July 11 2022 Release
-
autolens_workspace now has
advancedpackages which make navigation simpler for new users to find beginner scritps. -
Redesign of position based lens mass model resampling. This feature now uses a likelihood penalty term based API, which includes a new default approach which traces (y,x) coordinates of multiple images from the image plane to the source plane and decreases the likelihood based on how far part in the source-plane they are (as opposed to resampling the mass model). See this doc for a full descritipon (https://pyautolens.readthedocs.io/en/latest/general/demagnified_solutions.html).
-
If the position-based likelihood penalty term is not included in a fit using an
Inversion(e.g. pixelized source reconstruction) an exception is raised, because the fit will likely cause a demagnified solution. This can be disabled manually (see https://pyautolens.readthedocs.io/en/latest/general/demagnified_solutions.html). -
LightProfileOperated objects implemented, which are already convolved with the imaging dataset's PSF for modeling point source components in a galaxy (see https://github.com/Jammy2211/autolens_workspace/blob/release/scripts/imaging/modeling/advanced/light_parametric_operated__mass_total__source_parametric.py).
-
Numba is now an optional installation, see this doc page for a full description (https://pyautolens.readthedocs.io/en/latest/installation/numba.html).
-
Starting point API for starting an MCMC fit with walkers in certain positions or maximum likelihood estimator fit with a start point implemented (PyAutoLabs/PyAutoFit#562). The example tutorial script for this feature is not written yet.