0).

Fields that depend on the mask (noise map, array over-sampling) are not re-derived: replacing the mask requires replacing them too.

property grid: Grid2D#: The masked lens-plane grid, over-sampled with over_sample_size_lp.

dataset_from(data, noise_map=None)[source]#

An al.Imaging of data observed with this setup: masked with the setup mask and over-sampled with its over-sample sizes, following the aggregator’s reload order (build unmasked, apply_mask once, apply_over_sampling).

Parameters:

data – An aa.Array2D (masked or native) or an ndarray (native or slim).
noise_map – Defaults to the setup’s observed noise map.