Kepler-K2 SciCon V (March 2019)

Kepler-K2 SciCon V

People going from Penn State

Monday, March 4, 2019

Bill Borucki - Kepler was rejected 4 times before being accepted! Time #1 - worries that detectors wouldn’t work Time #2 - mission too costly Time #3 - haven’t demonstrated that can process 10,000 stars in an automated way (too much for manual analysis) Time #4 - haven’t demonstrated that can reach 10 ppm …. and after addressing these criticisms one by one the mission was selected and was an enormous success (and not just for exoplanets!)

Katelynn McCalmont - engineer’s view of the K2 mission; how to correct for the drifting; the drama at the end of the mission with watching the fuel tank pressure drop; running on fumes for C19. Managed to get all the science data off the satellite, though!!

Geert Barentsen - 21 scientific opportunities with Kepler and K2. The missions are done taking data, but the science is far from over!! https://arxiv.org/abs/1810.12554

Mia Lundkvist - asteroseismology of solar-type stars. Typical Gaia uncertainty on Kepler stellar radii are 5%; for asteroseismology is 3% (CKS is 12%).
Highlighted individual systems: Kepler-444, Kepler-10.
Ensemble studies: 1) Constrain the obliquity of exoplanet systems by comparing power in the m=0 vs. m=+/-1 modes for l=1. 2) Eccentricities from transit duration & knowing the stellar radius well (Van Eylen papers - eccentricity distribution is wider for single-planet systems). 3) Evaporation; looking at the desert (defined as 2-4 R_Earth and >650 times the flux incident at Earth (~4.5 day periods)). No asteroseismic stars in this region … Also the evaporation valley; with asteroseimic sample can determine the slope of the valley - is negative, consistent with evaporation (see next talk). We’re not done with asteroseismic analysis of Kepler data - there will be more targets that they can analyze the oscillations of.

Vincent Van Eylen - understanding planet formation through asteroseismology. Focusing on photoevaporation, as enabled by improved precision of asteroseismology (quotes 2%, versus spectroscopy + Gaia, quoted at 10%). Owen & Wu 2017 - envelopes between 1 and 10% are most stable to photoevaporation. Gets a negative slope using asteroseismic stellar radii -> photoevaporation Also location matches terrestrial core composition. Emptiness of valley suggests homogeneous core composition (but also limited sample size …). Reassess this with TESS, find a planet in the gap!

Hilke Schlichting - core-powered mass loss as an explanation for the evaporation valley (versus photoevaporation). Because the envelope acts as a thermal blanket while the core is accreting gas from the disk, the core doesn’t lose its residual formation heat until the envelope cools substantially (this takes Gyrs). Above envelope mass fractions ~10%, cooling dominated by envelope; below dominated by core. To keep those small envelopes around, the loss timescale > cooling timescale. Gupta & Schichting, submitted: slope is -0.1, about what Vincent found. Also looked at dependence on stellar properties, particularly mass (this mass loss mechanism depends on the bolometric luminosity, not on XUV like photoevaporation): valley is at larger radius around larger mass stars (slope as a function of M_star is ~0.3). This is consistent with Fulton & Petigura, 2018, and is the opposite of photoevaporation (because XUV stronger for low-mass stars, should be able to strip even larger cores, and slope with M_star would be negative … but from Gaidos: this isn’t quite true - the ratio of XUV-to-bolometric is higher, but not overall XUV, so can’t discount photoevaporation. Hilke points to Wu 2018 about trend between stellar and planet masses … hmmmmmmm.). Also looked at [Fe/H] … some dependence for planets with envelopes still because of the role of opacity in envelope constraction.

Travis Berger - precise characterization of Kepler stars and planets using Gaia DR2. Presented his paper (before that, Mathur et al. 2017 was the updated Kepler stellar properties catalog). He finds that the Kepler stellar sample consists of: Main sequence dwarfs (67%), Subgiants (21%), Red giants (12%). Using the Gaia data increases the Kepler stellar radii by 12% (median increase). Also sees binary sequence among stars. Among the planets, sees the evaporation valley; 74 planets in Mia’s defined evaporation desert. Also >100 planets of all sizes in habitable zone. Also cool inflated Jupiters … still young? In preparation: looking at masses and ages of the Kepler targets. Ages: distribution peaks at 3 Gyr.

BJ Fulton - revisiting the radius gap with DR2. Talking about the stellar mass dependence of the gap referred to by Hilke. Suspected that with Gaia, gap would be deeper and narrower than could measure with CKS (stellar radius uncertainties dilute the gap). Uses spectroscopic T_eff (rather than T_eff from colors as with Berger). Turns out gap looks nearly the same in depth/width in 1D, but cleans up in 2D (as function of period). Location of gap is consistent with cores made of a fraction of silicates and iron. Other than photoevaporation, other explanations for gap: core-heating (Hilke’s talk), pebble accretion, giant impacts, outgassing of a few %. Goes into stellar mass dependence. But what’s the cause of this dependence? Could be mass, metallicity, or age given correlations between properties in the Kepler target star sample.

Matteo Cantiello - magnetic fields in stars as revealed by Kepler. There’s a suppression of mixed modes in some stars - possibly due to magnetic fields.

Angela Santos - signatures of magnetic activity in seismic data of Kepler solar-type stars. Talking about 100-day timescales for the magnetic cycles. KIC 5184732; KIC 8006161 - find that these cycles are not constant period. The maximum shift in frequency (in micro-Hz) of these stars is higher for hotter stars, in agreement with theory. Also increases as metallicity increases. 60% of stars have these frequency shifts.

Ellianna Schwab Abrahams - fundamental properties of M-dwarfs in Kepler. Using Gaia and a color cut informed by M-dwarfs spectroscopically confirmed with SDSS. Also calculate own masses; find that the masses are ~25% larger than in the Kepler Input Catalog. Use this update radii and temperature. Also look at magnetic properties of M-dwarfs. Measured >800 new rotation periods based on McQuillan 2013; see bimodal distribution: most are slow rotators, but there are some rapid rotators. Comparing Kepler to M-Earth rotation rates (from Newton paper), see that the gap becomes wider for lower-mass stars. Also applying to K2. In total, confirmed > 30000 M-dwarfs in Kepler and K2.

