Dynamical Complexities in Planetary Systems, Near and Far

Wednesdays@NICO talk, February 22, 2017

Transcript

1. Dynamical Complexities in Planetary Systems, Near and Far Benjamin Nelson CIERA/NICO Data Science Scholar @exobenelson Wednesdays@NICO February 22, 2017 Wednesday, February 22, 17
2. Sun M V E M J S U N The Solar System Wednesdays@NICO February 22, 2017 Wednesday, February 22, 17
3. Sun M V E M J S U N Forming The Inner Solar System Grand Tack (2011) } - solves the “Mars problem” - explains asteroid belt dichotomy - brings water into inner SS Source: http://www.boulder.swri.edu/~kwalsh/GrandTack.html Wednesdays@NICO February 22, 2017 Wednesday, February 22, 17
4. Sun M V E M J S U N Forming The Inner Solar System Grand Tack (2011) } - solves the “Mars problem” - explains asteroid belt dichotomy - brings water into inner SS Source: http://www.boulder.swri.edu/~kwalsh/GrandTack.html Wednesdays@NICO February 22, 2017 Wednesday, February 22, 17
5. Sun M V E M J S U N Nice Model (2005) - explains Oort Cloud - explains Trojan systems - explains resonant TNOs - timing of Late Heavy Bombardment } Forming The Outer Solar System Wednesdays@NICO February 22, 2017 Wednesday, February 22, 17
6. Sun M V E M J S U N Nice Model (2005) - explains Oort Cloud - explains Trojan systems - explains resonant TNOs - timing of Late Heavy Bombardment } Forming The Outer Solar System Wednesdays@NICO February 22, 2017 Wednesday, February 22, 17
7. Sun M V E M J S U N Nice Model (2005) - explains Oort Cloud - explains Trojan systems - explains resonant TNOs - timing of Late Heavy Bombardment } Forming The Solar System Grand Tack (2011) } - solves the “Mars problem” - explains asteroid belt dichotomy - brings water into inner SS Wednesdays@NICO February 22, 2017 Wednesday, February 22, 17
8. Can we constrain formation histories of exoplanet systems? Wednesdays@NICO February 22, 2017 Wednesday, February 22, 17
9. Part 1: Individual Systems 55 Cancri Gliese 876 MCMC Bayesian model comparison Science! Methods! Wednesdays@NICO February 22, 2017 Wednesday, February 22, 17
10. How To Detect an Exoplanet: Radial Velocity Method Video source: h-p://www.eso.org/public/videos/eso1035g/ Wednesdays@NICO February 22, 2017 Wednesday, February 22, 17
11. How To Detect an Exoplanet: Radial Velocity Method Video source: h-p://www.eso.org/public/videos/eso1035g/ Wednesdays@NICO February 22, 2017 Wednesday, February 22, 17
12. How To Detect an Exoplanet: Radial Velocity Method Video source: h-p://www.eso.org/public/videos/eso1035g/ Jupiter’s RV signal on Sun: 12 m/s Earth’s RV signal on Sun: 0.1 m/s Today’s RV precision: ~1 m/s Wednesdays@NICO February 22, 2017 Wednesday, February 22, 17
13. Simple 1-planet System corner.py (pip install corner) Wednesdays@NICO February 22, 2017 Wednesday, February 22, 17
14. Complex 5-planet System corner.py (pip install corner) Wednesdays@NICO February 22, 2017 Wednesday, February 22, 17
15. Proposed displacement ( ) scaled Candidate for update Proposal i j k Radial velocity Using N-body Differential evolution MCMC (the data) (the model) (the method) Nelson, Ford, & Payne (2014) Wednesdays@NICO February 22, 2017 Wednesday, February 22, 17
16. Proposed displacement ( ) scaled Candidate for update Proposal i j k Radial velocity Using N-body Differential evolution MCMC (the data) (the model) (the method) Nelson, Ford, & Payne (2014) Wednesdays@NICO February 22, 2017 Wednesday, February 22, 17
