pip releases

pip release v0.6.0:

IMPORTANT:

If you have an existing config file you are using, you must add the following lines to the general.ini file:

log_file = output.log number_of_accepted_samples_between_output = 100

Interface:

• Code now outputs visualization of the current most likely model on the fly. Thus, you can check whether a fit is progressing well during the MultiNest analysis.

howtolens:

• New tutorial ‘chapter_2_lens_modeling/tutorial_7_masking_and_positions’, expands on masking an image and using positions to constrain the lens model.
• New tutorial ‘loading_and_preparing_data/preparing_data’ - describes the format observed data should be in and tools to convert data to that format.

autolens:

• The plane and ray_tracing modules now store grids as lists, and can produce multiple image_plane_images for each grid. This is a step towards fitting multiple images.
• The fitting module now fits lists of images, and produces lists of the model images, residuals, chi squareds etc. For single image fitting, non-plural properties (e.g. ‘model_image’, ‘residual’) can be used. This is another step towards multi-image fitting.
• New tools in the imaging.image.load_image_from_fits function for resizing images, recentering images, converting images from counts, etc.
• Added tools to create an image of a galaxy property (e.g. its surface density) and fit this quantity with another galaxy. Added galaxy phases to performt his fitting. Thus, intensities, surface densities, potentials and deflection angles can be fitted to one another. This will be used for automatic prior linking.
• Sensitivity fitter added, which allows one to quantify how sensitive we are to a change in lens model. Sensitivity phase added which performs this in a non linear search

Refactor:

• galaxy module separate from lensing module
• fitting module separate from lensing module
• inversion convolution moved to inversion module, in new file inversion.convolution.

pip release v0.5.4

• hotfix on plots of scatter points not showing.
• hotfix on incorrect model.info and model results display

pip release v0.5.3

Summary - Inversions are now fully implemented, including chapter 4 of how to lens. Galaxies are now named objects (as opposed to lists), which is reflect in their pass prior function).

Interface:

• Docker implementation now complete - see github for information on how to use.
• Galaxies now dictionaries and therefore named object.
• Small restructing of fitting / lensing / galaxy modules changing some imports and module names.

howtolens:

• Chapter 4 on inversions added.
• Renamed files to have non-numeric starts.

autolens:

• Plotting module now plots all coordinates in units of arcsecs or kpc, as opposed to always converting to pixels.
• Plotters now all have shared input allowing full customize of every figure.
• New customization includes ability to change point size of grid / mask and the colorbar.
• Refactored galaxy / galaxy_model modules from lensing into new module 'galaxy'. Added new module 'fitting' and separated lensing_fitting into its own module.
• Fixed issues with sim-link now outputting files or over-writing files of a MultiNest run.

pip release v0.4.3.0

Minor release - refactoring of plotting module. If Howtolens tutorials don't plot, grab this release.

pip release v0.4.3:

Summary - Main focus has been on improving visualization, so that is appears correctly in the howtolens tutorials and is tested so that we don't break the tutorials with code refactors.

Interface:

• Refactoring of plotting, which has changed the convention for plotting images of specific classes. Removed the 'individual' method such that every quantity of a class can now be plotted. For example, galaxy_plotters.plot_galaxy_individual(galaxy=galaxy, grid=grid, plot_intensities=True) is now just galaxy_plotters.plot_intensities(galaxy=galaxy, grid=grid).
• All of the above plotters can now take as input Matplotlib variables for controlling the figure size, title, axes labels etc.

howtolens:

• Updated plotters to reflect above changes and visualization now appears at reasonable sizes / aspect ratios.
• Added new tutorial 1_introduction/4_plane, which describes planes in more detial than before and introduces strong lensing concepts in a more detailed way.
• Extended 1_introduction/9_summary tutorial to more thoroughly describe the PyAutoLens user interface and data structures.
• 2_lens_modeling is now preloaded with configs to ensure fast non-linear searches.

autolens:

• unit tests of all plotting methods, finally!

pip release v0.4.2:

IMPORTANT:

• Coordinate convention has changed to (y,x), such that all coordinates (e.g. on a grid) are y first! This includes the centre=(y,x) of a light or mass profile.

UPDATE:

• Addresses bugs with plotters in the howtolens tutorials.
• Centralizes coordinate systems in the scaled_array module.
• Change tuple coordinate convention to (y,x), to match fits format. This includes the centre of a light and mass profiles.