Michael Gully-Santiago - Stellar Surface Inhomogeneities can bias transit-derived exoplanet densities. David et al. 2019 - a planet transiting a heavily spotted stars. Kepler can measure the variation due to star spots, but not the total amount of light that is lost by star spots. What is the typical ratio of total spot coverage to the Kepler amplitude (to get at this missing light)? Answer is unconstrained … anywhere from 1 to 100, depends on whether you assume spots are distributed symmetrically or not. So what happens if it’s really large? Nontrivial mapping from depth to Rp/Rstar: if the planet doesn’t cross any of the starspots, then it blocks a larger fraction of the bright light, and so the transit will appear deeper than it should given the planet’s size -> you will infer that the planet is larger than it is. Showed an example where there was a 4% difference in radius, and 12% difference in density. This is also wavelength-dependent: in optical, spots are very dark. But not as much a problem in IR … so can resolve this issue by looking at NIR; spectroscopy with IGRINS gives the T_eff of the spots, which gives you the spot coverage fraction.

Sharon Wang - RVxK2 - simultaneous Kepler photometry and RVs to help mitigate stellar jitter. Stellar jitter is the bottleneck for detecting Earth analogs. The idea is that RV variations and photometric variations are similar in timescale and amplitude (have used GPs trained on photometry to help with RV activity). Simultaneous observations happened for C16; ground-based used APF, IRFT, MINERVA, Keck, Magellan. Many aspects of stellar jitter; she talked particularly about oscillations. Usually mitigated in RVs by averaging, but doesn’t get down to 30 cm/s … so why don’t you model it instead? Yes!! Did this for a sub-giant; used GPs (celerite) with the same underlying model trained on photometry, only one more free parameter, get down to 0.8 cm/s. Doing this again with TESS.

Lisa Bugnet - FliPer, a tool to detect and characterize solar-like pulsators. Classifying all Kepler, K2, TESS stars by extending Fabienne’s flicker to the entire power spectrum. Using entire spectrum allows you to go to much lower logg using the granulation signal. Oscillators (solar-like acoustic modes, delta-scuti, cepheids, etc) also have unique power spectra. Uses a random forest for the classification between solar-like oscillations, classical pulsators, and binaries. Trained on 1600 stars; tested on 400; get ~90% accurate classifications for solar-like oscillations; ~80% for classical pulsators. Then applied to 136000 Kepler stars. Working on K2 and TESS.

Courtney Dressing - frequency of planetary systems in Kepler & K2. Overview talk; started with pre-Kepler radius distribution. Talked about radius distribution as a function of period, stellar type, metallicity (look again at Petigura et al. 2018 - updates earlier work). Her group is working on statistics with K2.

Gijs Mulders - Exoplanet Population Observation Simulator (EPOS - code available) to apply forward model of Kepler detection filter to output of theoretical simulations (also computes data-centric occurrence rates). Incorporates both Kepler data and RV data … check out Fernandes paper in press for occurrence rate of giant planets (from HARPS?); also compares to the Bern planet population synthesis models. Applying his EPOS code to the Bern models, he sees that the frequency of Jupiters and close-in super-Earths match-ish Kepler (using large boxes, within a factor of 2). Exception is sub-Neptunes - Bern model underpredicts abundance of them by factor of ~5. Also discusses multi-planet statistics, comparison between models and Kepler. See that sub-Neptunes in model too close to their stars. Also find that systems are dynamically cold (which produces more planet cores in model).

Tim Morton - probabilistic validation - the revolution enabled by Kepler. Probabilistic validation started with Kepler-9 d. Did not have a mass measurement. Then Kepler 22b, validated by BLENDER. Then his 2012 paper, predicting <10% false positive rate. Then computed false positive probabilities for entire Kepler catalog and validated >1200 new planets (2016 paper)

Marko Sestovic - planet occurrence rates around ultracool dwarfs (like TRAPPIST-1). Large gap in T_eff from TRAPPIST-1 to other M-dwarf planets. 611 ultracool stars observed by K2, 445 spectrographically confirmed. Model occurrence rate with an inhomogeneous poisson process that incorporates completeness via transit injections (can recover 6 of the 7 TRAPPIST planets, but couldn’t find planets around any of the other ultracool stars, so will only get upper limits). Conclude that finding TRAPPIST-1 planets was lucky: if 100% of stars have TRAPPIST-like systems, would have 73% chance of finding one. If 20%, have 23% chance; if 5%, 6.3% chance; if 1%, have 1.3% chance … for the completeness-corrected marginal radius distribution, get ~10 planets/star at 1 R_Earth, but that’s just because their entire sample is dominated by one system.

Christina Hedges - are there more planets in K2? Dotson et al. 2018: if apply Kepler-prime occurrence rates to K2, we haven’t found all of the planets that are there. Where are they hiding? Instrument systematics? Have only been searching for strictly periodic signals? Stellar variability? crowded field? multi-planet search biases? no homogeneous search yet? Highlights 3 specific systems. github.com/christinahedges/KepSciCon2019 - the Jupyter notebooks that her results are based on. Detrending codes used in various community efforts: K2SFF (Vanderburg; self-flat fielding), EVEREST (Luger; pixel-level decomposition), K2SC (Aigrain; using gaussian processes). None of these methods work for all stars. Example: K2-43b - short-term noise important, EVEREST does better than K2SFF. There’s another planet hiding in there. Another example: K2-168; another extra planet. Her suggestions: vary detrending method and parameters to ensure a complete search; remove stellar variability carefully; check harmonics for hiding resonant planets.

Tuesday, March 5

Sarah Ballard - overview of Multi-planet systems, including Kepler 9, 11, 32, 36, 19 & 46 (detections of non-transiting planets from TTVs), 20 & 42 (first Earth-sized planets), 186 & 62 (the habitable zone ~Earth-sized planets, just around smaller stars). She explicitly mentions the TTV vs. RV density difference as being interesting to get to the bottom of.

Chris Shallue - deep learning to find Earth analogs. (!!! He says that today’s neural networks are able to distinguish between Alaskan Malamutes and Siberian Huskies!!) Their detetion pipeline recovers more injected planets around Sun-like stars … because probing lower SNR, will have more false positives, so need automatic vetting … this is where the deep learning comes in (their vetting DNN is called AstroNet). Obtain 96% accuracy on signals from Kepler. Found two new planets: Kepler-80g and Kepler-90i. Have also applies AstroNet for K2, and discovered two new planets. AstroNet is also performing triage in MIT’s Quick-look pipeline. Long-period false positives are challenging because don’t have many of those in the training set.

Michelle Hill - Kepler Giant planets in the habitable zone (NOTE: their definition of giants are 3-25 R_Earth). Around G dwarfs, 6.5% +/-1.9; around K, 11.5% +/- 3.1; around M, 6% +/- 6 (versus ~1-20% for Earth-size planets in the habitable zone). Interested in this for habitable moons around these planets. Also looked at angular separation between giant planet and moon for direct imaging - need ~0.01-4 milliarcsec. Then looked at actual giant planets in habitable zones found via RVs, find many linear trends (so farther out companions) - how does this affect formation of potentially habitable moons?