17. Candidate for update i x y How Differential Evolution Works ter Braak (2006) Nelson, Ford, & Payne (2014) Wednesdays@NICO February 22, 2017 Wednesday, February 22, 17
18. Candidate for update i j k x y How Differential Evolution Works ter Braak (2006) Nelson, Ford, & Payne (2014) Wednesdays@NICO February 22, 2017 Wednesday, February 22, 17
19. Proposed displacement ( ) scaled Candidate for update i j k x y How Differential Evolution Works ter Braak (2006) Nelson, Ford, & Payne (2014) Wednesdays@NICO February 22, 2017 Wednesday, February 22, 17
20. Proposed displacement ( ) scaled Candidate for update Proposal i j k x y How Differential Evolution Works ter Braak (2006) Nelson, Ford, & Payne (2014) Wednesdays@NICO February 22, 2017 Wednesday, February 22, 17
21. i j k Proposal x y Efficient with correlated parameters ter Braak (2006) Nelson, Ford, & Payne (2014) Wednesdays@NICO February 22, 2017 Wednesday, February 22, 17
22. i j k Proposal x y Efficient with multiple modes ter Braak (2006) Nelson, Ford, & Payne (2014) Wednesdays@NICO February 22, 2017 Wednesday, February 22, 17
23. i j k Proposal x y Efficient with multiple modes Takeaway #1: number of Markov chains ≈ 3 x number of dimensions ter Braak (2006) Nelson, Ford, & Payne (2014) Wednesdays@NICO February 22, 2017 Wednesday, February 22, 17
24. i j k Proposal x y Efficient with multiple modes Takeaway #2: Boring methods papers can still get publicity. Just use a ridiculous acronym. ter Braak (2006) Nelson, Ford, & Payne (2014) Wednesdays@NICO February 22, 2017 Wednesday, February 22, 17
25. Beyond Random Walk MCMC Comparing the performance of these Python packages http://jakevdp.github.io/blog/2014/06/14/frequentism-and-bayesianism-4-bayesian-in-python/ http://andrewgelman.com/2015/10/15/whats-the-one-thing-you-have-to-know-about-pystan-and-pymc-click-here-to-find-out/ Goodman & Weare (2010) Foreman-Mackey+ (2013) Salvatier, Wiecki, & Fonnesbeck (2016) Hoffman & Gelman (2011) Carpenter+ (2017) Wednesdays@NICO February 22, 2017 Wednesday, February 22, 17
26. 55 Cancri Jupiter analog 55 Cnc A 0.01 0.1 1 10 e b c f d Distance (AU) transiting ~3:1 habitable zone? Wednesdays@NICO February 22, 2017 Wednesday, February 22, 17
27. 1. “e” should be roughly coplanar with other planets 3. “b” and “c” are dynamically coupled but not in resonance Dynamics in a Probabilistic Framework 8.63 ± 0.35 M⊕ (Winn+ 2011) 8.37 ± 0.38 M⊕ (Endl+ 2012) 8.09 ± 0.26 M⊕ (Nelson+ 2014) 2. Lower mass estimate for “e” 55 Cnc A e b c f d Nelson+ (2014) Wednesdays@NICO February 22, 2017 Wednesday, February 22, 17
28. 0.01 0.1 0.3 0.4 c b d Distance (AU) GJ 876 0.2 e 30 days 60 days 120 days 2 days Gliese 876 Wednesdays@NICO February 22, 2017 Wednesday, February 22, 17
29. 0.01 0.1 0.3 0.4 c b d Distance (AU) GJ 876 0.2 e 30 days 60 days 120 days 2 days Gliese 876 OGRE: h-p://github.com/dtamayo/OGRE Wednesdays@NICO February 22, 2017 Wednesday, February 22, 17
30. 0.01 0.1 0.3 0.4 c b d Distance (AU) GJ 876 0.2 e 30 days 60 days 120 days 2 days Planet or Stellar Activity? ? ? ? ? ? Wednesdays@NICO February 22, 2017 Wednesday, February 22, 17
31. 0.01 0.1 0.3 0.4 c b d Distance (AU) GJ 876 0.2 e 30 days 60 days 120 days 2 days Planet or Stellar Activity? ? ? ? ? ? Wednesdays@NICO February 22, 2017 Wednesday, February 22, 17
