Monday, 5 May 2014

The Proto-Milky Way during its First Billion Years

Here is a neat animation showing the accretion of gas onto the progenitor of a Milky Way like galaxy when the Universe was barely 10% of its current age. Note how anisotropic the gas flow -- colour coded according to its projected density -- onto the proto-galaxy is. The stellar component of the galaxy  is nursing the after-effects of a violent merger with another galaxy, with the cores of the two galaxies in an ongoing tussle as they mutually shred each other in their combined gravity. There has also been an enormous burst of star formation, which is driving very hot gas out of the galaxy and into its larger-scale environment. Pretty neat stuff…!

The stellar component of the proto-MW - reeling from the impact of a violent merger with another galaxy.

Thursday, 1 May 2014

Satellites as Probes of Cosmology

One of the topics my collaborators and I are particularly interested in is understanding what satellite galaxies can tell us about galaxy formation and cosmology. I'm lucky enough to work with some excellent people who are experts in real observations of the satellite populations of the Milky Way and Andromeda galaxies, and the idea is that we can use simulations of galaxies and make synthetic observations of their satellites to help us make sense of the real observations, i.e. what are they really telling us about how the Universe works. 

If you look at this little movie showing the spatial distribution of stars in and around an isolated Milky Way type galaxy, you can see distinct clumps -- satellites -- as well as thread-like streams looping over the galaxy, the tell-tale signature that the galaxy is cannibalising its population of satellites. 

This process of galactic cannibalism is well know -- the figure on the right is an artist's impression of what is happening, and some spectacular evidence has been found in the Sloan Digital Sky Survey's Field of Streams. However, it's very nice to see it occurring in one of my cosmological galaxy formation simulations, which are of moderate resolution. This suggests that we can study ensembles of galaxies -- spanning a range of masses, environments and assembly histories in different dark matter models and for different galaxy formation prescriptions -- and explore parameter space in a much less computationally expensive way than I had initially anticipated. My collaborators here at UWA, over at U Sydney and at UAM in Spain have a big grant to study this problem over the next 3 years -- and so if you want to know what we are learning… watch this space!

Wednesday, 30 April 2014

Baby Galaxies in the Early Universe

Over the last few weeks I've been working quite hard to get together a suite of galaxies formed in high resolution cosmological simulations, so that I can study the nitty-gritty of how gas gets into galaxies from the cosmic web, and how energetic feedback from supernovae (violent explosions as massive stars die) and growing super-massive black holes affects this picture. I really should have been focussing on finishing a few papers and preparing lecture notes for an advanced undergraduate course I've got to deliver in a couple of weeks -- but this was just too much fun. Here are some cool images…

This is an image centred on the galaxy at about half the age of the Universe, on a spatial scale of 100,000 light years. The galaxy will eventually end up with properties similar to our own Milky Way. Individual particles represent gas (coloured hues of red) and stars (coloured white) -- these are the particles that are evolved by a sophisticated bit of software called GADGET, run on a powerful supercomputer.

In this image I've ignored all of the stars and shown only the gas in the galaxy - note how it's concentrated in a thin rotationally supported disc, with fairly pronounced spiral features. There's also a hole in the centre of the galaxy - that's because the gas temperature and density in this region usually satisfies the conditions for it to become star forming...

 … while in this image I'm ignoring the gas and showing only the stars. The stars are distributed in a vertically extended disc with a central spheroid. If you squint carefully, you can see that some of the small clumps of stars surrounding the main galaxy -- satellites -- are being tidally disrupted, showing the characteristic "S-shape" as stars are stripped off in the gravitational field of the galaxy.

Finally, this is what the underlying dark matter halo looks like. Pretty featureless! Currently favoured theories of galaxy formation assume that galaxies form within scaffolding provided by some form of dark matter, whose precise properties remain elusive. 

One of the fun things you can do with simulations is look in detail at what happens over the history of the galaxy and make little movies. Here is what happens to the proto-Milky Way galaxy in the very early Universe -- during the first billion years after the Big Bang… It's a bit jerky -- I don't have the patience to make it pretty -- but it's fascinating to see what happens to the newly formed disc as it experiences close interactions with some lower-mass galaxies.

Now.. back to finishing those papers...