Kai Rodenbeck - revisiting exomoon around Kepler-1625 b (even different than the paper I discussed at the arXiv discussion a few weeks ago). Points out issue with enormous size of proposed moon. They ask: 1) can they reproduce findings of Teachey 2018? 2) How robust is this to detrending/stellar noise? Do injections with and without moon. Use four differen detrending functions and two cut-off conditions. There is 10-20% chance of detecting a moon if no moon was injected; 30-40% chance of detecting an injected moon. They find that the false positives were close to the properties of the injected transits, especially using the detrending method closest to what Teachey used.

Ashley Chontos - KOI-4: confirming Kepler’s first exoplanet detection. KOI-4 is listed as a false positive early on, but Robovetter reclassified it as planet candidate. Stellar properties were wrong - is actually a subgiant, so planet is Jupiter-sized. This planet was then confirmed via RVs (Kepler-1658 b). Combined fit shows massive hot Jupiter with mild eccentricity. Occupies a unique area of host radius vs. period space. Good target for observing orbital period decay! Over time span of Kepler, no evidence of period decay, which constrans the tidal Q of the star to be > 4e3.

Andrew Vanderburg - benchmark systems from K2. Starting with systems that are ameniable to ground-based follow-up: K2-3, many others that are part of Hubble follow-up survey. Brightest planet-hosting stars: planets around giants like HD73344. Particularly highlighted HD 106315 - doppler tomography used to assess spin-orbit alignment. Also TRAPPIST-1, K2-19 (very close to 3:2 resonance), WASP-47 with ultra-short period e, hot jupiter, smaller planet at 9 days, and long-period giant from RV - dynamical analyses show that has to have orderly architecture. Next start about extreme K2 systems: utra-short-period planets like K2-137 (P=4.3 hours; planet must be very elongated, like football, and heavily iron-enhanced), K2-141 (has phase curves indicating reflected light, or is very very hot blackbody). Also very long periods (single-transit systems): EPIC 248847494 b (P~10 years given transit duration), microlensing planet OGLE-2016-BLG-1190Lb that has K2 data. Young planets: planets detected in Hyades (Mann 2016), Taurus (De Martin 2019), Upper Scorpius (David 2016). Old planets: jupiters around evolved giants (Grunblatt). Disintegrating planets: K2-22 and WD 1145+017, add to the 2 found in the original Kepler mission. Also citizen science - find many single-transit events, also found K2_288 Bb (one transit found at beginning of data, which is usually clipped out), K2-138 (resonant chain).

Juliette Becker - dynamically determining planet parameters that are observationally ill-constrained. Conclusions: use a baseline prior incorporating that you’re more likely to see short-period planets transit; use dynamical constraints - they help!; and use legacy data. HIP 41378, 5-planet system, need to constrain outer two planets d and f (bright star, good multiplanet system for future observations, a transiting cold jupiter - RVs + transmission spectroscopy!!); observed in C5 and C18 (big gap in the middle). Also had data from KELT, WASP, and HAT (legacy data!!). Computed stability + used before-mentioned prior + knowledge that integer number of periods must have occurred, get a list of possible periods ordered by probability that those are the true period. f deep enough to be observed from ground - using these surveys, ruled out many shorter possible periods. e does not transit again in C18, so can rule out periods, but not as well constrained as d and f. Doing this really helps to guide follow-up observations.

Kevin Hardegree-Ullman - Spitzer follow-up of K2 targets (results from program PI Mike Werner, 68 observations between 2015-2018). Looked at 35 K2 planets. Targets span range in orbital period, radius, and stellar T_eff. Spitzer really helped lock down ephemerides for K2-3. K2-55 b - dense Neptune around high metallicity K star. HD3167 - good target for atmospheric observations. K2-288 Bb - very different transit depths between K2 and Spitzer, due to dilution of companion: less dilution at longer wavelengths, so planet likely to be orbiting the smaller star! K2-138 - 5, maybe 6 planets? Confirmed planet #6 with Spitzer!! Cool video by Matt Russo sonifying the resonance of this system; can hear the slight out-of-resonance today versus during formation (system-sounds.com). Many of these planets are sub-Neptune in size …

Joey Rodriguez - K2-226: a compact multiplanet system with some planets misaligned by ~10 degrees from the others (based on transit shape). 2 validated, 2 confirmed planets, 2 candidates. System is stable, but luckily so. There’s also a co-moving (from Gaia) stellar companion separated by 42 arcsec. There’s an 8.5-day planet candidate around the other star! Another example of this: TOI-125, misaligned by 15-20 degrees!

Fei Dai - ultra-short-period planets. Unique science with them: tidal evolution, thermal emission/surface reflectivity, star-planet magnetic interactions, atmospheric erosion. Look at hot Neptune desert as a function of stellar type: M-dwarfs are more effective at clearing out the desert (and mid-late M hosts even more) than GK dwarfs - points to photoevaporation over core-powered mass loss. Also look at composition of USPs - best chance of constraining the composition of Earth-sized planets in near future because devoid of H/He envelope, so less degeneracy. Do a uniform analysis of USP planets: used F>650 days as definition of USP instead of P<1 day. Applied GP regression to data. Error bars on radius and mass shrunk; better constraints on composition. Split into period bins - match with period distribution from Lee & Chang, 2017. Some connection to magnetic truncation of disk, secular theory … then looked at mutual inclination as a function of orbital distance (Dai+ 2018). Larger mutual inclinations for closer-in planets. Also larger mutual inclinations with larger period ratio (dynamically separated from outer system). Is this an explanation for the Kepler dichotomy? Weiss 2018 showed that the overabundance of single-transiting systems only present for periods < 3 days.

Wednesday, March 6

Marc Pinsonneault - galactic archeology with Kepler and K2. Discussing stellar abundances; now see that there’s great chemical diversity beyond just [Fe/H] and alpha-poor/enhanced. If care about [Si/Fe] of your planets, you need to take these trends/variance into account! Surprise: young alpha-rich stars (from combining Kepler asteroseismology with APOGEE: APOKASC-2). Still analyzing this data set (overlap is 6700 stars). When add in Gaia kinematics, get really rich view of Milky way: thin disk * thin disk have distinct kinematics; outer disk flares (so has perturbations from satellites). Also able to test models of stellar structure. Assess Gaia zero-point offset; about 50 microarcsec, but is color-dependent. Comparing Gaia and asteroseismology: systematics at 2% of stellar radii; mean agreement is 1%. State of the art: 0.5% in Teff, 1% in R, 2% in L. Still struggle to get ages for stars > 8 Gyr, even with all this info. Find systematic errors in red giant isochrones. Working on TESS now.

