looks strange” Learned - should have known: Can’t look for GW in DR2 (need residuals from epoch-astrometry, not motion fits) - not surprising: Gaia doesn’t share scanning solution (but it’s not too hard to extract; thanks @SihaoCheng!) - surprise: spin-2 spherical-harmonics are our friends (and emcee can calculate them quite well!) Accomplished - Many additions to simulated GW-detection pipeline - Realish scanning law for “observations” (almost done) - Incorporate model of Basic-Angle modulations (``) - Use real DR2 magnitudes and uncertainties - `` DR2 proper-motions, and positions Luke Zoltan Kelley LKelley@cfa.harvard.edu Harvard CfA ⇒ Northwestern CIERA Gaia Astrometric Gravitational Wave Detection Thanks So Much!! CCA, SOC, GAIA & Sprint Attendees
projects Computation of spectroscopic parallaxes (Eilers et al): understood individual outliers due to astrometric quality, photometric quality, variability Explored the identification of multiplicity from Gaia-DR2 and DR1-TGAS ◦ Goals: TGAS 2.0, long-period (>25 yr) binaries, dynamical masses from orbital fits ◦ People involved: Brandt, Faherty, Hernandez, Michalik, Mora, etc Intended to confirm/reject kinematic cluster members in chemical space ◦ Too little overlap with APOGEE, GALAH ◦ Learned about Hypatia (Hinkel et al.) - to be continued! Visiting the Netherlands? Say hi at ESTEC - standing invitation for all sprint participants! Email address is in the running notes of the sprint
APOGEE spectra? A (photometric) metallicity Map of the MW w/ N. Hinkel w/ Y. S. Ting YS Ting + shows can measure [Fe/H] from WISE+Gaia+2MASS+Panstarrs photemotry to ~0.20 dex (NN, Ting) and (PM, Hawkins) HYPOGEE: APOGEE+Hypatia (111 stars) [Eu/Fe] in GALAH+APOGEE: -0.14 (0.15 dex) [Eu/Fe] in Gaia-ESO+APOGEE: 0.02 (0.18 dex) Line identification underway w/ N Hinkel Maybe wrong?
profile of the stellar halo using BHB and K Giants from SEGUE (Xue et al.). • I used a non-parametric Bayesian scheme, along with a model for outliers. I take into account uncertainties in radial velocity, proper motion and distance. The model had about 120 free parameters. • While most N-body simulations suggest outer stellar halo to be radial ( >0) we find the MW halo to be non radial ( <0). Bullock & Johnston 2005 (Kafle et al. 2012)
for patchy/uncertain data - Chemical clustering on APOGEE spectra/GALAH abundances in Gaia overlap - Abundance trends in R, z with Gaia positions What didn’t: - Verifying clusters with chemical space alone - Distinguishing open clusters in chemical space with DBSCAN Next steps/new ideas: - Implement FOPTICS in Python (thanks: Joris de Ridder) - Look for chemistry of nearby young associations (thanks: Jonathan Gagné) - Action-angle space clustering (thanks: 4th floor library)
my (simulated) open clusters to be compared with observational data ✅ Used Boquan Chen’s clustering algorithm, adapted it for my simulations ✅ Learned about arcane topics such as parallaxes and proper motions ✅ Talked to a lot of people, got new ideas for my project ✅ Preached about AMUSE ✅ Tweeted a lot #GaiaSprint Did not feel judged about my usual excessive emoji use ✅ Wrote a script to run an open cluster moving through the galaxy and interacting with giant molecular clouds… ❌ ...to see if it would yield something similar to Semyeong’s object. It didn’t http://francisca.cr @franconchar franciscaconcha Let’s a k u p lu r !✨
set of 4,000 BHBs ---> Gaia catalogue of 400,000 Needs to be cleaned (a lot), but can be used to map halo in proper motions! Douglas Boubert (IoA, Cambridge -> Magdalen College, Oxford) What is next: Map these stars in velocity, add in GALEX and WISE, go to lower latitudes, extend to classifying quasars as well. What I learnt: 1. Only write code if you are sure that it isn’t in Bovy’s back-catalogue. 2. You are only as good as your internet connection to the Gaia archive.
