This website is home to data products from the application of Marginalized Analytic Dataspace Gaussian Inference for Component Separation (MADGICS) to decompose the public Gaia RVS spectra (most recently from Gaia DR3).

Please read our papers to learn more about the method and the catalog we produced. An introduction to MADGICS is provided in the catalog paper, though a more detail description is coming soon.

Title Reference
Measuring the 8621 Å Diffuse Interstellar Band in Gaia DR3 RVS Spectra: Obtaining a Clean Catalog by Marginalizing over Stellar Types Saydjari et al. (2022)
Marginalized Analytic Dataspace Gaussian Inference for Component Separation (MADGICS) Saydjari and Finkbeiner (in prep)
The catalog can be downloaded here (3.2 MB).


ewfloat64ÅDIB equivalent width
ew_errfloat64ÅDIB equivalent width uncertainty
lamfloat64ÅDIB center wavelength (stellar frame)
lam_errfloat64ÅDIB center wavelength uncertainty
sigfloat64ÅDIB width
sig_errfloat64ÅDIB width uncertainty
v_dust_lsrfloat64km/sDIB velocity in LSR
v_dust_lsr_errfloat64km/sDIB velocity in LSR uncertainty1
v_dust_lsr_COframefloat64km/sDIB velocity in LSR (of Dame CO map)
log10delchisqfloat64change in \(\chi^2\) from adding DIB component (log10)
dib_snrfloat64DIB signal-to-noise ratio (significance of detection)2
totchisqperdoffloat64total \(\chi^2\) per degree of freedom for MADGICS component model


1. This uncertainty does not include uncertainty in the stellar frame, which should be added in quadrature to obtain the true uncertainty.

2. This is a continuous quality metric that can be used as a cut to obtain cleaner catalogs.

Supplementary Values

In the main catalog file, we also provide a subset of values in the Gaia DR3 catalog, the Gaia DR3 DIB catalog, the Bailer-Jones distance catalog (CBJ), and the result of dustmap queries along the lines of sight. Any other value desired from external catalogs can be obtain by cross-match or using the Gaia source ID, which we have propagated.

gaia_source_idint64Gaia (e)DR3 source ID
linear_indxint64index of public RVS spectra, sorted on Gaia source ID
stellar_snrfloat64snr of the spectrum
rafloat64degright ascension
glonfloat64degGalactic longitude
glatfloat64degGalactic lattitude
Xfloat64kpcGalactic cartesian coordinate X
Yfloat64kpcGalactic cartesian coordinate Y
Zfloat64kpcGalactic cartesian coordinate Z
Ufloat64Galactic cartesian X-coordinate direction (unit direction vector)
Vfloat64Galactic cartesian Y-coordinate direction (unit direction vector)
Wfloat64Galactic cartesian Z-coordinate direction (unit direction vector)
parallaxfloat64masparallax of background star
parallax_errorfloat32masparallax of background star uncertainty
v_star_baryfloat32km/sradial velocity of background star (solar barycentric)
v_star_bary_errorfloat32km/sradial velocity uncertainty of background star (solar barycentric)
ebpminrpfloat32magE(BP-RP), color excess modeled by Gaia
sfdfloat64magreddening E(B-V) along line of sight, according to SFD dustmap
bayestar19float64magreddening E(B-V) at location of background star, according to Bayestar19 dustmap
dibew_gspspecfloat64ÅDIB equivalent width (GSP-Spec)
dibew_gspspec_uncertaintyfloat32ÅDIB equivalent width uncertainty (GSP-Spec)
dib_gspspec_lambdafloat32ÅDIB center wavelength (stellar frame) (GSP-Spec)
dib_gspspec_lambda_uncertaintyfloat32ÅDIB center wavelength uncertainty (GSP-Spec)
dibp2_gspspecfloat32ÅDIB width (GSP-Spec)
dibp2_gspspec_uncertaintyfloat32ÅDIB width uncertainty (GSP-Spec)
dibp0_gspspecfloat32DIB Gaussian amplitude (GSP-Spec)
dibqf_gspspecInt32GSP-Spec DIB pipeline quality factor
logchisq_gspspecfloat32GSP-Spec \(\chi^2\) (log) for stellar fit
r_med_geofloat32pcmedian geometric distance (CBJ)
r_lo_geofloat32pc16th-percentile geometric distance (CBJ)
r_hi_geofloat32pc84th-percentile geometric distance (CBJ)
r_med_photogeofloat32pcmedian photo-geometric distance (CBJ)
r_lo_photogeofloat32pc16th-percentile photo-geometric distance (CBJ)
r_hi_photogeofloat32pc84th-percentile photo-geometric distance (CBJ)
flag_cbjstringdistance inference flags (CBJ)

A small reproducibility repository (2.2 GB) is available from Zenodo. It provides all of the code, output summaries, and Jupyter notebooks which step through how the catalog was made and allow reproduction of every figure in the release paper. Individual files can be downloaded piecemeal from here. In addition, larger files containing all input spectra and output components are made available in the sources and out folders, respectively. Including these larger files, the repository is approx 500 GB, so care should be taken before starting a simple recursive download.

Please feel free to contact Andrew with any questions or for help using the catalog.

Andrew Saydjari
Catherine Zucker
J.E.G. Peek
Doug Finkbeiner