Giant Planets Transiting Giant Stars

My defense of my thesis topic: giant planets orbiting giant stars, and what we can learn about planet evolution from them.

Transcript

1. Giant Planets Transiting Giant Stars Samuel Kai Grunblatt Committee: Daniel Huber, Eric Gaidos, Andrew Howard, Rolf-Peter Kudritzki, Christoph Baranec, Jonathan Williams Institute for Astronomy, University of Hawaii— Manoa, Honolulu, HI
2. Stellar evolution in 30 seconds:
3. Stellar evolution in 30 seconds:
4. Planet detection in 20 seconds:
5. An Ode to
6. An Ode to Kepler
7. Distribution of Kepler host stars: mostly Sunlike Kepler Huber+ (2016)
8. K2: Extension to the NASA Kepler Mission high precision photometry from space (~30-100 ppm precision) 80 day observing campaigns, 30 minute cadence community chosen targets
9. Distribution of K2 host stars: more cool dwarfs, more giants Huber+ (2016)
10. Why should you care about planets orbiting giant stars?
11. Why should you care about planets orbiting giant stars? Most precisely characterized population of potential planet host stars: median M*, R* uncertainties 3.7%, 2.2% through asteroseismology of light curves alone! precise planet parameters can test stellar parameter estimation methods motivates new models to characterize stellar variability (Grunblatt+2015, Foreman-Mackey+2017, Grunblatt+2017, Jones+2018), allowing planet detection in stellar-variability-limited regime
12. Why should you care about planets orbiting giant stars? Most precisely characterized population of potential planet host stars: median M*, R* uncertainties 3.7%, 2.2% through asteroseismology of light curves alone! precise planet parameters can test stellar parameter estimation methods motivates new models to characterize stellar variability (Grunblatt+2015, Foreman-Mackey+2017, Grunblatt+2017, Jones+2018), allowing planet detection in stellar-variability-limited regime
13. To solve planet evolution mysteries. Can planets be inflated at late times? How? (Guillot+1996, Burrows+2000, Bodenheimer+2001, Lopez+2016, Grunblatt+ 2016, Grunblatt+2017) What about orbital dynamics? inspiral, circularization, engulfment timescales? (Villaver+2014, Fuller 2017, MacLeod+ 2018, Grunblatt+ 2018) How similar/different are the planet populations of main sequence and evolved stars? (Villaver+2007, 2009, Veras 2016, Jones+ 2016, Grunblatt+ in review) Why should you care about planets orbiting giant stars?
14. A Search for Giants Orbiting Giants with K2
15. Confirmed planets around red giant stars (2009) transit RV
16. Confirmed planets around red giant stars (2015) RV Kepler
17. Confirmed planets around red giant stars (2019) RV Kepler/K2 K2-97b K2-132b Kepler-56c Kepler-91b Kepler-432b Kepler-56b K2-161b
18. A Search for Giants Orbiting Giants with K2 ➤ >10,000 Low Luminosity Red Giant Branch (LLRGB) targets ➤ Transit detection limit: ~9 Rsun ➤ K2 limit for asteroseismology: 283 μHz (~3 Rsun) ➤ Temperature limits: 4500—5500 K 
 (avoids horizontal branch stars) Huber+ (2016)
19. Asteroseismology Radii: ≤3-5% Masses: ≤5-10% log(g): within 0.01 dex Grunblatt+ (2016)
20. Asteroseismology Radii: ≤3-5% Masses: ≤5-10% log(g): within 0.01 dex Grunblatt+ (2016)
21. Asteroseismology Radii: ≤3-5% Masses: ≤5-10% log(g): within 0.01 dex Grunblatt+ (2016) Asteroseismology Radii: ≤3-5% Masses: ≤5-10% log(g): within 0.01 dex
22. Asteroseismology Radii: ≤3-5% Masses: ≤5-10% log(g): within 0.01 dex Grunblatt+ (2016) Asteroseismology Radii: ≤3-5% Masses: ≤5-10% log(g): within 0.01 dex
23. Asteroseismic stellar parameters: Grunblatt+ (in prep.) Grunblatt+ (in review.)
24. Asteroseismic stellar parameters: 2546 LLRGB stars in K2 LLRGB stars Grunblatt+ (in review.)
25. Asteroseismic stellar parameters: 2546 LLRGB stars in K2 LLRGB stars Grunblatt+ (in review.) σΜ* = 3.7% σR* = 2.2%
26. Asteroseismic stellar parameters: 2546 LLRGB stars in K2 Grunblatt+ (in review.) median offset in R*: 3%, standard deviation: 10%. asteroseismology is precise AND accurate!