Result: finished 0, sniffed out or co-started >5. I am happy Learned: • that global kinematic disk maps can be made! (from Inno, Eilers, Hogg, ..) • how the disk heats (from Ting) • how to find massive stars (from Cantiello,Mora) • infinitely more from many others Took away: Gaia + Sprinters sowed 1000 scientific flowers. Let’s await the bloom The gently warping young disk (courtesy Laura Inno)
of gravitational objects by calculating their phase space volume in the Gaia data. Test this using M4 Problem: too far for radial position/velocity measurements to be accurate. But! It’s a spherical system, so just rotate the projected positions/velocities (this cannot increase the phase space volume). I calculate a mass within 10-20% of the accepted answer (with caveats) Accepted mass My best-fit mass Range of masses using range of concentration parameters of globular clusters Thanks to Katelin, Lina, Hamish, Oren, Colin, Adrian, and Hogg for help this week
Oren Slone Measurable with DR2 Unknown Radial + Vertical Forces And Disequilibria X R X Z Zero Disequilibria: X R = -K R X Z = -K Z Comparing Populations i and j, extracting time term T X Ri - X Rj = T Ri - T Rj X Zi - X Zj = T Zi - T Zj
Oren Slone Outlook: • Performing same analysis on tracers with different dynamical mixing timescales • Find regions that are relatively in equilibrium • Model time-dependent terms • Measure properties of the DM halo (triaxiality, radial profile, mass, local density) • Measure correlation of DM/acceleration with baryonic profile/acceleration • Compare with simulations of various DM models 10-13 km s-2
to find accelerating objects Wrapping up Gaia Sprint! LMC 4. PSA: Don’t forget radial velocity is in heliocentric frame SMC 1. Comoving Objects Collector: Group by patches of sky, only take objects with low variance on proper motion. Insights: 1. LMC is rotating and spinning 2. SMC also; but has satellites 3. I think we can use this to find streams 3. Lensing: Ken will talk about our ongoing analysis!
bear imprints of disruption and common envelope! • Binary fraction does not depend on [Fe/H] for wide binaries! • Learned about binary evolution, Gaia sensitivity limits, white dwarfs • Good discussion with Badenes, Price-Whelan, Rix, Liu, Widmark, Oh, Koposov ++ • Met many nice people!
sample to Gaia for: - Understanding astrometric flags (with Richard Galvez) - Comovers, i.e. triple systems (with Megan Bedell) - Cleaned 25pc sample way too much so killed all binaries Things I learned: - Tricks and quirks of the Gaia data - Brown dwarfs in Gaia too faint to find even fainter things around them for now :( Maybe next DR - My cleanish 25pc sample is way too clean! Need to relax quality cuts. Most importantly, quality cuts depend on the kind of objects you’re interested in. Thank you everyone! astrometric_excess_noise
selection for the AntiCenter Stream - to select targets from MSTO to RGB for stream modeling.. Saw that Monoceros and ACS appear as “distinct” kinematic components in mul vs mub space ACS looks very much like the remnant of a tidal tail from Sgr perturbation to disc. Had conversations with HWR and DH about outer disc (perhaps some stars in APOGEE fall on the structure or Monoceros - will need to discss further when I select all the M-giants in the field). Also some conversations with Colin Slater on Anticenter! Had conversations with Ron + Hamish about disc disequilibrium in simulations and data. Had problems with proper motion subtractions in the edge of the disc - will have to write own routines to figure what is going on. Was hoping to learn more about modeling, but will be in touch with members. Thanks for the invitation!!!
