Room Assignments for Thursday, August 17
Statistical Methods: 203
Computational Methods: 207
Observational Strategies: 107
Instrumentation Challenges: 108
Stellar Photospheres: 206
EPRV III will host five breakout sessions on a variety of topics. Here is a brief summary of the subject and scope of each session.
Computational Methods
Chair: Eric Ford
The Computational Methods breakout sessions will focus on improving computational methods for estimating the marginalized likelihood (aka “evidence”) of N-planet models based on radial velocity time series data. In particular, we will attempt to interpret the results of the EPRV3 Evidence Challenge, in which participants were asked to estimate the marginalized likelihood for zero, one, two and three planet models based on six simulated datasets. We will start by presenting a summary of the initial results and attempting to identify any “easy-to-fix” issues. Second, we will attempt to understand the reasons for the difference performance (e.g., precision, accuracy, bias, consistency) of the different methods and implementations and identify ways to improve their performance. Third, we will consider the computational cost and opportunities for improving speed and parallelism of methods and implementations that can provide a high-quality estimate of the marginalized likelihood. Finally, we will discuss whether and how we should organize the work to summarized the results of the EPRV3 Evidence Challenge for publication.
Meeting participants with experience and/or interest in statistical methodology and/or computation are encouraged to attend, even if they did not submit entries to the EPRV3 Evidence Challenge. We anticipate that the first session (and the final report in plenary session) will be of most interest to those who are tangentially interested, as days two and three are likely to become more technical with some participants engaging in coding and data analysis, while other participants discussing potential new/improved algorithms for future implementation.
Instrumentation Challenges in the 10 cm/s Era
Chairs: Arpita Roy, Sam Halverson, Ryan Terrien
The Instrumentation Challenges in the 10cm/s Era Breakout Session will be a broad foray into cutting edge solutions to long-standing problems in precision spectroscopy, as well as a deep dive into persisting concerns and technological hurdles. We will focus on comprehensive error budget driven instrument design, with an acute awareness of science requirements for tailored next-generation spectrometers. Both hardware and software/RV analysis issues will be addressed, so participants interested in all aspects of instrumentation are encouraged to attend. While we will pursue an overall narrative over the 3 days, feel free to attend the sessions that are of particular interest to you. To get the discussion started, please add questions and sub-topics that you think are important to current and future RV instrumentation on this Google Doc. Even if you cannot make it to some of the sessions, telling us both your overarching and detailed concerns will help make this a more well-rounded and useful conversation.
Community Forum: The path to 1cm/s instrumental precision
Statistical Methods
Chair: Rodrigo Díaz
The Statistical Methods breakout sessions will bring together exoplanet researchers and statisticians to discuss the state-of-the-art methods and algorithms used in the analysis of radial velocity time series. We plan to cover some of the more pressing aspects of the subject, such as using periodograms to detect signals, modelling of activity effects through non-parametric methods, and employing hierarchical Bayesian models to explore the underlying population of planets. The meetings will consist of a series of talks followed by open discussion, with the aim of producing a fruitful exchange of ideas and resources. People interested in “low signal-to-noise ratio” data analysis should find these sessions particularly interesting. Each day will address one of the topics, so partial attendance is welcome. On day 3, we will share the first part of the session with the Stellar Activity breakout session.
The planned schedule is:
Day 1 (Mon): Non-parametric models / Autoregressive models.
Day 2 (Wed): Hierarchical Bayesian models.
Day 3 (Thu): Periodograms (merged partially with the Stellar Activity session).
Observational Strategies
Chair: Jennifer Burt
The Observational Strategies breakout session will begin by recapping current target selection and scheduling approaches used by the world’s top RV facilities, and discussing the role of collaborative RV science in the era of K2 and TESS. We will then then focus on improvements that can be made to our observational efforts based on the community’s growing understanding of stellar activity, atmospheric effects, instrumental noise, and how best to model RV data. Each day will consist of a series of talks/tutorials followed by guided discussion among participants, and on day 2 we will host a shared talk series with the stellar activity group.
Notes for the week are linked here.
Our current schedule is:
Day 1:
Overview of current scheduling/observing approaches at top RV facilities
Discussion on the place of collaborative RV efforts in the era of K2 & TESS
Day 2:
Fabienne Bastien: Jitter & flicker
David Jones: Discussion of information needed for disentangling RV signals and stellar jitter
Johanna Teske: Results from experiment with variable timing/resolution PFS observations
Fabo Feng: Disentangling Keplerian signals from correlated noise in the Agatha framework
Ben Nelson: When non-Keplerian effects are important for modeling and detection
Day 3:
Sharon Wang: Summary & Discussion: Observational Strategies for Mitigating Telluric Contamination
Angie Wolfgang: Discussion on how observational choices affect population inference from the M-R diagram
Observing strategy challenge
Stellar Photospheres and Modeling Jitter
Chair: Heather Cegla
The breakout sessions on Stellar Photospheres and Modelling Jitter will cover a variety of topics. On Monday, there will be overviews of popular stellar models (SOAP 2.0 and StarSim 2.0), methods we can use to validate them (including the Göttingen solar telescope and the HARPS-N solar telescope). On Wednesday we will combine with the first half of the Observational Methods breakout session to discuss techniques to minimize the impact of stellar photospheres on radial velocity measurements, and on Thursday we will combine with the first half of the Statistical Methods session to discuss how stellar activity can impact periodogram analysis and our subsequent planet confirmations. Throughout each of these sessions we will also discuss a variety of stellar activity indicators to determine those best suited to move us towards 10 cm/s precision. Each session will also have ~30 min of open discussion time that may be used during the session or at the end; participants are encouraged to post questions in advance (or during) to the Google document here. A more in depth schedule, alongside a list of abstracts, for each day can be found here. If you are willing to take notes please email h.cegla@unige.ch. Most importantly, there will be a limited supply of Swiss chocolate for those who actively participate (questions posted on the Google document do count, but so do questions posed in person!).