28. Seeing Double with K2: Two Remarkably Similar Planets Orbiting Red Giant Branch Stars Grunblatt+ (2016, 2017)
29. Grunblatt+ (2017) Seeing Double with K2:…
30. Seeing Double with K2:… Grunblatt+ (2017) Rs = 3.85 +/- 0.13 R⊙ Ms = 1.08 +/- 0.08 M⊙ Rs = 4.20 +/- 0.14 R⊙ Ms = 1.16 +/- 0.12 M⊙
31. Seeing Double with K2:… Grunblatt+ (2017)
32. ] ] uncorrelated + correlated noise Seeing Double with K2:… Grunblatt+ (2017)
33. Gaussian Process Model Evaluate choice of mean function parameters and kernel hyperparameters with likelihood function. (for squared exponential case) log[L(r)] = 1 2 rT⌃ 1r 1 2 log|⌃| N 2 log(2⇡) r = v Ksin ⇣2⇡(t tc) Porb ⌘ ⌃ij = k(ti, tj) = h2exp  ⇣ti tj ⌘2
34. Gaussian process lightcurve fitting no GP Used squared exponential and simple harmonic oscillator GP kernel functions to account for granulation & oscillation noise
35. Gaussian process lightcurve fitting SE GP Used squared exponential and simple harmonic oscillator GP kernel functions to account for granulation & oscillation noise
36. Gaussian process lightcurve fitting SHO GP Used squared exponential and simple harmonic oscillator GP kernel functions to account for granulation & oscillation noise
37. Combined transit + GP models Grunblatt+ (2017) Simple Harmonic Oscillator GP Squared Exponential GP Used squared exponential and simple harmonic oscillator GP kernel functions to account for granulation & oscillation noise
38. Simple harmonic oscillator GP model traces stellar granulation & oscillation signals: estimate of νmax from time-domain! Grunblatt+ (2017) νmax, pipeline = 245.65 ± 3.51 μHz νmax, GP = 239.4 ± 1.8 μHz SHO model tells us about star, too
39. Grunblatt+ (2017) Seeing Double with K2:… Rs = 3.85 +/- 0.13 R⊙ Ms = 1.08 +/- 0.08 M⊙ Rp = 1.30 +/- 0.07 RJ Rs = 4.20 +/- 0.14 R⊙ Ms = 1.16 +/- 0.12 M⊙ Rp = 1.31 +/- 0.11 RJ
40. To solve planet evolution mysteries Can planets be inflated at late times? How? (Guillot+1996, Burrows+2000, Bodenheimer+2001, Lopez+2016, Grunblatt+ 2016, Grunblatt+2017) What about orbital dynamics? inspiral, circularization, engulfment timescales? (Villaver+2014, Fuller 2017, MacLeod+ 2018, Grunblatt+ 2018) How similar/different are the planet populations of main sequence and evolved stars? (Villaver+2007, 2009, Veras 2016, Jones+ 2016, Grunblatt+ in review) Why should you care about planets orbiting giant stars?
41. The Mystery of Planet (Re-)Inflation Grunblatt+ (2016, 2017)
42. Inflated Hot Jupiters are ubiquitous. Lopez & Fortney (2016)
43. Inflated Hot Jupiters are ubiquitous. maximum expected radius Lopez & Fortney (2016)
44. maximum expected radius highly inflated not inflated Inflated Hot Jupiters are ubiquitous. Lopez & Fortney (2016)
45. The Mechanism of Planet Inflation e.g., Bodenheimer+ (2001), Showman & Guillot (2002),
 Batygin & Stevenson (2010), Ginzburg & Sari (2016) Class I: planet interior inflated directly by increased stellar irradiation Class II: cooling delayed after planet formation e.g., Burrows+ (2000), Chabrier & Baraffe (2007), Leconte & Chabrier (2012), Wu & Lithwick (2013)
46. How to distinguish between Classes I and II? Class I: re-inflation Class II: no re-inflation
47. Are my K2 planets inflated? Lopez & Fortney (2016)
48. Are my K2 planets inflated? Yes. Lopez & Fortney (2016)
49. Are they re-inflated? avg incident flux on main sequence: K2-97b: 
 170 +140-60 F EPIC2287b: 190 +150-80 F Grunblatt+ (2017)