around post(-post) AGB stars… and failed to fail... so far? I tried a lot of stuff with varying levels of stupidity • constructing absolute magnitude incorrectly, like 3 times • selecting post-AGB stars from Gaia (hard to tell from reddened hot stars) • using catalogs of post AGB stars (dust likely way to close in & opaque) • trying to use the a_g_val (scary systematics) I learned about the HRD and WD cooling curves ~300, d < 1kpc Hot WDs ~1 Myr past tp 0.02 mag of reddening ~ 1% decrease in star counts Few M⊙ of dusty gas?? I remain skeptical
the completeness function from Ronald Drummel and Jan Rybizki. • Failed multiple times at getting the density distribution of the metal poor stars (all by myself!). • Joris De Ridder taught me how to query data (now I know that I hate it). • Jackie Faherty and Jonathan Gagne visualized my sample! • Keith Hawkins suggested ways to test the validity of the chemical abundances of my sample of stars. • Got faster MCMC code from Lachlan Lancaster. • Discussed new datasets of metal poor stars with Juna Kollmeier. • Discussed the density distribution with Vasily Belokurov. • Talked to the particle physicists: Mariangela, Matt, Neal, Ken, Katelin, Oren, and Hamish. • Did not manage to get a decent plot of the density distribution before we have to present the wrap up slides. R [kpc] M_sun/kpc^3 Slope = ?
extinction probabilities for each star and applied these to calculate absolute magnitudes, empirical radii and bolometric luminosities • Learned so much about dust extinction, data validation, data storage, integers and more! • Great discussions with so many people including Josh Peek, Eddie Schlafly, Boris Leistedt, Alcione Mora, Megan Bedell, Jonathan Gagne, Adrian Price Whelan, Rocio Kiman + it was lovely to meet everyone! Ellie Schwab Abrahams (AMNH, UC Berkeley)
G Period-Luminosity Relationship for RR Lyrae ◦ Use MG = 0.5 mag -- look for paper by Jill Neeley soon. ★ Collated 200+ RRL spectra into phase-curves in Temp, log(g), & [(Fe, Si, Al)/H] -- We can & will do RR Lyrae Chemistry in APOGEE! (left) ◦ Bounced it to students at UVa to finish polishing the line list :-). ★ Cross Match OGLEIV (Sgr + Bulge) with Berry’s master Gaia RRL Catalog. Sgr OGLE IV RRL ★ Use 1% mean distance to Sgr from SMHASH (MIR RRL) an assume that the scatter in mag = line-of-sight depth. Coarsely, agrees with what we get using MIR PL. ★ Convert PM to Vtan for each star. ★ Remove “bulk” motion (just the PM components) Super Preliminary
I did this week: • Completed my catalog of subdwarfs type sdM6 and later • Updated absolute magnitude relations for JHK vs SpT • Started to play around with the sample • Found an awesome spot to work all week! Post-Gaia
pleasure to meet the authors of my favourite papers -- Adapted a radial migration model (##) from p([Fe/H], age | R_gal, model) to p([Fe/H], age, R_gal | model) → can now predict a density profile. to plug in a spatial selection function. -- Received tips for APOGEE selection function from Melissa N. -- Was stuck for 24h on a 4D integral connecting model predictions to selected data, until Dan F.M. pointed out it should be a 3D integral! -- Working on a 3D integral.
parallaxes with APOGEE and Eilers ◦ (side conversation: radial-velocity binaries with Price-Whelan and Casey) ◦ (side conversation: velocity-marginalized parallax likelihoods with Leistedt) • I learned: L1-regularized linear regression is incredible: We beat full spectral modeling in spectroscopic-parallax precision, and our data-driven model is literally a projection of the spectral (and photometric) data onto a single linear basis vector! ◦ Take that, deep learning! ◦ Model is interpretable and uncertainties can be propagated; not true for deep learning! • Now we can: Make precise maps of the Milky Way in abundances and kinematics. ◦ (out to 30 kpc from the Galactic Center; much further than red-clump stars)
structure in the velocity distribution, and potential causes for it - e.g. Spirals, phase mixing after passage of dwarf / merger. Finished this paper On (arXiv tomorrow!) Then I (mostly) learned actions! (Thanks Wilma & Adrian) And, with 4th floor library: Action space clustering (credit: Natalie P-J) Searching for resonance conditions in action / angle / frequency space (credit: Ted)
& 46 kpc away)? • Giant stars hosting sub-Neptune planets? • Matthew Wilde told me he also see similar stuff for Kepler Could there be potential metallicity offset between Exoplanet host and non-host? Done: X-matched Gaia x K2 x APASS x 2MASS for M dwarfs
goose chase :/ Talked to very interesting people \o/ Downloaded ~10K Gaia Cepheids :O Downloaded ~50 HST Cepheids :) Moved from Python 2.7→3.6 >:( Discovered Astropy’s awesomeness \o/ Matched all but 6 Cepheids… O.o
with 5d information in Gaia 2.Chemo-Dynamical Tagging with APOGEE-Gaia (Refound open and globular clusters and some features) 3.Fe-Dynamical Tagging with LAMOST-Gaia (locally) 4. Did not have time to apply the SNN algorithm to Gaia DR2 beyond Orion. To be continued…….