Dennis Stello - galactic archeology with K2 (asteroseismology + spectroscopy of giant stars). Comparing distributions of stellar mass and radii with galactic models. Not a good match in mass from Kepler-prime, but also complex selection function. With K2 could select stars themselves. Use 4 campaigns, 2 at low galactic latitude, 2 at higher. Discrepancy remains, and is worse at high galactic latitude (model predicts lower mass stars). If increase [M/H] of thick disk, solves much of the problem. Extending analysis to all campaigns; speed up in analysis time enabled by automated image recognition using deep learning. Turned power spectra plot into image; once trained, finds oscillations in all stars from one campaign in 10 seconds! Also identified same features that the expert would. Also tested as a function of observation baseline - on TESS timescales, still 93% accurate! Super-human performance there.

Jie Yu - Astereoseismolgoy of 20,000 red giants observed by Kepler. Can see difference between red clump and a 2nd clump in nu-max vs. delta-nu space. Helium core-burning stars really stand out in radius, mass distributions (~11 R_sun). For low-luminosity red giants, clear trends with mass and metallicity. For high-luminosity red giants, can clearly distinguish between different l for a given m; also good match with theoretical predictions. Also looked at whether dominant l modes are radial or non-radial - depends on m.

Rafael Garcia - comprehensive kepler red giant catalog (and all the nasty details therein). This catalog serves as the input into searches for seismic oscillations; aim is to avoid any selection bias. Uses neural net from Hon 2019, which incorporates FliPer (presented earlier) to identify giants. Also processed the lightcurves in a asteroseismic-friendly way; available on MAST.

Dan Huber - asteroseismic age for galactic halo. Question: how did the halo form? monolithic collapse at early times, accretion from satellites, ejection from disk? Ages are key to answering this! Serendipity: about 1000 K,M dwarfs actually giants (from seismology) - tend to be fainter, from 14-16 magnitude, so are actually quite far away (up to 20 Kpc). With this sample, get spectra to test if they’re actually a part of the halo. Able to extend the APOKASC sample to larger height above galactic plane and lower metallicity - doubled the sample if distant halo giants Kepler observed. Soooo what are the ages? Still working on answer for whole sample, but from some stars, have <10 Gyr, so maybe satellite accretion? Still work to do on systematics, presence of blue stragglers, etc.

Adam Kraus - planets in binary star systems. Kepler is mostly unbiased for stellar multiplicity in its planet search (compared to other searches that exclude binaries from the target list). Kraus et al. 2016 result: KOIs that are close binaries (< 50 AU) are lower by factor of 3 from the field population (this about 20% of all stars). Then look for proper motions to distinguish faint companions from background stars and resolve orbital motions for components in candidate binaries using NIRC2 + Gaia; example: KOI-3444, can see radial changes in separation between A and B (maybe binary is edge on -> aligned with transiting planet?!)! Most candidate binaries are clustered around 0 proper motion. Most binaries are closer to edge-on than face-on (based on elongation of distribution around 0 proper motion), so suggests that the stellar and planetary orbits are dynamically connected. Other groups are working on control (non-KOI) samples.

Rachel Matson - speckle imaging to detect Kepler unresolved binaries. Observe color (692 and 880 nm). Summary of other results: Furlan 2017 - ~30% of host stars have companions < 4”; Hirsch 2017 - planet radii underestimated by 1.17 if around primary, by 1.65 if around secondary; Bouma 2018 - occurrence rates of planets < 2 R_E could be overestimated by as much as 50%. Also Johanna’s paper!! Matson 2018: simulated binaries from Raghavan 2010, applied detection completeness from speckle survey; are they finding any more or less binaries around KOIs? Same amount as field stars (40-50% , when corrected for completeness). Has nice plot in latest submitted paper (arXiv: 1811.0218) about mass ratio vs. measured V magnitude; speckle detection can get down to early M dwarfs for contrast of 10 magnitudes (10^-4, i.e. transit depth of Earth-sized planets).

Nicole Hess, presented by Horch - speckle images to identify bound stellar companions to Kepler exoplanet hosts. Horch 2014, Matson 2018 (simulated from field population of binaries); Hirsch 2017 - gravitationally bound via color-magnitude diagram. Also common proper motion as another test (using speckle imaging) - Wittrock 2016 as a specific example of one binary. Done this for Kepler-1, 98, 13, 14, 449, 984, 450. They agree with Hirsch 2017 for most systems tested.

Wei Zhu (started in microlensing) - Many Kepler planets have distant companions. His 2018 paper looking at architecture of inner systems, especially as function of e and i. Get his own occurrence rate (30% of stars host planets) and multiplicity (average of 3 planets per system) and says no Kepler dichotomy (check methodology of this paper …). Interested in farther out in the system because massive planets will dominate the mass and angular momentum budget of the system. Look at super Earth-cold Jupiter relations (Zhu & Wu, 2018) - claim 13 of kepler systems have cold jupiters, higher than over all probability of cold jupiters (10%). Does NOT take into account detection efficiency; says Marta Bryan’s paper, which does take into account efficiency, confirms this result. (Un)popularity of solar system: we have no super-Earths, like 70% of systems, but does have cold Jupiter (like 10% of systems); probability of no super-Earth AND cold jupiter, is ~1% (um check the conditional probabilities of this). Looking toward combining TESS (inner systems) with Gaia (outer systems), assess how much the correlation between them remain. Theoretical interpretation: the fact that super-Earths and cold Jupiters often coexist, they know that they don’t compete for planetary building blocks. Also formation of cold Jupiters require more stringent conditions.

Chris Lintott (remote talk) - citizen science with Kepler and K2. He came to this from Zooniverse, was involved in building Planet Hunters. He thinks the success of Planet Hunters comes from the simplicity of the call to action (find planets). He wasn’t convinced that they were going to beat computer algorithms at finding planets - he was pleasantly surprised to be wrong, they have found many systems that slipped through the pipeline!! Often transits were missed because of large TTVs, very large stellar variability. Poor completeness below transit depths of 0.1% (Meg Schwamb’s 2012 paper), but the PH planet properties are fundamentally different than those found via automated algorithms: planet hunters found 10% of Kepler planet candidates > 100 days, 50% at > 600 days (Schmidt 2014, Wang 2015). Also Boyajian’s star discovered by PH. Also K2-138. Also exocoment transits around white dwarf - the citizen scientist co-author downloaded ALL >200,000 Kepler light curves and looked at them by eye to find this one signal! Talk about enthusiasm!! Citizen scientists can interpret graphs! Don’t underestimate the public. Planet hunters with TESS: 15,000 lightcurves looked at by 8-15 people; ~80% are single-transits; others are around unusually variable stars. About half became TCEs. This effort also helped mature the Kepler pipeline - identify additional instrumental effects, understand why pipeline didn’t pick those up, etc. Both human and machine learning are valuable - humans to give the training set, also can tell the machine learning people to ignore the edge cases, which makes their job easier. Paper by Wright (??) on this: combination of both performs better than either individually.

