lenstronomy: looking back on 2020 and an outlook for 2021

[Simon Birrer]

Dear lenstronomy user and developer community,

I would like to look back on the year and provide an overview of the activities and milestones that were achieved, thanks to many of you as active users and developers!

I also would like to provide an outlook to the next year outlining the goals and direction of the development I anticipate for next year.

2020 in review:

lenstronomy has been successfully used in 35 publications (published or submitted in 2020) which is more than a doubling relative to the previous year. (https://github.com/sibirrer/lenstronomy/blob/master/PUBLISHED.rst)

lenstronomy was employed for a wide range of tasks, from measuring the Hubble constant, quantifying dark matter substructure, automatized lens modeling, source reconstruction, galaxy-quasar host decomposition, galaxy morphology analyses, forecasting lensed gravitational wave events, generation of large training sets for neural network application in both lens modeling and lens finding, to enhance lensing maps in large scale simulation products, simulating and modeling lensed supernovae, and probably used for applications I am not even aware of.

In 2020 the number of direct contributors to the lenstronomy source code has more than doubled (https://github.com/sibirrer/lenstronomy/graphs/contributors).

Among the milestones in the lenstronomy developments of 2020 are:

- astropy affiliation (https://www.astropy.org/affiliated/)

- Improved and additionally added jupyter notebooks in the lenstronomy extensions (https://github.com/sibirrer/lenstronomy_extensions/tree/master/lenstronomy_extensions/Notebooks)

- Pre-configured simulation settings for different current and upcoming imaging surveys

- integration with the SLIT algorithm for source reconstruction (https://github.com/aymgal/SLITronomy)

- removed support for Python 2

- LightCone module

- re-design of the GalKin module

- updated Analysis module

- multiple new lens models (EPL, core profiles, multi-pole etc) and more consistent naming conventions

Many of these added features and functionalities were thanks to new contributors - many many thanks! Also, reporting of bugs, raising questions and providing suggestions greatly improved the stability and usability of many modules - many many thanks!

Overall, the major design pillars of lenstronomy proved robust in its development and only small backwards incompatible changes needed to be made over the course of this year.

Over the course of last year, 10 affiliated packages extending the functionality of lenstronomy in specific domains were made publicly available by the community. These packages range from adding novel source reconstruction algorithm to lenstronomy, providing (gravitational) wave propagation solutions, managing large training set generation, integrated BNN inferences, facilitating automatized lens modeling, facilitating cluster source reconstruction, as well as hierarchical inferences of population level lens and cosmological parameters.(https://github.com/sibirrer/lenstronomy/blob/master/AFFILIATEDPACKAGES.rst)

This ecosystem of different packages greatly enhance the use-cases, provide science analysis level products and demonstrate the modular buildup of lenstronomy to enable diverse applications. These packages are highly impacted by the expertise of the domain specific application and also reduce the gap between the lenstronomy software and specific scientific analyses. These packages are written in high-quality code and they have the potential to become, by itself, heavily influential and popular software products. Many many thanks to all the developers to put out their packages open-source and making use of lenstronomy as part of their functionality!

Outlook to 2021:

I define three milestones for lenstronomy in the next year to consolidate the recent success, keep the momentum in the community, and to enable the best possible science for the user community.

continuity:

with the ever more complex analyses tasks the user community makes use of lenstronomy, and more and more packages with vital dependencies to lenstronomy appearing, backwards compatibility becomes ever more critical. Backwards incompatible changes are only going to be made in a sequence of a first release with depreciation warnings (still compatible), followed by a pre-release, and only after a brace period the changes get activated in the PyPi release version.

At the current stage, I do not anticipate backwards-incompatible changes.

ensuring (academic) credits for developers and contributors:

The original lenstronomy release publication came out in Spring 2018 (Birrer & Amara 2018). In the almost three years since, very valuable contributions to the development of lenstronomy has been made by more than a dozen additional developer. To ensure citation credits for the use of lenstronomy is extended to reflect the development since the last publication, I plan to a JOSS paper highlighting the high-level role of lenstronomy with an extended author. I currently expect a submission in April or May.

facilitating an ecosystem of inter-related software:

I anticipate an increased usage and development of software and wrappers making use of lenstronomy for ever more complex tasks, such as end-to-end inference pipelines. There is great potential for community benefits when these packages are as compatible and complementary as possible. I encourage active engagement and communications among different developer and the wider user community. This mailing list and in particular the lenstronomy Slack channel can serve as a platform for these discussions and exchanges. I am very happy to hear your feedback and ideas how we can actively foster a community and to enhance the impact of each of us participating and contributing.

Many thanks to you! lenstronomy would not be what it is without an active user and developer community. I wish you a relaxing holiday break, a good start in 2021 and hope the next year will have less bugs to be fixed and more science to be done :)

Happy Xoding!

Simon

PS:

- If you have a publication using lenstronomy, feel free to advertise it on the Slack channel and make sure I add it on the list of published work :)

- If you have a package or wrapper using functionalities of lenstronomy, please reach out to me and I am happy to advertise it more broadly to the community :)

- If you have colleagues who don’t know about lenstronomy but might be interested in using it, spread the word :)

- If you have colleagues using lenstronomy but are not on this mailing list, suggest them to join :)

- If you like lenstronomy or a feature of an affiliated package, you might want to give it a *star* on GitHub (it enhances the visibility) :)