50. typical incident flux range for 1.3 RJ planets current incident flux: K2-97b: 
 900 +200-150 F EPIC2287b: 850 +250-140 F current incident fluxes Are they re-inflated? Grunblatt+ (2017)
51. Need to delay cooling by different rates in every case to explain observed planet radii. ] different delayed cooling models Are they re-inflated? Grunblatt+ (2017)
52. delayed cooling ] re-inflation Grunblatt+ (2017) Is K2-97b re-inflated? Probably. Data implies significant post-MS planet heating. But how? K2-97b: skg3@hawaii.edu @skgrunblatt
53. delayed cooling ] re-inflation Grunblatt+ (2017) Is K2-97b re-inflated? Probably. Kepler-422b: 1.15 Msun 0.43 MJ 7.89 days Data implies significant post-MS planet heating. But how? K2-97b: skg3@hawaii.edu @skgrunblatt
54. delayed cooling ] re-inflation Grunblatt+ (2017) Is K2-97b re-inflated? Probably. Kepler-422b: 1.15 Msun 0.43 MJ 7.89 days Data implies significant post-MS planet heating. But how? K2-97b: skg3@hawaii.edu @skgrunblatt Berger+ (2018)
55. To solve planet evolution mysteries Can planets be inflated at late times? How? (Guillot+1996, Burrows+2000, Bodenheimer+2001, Lopez+2016, Grunblatt+ 2016, Grunblatt+2017) What about orbital dynamics? inspiral, circularization, engulfment timescales? (Villaver+2014, Fuller 2017, MacLeod+ 2018, Grunblatt+ 2018) How similar/different are the planet populations of main sequence and evolved stars? (Villaver+2007, 2009, Veras 2016, Jones+ 2016, Grunblatt+ in review) Why should you care about planets orbiting giant stars?
56. To solve planet evolution mysteries Can planets be inflated at late times? How? (Guillot+1996, Burrows+2000, Bodenheimer+2001, Lopez+2016, Grunblatt+ 2016, Grunblatt+2017) What about orbital dynamics? inspiral, circularization, engulfment timescales? (Villaver+2014, Fuller 2017, MacLeod+ 2018, Grunblatt+ 2018) How similar/different are the planet populations of main sequence and evolved stars? (Villaver+2007, 2009, Veras 2016, Jones+ 2016, Grunblatt+ in review) Why should you care about planets orbiting giant stars?
57. Grunblatt et al. (2017) Seeing Double with K2: what can RVs tell us? Mp = 0.48 +/- 0.07 MJ Mp = 0.49 +/- 0.06 MJ
58. Grunblatt+ (2018) They’re eccentric! (e~0.3)
59. They’re eccentric! (e~0.3) Grunblatt+ (2018)
60. Do Close-In Planets Orbiting Evolved Stars Prefer Eccentric Orbits? Grunblatt+ (2018)
61. Hot Jupiters and Cool Stars Villaver, Livio, Mustill & Siess (2014) skg3@hawaii.edu @skgrunblatt
62. Hot Jupiters and Cool Stars Villaver, Livio, Mustill & Siess (2014) skg3@hawaii.edu @skgrunblatt Villaver+ (2014)
63. Hot Jupiters and Cool Stars Villaver, Livio, Mustill & Siess (2014) skg3@hawaii.edu @skgrunblatt Villaver+ (2014)
64. Do close-in giant planets orbiting evolved stars prefer eccentric orbits? Grunblatt+ (2018)
65. Do close-in giant planets orbiting evolved stars prefer eccentric orbits? e = 0.15 +0.08-0.04 Grunblatt+ (2018)
66. Do close-in giant planets orbiting evolved stars prefer eccentric orbits? e = 0.06 +0.02-0.01 e = 0.15 +0.08-0.04 Grunblatt+ (2018)
67. Do close-in giant planets orbiting evolved stars prefer eccentric orbits? e = 0.06 +0.02-0.01 e = 0.15 +0.08-0.04 Grunblatt+ (2018) ⌧ins = Porb ⇣2Q⇤ 27⇡ ⌘⇣M⇤ Mp ⌘⇣ a R⇤ ⌘5 ⌧circ = Porb ⇣2Qp 63⇡ ⌘⇣Mp M⇤ ⌘⇣ a R⇤ ⌘5
68. Do close-in giant planets orbiting evolved stars prefer eccentric orbits? e = 0.06 +0.02-0.01 e = 0.15 +0.08-0.04 Grunblatt+ (2018) K2-97b: ⌧ins . 2 Gyr ⌧circ ⇠ 5 Gyr
