October General Meeting Report
Speaker: Dr Chris Fluke, Swinburne University of Technology
Topic: Computational Cosmology
We are used to living in an ordered, mechanical Universe; everything in its place and a place for everything. In such a world every effect has a cause and for every action there is an equal and opposite reaction. If something moves or if it evolves it can be modelled and understood. In that respect nothing has changed much over the last 5,000 years. We theorise, we hypothesise and make our understanding fit reality. For thousands of years though the world was understood to be static, everlasting from eternity to eternity. That’s the way it seemed and that’s the way it made sense. The few anomalous Wanderers across the sky could be explained by divine attributes or with clockwork mechanisms and only served to confirm the eternal nature of the given background of Earth, Heaven and you-know-what below. Whatever did not fit into the four elements: Earth, Water, Fire and Air became quintessence, the stuff beyond our five senses, or was attributed to the supernatural. Advances in technology changed all that. As measurements became more and more refined, theories and hypotheses could be tested and proven or disproved. Rocks were no longer eternal, stars and galaxies moved and space became a malleable commodity.
When Copernicus in 1543 first tried to promote his hypothesis of the solar system with his six books “On the Revolutions of the Heavenly Spheres” it was not an instant hit. While his formulation of the sun at the centre of the universe is today considered one of the most important scientific hypotheses in history, he was not the first to do so. The system Copernicus proposed was experimentally no better than Ptolemy's proven Earth centred model detailed in his Almagest (The Great Book, or The Great Treatise) 1500 years earlier, in 148AD. To match the accuracy of existing data Copernicus was forced to introduce even more Epicycles into the planetary orbits than the Greek had used. He was acutely aware that he could not present any observational "proof" in his manuscript, and relied instead on arguments about what would be a more complete and elegant system. Few astronomers were convinced though by the theoretical Copernican system, not until Galileo and Johannes Kepler championed it and improved it with new observational data in the early 17th century. It was only then that it came to mark the starting point of modern astronomy and of modern science, encouraging young astronomers, scientists and scholars to take a more sceptical attitude toward established dogma.
Dr Chris Fluke is one of those young scholars. An astronomer with the Centre for Astrophysics & Supercomputing at Swinburne University of Technology he looks at the distribution of matter in the Universe by modelling the effect it has on radiation, and uses advanced visualisation techniques on astronomical datasets in order to provide new insights and understanding of the cosmos. Faced with the embarrassing fact that the visible stuff we thought made up all of our universe accounts for at most 4% of cosmic gravity, it becomes vital to develop models that help predict how those visible four percent are interacting with themselves and with the invisible 96% background. It is a big job whichever way you look at it. But astronomers are used to big numbers. They regularly deal in millions and billions. The visible universe is now 13.7 billion years old and contains some 10 billion galaxies like our Milky Way Galaxy, which is itself made up of some 200 billion stars. Did it really all start with a Big Bang? You have to think big for working with such numbers; big computers that is. Demonstrating computer simulations of gravitational interaction between multiple stars and many-body star-systems Chris was able to model exotic galaxy configurations, like the NGC 250 or Antennae Galaxies, forming on the screen as the result of a collision between two individual spiral galaxies. One obvious advantage of such computer simulations is that a process that would normally take billions of years can be sped up to happen within minutes. The other is, observational astronomy has to be content with the sky as it is, a sample of one, whereas computational astronomy can have many versions, make predictions and test the results and repeat it with different parameters at will. A surprising result of several such test runs is that such collisions invariably seem to destroy the galaxies’ spiral structure creating mechanism, leaving behind a randomly oriented, gravitationally bound collection of stars. This would suggest that most (if not all) elliptical or irregular galaxies are the result of collisions between spiral galaxies. On the downside is the huge computing (number crunching) power needed for meaningful, realistic results. In practice a compromise has to be made between the number of points and individual interactions used in the computations, and a practical time frame. A Teraflops computer takes about 24 hours for just one step in a one million-body computation. The computational capacity of computers is measured in Floating Point Operations Per Second (FLOPS). This has been steadily on the increase from year to year. The original Supercomputer, the Cray-1, in 1976 was capable of 80 Megaflops. In fewer than 30 years since then, the computational speed of supercomputers has jumped a million fold, and the race is on to create a Petaflops (1015) machine by the year 2007. Depending on the outcomes of simulations on current Teraflops computers, solving big number N-body computations, computational cosmology may eventually be able to decide for us which Big hypothesis will accompany the Big Bang and predict the Big End. Is it the Big Crunch, where everything finishes up in a virtual Black Hole? The Big Chill where expansion continues and nuclear reactions virtually cease? The Big Rip (or the Big Split) where Einstein’s Cosmological Constant is not constant but increases gradually to infinity to tear apart every galaxy, every star and finally every atom until space is virtually empty? Or is it something else Big altogether? Computational cosmology displays can provide astronomers with new ways of exploring the terabyte and petabyte datasets that are now regularly being produced from all-sky surveys, high-resolution computer simulations, and Virtual Observatory projects. It is in our human nature to want to know the beginning and the end fate of our environment, even if the first was 13 billion years in the past and the second could be 55 billion years in the future.
Computers as a tool for science have come a long way. The name that was once used to define a group of people computing mathematical solutions with pen and paper and slide-rule, is today an ubiquitous tool in almost every aspect of human life. Sometimes though, when everything seems to go wrong, don’t you secretly wish you could fall back on that simple philosophy left over from an age when, if reason failed to produce a logical explanation, you could always invoke the supernatural?
The vote of thanks was given by Lockie Cresswell with the traditional N body object in a brown paper bag.