Stream fanning

streamfanning In my first year at Columbia I worked with grad student Sarah Pearson on an idea that Kathryn Johnston had while trying to find an orbit for Palomar 5 in a Law & Majewski potential. Wait what? Who’s Sarah, who’s Palomar 5, and what is a Law & Majewski potential?

Palomar 5 is a globular cluster in the halo of our Galaxy, the Milky Way. It is about 12 billion years old and consists of roughly 30,000 stars. The star cluster can be seen within the footprint of the Sloan Digital Sky Survey. But even more fascinating is that we can also see a stream, consisting of at least as many stars, stretching out from the cluster along its orbit. This stream – there are actually two, one in the leading direction and one in the trailing direction – spans about 23 degrees on the sky, while being on average half a degree wide. That’s about the size of 50 full moons!

In the above figure, Palomar 5 can be seen in the blue density contours as a blob at roughly RA = 229 deg and Dec = 0 deg. The stream stretches from the upper left to the lower right. There’s lots of random fluctuations all over this density map, but we made sure that the stream is really only along this diagonal, and that the rest is just noise. Red points in the figure show the N-body model that comes closest to the observed shape of the stream. That’s not close AT ALL you may say, and you’re right. What Sarah used to generate this model is a very specific form of the mass distribution within the Milky Way and its dark matter halo. This form was proposed by David Law and Steve Majewski as they modeled the Sagittarius stream – a similar stream to the one of Palomar 5, but much further out in the halo and from a different kind of progenitor, a dwarf galaxy.

The Law & Majewski potential, as we call it, looks like an American football, and its oriented such that it causes all stars, star clusters and dwarf galaxies to be on weird orbits within the Galaxy halo. Sarah found that the orbits are in fact so weird that it’s impossible to find a thin and curved stream like the Palomar 5 stream within this potential. The streams don’t stay thin and long, but instead they fan out. From this we concluded that a) the Law & Majewski potential must be wrong within the inner part of the halo of the Galaxy, and b) that whenever we see a thin and curved stream in the sky, we can rule out certain classes of (weirdly shaped) potentials. As a consistency test, Sarah tried the same with a simple spherical dark matter halo – and it was super easy to get perfect models of Palomar 5.

Given the fact that this was Sarah’s first-year project, and that she was still taking lots of classes at the time, we’re all stoked that she got this substantial paper out so quickly. Go check out her website, she’s up to great things.

With the AAS in Seattle


The year 2015 started off with a trip to Seattle to attend the 225th meeting of the American Astronomical Society (AAS). I just became a member of the AAS (for which I’m grateful to my sponsors Kathryn Johnston and Jerry Ostriker!), and so this was the first time for me to attend one of these ginormous, annual meetings. When I first looked at the program, I was impressed (not to say overwhelmed) by the number of attendees, the variety of splinter meetings, and by the quality & quantity of career opportunities. This impression didn’t change much throughout the week.

About 3000 astronomers from all over the world (but mostly the US) come to the traditional winter meeting. All fields, subfields, and niches of astronomy are present. So whatever you’re looking for you’ll find it there. On top of that, many companies, institutions and media representatives show up at this meeting. It’s the one time of the year where astronomy makes big news and big business.

Now, usually a conference has about 100 participants from one discipline of which maybe 50 will give a talk of about 20-30 min. At the AAS meeting, talks are only 5 minutes, and the sessions are very mixed. Hence, it’s much more favorable to present a poster – which I did (see above). I haven’t presented a poster in years, so I was skeptical in the beginning. But now after this meeting I’m a big fan of posters! The interactions you have while standing in front of your poster are much deeper and more fruitful than what you get from a talk. Without the big audience listening, people can and will ask dumb questions, which turn out to be not dumb at all. They make you realize what you can’t realize during a talk – how many people in the audience don’t get what you’re doing because you didn’t bother to explain.

Besides the poster, I enjoyed many helpful career sessions, a packed (!) impostor-syndrome session, a memorable astronomer party, and lots of great food and coffee with amazing colleagues and friends. The whole event is really more of a big party than a conference. Thus, I’m looking forward to the next winter meeting in Florida – even though beautiful and coffee-loving Seattle is hard to beat as a venue!

Tidal streams in an evolving dark matter halo


2014 has been a great scientific year for me, in which I had the opportunity to contribute to many exciting projects. I’m very fond of the paper that came out of my collaboration with Ana Bonaca and Marla Geha, which we started back in 2013 when I was at Yale. Ana, who is a PhD student at Yale and who got famous for discovering the Triangulum stream in the southern part of the Sloan Digital Sky Survey, put a lot of effort into this project, and I’ve learned a lot from working with her.