Thursday, March 7

Ruth Angus - Rotation periods + stellar activity in clusters with Kepler. Python code Starry by Rodrigo Luger does the inverse problem: produces a probabilistic map of star spots given an integrated lightcurve (??? how does it account for degeneracies?!?!). Three approaches: 1) Periodograms not so good for identifying rotation periods - signal is not sinusoidal, and is sensitive to harmonics. 2) Autocorrelation functions are more model-independent, and work better to identify true period. 3) Fit a GP to the lightcurve (her code starrotate, builds on DFM’s exoplanet and celerite packages). Also mentions TESS Roulette by Megan Bedell to get a sense of how often stellar rotations appear. McQuillan+ 2014 rotation periods applied to Gaia: see the correlation with stellar type, and some gradient across the width of the HR diagram (age - slow down by magnetic breaking). See a gap in period vs. stellar mass, especially for low-mass stars. Corresponds to 6 Gyr isochrone, but also traces out a constant Rossby number = 0.5 (rotation period/convective overturn timescale). Stellar rotation periods have been used by tons of subfields - keep doing it for K2 and TESS! Then talks about gyrochronology - showed simulation; in theory, after stars reach the main sequence, knowing the stellar mass and rotation period will give you an age. Meibom, van Saders - older astereoseismic stars fall below the Skumanich period-age relation from the seventies (rotating more rapidly that would be expected from relation). Can be solved if magnetic breaking stops at some point. van Saders also trying to model full distribution in McQuillan catalog. Ruth’s gyrochronology Python package: stardate. Stellar activity via flares with Kepler: Davenport shows that flare frequency & flare energy are related to rotation period (higher rotation period, more flare energy at a given flare frequency). Also Ben Montet’s work on longer magnetic cycles: spot dominated for faster rotation periods; faculae dominated at slower rotation periods.

Jason Curtis - building precision stellar clocks with Kepler and Gaia: calibrating gyrochronology with clusters. Gaia helps identify cluster members, which means that the rotation period-Teff relations are very tight at a given age. The issue: no gyrochronology model explains all the data (Hyades can’t even be fit with a Praesepe model). For NGC6811 - able to extend down to M0 with K2. Issue: K-dwarfs don’t slow down! Possibility: there’s a color-dependent magnetic breaking index, but would take K-dwarfs longer than age of the Universe to reach their observed periods! So there’s a time-variable magnetic braking index. Ruprecht 147 really helps to get to lower stellar masses; combine with NGC 6811 to get 2.5 gyrochrone; can constrain this timescale for stalling in the magnetic breaking. This stalling constraint also works for NGC 752. Talked about a stellar stream. What’s with the gap in the mass-rotation period space? Older clusters cross the gap, so it can’t be created by one event in time.

Beate Stelzer - rotation-activity-age relation of M-dwarfs. Look at X-rays. How to constrain the evolution of X-ray luminosity for > a few 100 Myrs? Use M-dwarf companions to white dwarfs. Also looking at flares in K2; find that residual noise in detrended lightcurves that are not at the time of flares is correlated with activity - drastically decreases above 10-day rotation periods. Looking at L_X vs. p_rot relation for M dwarfs (saturation regime + linear regime that I looked at with CHAMP > 10 years ago); new M-dwarfs from K2 at longer periods follow this relation, but also just have some upper limits, so there’s a large spread. Also first constraints on X-ray luminosity and temperature for M-dwarfs in Kepler.

Lauren Doyle - rotational phase distribution of stellar flares on M-dwarfs with K2. Stars later than M4 are fully convective and don’t have a tachocline, but do have magnetic activity, so they must generate their magnetic field in a different way than the Sun. K2 short-cadence data on 31 M-dwarfs; calculate rotation periods, identified flares by eye, calculate energies from multi-wavelength fluxes from PanStarrs. Large flares seen at maximum amplitude of rotational modulation, but flares exist at all phases. Added 18 M-dwarf flare stars from C10-18. With updated data, no preference for any rotational phase, regardless of flare energy. First results from TESS: 150 low-mass stars from Sectors 1-3 (triples sample from K2!); again, find flares are randomly distributed, not even as function of spectral type or rotation period. SO, no correlation between flares and location of star spots. Note: TESS bandpass is redder, so is less sensitive to flares.

Joshua Reding - How hardware failures led to the discovery of the most rapidly rotating isolated white dwarf. They are end stage of 97% of stars; 97% of those are non-variable. Variable ones are caused by binarity, g-mode pulsations, magnetic spots. In Kepler-prime, only 20 white dwarfs, but in K2 observed >2000! SDSS J1252-0234 is the most rapidly rotating WD: ~ 0.68 M_Sun, Teff = 8500K, logg=8. Featureless spectrum (no deep Balmer lines of a H/He white dwarf), maybe produced by extremely strong magnetic field. 7 days into observing this target, that Kepler CCD failed, so got tentative periodic signal that followed up with ground-based spectrograph … whose dispersion element was working, so were forced to do photometry. Typical WD rotation rates are hours to days … this one is 5 minutes!! Also have multi-color photometry from McDonald; g amplitude larger than i. So why variable? Only one peak in SD, and not over-luminous in color-magnitude sequence, so not binary. Velocities of g-mode pulsations would be 0.1c - way too fast; also doesn’t explain color difference in variability amplitude! So maybe magnetic spots! Could be merger remnant …

