Still from a simulation of individual galaxies forming, starting at a time when the Universe was just a few million years old. Credit: Hopkins Research Group, Caltech
Caltech researchers use deep learning and supercomputing to identify Nyx, a product of a long-ago galaxy merger.
Astronomers can go their whole career without finding a new object in the sky. But for Lina Necib, a postdoctoral scholar in theoretical physics at Caltech, the discovery of a cluster of stars in the Milky Way, but not born of the Milky Way, came early with a little help from supercomputers, the Gaia space observatory, and new deep learning methods.
Writing in Nature Astronomy this week, Necib and her collaborators describe Nyx, a vast new stellar stream in the vicinity of the Sun, that may provide the first indication that a dwarf galaxy had merged with the Milky Way disk. These stellar streams are thought to be globular clusters or dwarf galaxies that have been stretched out along its orbit by tidal forces before being completely disrupted.
The discovery of Nyx took a circuitous route, but one that reflects the multifaceted way astronomy and astrophysics are studied today.
Necib studies the kinematics or motions of stars and dark matter in the Milky Way. If there are any clumps of stars that are moving together in a particular fashion, that usually tells us that there is a reason that theyre moving together.
Since 2014, researchers from Caltech, Northwestern University, UC San Diego and UC Berkeley, among other institutions, have been developing highly-detailed simulations of realistic galaxies as part of a project called FIRE (Feedback In Realistic Environments). These simulations include everything scientists know about how galaxies form and evolve. Starting from the virtual equivalent of the beginning of time, the simulations produce galaxies that look and act much like our own.
Concurrent to the FIRE project, the Gaia space observatory was launched in 2013 by the European Space Agency. Its goal is to create an extraordinarily precise three-dimensional map of about one billion stars throughout the Milky Way galaxy and beyond.
The FIRE and FIRE-2 simulations follow the region that will become a single galaxy by the present time, tracing the evolution of dark matter and gas, which eventually turns into stars. Credit: Hopkins Research Group, Caltech
Its the largest kinematic study to date. The observatory provides the motions of one billion stars, she explained. A subset of it, seven million stars, have 3D velocities, which means that we can know exactly where a star is and its motion. Weve gone from very small datasets to doing massive analyses that we couldnt do before to understand the structure of the Milky Way.
The discovery of Nyx involved combining these two major astrophysics projects and analyzing them using deep learning methods.
Among the questions that both the simulations and the sky survey address is: How did the Milky Way become what it is today?
Galaxies form by swallowing other galaxies, Necib said. Weve assumed that the Milky Way had a quiet merger history, and for a while it was concerning how quiet it was because our simulations show a lot of mergers. Now, with access to a lot of smaller structures, we understand it wasnt as quiet as it seemed. Its very powerful to have all these tools, data and simulations. All of them have to be used at once to disentangle this problem. Were at the beginning stages of being able to really understand the formation of the Milky way.
A map of a billion stars is a mixed blessing: so much information, but nearly impossible to parse by human perception.
Before, astronomers had to do a lot of looking and plotting, and maybe use some clustering algorithms. But thats not really possible anymore, Necib said. We cant stare at seven million stars and figure out what theyre doing. What we did in this series of projects was use the Gaia mock catalogues.
The Gaia mock catalogue, developed by Robyn Sanderson (University of Pennsylvania), essentially asked: If the FIRE simulations were real and observed with Gaia, what would we see?
Necibs collaborator, Bryan Ostdiek (formerly at University of Oregon, and now at Harvard University), who had previously been involved in the Large Hadron Collider (LHC) project, had experience dealing with huge datasets using machine and deep learning. Porting those methods over to astrophysics opened the door to a new way to explore the cosmos.
At the LHC, we have incredible simulations, but we worry that machines trained on them may learn the simulation and not real physics, Ostdiek said. In a similar way, the FIRE galaxies provide a wonderful environment to train our models, but they are not the Milky Way. We had to learn not only what could help us identify the interesting stars in simulation, but also how to get this to generalize to our real galaxy.
The team developed a method of tracking the movements of each star in the virtual galaxies and labelling the stars as either born in the host galaxy or accreted as the products of galaxy mergers. The two types of stars have different signatures, though the differences are often subtle. These labels were used to train the deep learning model, which was then tested on other FIRE simulations.
After they built the catalogue, they applied it to the Gaia data. We asked the neural network, Based on what youve learned, can you label if the stars were accreted or not?' Necib said.
