This repository is a compilation of third-party catalogs, not an original discovery list. Public download is disabled. Access is granted only to approved project collaborators.

I understand that this is a homogenized compilation, not an original catalog; that scientific credit belongs to the source papers; that I will not copy or redistribute these files without written permission from the OneAstronomy maintainers; and that any approved use remains subject to the licenses and terms of the original catalogs.

OneAstronomy Strong Lens Master Catalog (v1)

Unified lens and non-lens tables compiled for the OneAstronomy strong-lensing workflow (2026).
This release is a homogenized collection of public catalogs. Candidates found by our own search will be added later.

The repository now includes catalog tables and Legacy Survey image cutouts (FITS + JPEG).

Where the files are

jady-zhao/OneAstro_StrongLens
├── data/                              # master tables
│   ├── lens_systems.csv               # 42,834 lens / candidate systems
│   ├── nonlens_systems.csv            # 62,858 explicit non-lenses
│   ├── field_dictionary.csv           # column definitions
│   └── merge_report.json
├── cutouts/legacy_survey/             # individual cutouts already on the Hub
│   └── {layer}/{L0|L1|L2}/{fits|jpg}/{uid}_ra..._dec....{fits|jpg}
│       layer = ls-dr11-south | ls-dr9-north | ls-dr11
│       non-lenses live under cutouts/legacy_survey/nonlens/...
└── cutouts/archives/                  # remaining cutouts as zip shards
    └── {layer}_{tier}_{fits|jpg}[_part_XX].zip

How to use the images:

Individual files (mostly L1/L2, plus a first batch of non-lens L0 FITS):
cutouts/legacy_survey/.../{uid}_ra..._dec....fits or .jpg
Join to the tables with uid.
Zip shards (remaining L0 / non-lens, to stay under Hub directory-size limits):
cutouts/archives/*.zip
Each zip is one folder of FITS or JPEG cutouts; filenames still contain uid.
Not in this repo: row-level source_records.csv (~62 MB). Ask if you need it.

Example download of one cutout:

from huggingface_hub import hf_hub_download

path = hf_hub_download(
    repo_id="jady-zhao/OneAstro_StrongLens",
    repo_type="dataset",
    filename="cutouts/legacy_survey/ls-dr11-south/L2/fits/<uid>_ra..._dec....fits",
)

What's in this release

File	Rows	Description
`data/lens_systems.csv`	42,834	Unique strong-lens / candidate systems
`data/nonlens_systems.csv`	62,858	Explicit non-lens / contaminant systems
`data/field_dictionary.csv`	—	Column definitions
`data/merge_report.json`	—	Build statistics (2026-08-24)
`SOURCES_AND_CITATIONS.md`	—	Required reading: provenance & papers to cite
`citations.bib`	—	BibTeX for source catalogs
`catalog_provenance.csv`	—	`catalog_id` → citation mapping + record counts
`cutouts/legacy_survey/`	—	Per-object Legacy Survey FITS/JPEG (uploaded as individual files)
`cutouts/archives/`	—	Remaining cutouts packed as zip shards (`*_part_XX.zip`)

Access

This compilation is not openly downloadable.

The dataset card is public so collaborators can find the project.
Data files are gated. Download is available only after the maintainers approve your access request.
Do not treat this table as an original discovery catalog.

Collaborators with approved access can then load:

from datasets import load_dataset

lens = load_dataset("jady-zhao/OneAstro_StrongLens", name="lens", split="train")
nonlens = load_dataset("jady-zhao/OneAstro_StrongLens", name="nonlens", split="train")

How the catalog was built

Ingest public discovery / spectroscopic / meta catalogs into uniform source_records.
Split sample_class = lens | nonlens (explicit negatives only).
Spatially cluster within lens_type at 2 arcsec (union-find).
Aggregate to one row per uid (SL2026_* lenses, NL2026_* non-lenses).
Drop non-lens records that fall within 2″ of a known lens of the same type.

Full workflow notes: see project DATASET_DOCUMENTATION.md (copied as DATASET_DOCUMENTATION.md in this folder).

Status and confidence tiers (lens table)

Field	Values	Use
`status`	`confirmed` / `likely` / `candidate` / `unknown`	Evidence strength (strongest across sources)
`confidence_tier`	`L2` / `L1` / `L0`	Training / validation split

Approximate counts (lens systems): L0 35,197 · L1 7,213 · L2 424 · confirmed 630.

Important: candidate means a positive selection by at least one catalog (weak evidence), not a non-lens.

Citation policy (please read)

If you use this compilation, please cite:

This dataset (Hugging Face / forthcoming DOI).
Every original catalog that contributed systems you actually use (see SOURCES_AND_CITATIONS.md).
The sources column on each row lists contributing catalog_ids.

This product is a compilation and homogenization. Scientific credit belongs primarily to the discovery papers and survey teams listed in the provenance files.

License and access

This product is a compilation, not an original discovery catalog. It is not released under an open license. Public download is disabled.

Compilation metadata and aggregation: restricted; no redistribution.
Underlying catalog values: still governed by the original papers and archives.
Files: gated; access by maintainer approval for collaborators only.

If you are a project partner, request access on this page and wait for approval. Cite the original catalogs listed in SOURCES_AND_CITATIONS.md, not this compilation as a primary source.

Contact / project

OneAstronomy — strong gravitational lensing workstream (2026).

Downloads last month: 131

Total file size:

12.4 GB