deeplenstronomy 0.0.2.1

wrap lenstronomy for efficient simulation generation

Welcome to deeplenstronomy!

deeplenstronomy is a tool for simulating large datasets for applying deep learning to strong gravitational lensing. It works by wrapping the functionalities of lenstronomy in a convenient yaml-style interface, allowing users to embrace the astronomer part of their brain rather than their programmer part when generating training datasets.

Installation

With conda (Recommended)

• Step 0: Set up an environment. This can be done straightforwardly with a conda installation:

conda create -n deeplens python=3.7 jupyter scipy pandas numpy matplotlib astropy h5py PyYAML mpmath future
conda activate deeplens

• Step 1: pip install lenstronomy
• Step 2: pip install deeplenstronomy

With pip

• Step 1: pip install deeplenstronomy

Getting Started and Example Notebooks

Start by reading the Getting Started Guide to familiarize yourself with the deeplenstronomy style.

After that, check out the example notebooks below:

Notebooks for deeplenstronomy Utilities

• Creating deeplenstronomy Configuration Files
• Generating Datasets
• Visualizing deeplenstronomy Images
• Utilizing Astronomical Surveys
• Defining Your Own Probability Distributions
• Using Your Own Images as Backgrounds
• Simulating Time-Series Datasets

Notebooks for Applying deeplenstronomy to Machine Learning Analyses

• Using deeplenstronomy for Active Learning
• Using deeplenstronomy for Classification and Regression

Notebooks for Suggested Science Cases

• A Walkthrough of Using deeplenstronomy for Science

API Documentation

deeplenstronomy is designed so that users only need to work with their personal configuration files and the dataset generatation and image visualization functions. However, if you would like to view the full API documentation, you can visit the docs page.

Citation

If you use deeplenstronomy in your work, please include the following citations:

@article{deeplenstronomy,
  doi = {10.21105/joss.02854},
  url = {https://doi.org/10.21105/joss.02854},
  year = {2021},
  publisher = {The Open Journal},
  volume = {6},
  number = {58},
  pages = {2854},
  author = {Robert Morgan and Brian Nord and Simon Birrer and Joshua Yao-Yu Lin and Jason Poh},
  title = {deeplenstronomy: A dataset simulation package for strong gravitational lensing},
  journal = {Journal of Open Source Software}
}

@article{lenstronomy,
    title     =   "lenstronomy: Multi-purpose gravitational lens modelling software package",
    journal   =   "Physics of the Dark Universe",
    volume    =   "22",
    pages     =   "189 - 201",
    year      =   "2018",
    issn      =   "2212-6864",
    doi       =   "10.1016/j.dark.2018.11.002",
    url       =   "http://www.sciencedirect.com/science/article/pii/S2212686418301869",
    author    =   "Simon Birrer and Adam Amara",
    keywords  =   "Gravitational lensing, Software, Image simulations"
}

Contact

If you have any questions or run into any errors with the beta release of deeplenstronomy, please don't hesitate to reach out:

Rob Morgan
robert [dot] morgan [at] wisc.edu

You can also message me on the DES, DELVE, LSSTC, deepskies, or lenstronomers Slack workspaces