32. What is the evidence for n planets? p(d|M) = Z p(✓|M)p(d|✓, M)d✓ The probability of my dataset d being generated from some model M parameterized by θ... fully marginalized likelihood (FML) prior likelihood To choose between two competing models M1 and M2, take the ratio of their FMLs... = p(d|M2) p(d|M1) Bayes factor Wednesdays@NICO February 22, 2017 Wednesday, February 22, 17
33. Nested sampling Multinest: https://ccpforge.cse.rl.ac.uk/gf/project/multinest/ PyMultinest: https://github.com/JohannesBuchner/PyMultiNest DNest: https://github.com/eggplantbren/DNest4 Thermodynamic Integration: http://dan.iel.fm/emcee/current/user/pt/ Importance Sampling: https://github.com/benelson/FML Efficient ways to compute FMLs FML = Z p(✓)p(d|✓)d✓ = Z p(✓)p(d|✓) g(✓) g(✓)d✓ [ FML ⇡ 1 N X ✓i ⇠g(✓) p(✓i)p(d|✓i) g(✓i) Seth Pritchard (UT San Antonio) Wednesdays@NICO February 22, 2017 Wednesday, February 22, 17
34. ~3 x 103 ~107 And the best Gliese 876 model is... GJ 876 c b d e GJ 876 c b d e GJ 876 c b d e GJ 876 c b d e Bayes Factor = = = f ~1031 Nelson+ (2016) Wednesdays@NICO February 22, 2017 Wednesday, February 22, 17
35. Coplanar and Chaotic 1. strong upper limits on 3-dimensional orbital architecture 2. measured orbital chaos on the timescale of ~10 years Nelson+ (2016) Wednesdays@NICO February 22, 2017 Wednesday, February 22, 17
36. Forming Chaos in Gliese 876: Disk Turbulence Batygin, Deck, & Holman 2015 Wednesdays@NICO February 22, 2017 Wednesday, February 22, 17
37. Dempsey & Nelson, in prep. Adam Dempsey (NU CIERA) Forming Chaos in Gliese 876: No Disk Turbulence Wednesdays@NICO February 22, 2017 Wednesday, February 22, 17
38. Part 2: Planet Populations Super Earths Hot Jupiters Planet Pairs Hierarchical Bayes More FMLs (FML...) Science! Methods! Wednesdays@NICO February 22, 2017 Wednesday, February 22, 17
39. Kepler in a nutshell Credit: NASA/Kepler Mission/Dana Berry Wednesdays@NICO February 22, 2017 Wednesday, February 22, 17
40. Kepler in a nutshell Credit: NASA/Kepler Mission/Dana Berry Wednesdays@NICO February 22, 2017 Wednesday, February 22, 17
41. How To Detect an Exoplanet: Transit Method Wednesdays@NICO February 22, 2017 Wednesday, February 22, 17
42. The Kepler Yield Credit: Alex Parker (SWRI) Wednesdays@NICO February 22, 2017 Wednesday, February 22, 17
43. The Kepler Yield Credit: Alex Parker (SWRI) Wednesdays@NICO February 22, 2017 Wednesday, February 22, 17
44. Super-Earths / Mini-Neptunes 1 RE 4 RE 1.4 1.0 2.0 2.8 4.0 5.7 8.0 11.3 16.0 22.6 Earth Earths Super- Earths Small Neptunes Large Neptunes Giant Planets 85 Fressin+ (2013) Wednesdays@NICO February 22, 2017 Wednesday, February 22, 17
45. 1 RE 4 RE Rogers (2015) Properties of Super-Earths Most 1.6 RE Planets Are Not Rocky Wolfgang, Rogers, & Ford (2016) A Probabilistic Mass-Radius Relationship Wednesdays@NICO February 22, 2017 Wednesday, February 22, 17
46. Hot Jupiter Formation Disk migration Eccentric migration Two main camps for explaining these objects Before: After: Wednesdays@NICO February 22, 2017 Source: http://jila.colorado.edu/~pja/planet_migration.html Wednesday, February 22, 17
47. In terms of “x”... x ⌘ a aRoche actual planet-star separation distance at which star’s gravity will destroy planet* Wednesdays@NICO February 22, 2017 *not exactly, e.g., Valsecchi, Rasio, & Steffen 2014 Wednesday, February 22, 17