heavier than 2 * 10^7 solar masses with a scale radius of 0.4 pc and line-of-sight distance of 2 kpc would have shown up as an outlier larger than any measured test statistic Astrometric, weak, gravitational lensing by completely dark, compact, dark matter clumps in the LMC foreground
to calibrate RR Lyrae color First action distribution in my life Learnt useful knowledge of Gaia, lightcurve data, RR Lyrae property, dynamics Co m=4 m=2 Assume Flat rotation curve Omega_p ~ 1.5 R_co ~ 5.3 kpc Thanks to Wilma and Johanna JR Lz
tonne of stuff this week, and formed new collaborations. • Inferring binarity for ~10 million stars in Gaia: detection, characterisation (period, semi-amplitude, inclination). • Built a non-parametric model of binarity across the H-R diagram using astrometry, radial velocity (absence/scatter), colours, systematics, photometric variability, etc. (~210 million parameters) • Found a few candidate stellar-mass black hole binary systems. • Used Gaia data to kill off multi-modality in APOGEE binary solutions. • Made a Python tool to query Gaia forecast tool. Through collaboration and discussions with Badenes, El-Badry, Foreman-Mackey, Hogg, Holl, Michalik, Hernandez, Koposov, and Price-Whelan. Slide design by Adrian Price-Whelan
awesome visualizations with AMNH and Jonathan Gagné, Jackie Faherty, & Keith Hawkins. Young moving groups -- hint of a chemical stream? Still working on: HYPOGEE...111 stars I also am using (tried? still trying?): Elements = [-0.25, 0, 0.25] Fe, C, O, Ti, Mg, Na, Al [Eu/Fe] in GALAH+APOGEE: -0.14 (0.15 dex) [Eu/Fe] in Gaia-ESO+APOGEE: 0.02 (0.18 dex)
ICCUB, DPAC-CU9) • I used J. Gagné’s selection of YLAs to re-compute dynamical ages via traceback analysis (backward orbital integration). • For most of the YLAs we are able to provide an estimated age or a range, improving GDR1-TGAS results. • For TW-Hya, we now have 22 members and give an age estimate of 4-11Myr. TW - Hya, estimated age from literature: 3-15Myr Not available using GDR1-TGAS __ observed __ estimated ... errors
check if it’s associated with Pal 5’s leading arm My project on Pal 5: Sarah Pearson (Columbia University) As predicted by my simulations of Pal 5 in a barred Galactic potential CMD of red blob region CMD of Pal 5 and blob CMD of Pal 5 (shifted) to closer distance and the blob Pal 5 in DR2:
from APOGEE spectra + 3 colors • distances to ~50,000 RGB stars out to ~30 kpc with < 10% uncertainties In collaboration with David W Hogg, Melissa Ness, Hans-Walter Rix, Daniel Michalik & Sven Buder! Thank you!!