Andrew Mann - tracing planetary evolution with K2 clusters. How does exoplanet density changes over time? Could arrive at the same final M-R relation, but via a different path. Observe Praesepe (650 Myr), Hyades (650 Mr - secular timescales), Pleiades (125 Myr - photoevaporation), Taurus-Auriga (0-5 Myr), Upper Scorpius (11 Myr - formation) with K2 via the ZEIT survey (Zodical Exoplanets in Time). Complications in observing young planets around young stars: flares (more common), star spots (1-2 orders of magnitude larger, also faster evolution of spots), debris disks, stellar rotation periods comparable to planet orbital timescales. Uses notch filter (window of ~30 data points, fit a notch - also median detrends the lightcurve. HIGHLY sensitive to false positives. LP358-548bcd: 2-planet system in the Hyades. HD283869 b: single planet in the Hyades. Have doubled the sample of young planets to 20 palnets, almost all new ones below 6 R_Earth (had none before K2). Has injection/recovery tests - recover most > 1 R_Earth for P<30 days. Planet radii are larger at young ages, not an effect of detection biases. Also see plenty of eclipsing binaries in these clusters; there’s disagreement between observations and binary star models. Taking a deeper look at K2-25b and K2-33b: transits in Ly-alpha (assume fully ionized . . . some difficulty in interpreting lack of transits in Ly-alpha). Flat transmission spectrum for these. Also some constraints on eccentricity . . . constrain to be > 0.2 (but how well do we really know the sizes of these young stars?!?!?!) Better test from RVs: confirm this with HPF and infrared doppler instrument on Subaru. Initial planet occurrence rates as a function of age. With TESS: THYME survey (TESS Hunt for Young Moving-group Exoplanets. Have thought about Gully’s warning from Monday about stellar spots biasing transit depths … could partly (but not fully) explain why Spitzer transit depths are smaller than Kepler. Also, not much variability as function of rotation phase as you’d expect if this effect was significant.

Ann Marie Cody - Young stars in the Time Domain. Variability has long been associated with young stars (essentially all T-tauri stars are variable). Disks around YSOs … we cannot easily probe the inner 1 AU of disks, but we can map out structure and dynamics with time domain studies. MOST, CoRoT, Spitzer, and K2 have all contributed to these time-domain studies. Classifying lightcurves by eye: dippers (sharp downward), symmetric variability, and bursters (sharp upward) vs. periodic, quasi-periodic, and aperiodic (physically, it’s accretion, occultation, star spots, and binaries). Also came up with a statistical classification of these, agrees with by-eye classifications. ALMA lets us test classification as a function of disk inclination. 13% of YSOs show bursting, with brightening events up to factor of 10 on day-week timescales, with dynamics governed by both accretion rate and inner disk strength; inner edge of disk contains axial asymmetries, with dynamic changes on <1 week timescales, sometimes misaligned with resolved outer disk.

Eric Gaidos - what orbits a young dipper star in Taurus? Hard to understand planet formation because we have gaps in both time (clusters vs. field stars) and space (image outer disks vs. observed planets very close in). Obscuration of disk (about edge-on) is thought to be what is causing the episodic dipper stars like EPIC 204638512 … but for this one, the outer disk is face-on!! Mechanisms for moving disk material in front of star: winds, polar-orbit exocomets (Ansdell 2019). Also have periodic dippers - more regular than earlier discussed, wrong shape for a transit; dips are deeper at bluer wavelengths, so likely structures in dust very close to star. Gaidos 2019 - no sodium in the tail of evaporating planet K2-22 b. TESS is also discovering dippers.

Lauren Venuti - star-disk dynamical interactions in the Lagoon Nebula, which is a complex HII region and star forming site; contains NGC 6530 (2 Myr); contains ~3000 pre-main sequence stars (including 60-70 OB stars), with disk fraction of ~50%. K2 C9 observations; light curves extracted for 323 confirmed members so far. Also many auxiliary observations: photometry, simultaneous H-alpha time series, Spitzer simultaneous monitoring, speckle imaging for a fraction. Lots of variable behavior like what Ann Marie Cody discussed; classified their sample. Previous work shows that different light curve types match predictions for distinct star-disk interaction modes: quasi-periodic dipper comes from stable, funnel-flow accretion; burster/stochastic are unstable accretion in intense, short-lived bursts. For Lagoon: objects with irregular light curves more likely to have large UV excesses; periodic variables correspond to colors of young stars without disks.

Samuel Grunblatt - planets transiting evolved stars. Can help solve mystery of planet inflation: is planet interior inflated directly by increased stellar irradiation, or by delaying cooling after planet formation? Spoiler: highly irradiated Jupiters around evolved stars (K2-97 b, Kepler-422 b) are inflated (by 2-sigma and 1-sigma, respectively), so it has to be the first one (they weren’t highly irradiated when star was on main sequence). How do they become re-inflated? Effect of tides must also be taken into account: planets circularize and in-spiral due to increase in importance of stellar tides once star evolves, creating population of moderately eccentric (e~0.2) Jupiters during this process. Also want to understand stellar variability and its influence on planet property measurements and detection. Occurrence rate of planets with 0.2 AU of low-luminosity red giant branch stars. >10,000 targets of which 2500 were < 9 R_sun, 4500-5500 K, had detectable oscillations to accurately characterize stellar (and planetary) properties. For 3.5-10 days: 0.5% +/- 0.3% for 1-2 R_Jupiter, < 0.2% for 0.5-1 R_Jupiter, <2% for 0.2-0.5 R_Jupiter … so occurrence rate seems to increase with radius around evolved stars, in contrast to decreasing occurrence with radius for main-sequence stars (but possibly mismatch in period ranges considered?) … continue this with TESS!!

Lauren Weiss - planetary system architectures (overview). Kepler has changed how with think about planetary system architectures and formation: found circumbinary planets; misaligned planetary systems; compact multi-planet systems. Only have time to talk about compact multis … compactness has inspired new-ish theories of planet formation: type 1 migration and pebble formation. Test these theories by finding patterns in planetary architectures. So far, host star masses, metallicities, vsini, logg are indistinguishable for singles vs. multis (Xie+ 16 (LAMOST), Weiss+ 18b). Kepler dichotomy - Ballard & Johnson 2016 (yes), Gaidos 2016 (no). Singles and multis have similar distributions in Rp, P; both have the evaporation valley, so perhaps both singles and multis arrived in their present location in the first 100 Myr (maybe). Eccentricity information in transit duration … based on this, Xie 2016 says that singles have higher eccentricities than multis; also Mills finds this using CKS stellar parameters, and additionally finds that singles with the highest eccentricities are around high-metallicity stars -> giant planet perturbers? Mutual inclinations: < 3 degrees (Fang & Margot, 2012). But also the shortest-period planets have the highest mutual inclinations; not completely a transit geometry completeness effect (Dai+ 2018). So likely that singles had “dynamically exciting lives”. Then talk about peas in a pod using systems with >= 4 planets in CKS system (Weiss 2018a): intra-system, have similar sizes; also have similar spacing from system-to-system … but not necessarily in mean-motion resonances. So question: did these systems start as peas in a pod? Regular intervals in semimajor axis and similar masses naturally an outcome of oligarchic growth … so what about the masses of these systems? Millholland 2017 - yes for TTVs. What about TESS? so far, peas in a pod prevail, based on info currently on TFOP.

