README.md

Description of the update of the lenstronomy

This repository contains modifications of lenstronomy based on the late August 2021 version of the original lenstronomy https://github.com/sibirrer/lenstronomy .

The purpose of this modification is to analyze the noise of the ALMA data by studying the non-diagonal noise pixel-pixel covariance matrix and thus help to fit the lens and source model of a gravitational lens picture.

Here is a list of the updates based on the original lenstronomy:

• 1. Add 'primary_beam' to the data_kwargs. Allow multiplying a primary beam after an image is generated and before get convolved by PSF in both ImageModel and fitting.
• 2. Add two more methods to calculate the log_likelihood for interferometry data using the natwt PSF and the corresponding covariance matrix. Use kwargs_data['likelihood_method'] to tell the fitting program which method to use. For details reading the comments in lenstronomy/Data/imaging_data.py .
• 3. Add 'use_linear_solver' to the data_kwargs, which (must be used together with kwargs_constraints = {'linear_solver':False/True}) controls whether we perform a linear solver to find the optimal 'amp's in fitting. Note that if the kwargs_data['likelihood_method'] == 'natwt_special', the linear solver is a different one designed for the natwt PSF. If kwargs_data['likelihood_method'] == 'diagonal' or 'eigen', the dafault linear solver of lenstronomy will be used.

Some example jupyter notebooks:

• Noise covariance matrix, eigen values and eigen modes -- a square sampling region