The question we asked was simple: “What is the effect of an evolving galactic dark matter halo on the formation and evolution of tidal streams!?” – motivated by the fact that in all tidal stream research we simplify the galactic halo by assuming that it has a simple analytical form and that it does not evolve with time. It was obvious to everyone that this has to be an oversimplification, as dark matter halos are supposedly made of dark matter particles, which form bound substructures on all length scales. That is, a galactic dark matter halo contains hundred thousands of subhalos orbiting within the main halo, which themselves are substructured. But due to computational difficulties, investigations have either focussed on tidal streams or on the structure of dark matter halos, not on both at the same time.

Thus, answering our question was tricky, and it required us to run a >1 billion particle simulation of a live-forming dark matter halo. Together with Jürg Diemand from the University of Zurich, we re-computed the Via Lactea II simulation (one of the highest-resolution N-body simulations of a Milky Way-sized dark matter halo) from a snapshot 6 billion years in the past to the present day. While doing so, we simultaneously generated tidal streams from >10,000 “cluster particles”, which we randomly inserted into the simulation such that they covered a wide range of possible orbits within the main halo.

Over the course of 6 billion years of simulation, each of the cluster particles generated a streakline-ish tidal stream by constantly releasing stream particles into the galactic halo. The same we did for an analytic galaxy halo that did not evolve with time. Two examples of streams forming in the evolving and in the non-evolving galaxy halos are shown in the figure above. It is immediately apparent that the streams in the “lumpy & evolving” halo are much wider and more dispersed than in the “smooth & static” case.

Instead of asking what is causing these differences in detail, we asked ourselves, how this would affect our goal of measuring the weight of a galaxy by modeling the streams. Ana took 256 random streams from both data sets (evolving halo and analytic halo) and modeled each of them with our Fast-Forward method – attempting to recover the mass and shape of the dark matter halo. It turned out that, while this was easily possible for the analytic, static halo, it can be fatal for the evolving halo.

Streams have millions of encounters with the dark matter subhalos while they are orbiting through the main halo. These encounters can be weak and negligible, or they can be strong and even deflect the orbit of the whole stream. Moreover, the encounters can be disruptive, and punch holes into the dynamically cold streams, or merely puff them up a bit. All these effects together make our recovery approach prone to biases, i.e. errors in the interpretation of the underlying mass and shape of the galactic dark matter halo. We may overestimate the mass of the galaxy by up to 50%, Ana found!

However, there’s still hope for us! By combining several streams, we will be able to correct for all the mistakes we could possibly make. So there’s our main motivation for observing more streams on the sky, and finally modeling them all together. It seems to be the only way to accurately measure the weight of a galaxy like our Milky Way.

Streams, streams, streams


For my presentation at the Gaia Challenge in Heidelberg, I made a quick ADS search for publications on tidal stream observations since their first discovery in 1995 by Carl Grillmair. Although tidal streams had been theoretically predicted before 1995, Grillmair showed for the first time that some star clusters have significant amounts of stars outside their tidal radii. At about the same time, the Sagittarius (Sgr) dwarf galaxy was discovered by Rodrigo Ibata, and people started finding patches of stars that belong to the stream emanating from this galactic satellite everywhere in the halo of the Milky Way. It took 6 more years until Michael Odenkirchen showed with commissioning data of the Sloan Digital Sky Survey that the globular cluster Palomar 5 has a coherent stream emanating from its Lagrange points out into the tidal field of the Milky Way. This publication also marked the onset of survey since in astronomy. Since 2000/2001 the rate of papers presenting new observational results on streams in the Galactic halo or around other galaxies grows exponentially. While this growth was initially driven by studies on the Sgr stream (red cumulative curve above), the focus is now shifting towards fainter streams like Palomar 5, NGC 5466 or GD-1 (blue curve). It’s a really exciting time to work in this field!

Gaia Challenge 2014


End of October, I was in Heidelberg to attend the Gaia Challenge workshop at the Max Planck Institute for Astronomy (MPIA). This workshop series was initiated last year by Justin Read, Mark Gieles and Daisuke Kawata at the University of Surrey (Guilford, south of London). This year we came together again, but this time the meeting was organized by Glenn van de Ven, who is a research-group leader at the MPIA. Glenn booked the Haus der Astronomie (house of astronomy) for us, which is a public outreach building on the premises of MPIA. The cool thing about it is that it’s shaped like a spiral galaxy (see photo above). The bulge of the Haus der Astronomie is a large auditorium and the spiral arms consist of seminar rooms and a child day care center. It’s quite a unique place!

About 80 participants came together to work on modeling problems related to the Gaia mission. We formed five groups with significant overlap between each other:

  1. spherical & triaxial systems (like dwarf galaxies),
  2. disks (meaning mainly the Galactic disk),
  3. streams & halo stars (everything which is not in the Galactic bulge or the disk),
  4. collisional systems (i.e. star clusters),
  5. astrophysical parameters (covering everything non-dynamical Gaia is going to measure).