The model ranked how confident it was that a star was born outside the Milky Way on a range from 0 to 1. The team created a cutoff with a tolerance for error and began exploring the results.
This approach of applying a model trained on one dataset and applying it to a different but related one is called transfer learning and can be fraught with challenges. We needed to make sure that were not learning artificial things about the simulation, but really whats going on in the data, Necib said. For that, we had to give it a little bit of help and tell it to reweigh certain known elements to give it a bit of an anchor.
They first checked to see if it could identify known features of the galaxy. These include the Gaia sausage the remains of a dwarf galaxy that merged with the Milky Way about six to ten billion years ago and that has a distinctive sausage-like orbital shape.
It has a very specific signature, she explained. If the neural network worked the way its supposed to, we should see this huge structure that we already know is there.
The Gaia sausage was there, as was the stellar halo background stars that give the Milky Way its tell-tale shape and the Helmi stream, another known dwarf galaxy that merged with the Milky Way in the distant past and was discovered in 1999.
The model identified another structure in the analysis: a cluster of 250 stars, rotating with the Milky Ways disk, but also going toward the center of the galaxy.
Your first instinct is that you have a bug, Necib recounted. And youre like, Oh no! So, I didnt tell any of my collaborators for three weeks. Then I started realizing its not a bug, its actually real and its new.
But what if it had already been discovered? You start going through the literature, making sure that nobody has seen it and luckily for me, nobody had. So I got to name it, which is the most exciting thing in astrophysics. I called it Nyx, the Greek goddess of the night. This particular structure is very interesting because it would have been very difficult to see without machine learning.
The project required advanced computing at many different stages. The FIRE and updated FIRE-2 simulations are among the largest computer models of galaxies ever attempted. Each of the nine main simulations three separate galaxy formations, each with slightly different starting point for the sun took months to compute on the largest, fastest supercomputers in the world. These included Blue Waters at the National Center for Supercomputing Applications (NCSA), NASAs High-End Computing facilities, and most recently Stampede2 at the Texas Advanced Computing Center (TACC).
The researchers used clusters at the University of Oregon to train the deep learning model and to apply it to the massive Gaia dataset. They are currently using Frontera, the fastest system at any university in the world, to continue the work.
Everything about this project is computationally very intensive and would not be able to happen without large-scale computing, Necib said.
Necib and her team plan to explore Nyx further using ground-based telescopes. This will provide information about the chemical makeup of the stream, and other details that will help them date Nyxs arrival into the Milky Way, and possibly provide clues on where it came from.
The next data release of Gaia in 2021 will contain additional information about 100 million stars in the catalogue, making more discoveries of accreted clusters likely.
When the Gaia mission started, astronomers knew it was one of the largest datasets that they were going to get, with lots to be excited about, Necib said. But we needed to evolve our techniques to adapt to the dataset. If we didnt change or update our methods, wed be missing out on physics that are in our dataset.
The successes of the Caltech teams approach may have an even bigger impact. Were developing computational tools that will be available for many areas of research and for non-research related things, too, she said. This is how we push the technological frontier in general.
Reference: Evidence for a vast prograde stellar stream in the solar vicinity by Lina Necib, Bryan Ostdiek, Mariangela Lisanti, Timothy Cohen, Marat Freytsis, Shea Garrison-Kimmel, Philip F. Hopkins, Andrew Wetzel and Robyn Sanderson, 6 July 2020, Nature Astronomy.DOI: 10.1038/s41550-020-1131-2
View original post here:
- Rotational spectra of isotopic species of methyl cyanide, CH_3CN, in their ground vibrational states up to terahertz frequencies - November 8th, 2009 [November 8th, 2009]