New connections! (Dank U Santiago and the Leiden Sterrewacht group ;) & Co. ◦ Discussions on stellar perturbations on Solar System, and brown dwarfs ◦ Start astrometric reductions from stellar catalogue • I didn’t yet do that ◦ Looking forward to DR3 and next GaiaSprint (this was my 1st one) & more Solar System Science? (SSO rocks!) Thanks to NYC GaiaSprint organizers and Flatiron inst. firstname.lastname@example.org
truly amazing (thanks Elena D'Onghia) though very difficult (thanks (?) Vasily Belokurov) ◦ Found that one suggested leading stream star is in the disk (hence discarded) • people seem extremely interested using the variable star samples (especially RR Lyrae) ◦ Our partially disjoint classifier and detailed analyses tables successfully confused people, so we provided above link to get unconfused) ◦ Many people hack first / read later (is reading papers overrated?) • That it was very useful&pleasing to contribute in people’s understanding of the data • This sprint was the best scientific experience I had (so far ;)
truly amazing (thanks Elena D'Onghia) though very difficult (thanks (?) Vasily Belokurov) ◦ Found that one suggested leading stream star is in the disk (hence discarded) • people seem extremely interested using the variable star samples (especially RR Lyrae) ◦ Our partially disjoint classifier and detailed analyses tables successfully confused people, so we provided above link to get unconfused. ◦ Many people hack first / read later (is reading papers overrated?) • That it was very useful & pleasing to contribute in people’s understanding of the data • This sprint was the best scientific experience I ever had! • Flatiron Institute: many thanks for the travel support, great facilities and super tasty food
in Galactic longitude coord; mostly showing “normal” disk behavior + Mon + ACS + EBS + Etc. No big surprises(?) Lots of thanks to Chervin Laporte for the ideas and for doing a more proper job Colin Slater (University of Washington) MSTO stars, 20.5 > G > 19, low parallax
Jupiters to ~340 for which I could calculate UVW velocities (!) • Learned many things that made it all that easier (thanks Adrian and others!) • Future: ◦ Disentangle metallicity-age effect ◦ Select a better control sample to do that ◦ Extinction correction ◦ Better MS selection
Milky Way galaxy with a buckling bar of 5 kpc. Stars selected within 1 kpc from the Sun. At the time of the buckling of the bar there is an empty region in JR- Lz (thanks Wilma) Test to look for corotation in Action Space: we can look at Perseus, Outer Arm Vr Vphi See Chao Liu’s plots eccentricity
people, learned A TON. Special thanks to Andy Casey, Adrian Price-Whelan, Kareem El-Badry, and Josh Peek! • Stellar multiplicity with Gaia RV dispersions: doable but needs more work. Andy Casey’s approach is more powerful (but potentially harder to interpret). • Many cool systems (bunnies) in the censored zone!
• Looked at kinematics + stellar age + composition connections with solar twins • Found co-moving Gaia sources in the Kepler field • Fun science with excellent people! One cool result: (with Ben Montet (UChicago)) This star is 4 Gyr old (from isochrones + Gaia plx) These stars are comoving in Gaia DR2! This star has a Hot Jupiter and a 4-day rotation period (from Kepler PSF modeling)
a few odd sources which I’d like to investigate, explained the astrometry flags, good some ideas of what could be helpful to try improving, enjoyed seeing how Gaia data gets used in very imaginative and surprising ways. Got many new ideas of things to look at… so may be DR3 will get delayed a bit :-) I did that: Worked with Jackie, Tim, Daniel and Alcione on possible binaries found comparing DR1/DR2 proper motions, investigated DR2 pm systematics in the RF.
do my proposed project! - I learned stuff about actions and angles (thanks to Wilma Trick: Action Woman), learned stuff about the bar (and some bars) from Jason - and made a .gif, because thats what we do now, right? (thanks Adrian) - Once again, had a great week with all you great folks.
measured their tangential motion. Looked at lots of awesome Sag plots by Denis and Vasily. Failed: detecting fainter streams in proper motion space. But at least now I’ve learned how to do it. And learned a lot from many awesome people!
(?helpful?) thoughts for projects with Sarah Pearson, David Hendel, Chervin Laporte; Ana Bonaca • Appreciated being surrounded by exciting science; • Interrupted Wilma too many times as she imparted great action-intuition; THANKS: • Gaia team • Sprint leaders - Hogg! Ana! Adrian! ….. • Jackie for AMNH immersion • And fellow sprinters/ramblers…...
Distances to ALL SOS RRLyrae computed with PMG from R. Beaton @ Gaia Sprint (available online) • In progress: testing RRL catalogue from Classifier Table (Berry Holl) -> 195,780 (RR*) matched to compilation of (~12) RRLyrae surveys from literature -> advantage: sky uniformity, at cost of higher contamination SOS RR (~120K, astro_excess_noise<2) with (lots of) help from Berry Holl and Joris de Ridder