RGES-PIT/FalsePositives

Light curves generated by Alexander Stephan and RGES PIT WG6.

The following note is copy pasted from Alex Stephan's WG6 improved_noise folder README.

A given fits-file contains the resampled and filter-translated lightcurves of a variable
source. The original data was taken from OGLE and UKIRT, then resampled with Gaussian
process fitting, and finally interpolated to match OGLE V and I filters and UKIRT H and K
filters with the Roman filters planned to be used for the microlensing survey. An exception are flares, which are taken from TESS data.
Additionally, uncertainties are applied based on estimated signal-to-noise levels for Roman. 
The Roman filters are: 
F146 (wide filter with H band overlap) on a short cadence (about 12.1 minutes)
F087 (bluer filter with I band overlap) on a long cadence (about 6 hours)
F213 (redder filter with K band overlap) on a long cadence (about 6 hours)
F087 and F213 are planned to be offset by 3 hours. 
Timestamps for the lightcurves are included in the fits-files.

The added noise includes both photometric noise as well as noise from blending/crowding.

As the OGLE and UKIRT data is often too bright for Roman’s sensitivity range, the lightcurves have been shifted in magnitude to fall within the 15th to 24th magnitude range.

Long Period Variables (LPV)
RR Lyraes (RRLYR)
Heartbeat Stars (HB)
Eclipsing Binaries (ECL) -- note that these don't have F213 data 
Ellipsoidal Binaries (ELL)
Delta Scutis (DSCT)
Cepheids, type 1 and 2 (CEP, T2CEP)
M-dwarf flares with rotational variability (FL)
Dwarf novae/Cataclysmic variables (CV)

Light curves are organized by variability type. To download light curve data:

import duckdb
from huggingface_hub import HfFileSystem
con = duckdb.connect()
con.register_filesystem(HfFileSystem())
lc = con.execute("""
    SELECT name, JD, mag, mag_error
    FROM 'hf://datasets/RGES-PIT/FalsePositives/fl.parquet'
    LIMIT 10
""").df()
print(lc)

To inspect the metadata, see meta.parquet.