- Cosmological parameter extraction and biases from type Ia supernova magnitude evolution - November 8th, 2009 [November 8th, 2009]
- Continuous monitoring of pulse period variations in Hercules X-1 using Swift/BAT - November 8th, 2009 [November 8th, 2009]
- Constraining the ortho-to-para ratio of H{_2} with anomalous H{_2}CO absorption - November 8th, 2009 [November 8th, 2009]
- A photometric and spectroscopic study of the new dwarf spheroidal galaxy in Hercules - Metallicity, velocities, and a clean list of RGB members - November 8th, 2009 [November 8th, 2009]
- Luminosities and mass-loss rates of SMC and LMC AGB stars and red supergiants - November 8th, 2009 [November 8th, 2009]
- Electron beam – plasma system with the return current and directivity of its X-ray emission - November 8th, 2009 [November 8th, 2009]
- The propagation of the shock wave from a strong explosion in a plane-parallel stratified medium: the Kompaneets approximation - November 8th, 2009 [November 8th, 2009]
- Analysis of hydrogen-rich magnetic white dwarfs detected in the Sloan Digital Sky Survey - November 8th, 2009 [November 8th, 2009]
- Letter: Centaurus A as TeV \gamma-ray and possible UHE cosmic-ray source - November 8th, 2009 [November 8th, 2009]
- Young pre-low-mass X-ray binaries in the propeller phase - Nature of the 6.7-h periodic X-ray source 1E 161348-5055 in RCW 103 - November 8th, 2009 [November 8th, 2009]
- Radiative rates and electron impact excitation rates for transitions in Cr VIII - November 8th, 2009 [November 8th, 2009]
- Solar granulation from photosphere to low chromosphere observed in Ba II 4554 Å line - November 8th, 2009 [November 8th, 2009]
- Does the HD 209458 planetary system pose a challenge to the stellar atmosphere models? - November 8th, 2009 [November 8th, 2009]
- Effect of asymmetry of the radio source distribution on the apparent proper motion kinematic analysis - November 8th, 2009 [November 8th, 2009]
- Destriping CMB temperature and polarization maps - November 8th, 2009 [November 8th, 2009]
- Search for cold debris disks around M-dwarfs. II - November 8th, 2009 [November 8th, 2009]
- Precise data on Leonid fireballs from all-sky photographic records - November 8th, 2009 [November 8th, 2009]
- An X-ray view of 82 LINERs with Chandra and XMM-Newton data - November 8th, 2009 [November 8th, 2009]
- Radio observations of ZwCl 2341.1+0000: a double radio relic cluster - November 8th, 2009 [November 8th, 2009]
- Candidate free-floating super-Jupiters in the young \sigma Orionis open cluster - November 8th, 2009 [November 8th, 2009]
- The metallicity gradient as a tracer of history and structure: the Magellanic Clouds and M33 galaxies - November 8th, 2009 [November 8th, 2009]
- XMMSL1 J060636.2-694933: an XMM-Newton slew discovery and Swift/Magellan follow up of a new classical nova in the LMC - November 8th, 2009 [November 8th, 2009]
- The inner rim structures of protoplanetary discs - November 8th, 2009 [November 8th, 2009]
- The solar Ba{\sf II} 4554 Å line as a Doppler diagnostic: NLTE analysis in 3D hydrodynamical model - November 8th, 2009 [November 8th, 2009]
- Magnetic evolution of superactive regions - Complexity and potentially unstable magnetic discontinuities - November 8th, 2009 [November 8th, 2009]
- Low-mass protostars and dense cores in different evolutionary stages in IRAS 00213+6530 - November 8th, 2009 [November 8th, 2009]
- PMAS optical integral field spectroscopy of luminous infrared galaxies - I. The atlas - November 8th, 2009 [November 8th, 2009]
- First AGILE catalog of high-confidence gamma-ray sources - November 8th, 2009 [November 8th, 2009]
- Radiative hydrodynamics simulations of red supergiant stars - I. interpretation of interferometric observations - November 8th, 2009 [November 8th, 2009]
- Extrasolar planets and brown dwarfs around A–F type stars - VII. \theta Cygni radial velocity variations: planets or stellar phenomenon? - November 8th, 2009 [November 8th, 2009]
- Cosmic rays and the magnetic field in the nearby starburst galaxy NGC 253 - II. The magnetic field structure - November 8th, 2009 [November 8th, 2009]
- Physical structure and water line spectrum predictions of the intermediate mass protostar OMC2-FIR4 - November 8th, 2009 [November 8th, 2009]
- The bright galaxy population of five medium redshift clusters - II. Quantitative galaxy morphology - November 8th, 2009 [November 8th, 2009]
- Dust in brown dwarfs and extra-solar planets - II. Cloud formation for cosmologically evolving abundances - November 8th, 2009 [November 8th, 2009]