Wednesday, 12 March 2014

Dark Matters

Our standard model for dark matter, Cold Dark Matter,
predicts that dark matter haloes of all masses - clusters,
galaxies and dwarfs - should be swarming with
substructures, remnants of the merging hierarchy.
I spent yesterday at the Australian Astronomical Observatory in Sydney, at what is quaintly called a Town Hall meeting, to build community support for a very cool new galaxy survey known as WAVES, which might be carried out on the planned 4MOST instrument on the ESO VISTA telescope. In essence, this was a get-together of various members of the great and good of Australian astronomy research to discuss how great 4MOST will be for their research programmes - ranging from AGN to cold gas in galaxies to surveys to constrain the dark sector of dark matter and dark energy. I'm part of the design team at ICRAR/UWA - alongside my colleagues Simon, who has made the WAVES dream a real possibility, Aaron and Martin. I attended the meeting to talk about my particular interest in testing the nature of dark matter - which is one of the central objectives of the survey.

Observational evidence for some form of non-baryonic (i.e. exotic) dark matter is now well established; what is less well established (i.e. it's not) is its physical nature, by which I mean its identification with one of a myriad of fundamental particles. The model for dark matter that is most widely favoured by the community is cold dark matter (CDM); cold because the CDM particle wasn't in thermal equilibrium when it decoupled in the first few minutes after the Big Bang and it moved at non-relativistic speeds. The CDM model has been kicking around for close to three decades and it's had its ups and downs. Pretty much everything we have learned about galaxy formation (from a theoretical perspective) has been within the framework provided by the CDM model. CDM makes a couple of key predictions -- first, that the massive virtualised structures that are the end-point of gravitational collapse, so-called dark matter haloes, have divergent central dark matter densities; and second, the CDM haloes contain a wealth of small-scale structure, remnants of their hierarchical assembly histories. 

How we properly test observationally whether or not the CDM is an accurate model of the Universe has bugged me for some time, and in a paper I published last year I looked into how we  might go about doing this. I argued that, although CDM haloes have divergent central densities, this is not a good test because astrophysical processes can erase these dark matter cusps. Rather it's better to go after dark matter substructure, the prediction that is the signature of CDM. I ran a series of supercomputer simulations of model universes, comparing a fiducial CDM universe with warmer versions of dark matter -- WDM models. WDM models suppress the formation of low-mass dark matter haloes because, in the jargon, they have reduced power on small-scales. This means that WDM free-streams in the early Universe because the WDM particle was in thermal equilibrium when it decoupled, erasing small-scale matter density perturbations. The bottom lines is that reduced small-scale power in the early Universe broadly translates into fewer low-mass dark matter haloes (and substructures) in the present-day Universe. 

How can this be tested? I argued in the paper - and during my talk - that looking at clustering of low-mass galaxies, minor merger rates and evidence for disturbances might be a good way to test the veracity of the CDM model. These kinds of statistics may not be so sensitive to our understanding of how low-mass galaxies form (another challenging problem) and they can be robustly measured in observational data. I also suggested that the signal probably requires a new approach to analysing and comparing simulations with observations - rather than running a simulation and a best-bet galaxy formation model, making mock catalogues and comparing, we will need to apply advanced statistical techniques to isolate regions of parameter space in the models that are allowed by the data. Fortunately, we have just hired a very clever postdoc to work on precisely this problem… watch this space!

Sunday, 24 July 2011

Work-Load, or is it Work-Life, Imbalance?

It's funny how, when preoccupied and lost in thought, you can easily mis-hear. I have been a bit troubled by how inefficient the code I use to run simulations -- GADGET2 -- has been performing, and some simple analysis showed that it simply does not balance its work-load well. GADGET2 is a parallel code, which means it splits problems across different processors on a supercomputer to solve the problems more quickly. Ideally the way the problem is split should be optimised, so that each processor does roughly the same amount of work. If it's not, then some processors will do more work than others, and the time it takes to solve the problem will increase accordingly. This is not a good thing. In this case the jargon is that the code suffers from work-load imbalances.

One of my work-load imbalanced galaxy clusters...
I was so pre-occupied with this problem and my seeming powerlessness to resolve it that I was completely thrown by a colleague thanking me for my document on work-load balance when I went to make a cup of tea this morning. Work-load balance? How could he have known? I had been writing e-mails about work-load imbalances, but not to him. Or had I? Momentary confusion followed by a clarification and it all made sense. We had discussed work-life balance in astronomy a few weeks ago and I'd passed on a document on the topic that was based on a meeting I'd attended a couple of years ago. Surely this was an instance where work-life balance had been lost? Was I really keeping my work in perspective? The thing is, I'm not sure I know how to work any other way -- becoming completely absorbed in a problem such that it gobbles up my waking moments. This is not a good thing.  

So... how do I improve my work-load, sorry, work-life imbalance? Maybe I need to turn some of the brain power I expend on my work on solving my work-life imbalance conundrum...