I had the fun role of being the coordinator of the streams & halo stars group. Together with Andreaa Font, I organized our daily meetings, lead the discussions in the group, focussed our modeling efforts, and summarized our progress. Our mission is to identify streams (i.e. groups of stars that are coherent in phase space) in survey data like Gaia’s first data release (expected early 2017) and then find fitting dynamical models for it. Sounds simple, but it’s absolutely not. With still 2.5 years to go until Gaia data release 1, we are still struggling with the most basic problems. But the workshop gives the stream & halo stars community a great forum to come together and figure out how to address these issues.

The concept of the workshop is to pose challenges in form of numerical simulations (which can be downloaded from the Gaia Challenge wiki). These we take as idealized survey data. The modelers apply their respective methods to this idealized data and try to recover the underlying (known) model parameters of the simulation. For the streams group this means that we make a numerical simulation of a dissolving Galactic satellite, like Palomar 5 or the Sagittarius dwarf galaxy, and let it form a stream. The modelers then get the stream stars’ positions and velocities and have to figure out what the parameters of the Galactic model are, which was used in the simulation. A first attempt of collectively solving one specific challenge is the Palomar 5 challenge that I posted. Several people are trying to model it right now, and we are collectively writing a paper about the results using github. It’s all open science, and everybody is invited to join!

Juggling with black holes


Last Friday I gave a public lecture on gravitational dynamics at the Columbia Astronomy Department Stargazing & Lecture Series. The talk was really well attended, and I was surprised to see so many people interested in hearing about astronomy on a Friday night. The big auditorium in Pupin Hall felt packed to me (my perception may have been biased, though).

The title of my talk was Juggling with Black Holes, which is probably why so many people came in the first place. Who doesn’t like black holes?! Especially now that Interstellar is in theaters! I mainly focussed on explaining what a black hole is, and how black holes interact with stars and other black holes – causing slingshots, mergers and runaways. With a few simple animations, I wanted to show the beauty of pure gravitational dynamics.

Afterwards I was confronted with tons of questions, some of which were challenging, others were just outspokenly amazing. Best one: “Can we use black holes to generate energy?” This question came from a super bright high school student in a separate session after the main lecture. The separate meeting was organized by Project Rousseau – a non-profit organization, which helps underprivileged students to develop their full potential.

I’m not kidding when I say that this was the most rewarding experience of my career. I had no idea how beneficial public outreach can be. If you have the chance to reach out, do it!

Signs of hidden matter

Comparison of different observational data (rows) to three different cluster models (columns) of the globular cluster NGC 6624. The observed accelerations of a low-mass X-ray binary and of three pulsars (bottom row) tell us that the density of matter (top row) in the center (small radius R) has to be higher than what we see in the form of stars.

When we look into the night sky we are fooled by the bright and sparkling stars, and we think they make up most of the matter in the Universe. Among astronomers that’s commonly believed to not be true, since observations tell us that a significantly larger amount of mass fills the Universe in the form of dark matter. What dark matter is, and whether it exists at all, is difficult to answer and no one has found a definite answer yet. However, there are other forms of matter that hide from our naked eyes, or even from our large armada of telescopes. Planets, brown dwarfs, low-mass stars, neutron stars, white dwarfs and black holes are all emitting barely any electromagnetic radiation (a.k.a. light). Thus, only in very few, and nearby cases we can actually observe these objects. For most parts of the Universe, this hidden matter remains unseen and we have to add a certain amount of dark mass to the mass we see in stars based on the best of our knowledge.

It’s important to note at this point, though, that this hidden matter is by far not enough to explain the mass discrepancy in the Universe that led us to the hypothesis of dark matter. However, in certain parts of the Universe, hidden matter in the form of black holes, neutron stars and white dwarfs can make up a large fraction of the local matter density. One of these places is the Milky Way globular cluster NGC 6624. According to a paper I published recently with Miklos Peuten, Michael Brockamp and Pavel Kroupa, the center of this cluster should be filled to the brim with these dark remnants, as we call them. One third of all matter in this star cluster is hidden from our deepest observations, and dwells in the cluster’s center. There it causes such a strong gravitational acceleration on the neighboring stars, and especially on a low-mass X-ray binary called 4U 1820-30, that we can measure how these objects got accelerated over the past 20 years.

Space telescopes and radio antennae have monitored 4 special objects (the above mentioned X-ray binary and 3 pulsars) in the cluster over decades, and they tell us what the gravitational field in the cluster is like. By comparing this inferred gravitational field with the mass distribution in form of the stars, we can deduce that we’re seeing only 2/3 of the mass of the star cluster. Our combination of observations and modeling of this cluster allows us to exclude that this hidden matter is in the form of a huge black hole with several thousand times the mass of the Sun – although this was the kind of hidden object we were originally looking for. Our new interpretation of the observational material is no less spectacular, as dark remnants are usually not believed to be dynamically important in star clusters. We show that in fact they are, and that in some parts they can even entirely dominate the dynamics of the stars we see!