48. In terms of “x”... x ⌘ a aRoche actual planet-star separation distance at which star’s gravity will destroy planet* x can’t be less than 1! x can’t be less than 2! Eccentric migration Disk migration *not exactly, e.g., Valsecchi, Rasio, & Steffen 2014 Wednesdays@NICO February 22, 2017 Wednesday, February 22, 17
49. What do the data show? Network-based Frequency Analysis http://csun.uic.edu/codes/NFA.htm l Wednesdays@NICO February 22, 2017 Nelson, Ford, & Rasio (in review) Wednesday, February 22, 17
50. Modeling Two Overlapping Populations probabilistic graphical models with daft http://daft-pgm.org/ Wednesdays@NICO February 22, 2017 Nelson, Ford, & Rasio (in review) Wednesday, February 22, 17
51. Sampling a 300+ dimensional-space Hamiltonian Monte Carlo arXiv: 1701.02434 + Sampler Hoffman & Gelman 2011 Wednesdays@NICO February 22, 2017 Wednesday, February 22, 17
52. data model method Wednesdays@NICO February 22, 2017 Wednesday, February 22, 17
53. Different Datasets Different Inferred Populations RV and Kepler data are well modeled with a single population consistent with an eccentric migration history. HAT and WASP data are better explained with a mixture of two populations*: ~2/3 coming from eccentric migration ~1/3 coming from disk migration Project available at: https://github.com/benelson/hjs-with-stan Nelson, Ford, & Rasio (in review) Wednesdays@NICO February 22, 2017 * caveats include completeness within surveys, burn-in/convergence, model assumptions (e.g. two components, hard edges of truncated power law), no constraints via other data (e.g. spin-orbit alignment, metallicity, stellar age, multiplicity/companions) Wednesday, February 22, 17
54. Future Exoplanet Mission: TESS Wednesdays@NICO February 22, 2017 Expected to find 10,000+ Hot Jupiters (Sullivan+ 2015) Wednesday, February 22, 17
55. Period Ratios of Kepler Planet Pairs 1 2 3 4 5 6 7 8 Period Ratio PDF Steffen & Hwang (2015) Wednesdays@NICO February 22, 2017 Wednesday, February 22, 17
56. Period Ratios of Kepler Planet Pairs 1 2 3 4 5 6 7 8 Period Ratio PDF Steffen & Hwang (2015) Wednesdays@NICO February 22, 2017 Wednesday, February 22, 17
57. Kepler shows lots of planet pairs with period ratios near 2 and 2.17 Did people miss these planets in the RV data? Wednesdays@NICO February 22, 2017 Wednesday, February 22, 17
58. Wednesdays@NICO February 22, 2017 Wednesday, February 22, 17
59. Wednesdays@NICO February 22, 2017 1-planet 2-planet (near 2:1) 2-planet (near 2.17:1) Models Compared Wednesday, February 22, 17
60. Wednesdays@NICO February 22, 2017 Strong Evidence for ~30 New Planets PRELIMINARY RESULTS Boivert, Nelson, & Steffen (in prep.) Wednesday, February 22, 17
61. A lot of the work presented here was a product of... ASTRO Program Working Group I: Uncertainty Quantification and Astrophysical Emulation Working Group II: Synoptic Time Domain Surveys Working Group III: Multivariate and Irregularly Sampled Time Series Working Group IV: Astrophysical Populations Working Group V: Statistics, computation, and modeling in cosmology Information: https://www.samsi.info/ Wednesdays@NICO February 22, 2017 Wednesday, February 22, 17
62. Thank you! Wednesdays@NICO February 22, 2017 Wednesday, February 22, 17
63. Wednesdays@NICO February 22, 2017 Wednesday, February 22, 17
64. Wednesdays@NICO February 22, 2017 Wednesday, February 22, 17
65. Wednesdays@NICO February 22, 2017 Wednesday, February 22, 17