- The quiet Sun magnetic field observed with ZIMPOL on THEMIS - I. The probability density function - November 8th, 2009 [November 8th, 2009]
- Complexity in the sunspot cycle - November 8th, 2009 [November 8th, 2009]
- Properties and nature of Be stars - 26. Long-term and orbital changes of \zeta Tauri - November 8th, 2009 [November 8th, 2009]
- The massive Wolf-Rayet binary LSS 1964 (=WR 29) - II. The V light curve - November 8th, 2009 [November 8th, 2009]
- Supernova progenitor stars in the initial range of 23 to 33 solar masses and their relation with the SNR Cassiopeia A - November 8th, 2009 [November 8th, 2009]
- The Hertzsprung-Russell Diagram of Star Clusters - November 8th, 2009 [November 8th, 2009]
- Table of the 10 Brightest stars within 10 Parsecs of the Sun - November 8th, 2009 [November 8th, 2009]
- The Hertzsprung-Russell Diagram of the Nearest Stars - November 8th, 2009 [November 8th, 2009]
- Magnitude and Color in Astronomy - November 8th, 2009 [November 8th, 2009]
- Stellar Types - November 8th, 2009 [November 8th, 2009]
- Brown Dwarfs - November 8th, 2009 [November 8th, 2009]
- Spotting the Minimum - November 8th, 2009 [November 8th, 2009]
- The Structure and Evolution of Brown Dwarfs - November 8th, 2009 [November 8th, 2009]
- No Bang from the Big Bang Machine - November 8th, 2009 [November 8th, 2009]
- The Sizes of the Stars and the Planets - November 8th, 2009 [November 8th, 2009]
- An Implausible Light Thrust - November 8th, 2009 [November 8th, 2009]
- the Masses of Degenerate Objects - November 8th, 2009 [November 8th, 2009]
- Degeneracy Pressure - November 8th, 2009 [November 8th, 2009]
- Introduction to Degenerate Objects - November 8th, 2009 [November 8th, 2009]
- The Radii of Degenerate Objects - November 8th, 2009 [November 8th, 2009]
- The Inevitability of Black Holes - November 8th, 2009 [November 8th, 2009]
- Scientific Pig-Out - November 8th, 2009 [November 8th, 2009]
- The Neutrino Cooling of Degenerate Dwarfs - November 8th, 2009 [November 8th, 2009]
- The Neutrino Cooling of Neutron Stars - November 8th, 2009 [November 8th, 2009]
- Overview of Supernovae - November 8th, 2009 [November 8th, 2009]
- Energetics of Thermonuclear Supernovae - November 8th, 2009 [November 8th, 2009]
- Thermonuclear Supernovae - November 8th, 2009 [November 8th, 2009]
- Nuclear Reactions in Thermonuclear Supernovae - November 8th, 2009 [November 8th, 2009]
- Core-Collapse Supernovae - November 8th, 2009 [November 8th, 2009]
- Neutrinos and SN 1987A - November 8th, 2009 [November 8th, 2009]
- Revealing the sub-AU asymmetries of the inner dust rim in the disk around the Herbig Ae star R Coronae Austrinae - December 13th, 2009 [December 13th, 2009]
- Probing the dust properties of galaxies up to submillimetre wavelengths - I. The spectral energy distribution of dwarf galaxies using LABOCA - December 13th, 2009 [December 13th, 2009]
- On the physical origin of the second solar spectrum of the Sc II line at 4247 Å - December 13th, 2009 [December 13th, 2009]
- On detecting the large separation in the autocorrelation of stellar oscillation times series - December 13th, 2009 [December 13th, 2009]
- Imaging the spotty surface of Betelgeuse in the H band - December 13th, 2009 [December 13th, 2009]
- Chandra observation of Cepheus A: the diffuse emission of HH 168 resolved - December 13th, 2009 [December 13th, 2009]
- A planetary eclipse map of CoRoT-2a - Comprehensive lightcurve modeling combining rotational-modulation and transits - December 13th, 2009 [December 13th, 2009]
- The chemical composition of carbon stars. The R-type stars - December 13th, 2009 [December 13th, 2009]
- Flow instabilities of magnetic flux tubes - IV. Flux storage in the solar overshoot region - December 13th, 2009 [December 13th, 2009]
- Fragmentation of a dynamically condensing radiative layer - December 13th, 2009 [December 13th, 2009]
- Temporal variations of the CaXIX spectra in solar flares - December 13th, 2009 [December 13th, 2009]
- Deuterium chemistry in the Orion Bar PDR - “Warm” chemistry starring CH_{2}D^+ - December 13th, 2009 [December 13th, 2009]
- Metal abundances in the cool cores of galaxy clusters - December 13th, 2009 [December 13th, 2009]
- The nature of the X-ray binary IGR J19294+1816 from INTEGRAL, RXTE, and Swift observations - December 13th, 2009 [December 13th, 2009]
- Relating basic properties of bright early-type dwarf galaxies to their location in Abell 901/902 - December 13th, 2009 [December 13th, 2009]