GalactICS for non-Astronomers

Recently I published a paper on a code that myself and my collaborators have built called GalactICS (which stands for Galaxy Initial ConditionS).  And, while anyone can read the paper itself on arXiv, it’s not particularly friendly to non-astronomers (mainly because it’s written in science-ese). So I decided to write up a little post talking about the paper itself and what the code does.

Before going into the details of the paper and the code, it’s worth giving a bit of background information.  Much of my research focuses on ’N-Body’ simulations.  N-Body simulations are built on the ‘spherical cow approximation’.  The idea is to take something very complicated, like a cow, and treating it like something relatively simple, like a sphere.  The reason for making such a simplification is that, if you need to consider how millions of cows behave, it may be that the individual characteristics of each cow don’t matter too much  and a ‘spherical cow’ may be good enough to model the entire population.

N-Body simulations take a large number of particles and evolve them using some straightforward physical laws.  For instance, in a simple cosmological simulation, the entire Universe is represented by millions of particles that then move under gravity.  A more complicated simulation might separate out the different types of matter, like stars and gas and dark matter.  And even more complicated ones will try to deal with supernovae or star formation or supermassive blackholes.  It is still quite complicated, but it’s always worth remembering that, in these types of simulations, a single particle might represent millions of stars.

My current research has been focused around uses for ‘tailored galaxy simulations’.  Rather than running a cosmological simulation of the entire Universe, I’ve been simulating individual or pairs of galaxies.  The idea behind these sorts of simulations is to start off by generating some galaxy (or galaxies) with a set of very specific properties.  These properties include things like the amount of dark matter, the level of random motions in the disk, or, for pairs, the starting point and orientations of the two galaxies.  Then, you let the system evolve and see what happens.  This is a little bit different than a cosmological simulations.  In cosmological simulations there are thousands of galaxies to choose from, and they may end up being more realistic because the way they are formed is closer to how nature does it…but there is no guarantee that there will be a galaxy with the properties that I’m looking for in it.

Tailored simulations are useful for a variety of different science projects.  They can be used to try to model real galaxies.  By comparing a bunch of test galaxy simulations to observations, the very best fitting one can be found.  And then, using that simulation, we can figure out the dark matter content of the galaxy, how it’s evolving, and what dynamical processes are occurring.  Tailored simulations can also be used to see what are the dominant causes of interesting phenomena.  For instance, a lot of spiral galaxies (around 50% or more) contain bars.  It turns out that the ratio of dark matter to disk matter can cause a bar to form out of an initially circular disk.  Of course, an interaction with another galaxy can also drive bar formation.  So, a question that tailored simulations can explore is which of these two mechanisms is responsible for making most of the bars in the Universe (other than people enjoying beer).  These are just two quick examples, but there are many other uses for tailored simulations.

And this brings me to GalactICS and my recent paper.  You see, there are three main parts to a simulation; making the initial conditions, evolving the system, and analyzing the output.  GalactICS is a tool to make those initial conditions.  It turns out that making a galaxy is a difficult process.  If you just throw down a bunch of particles randomly in space and simulate them, the system will either collapse under gravity or explode apart if the the velocities aren’t quite right.  Since galaxies don’t do either of those things, that approach is probably not the best one.  Instead, GalactICS carefully builds up each component of a galaxy; the bulge, the disk(s), the dark matter halo, and, for the first  time in the code, a gas disk.  It’s set up so that everything is initially balanced and in equilibrium.  Then it will evolve in a ‘normal’ way and the results will be scientifically useful.

Now GalactICS is not a new code.  It goes back to a paper published in 1995.  But it’s been improved a lot since then.  One of the key players in that improvement is my Ph. D. supervisor, Prof. Widrow.  Which is why it’s no surprise that he is one of the people I worked closely with for making this new version of the code.  We ended up with two big improvements over the old versions.  Firstly, GalactICS can now make galaxies with ‘thin’ and ‘thick’ disks.  That is, rather than our spiral galaxies just having a single disk, there can be two of them, with different properties stuck together.  This actually matches what we see in the Milky Way and other nearby galaxies.  Secondly, and, in my opinion, much more importantly, the new version of GalactICS can make gas disk components.

Gas disks are present in all spiral galaxies.   They contain the material that will eventually form into stars and are a very important part of a galaxy.  But, most tools for making tailored simulations don’t include gas disks.  As difficult as it is to build a spiral galaxy that is in equilibrium, it is an order of magnitude more difficult to build one with a gas disk.  The reason why it’s so difficult goes back to the spherical cow.  In terms of the initial conditions, stellar components, like the disk and bulge, as well as the dark matter halo, only really care about gravity.  So, when you build your object you only need to figure out how to balance the motions of the various particles against the gravitational force.  But gas is different.  Gas cares about properties like pressure and temperature and physical forces that don’t really affect the other components.  It makes them much more complicated, and very difficult to build.

But, we’ve found a way to do this!  A group of researchers released a paper in 2010 on how to build an N-body gas disk in an ‘analytic’ potential (that means that they use a mathematical function to deal with non-gas components instead of generating a bunch of particles).  We were able to adapt their approach and merge it with how GalactICS builds the other galaxy components.  I won’t bore you with the details on how this works, but we were very successful.  We reproduced some of their test models and then we moved on to one of my favorite galaxies to simulate; the Milky Way.  Our final model actually looks quite close to our galaxy.  It evolves a bar like the one we observe, about the same number of spiral arms, and matches many other observations.  I’m very happy with how this all turned out.

So, that’s my more recent paper.  My collaborators and I developed a new method for building galaxies that can be used in tailored simulations.  Since then, we’ve used these model galaxies to look at our home, to model other galaxies, to test observational algorithms, put them in VR, and are throwing pairs of them together to see what happens.  It’s really quite exciting and a very fun toy to play with.  And, if you want, you can grab a copy of it to play with just over here.  Have fun!