Jack Lissauer - architectures of planetary systems using all 4 years of Kepler data (3rd paper in the series - more data, better stellar parameters now). Part 1: period ratios & 3-planet resonances. Use all planet candidates, not validated planets, to limit biases as much as possible. In the future will look at distribution of planetary properties as a function of multiplicity. Addresses how period ratio distribution looks different as change histogram bin width/phase … so use kernel density estimators (showed as a function of smoothing length). So take numerical derivative of CDF (so, the PDF … as a function of number of points used in calculation of slope). Dips and peaks near MMRs almost cancel out (only slight net excess of planet pairs near MMRs); most Kepler planets not near resonance. 3-body resonances: f-3-body = p*f_in - (p+q)*f_mid + q*f_out. KOI-500 has 2 3-body resonances. There is an excess of 3-body resonances with small p and q’s, but overall they are rare. Most 3-body resonances have both inner & outer pairs just wide of first-order 2-body resonances. 3-body resonances are too weak to be ‘sticky’ without 2-body MMRs.

Darin Ragozzine - getting more out of information-rich Kepler Multis that show TTVs. Borsato+ 2019: no fundamental tension between RV and TTV for Kepler-9. Mackenzie Kane (paper on arXiv today) - demographics of TTV systems (by-eye classification of strength of TTVs plus Fourier diagrams, etc). Areas of improvement: have an intermediate step of the TTV catalog step; small planets sometimes don’t have transit times; construct TTV catalog from short cadence data (doesn’t currently exist!). Address all these by photodynamical modeling - fit straight to lightcurve, like that applied to Kepler-444 by Mills & Fabrycky 2017. Working on making this public tool: PhoDyMM. Did this for Kepler-18; improved model for mass-eccentricity degeneracy and uses short-cadence data. Will be applying PhoDyMM to all the Kepler multis to do a completeness-corrected M-R relationship.

Sarah Millholland - Obliquity Tides’ role in the planet period ratio distribution. Pile-up wide of mean motion resonances … convergent orbital migration can capture planets into MMR. Tidal evolution in MMR works to reproduce this pile-up, but requires extra source of dissipation. Possible source of this: obliquity tides, because tidal dissipation is enhanced at non-zero obliquity. So, how to obtain and maintain large obliquities? Maintained if planets are trapped in secular spin-orbit resonance: there’s an outer planet with slightly misaligned orbit, which causes precession of the orbits; bulge on star causes the spin-axes to precess. The resonance happens when orbital precession frequency = spin precession frequency - this is a Cassini state. Note: spin precession frequency NOT very sensitive to Love-number. Planet-planet interactions are not only way to maintain this obliquity - can also be induced by circumstellar disk. Typical systems are susceptible to this spin-orbit resonance. Considered a set of 55 Kepler systems with masses measured via TTVs (Hadden & Lithwick, 2017) … 85% of these systems could be in these resonances. Describes her fiducial simulations of this effect. This can help explain why the pile-up at 2:1 is more extreme than 3:2, why systems with the smallest orbital periods have the largest offsets from resonance (Dielisle 2014), and why planets in systems with P2/P1 wide of resonance have larger radii (obliquity tides can provide source of heat to inflate the planets. Also maybe explain rapid orbital decay of WASP-12 b (Millholland & Laughlin, 2018), and anolaous features in thermal phase curves like westward hot spot offset of CoRoT-2b. Another prediction: these Kepler multis should not have satellites, because if they did they couldn’t uphold the spin-orbit resonance.

Miranda Herman - revisiting Kepler’s long-period transiting planets (with Yanqin Wu). They are hard to find … so planet hunters! But it’s hard to characterize detection efficiency of these approaches. So DFM’s 2016 paper on automated search … she extends this to dimmer planets and includes revisions for stellar radii from Gaia. Does injection/recovery tests. New occurrence rate: 0.7 planets per Sun-like star for 0.3-1 R_J and 2-10 yr. New radius distribution: assumes power-law, no period dependence -> power-law index is -1.6, fit to 3 radius bins over the aforementioned range. In line with microlensing studies: cold neptunes ~4 times more common than cold jupiters. Correlation with inner systems: 5 of the planet candidates have inner transiting companions. Average mutual inclination in these systems must be low (< 4 degrees). Find that probability of super-earth given presence of outer planet is ~90%.

Friday, March 8

Barbara Endl - Asteroseismology of white dwarfs. Found outbursts and oscillations for several WD. Different models have different modes; fits are model-dependent. Calibrating model grids for ~20 DAV WDs: significant differences (2 dex in mass abundance of H, 300K in T_eff, 0.1 M_Sun in mass) for the 2 model grids she considers.

Roberto Szabo - classical pulsating Variables in the Kepler/K2 era (mainly talking about RR Lyrae) They are horizontal branch stars, He core H burning, 0.5-1.5 amplitude variables over 0.2-1 days. Pre-Kepler, had characterized simple pulsation plus some nuisance effects. Now we’ve seen high-order resonances, high radial overtons, period doubling - these stars are not boring. There were 50 RRLyr in the Kepler field, but would expect to find more - the Kepler Pixel Project. Also finding more in C10-19 of K2; designed to be a statistical mission. Final data release of light curves will contain > 500 RRLyr! Cepheids - only 1 classical Ceph in Kepler field; detected granulation on this star. In K2, hundreds of classical Cepheids spanning many types.

Katrien Kolenberg - RRLyr in a new light with Kepler. Were originally discovered 120 years ago, and RRlyr are present in globular clusters. Blazhko Effect: the amplitude of the variability is modulated on week to month-long timescales. Lightcurves do not look sinusoidal at all. In K2, the RRLyr she was studying was saturated -> custom aperture. Period doubling newly observed with Kepler: every other cycle is at higher flux than the intermediate ones - only seen in stars that exhibit the Blazhko effect. To explain this, they get additional ground-based data: color photometry, spectroscopy with high time resolution simultaneously with K2. Their spectral lines move and change shape over the course of their variability cycle. The star does not monolithically expand/contract: outer layers contract after inner layers. Also see some chaos in the variability amplitude! Did nonlinear asteroseismology of RRLyrae stars. New models for these stars include radial resonance (connected to period doubling) and shock models. Could possibly extend these models to Cepheids.

David Ciardi - Follow-up observation programs as a legacy of Kepler and K2. FOP became more than getting RVs: also characterize the star, vet astrophysical false positives, validate Kepler’s planet candidates. Enormous task. Kepler was funded by the project, K2 follow-up was totally community-driven, which evolved from competition to collaborative. Furlan, Ciardi+ 2018 - set of 615 “standard” KOIs to compare parameters across techniques. Background binaries - which star does it orbit? Ex: KOI2626 is a triple, so have underestimated radius by factor of 2. Furlan 2017: ~100% of KOIs observed with at least one high-resolution imaging technique. True legacy of Kepler/K2 FOP: social change so that now with TESS FOP, engaged community from the beginning that is open to everyone: currently almost 300 community members involved.