69. Do close-in giant planets orbiting evolved stars prefer eccentric orbits? Grunblatt+ (2018) ?
70. To solve planet evolution mysteries Can planets be inflated at late times? How? (Guillot+1996, Burrows+2000, Bodenheimer+2001, Lopez+2016, Grunblatt+ 2016, Grunblatt+2017) What about orbital dynamics? inspiral, circularization, engulfment timescales? (Villaver+2014, Fuller 2017, MacLeod+ 2018, Grunblatt+ 2018) How similar/different are the planet populations of main sequence and evolved stars? (Villaver+2007, 2009, Veras 2016, Jones+ 2016, Grunblatt+ in review) Why should you care about planets orbiting giant stars?
71. To solve planet evolution mysteries Can planets be inflated at late times? How? (Guillot+1996, Burrows+2000, Bodenheimer+2001, Lopez+2016, Grunblatt+ 2016, Grunblatt+2017) What about orbital dynamics? inspiral, circularization, engulfment timescales? (Villaver+2014, Fuller 2017, MacLeod+ 2018, Grunblatt+ 2018) How similar/different are the planet populations of main sequence and evolved stars? (Villaver+2007, 2009, Veras 2016, Jones+ 2016, Grunblatt+ in review) Why should you care about planets orbiting giant stars?
72. Giant Planet Occurrence Within 0.2 AU of Low Luminosity Red Giant Branch Stars Grunblatt+ (in review.)
73. How to calculate occurrence: 1/pj = a/R*, n*,j = number of stars searched where SNR(transit) > some threshold (determined by injection/recovery here) Made grid of periods and planet radii, and for each scenario, calculated whether SNR > SNRthreshold for every star in our sample to find n*,j Howard et al. (2012)
74. LLRGB Injection/Recovery Grunblatt+ (in review.)
75. LLRGB Planet Occurrence Grunblatt+ (in review.)
76. LLRGB Planet Occurrence Grunblatt+ (in review.)
77. LLRGB Planet Occurrence Grunblatt+ (in review.)
78. LLRGB Planet Occurrence Grunblatt+ (in review.) planet occurrence decreases with radius around MS stars…
79. LLRGB Planet Occurrence planet occurrence decreases with radius around MS stars… …but seems to increase with radius around evolved stars! Grunblatt+ (in review.)
80. LLRGB Planet Occurrence re-inflation by irradiation Grunblatt+ (in review.)
81. LLRGB Planet Occurrence re-inflation by tidal migration Grunblatt+ (in review.)
82. LLRGB Planet Occurrence Grunblatt+ (in review.)
83. To solve planet evolution mysteries Can planets be inflated at late times? How? (Guillot+1996, Burrows+2000, Bodenheimer+2001, Lopez+2016, Grunblatt+ 2016, Grunblatt+2017) What about orbital dynamics? inspiral, circularization, engulfment timescales? (Villaver+2014, Fuller 2017, MacLeod+ 2018, Grunblatt+ 2018) How similar/different are the planet populations of main sequence and evolved stars? (Villaver+2007, 2009, Veras 2016, Jones+ 2016, Grunblatt+ in review) Why should you care about planets orbiting giant stars?
84. What’s next?
85. Better statistics with TESS! more stars=more opportunities for short period, small planets
86. Better statistics with TESS! longer baselines=more complete for longer period planets
87. Better statistics with TESS! Grunblatt (2018)
88. >10x as many LLRGB planets (Campante+ 2016, Barclay+ 2018) Grunblatt (2018)
89. Test re-inflation dependencies on star, planet properties Campante+ (2016), Barclay+ (2018), Grunblatt (2018)
90. Campante+ (2016), Barclay+ (2018), Grunblatt (2018) Test migration dependencies on star, planet properties
91. Foreman-Mackey+ (2017) Kepler TESS Better statistics with TESS: 100x as many targets! ~103 targets ~105 targets: 100x increase!
