In the decade the consortium has been working with SGI, the nature of its investigators' research has changed completely. "The questions we're trying to answer now are much more precisely defined," explains Dr Paul Shellard from the Department of Applied Mathematics and Theoretical Physics (DAMTP) at Cambridge University.

"Ten years ago we were working with early universe theories, but cosmology has since become a much more quantitative science. We're having to make theoretical predictions of comparable quality and precision to the observations provided by the latest satellites and telescopes; and the observations themselves require much more computing power to analyse. That's why computers are so important. You can't analyse the observations without folding in your theory to see if it matches what you're seeing; and the observations themselves are huge datasets that need to be analysed to test the likelihood your theory is correct."

The global cosmology community is a highly competitive environment, where investigators are looking to solve problems quickly in a rapidly moving field. "Our large set of investigators have unique interdisciplinary strengths with which to tackle the big set of problems we are trying to solve," continues Paul Shellard. "These range from predicting the creation of the early universe, and running field theory simulations a fraction of a nanosecond after the Big Bang, to people investigating the cosmic microwave background (CMB) and trying to find correlations in the distribution of galaxies.

"Dark energy is an extremely large and complex problem that has globalised cosmology. We're all striving to solve this fundamental problem, we need big resources to do so, and whoever gets those resources will win," continues Bob Nichol.

"We've known for a very long time that only a very small part of the universe is made up of 'normal' matter, and since the 1930s we've also known about dark matter - so called because it doesn't shine. The only way it interacts with us is through its force of gravity, and we know it must be there because if it wasn't, our galaxy would literally spin apart.

"From our calculations of the universe's energy budget, around 30% is made up of matter (dark and normal), and the revolution has been our realisation that the remainder is in a pure form of energy associated with 'empty' space. This dark energy isn't associated with matter, but is affecting the expansion of the universe, and that's why we know it's there.

"The conundrum is, why is it there and why does it make up 70% of the universe? And the problem is that if you ask a theoretical physicist how much dark energy they'd predict, their answer is 10¹²⁰ larger than the amount we've measured! So the cosmological community is now trying to understand some of the fundamental properties of dark energy to see why this is the case. To answer that we have projects such as the Dark Energy Survey which will enable us to image the sky to much greater distances than ever before.

"To bring this back to COSMOS, the irony of the Dark Energy Survey is that it can only see one universe. But according to the concept of cosmic variance, the universe we see is just one possible outcome of the random processes that led to its creation. So what cosmologists are constantly fighting is that we only have one universe to look at, but if the random processes had turned out differently, our universe would look different.

"With the CMB, for example, as photons work their way across the universe, their signal gets degraded by contaminants along the line of sight. There are a lot of galaxies between us and the other side of the universe, so for the Planck experiment there's a contaminating signal from galaxies that are undergoing massive bursts of star formation, and have a lot of dust which changes the photons' energy.

"Planck would really like to get rid of all of those contaminants so it can get a pristine picture of what's happening in the earliest point in the universe. So understanding dust physics within galaxies is absolutely critical if you want to clean the signal to understand what's happening. Our radiative transfer codes do that, and we'll be using COSMOS to undertake the high resolution parameter studies needed to achieve this.

"In dark energy meanwhile, a lot of work is based around very high resolution, large scale N-body simulations, which focus solely on gravity. To enable these to be linked to the baryonic world, we're combining gravitational physics, the internal physics of star formation within galaxies, and the chemical evolution history of gas and stars, within these simulations. This allows us to predict the amount of radiation in different wavebands, and tune this to match the waveband of a particular experiment. In the case of the Dark Energy Survey, for example, there are a series of filters that will be used to probe different populations, so we can convolve our simulations with those filters to give a more self-consistent picture of what the survey will actually observe, and assist understanding of selection effects within the simulations themselves.

"Our third main area is galactic archaeology. Rather than looking at statistical samples within a simulation, we're trying to understand the internal physics of our own galaxy in unprecedented detail. The UK has a major role in RAVE (the Radial Velocity Experiment) in which we're measuring the kinematics, velocities and chemical fingerprints of over a million stars over the next five years.

"We have one of the world's most sophisticated chemical evolution codes that allows us to track the histories of hundreds of elements and isotopes using any range of star formation and assembly histories, and we'll be using COSMOS to make as clean and sophisticated a model of the Milky Way as we can. This will give us a unique forensic history of our galaxy – and also tell us a lot about how galaxies as a whole were formed."

"At the time we were discussing the latest upgrade, the COSMOS consortium were not due to receive any significant public funding for anything up to a year. We therefore looked at how we could turn this to our advantage and find an intermediate solution," explains SGI UK Sciences Manager, Simon Appleby. "SGI provided evidence that in terms of power costs, floor space and performance, the Altix 4700 was considerably more efficient, in all areas. This allowed a very strong, compelling business case to be presented not just to the DAMTP at Cambridge, but to all the UK COSMOS consortium members.

"The result was that within three weeks, SGI were able to upgrade the COSMOS machine to a uniquely scalable, highly cost/performance efficient solution, based on the very latest dual-core Intel® Itanium® 2 processors. The Altix 4700 has also subsequently been expanded with extra CPUs and memory at the request of further COSMOS members. This will put the whole of the UK COSMOS consortium in an even better position, not only in terms of quicker time to discovery, but in allowing them to secure additional funding both now and in the future."

FURTHER INFORMATION
www.sgi.com/altix