David Latham - Contributions from HARPS-N to the M-R Diagram for Kepler/K2. Look out for Li Zeng’s paper to PNAS (under review) - gap in M-R space? First recon spectroscopy to determine if EPRVs are feasible for a given target (i.e. rule out binaries) using TRES on FLWO at Mount Hopkins; did ~3000 observations for Kepler/K2. Masses using HARPS-N: 4000 observations of 80 stars (2/3rds for Kepler, 13 for K2); 22 published papers with 41 masses (31 better than 3-sigma mass measurements), 18 more coming. Can maybe see gap in M-R space given HARPS data. Correcting for stellar hitter is the main challenge; non-transiting planets can distort results; solar observations are helpful (help understand instrument performance).

Eric Petigura - Metal-rich stars host a greater diversity of planets (his 2018 paper). Metallicity as a link between initial conditions in disk and final outcomes in star/planet. Gonzalez 1997, Fischer & Valenti 2005, Buchhave+ 2012. Looking at the radius-period-metallicity space; computing occurrence rates in this 3-D space. Stellar sample: photometric metallicities were low precision (0.4 dex) and had prior based on solar neighborhood stars. LAMOST metallicities improved precision (0.1 dex) with mean [Fe/H] = 0 … so Kepler field is not metal-poor. Small and distant planets form at all metallicities, while the presence large and close-in (between 1 and 10 days) planets are enhanced by higher metallicity. See that all sized planets have a correlation with metallicity, but that the trend steepens with larger planets. Break in occurrence as a function of period that differs for planets in different radius bins. Planet-metallicity correlation possible for both in-situ formation (if inner disk varies in metallicity) and high eccentricity migration. Distinguish between these by continuing to measure eccentricities and spin-orbit alignments.

Cintia Fernanda Martinez - independent spectroscopic analysis of CKS sample: slope in small planet radius gap. Uses MOOG code (assumes LTE) for the spectroscopic analysis; Teff and logg derived from equivalent widths of FeI and II lines. Also uses Gaia parallaxes to constrain stellar radii, masses. Derived radii have median internal uncertainty of 2.8%. Teff: Good agreement with CKS; small trend with Teff when compared to Buchhave, significant offset from APOGEE (that team still working on Teff scale for dwarfs). logg: small offset for CKS, good agreement with Buchhave. Also compared with asteroseismology - Rstar is very consistent with Huber 2017, Silva-Aguirre 2015, Serenelli 2017. Good agreement in R_star with CKS, but MOOG is 1.5% larger. Reproduces radius gap. Extending to period, have period dependence of gap in agreement with Van Eylen 2018 (R_gap ~ P^-0.1).

Eric Mamajek - closer look at habitable zone terrestrial-sized planet candidates. Definition of habitable zones based on 1-D atmospheres. Gaia helps refine the planet parameters. Has compiled a list of candidate “small temperate planets” from the literature. One example: Kepler-1512. Gaia significantly changed stellar properties - 6 times farther away now. Gaia in general made Kepler’s habitable zone candidates larger and more highly irradiated - out of a couple dozen, only 2 remain in the habitable zone.

Ian Crossfield - atmospheric characterization of Kepler/K2 planets. Atmospheres are hazier around the cooler planets, as found by Crossfield & Kreidberg 2017 - meta-analysis of six low-mass planets (but all > 6 M_Earth). Matches predictions by Caroline Morley. But only 6 planets … want to expand this sample with new low-mass atmospheric observations: GJ 9827 d, HIP 41378b, HD 3167c, HD 106315c, all found by K2, try to extend sample to lower masses. Observations are 12 done. HD 3167 c has some hints of an atmospheric signal. Also first molecular features in HD 97658 b? Falls right on H2O amplitude vs. planet equilibrium trend found in 2017. Looking toward TESS: issues with uncertainties on transit ephemeris; have Spitzer time to address that issue. They find the secondary eclipse of TESS planet HD 202772 b - very deep!!

Melinda Soares-Furtado, poster contest winner - Crowded K2 fields - lots remains to be discovered. They use image subtraction (K2imsub) to search for periodic variability. Outlines her approach to this. Classifies variables and assesses cluster membership. Now have catalog of 1151 variables; 45% are new detections (new ones have lower amplitudes, longer periods). Found 2 new planet candidates, second pulsation mode in a delta Cepheid variable.

Jon Zink, poster contest winner - accounting for incompleteness due to transit multiplicity. Where’s Waldo analogy lolol. During injection & recovery, noticed that existing planets were sometimes lost when the injected transits were recovered. The second signal is harder to detect in your lightcurve that the first. Effect is about 5% at higher SNR, but at detection threshold, it’s a 20% effect!!! Fit marginal radius, period distributions as a function of (observed) multiplicity. Detection efficiency as a function of multiplicity can account for the Kepler dichotomy (???).

Andras Pal, presented by Robert Szakats - Solar system observations with K2 - lots of solar system objects in the ecliptic. In long cadence data, main belt asteroids are fast (5-8 pixels/frame); Jupiter trojans are 2-3 pixels/frame; centaurs, TNOs are 0.5 pixels/frame. Have specialized target pixel files in boomerang shape to account for this. Found some very faint things. Looking for variations in the lightcurves to get out rotation periods of these objects. Future: want to look at the bulge (C9) and TESS.

Jessie Dotson - More solar system observations with K2. With K2, observed 7 moons (around Saturn & Uranus, and Triton) and 370 small bodies (Jupiter trojans and TNOs, mostly). And also Pluto + Charon in C7!! Boomerang postage stamps are a hack - have to reassemble them. Amplitude of their lightcurve is smaller than predicted by static frost model, possibly due to seasonal volatile transport. SO many serendipitous TNOs when observed Uranus!! Serendipitous photometry of 608 main belt asteroids; 90 have measured rotation periods. What can we learn from lightcurves of small bodies? Amplitude gives aspect ratio; rotation period gives dynamical history, binarity, maybe structure. Also get spin axis and shape model. But observational biases limit population studies.

Andrea Fortier - CHEOPS (CHaracterizing ExOplanet Satellite) Mission. 30 cm effective aperture. Single-target pointing. ESA Small Mission. Have characterized sensitivity: SNR = 5 for Earth-sized transit. Earth polar orbit, flying around terminator to minimize stray light. Launch late 2019; nominal mission 3.5 years. Guest Observer program is now open (proposals due mid-May).