Data Challenge Final Jamboree

Friday April 30th, online, public, recorded

During this 3 ½ hour "Final Jamboree" event, we presented the challenge, its organization, summarized how the data were simulated and showed our in-house analysis for planet c.

We explained which metrics were applied to the participants' entries to rank them.

Finally, we announced the winners/prizes and discussed the lessons learned, the the legacy tutorial as well as heard the 3 finalist teams give their feedback!

About the Roman Exoplanet Imaging Data Challenge

The ROMAN Exoplanet Imaging Data Challenge ran for 8 months in 2019-2020. Organized by the Turnbull Science Investigation Team (SIT), this data challenge was a unique opportunity for exoplanet scientists of all backgrounds and experience levels to get acquainted with realistic Roman CGI (coronagraphic) simulated data with a new contrast regime enabling to unveil giant planets in reflected light.

Participants had to recover the astrometry of an exoplanetary system combining precursor radial velocity data with up to 6 coronagraphic imaging epochs throughout the mission. They had to perform accurate orbital fitting and determine the mass of any planet hidden in the data. It involved PSF subtraction techniques, post-processing, and other astrophysics hurdles to overcome such as contamination sources (stellar, extragalactic, and exozodiacal light).

We are now ready to announce the results to the community, present our in-house analysis, and discuss lessons learned and paths forward to best exploit scientifically the data coming out of Roman CGI. For that, we welcome you to attend our 3 ½ hour event on Friday, April 30 2021. The event was to be publicly live-streamed on STScI Live Science Events' channel (no need to have Facebook account). However, if you would like to actively participate in the discussion and/or the Q&As, please register and we will send you a link to connect.


Video of Session 1

11:00-12:00 Presentation (Chair: Margaret Turnbull)

  • Welcome message & motivations for this challenge - Margaret Turnbull

  • Data simulations (imaging/radial velocity) and architecture of the hidden exoplanetary system - Neil Zimmerman

  • Organization, steps, tutorials, legacy dataset & tutorial notebooks - Julien Girard

        • Source identification hurdles and "in-house" analysis tour for one planet - Ell Bogat

  • Q&A

12:00 - 12:45 Results (Chair: Margaret Turnbull)

  • Prizes: what and why - Julien Girard

  • Metrics and astrometric analysis of the results - Ell Bogat

  • Results and winners - Julien Girard

  • Q&A

30 min BREAK

Video of Session 2

1:15 - 2:00 Feedback and Lessons learned from finalist teams (Chair: Neil Zimmerman)

  • Introduction of the teams and lessons learned exercise - Julien Girard

  • Team ▼ TRIANGLE with a focus on photometry, phase curve and albedo

  • Team ♦︎ DIAMOND with a focus on how to apprehend CGI reflected light data from previous experiences with ground based data

  • Team ⬟ PENTAGON with a focus on orbital fitting (integrating RV) & post-processing

  • Q&A

2:00- 2:30 Discussion & wrap up (Chairs: Sergi Hildebrandt, Julien Girard and the Team)

  • How to prepare to real CGI data?

  • The increasing role of post-processing for the real, less idealistic data and/or the most recent observing scenarios with higher fidelity model uncertainty factors.

  • Other CGI modes (prism spectroscopy)

  • The star shade rendezvous

Contact: Julien Girard -

illustration of Nancy Grace Roman: Jessica Johnson