Background

The scientific applications of strong gravitational lenses are broad, but they are currently limited by small sample sizes; ~1000 strong lenses are known overall, and specific applications typically use limited small samples. LSST’s sensitivity, image quality, survey volume and cadenced observations will be transformational in this field, enlarging samples by factors of ~100 and allowing us to cherry-pick systems best suited to particular science goals.  LSST will discover ~100 000 galaxy scale lenses, 1000s of group- and cluster-scale lenses, ~10 000 strongly lensed quasars and 500 lensed type Ia SNe. This increase creates tremendous opportunity but also poses significant challenges that need research and development to enable the scientific return from LSST’s strong lenses systems for any given science objective. For example:

• How do we find strong lenses in LSST data? With only 1 in 104 galaxies producing lensed images of a background galaxy, and remaining in a regime of high false-positive rates, amassing large samples of strong lenses that are complete and pure continues to be challenging, and will be even more so considering Rubin’s data volume. Identification of variable and non-variable lensed sources, built on sophisticated classification tools from machine learning and citizen science, play major roles in this strategy and developing the infrastructure to support this is a pressing goal. 

• How do we accurately and efficiently analyse large samples of lenses for key science goals from LSST data? Do we have in place efficient software to implement statistical and machine learning methods tailored to the analysis of LSST data?  How well do we understand the systematics? Ensuring LSST specific tools exist and work on the LSST archive environment is crucial for extracting early LSST science.

• How do we create reliable lens models for >104 LSST strong lenses? How accurate is automated lens modelling and what are the limitations? What fraction of LSST strong lenses will be edge/exotic cases and how will we identify and model them? Accurate lens modelling is often a time consuming and careful task, comparing lens modelling methods and understanding their limitations is an important consideration when assessing the accuracy of scientific results from strong lenses.  

• What is the best means of acquiring supplementary data on large samples of strong lenses?  Which science cases can be met with LSST data alone and which will require comprehensive follow-up of the lensed LSST sources? With such large samples from LSST, and a wide variety of science objectives, it is imperative that a co-ordinated international follow-up strategy is developed such that this can be done in an efficient way.

• How does SL enhance the scientific priorities of other SCs and within Rubin? What can SL teams learn from the experience, planned development and analysis ongoing in other SCs and other cross-collaboration groups (e.g. Deblending, photo-z, SED fitting) to optimise processing, deliverables and analysis? What additional infrastructure, software and data requirements might be required to achieve their goals?  How can SL teams take advantage of broker and TOM services developed to enable follow-up programs, and what additional functionality will they require?  How will SL teams run their analysis software within the framework of the Rubin Science Platform? What additional storage and processing requirements are needed beyond the RSP?

This LSST focussed strong-lensing workshop will address these issues by bringing together strong lensing experts from the LSST Strong Lensing Science Collaboration (SLSC), and the DESC Strong Lensing Working Group (DESC-SLTT) together with other SCs and working groups for which strongly lensed sources are relevant such as the Galaxies, AGN and Transients and Variable Stars SCs, the DESC Dark Matter and DESC Blending groups. Representatives from the DM will provide key input and individuals from the Informatics and Statistics SC (ISSC) will contribute expertise in the development of tailored analysis methods, to spur collaboration on data science challenges. While working closely together, these groups have not had a face-to-face joint strong lensing meeting since the formation of these collaborations. We would like to encourage discussion between the SCs and groups such that we can coordinate all of our scientific priorities, activities and development of tools in the LSST environment to maximise the return of Strong Lensing Science and tools to the LSST community. Given the large variety of applications of strong lensing over many fields, the step change from detailed studies of individual systems to statistically large samples requires significant development of tools for lens finding, modelling and analysis. This workshop will provide an overview of different science drivers, encourage synergies between approaches and to consolidate our efforts to maximize the exciting science from strong gravitational lenses.