92. Foreman-Mackey+ (2017) Kepler TESS Better statistics with TESS: 100x as many targets! ~103 targets ~105 targets: 100x increase! real TESS data from eleanor(Feinstein+ 2019) first new TESS planet around an oscillating star: TOI-197 (Huber+ 2019)
93. Foreman-Mackey+ (2017) Kepler TESS Better statistics with TESS: 100x as many targets! ~103 targets ~105 targets: 100x increase! real TESS FFI data from eleanor (Feinstein+ 2019)
94. Better statistics with TESS: new stellar variability models Grunblatt+ (in prep.)
95. Better statistics with TESS: more RV followup possible
96. Better statistics with TESS: funded in TESS GI Cycle 2!
97. Summary: Giant planets orbiting giant stars reveal planet evolution mysteries. planets can be re-inflated at late times (Grunblatt+ 2016, 2017). planets inspiral faster then they circularize during RGB evolution (Grunblatt+ 2018). warm/hot Jupiters are roughly equally common around MS and LLRGB stars. evolved system planets appear larger overall (Grunblatt+ in review). future is extremely bright with TESS: hundreds of planets transiting giant stars predicted to be detected! most well-characterized population of potential planet host stars: allow us to test new stellar variability models!
98. Thanks!
99. Extra Slides
100. LLRGB Planet Occurrence Grunblatt+ (in review.)
101. Stellar Radius Determination with Parallaxes Parallax gives distance. Photometry + reddening map gives color, combining with distance gives absolute magnitude (proxy for luminosity). Color-temperature relation gives Teff.(Gonzalez Hernandez & Bonifacio 2009) Luminosity and temperature give radius via Stefan— Boltzmann equation. Parallax gives distance. Photometry + reddening map gives color. Surface brightness- color relation gives angular diameter. (Graczyk+ 2018) Angular diameter + distance gives radius. isoclassify method SBCR method
102. Why are K2-97b, K2-132b so similar? Short answer: probably survey bias * intrinsic planet occurrence 0.01 0.10 1.00 10.00 planet mass (Jupiters) 0.0 0.2 0.4 0.6 0.8 1.0 survey bias factor
103. Why are planet radii not as well constrained as stellar radii? Different lightcurves treat systematics differently. Somewhat accounted for by GP Grunblatt+ (2016)
104. Asteroseismic stellar parameters: really accurate? Calibrated asteroseismic relations with interferometry (~5% agreement), eclipsing binaries (5-10% agreements). Soon, calculating bolometric fluxes and radii from spectra + Gaia parallaxes for >5% precision. ~500 spectra in hand now. Grunblatt+ (in prep.)
105. LLRGB star variability: limited by RV jitter? RV jitter for evolved stars poorly understood. K2-97, K2-132 have low jitter (~3 m/s) and clear Keplerian signal (~50 m/ s). Other LLRGBs may have much more jitter. Grunblatt+ (in prep.)
106. What is a Gaussian process estimator? A Gaussian process (GP) estimator is a nonparametric estimator of time-series data described by a kernel function and its hyperparameters.
107. A Gaussian process (GP) estimator is a nonparametric estimator of time-series data described by a kernel function and its hyperparameters. ⌃ij = k(ti, tj) = h2exp  ⇣ti tj ⌘2 Simplest kernel function: Squared exponential (SE) What is a Gaussian process estimator?
108. What is a Gaussian process estimator? A Gaussian process (GP) estimator is a nonparametric estimator of time-series data described by a kernel function and its hyperparameters. ⌃ij = k(ti, tj) = h2exp  ⇣ti tj ⌘2 Simplest kernel function: Squared exponential (SE)
109. What is a Gaussian process estimator? A Gaussian process (GP) estimator is a nonparametric estimator of time-series data described by a kernel function and its hyperparameters. ⌃ij = k(ti, tj) = h2exp  ⇣ti tj ⌘2 covariance matrix Simplest kernel function: Squared exponential (SE)
110. What is a Gaussian process estimator? A Gaussian process (GP) estimator is a nonparametric estimator of time-series data described by a kernel function and its hyperparameters. ⌃ij = k(ti, tj) = h2exp  ⇣ti tj ⌘2 covariance matrix now with off-diagonal terms! covariance matrix Simplest kernel function: Squared exponential (SE)
111. What is a Gaussian process estimator? A Gaussian process (GP) estimator is a nonparametric estimator of time-series data described by a kernel function and its hyperparameters. ⌃ij = k(ti, tj) = h2exp  ⇣ti tj ⌘2 other options: periodic h2exp  sin2[⇡(ti tj )/✓] 2w2 quasiperiodic h2exp  sin2[⇡(ti tj )/✓] 2w2 ⇣ ti tj ⌘2 or maybe something more physically motivated… Simplest kernel function: Squared exponential (SE)
112. • Simplest kernel: squared exponential. Described by: • where (h, λ) are the hyperparameters: parameters of the kernel. Kernel function basics Roberts+ (2012) λ = 0.1 λ = 1 λ = 10 ⌃ij = k(ti, tj) = h2exp  ⇣ti tj ⌘2 h t