George Lake is professor for astrophysics at the University of Zurich (UZH). At his visit to the hpc-ch booth at SC11 we interviewed him about his research activities. George told us that at UZH there are different research groups working on dark matter, large scale structure of the universe and galaxy and clusters of galaxies. Most of the matter in the universe is dark matter, a mysterious component we still do not understand but it drives the process of galaxies formation and evolution. The researchers at UZH recently discovered that dark matter is clustered in little lumps that are present even in the disk of our galaxy. Currently they are studying both novel observational approaches to detect these lumps and their theoretical consequences on the process of galaxy and galaxy cluster formation.
For many years George has been involved in supercomputing . He was first one of the leading scientists for a large NASA program, HPCC/ESS (High Performance Computing and Communications for Earth and Space Sciences), then he was chief scientist at the ARSC supercomputing center. The NASA was the one where Tom Sterling initiated the Beowulf cluster project. Originally the Beowulf was a gigaflop workstation. The idea was to create a personal supercomputer setup on your desk. George was the first to say “workstations heck, I’m building a server”.
George also boosted the research and supercomputing environment in Switzerland. The general envornment in Switzerland has very balanced investments. We have a national supercomputing centre in Manno (Ticino) and the university has an own cluster. There are good investments in research groups to develop next generation software to use next generation systems (HP2C initiative). It is a confluence of things that the US is tried to achieve for many years and the Swiss have accomplished.
The molecular scientist Jürg Hutter, Professor at the University of Zurich, began developing CP2K about ten years ago in collaboration with the research group led by Michele Parrinello, Professor of Computational Science at ETH Zurich and the Università della Svizzera italiana.
The CP2K program now consists of nearly a million lines of code and is continuously being developed by international teams. It uses many different algorithms and complex numerical structures. It is hard to target these to current computer architectures, so in the last two years Hutter and his team have been focussing on optimising the program for use on the still new computer architectures that are based on multicore processors, where there are several processors on one chip, or graphic processors (GPUs).
A structured two-dimensional surface. These so-called nanomeshes can be used as a type of template for developing new materials.
CSCS dedicates a longer article to CP2K with an interview with Jürg Hutter and Joost VandeVondele, senior research assistant in Hutter´s team. Hutter and VandeVondele explain the many-sided applications of CP2K in science and industry like, the exploration of new energy sources as the solar cells known as Grätzel cells.
Lucio Mayer and his team simulated for the first time worldwide the birth of the Milky Way. The results stimulated many discussions in online forums. Lucio answers our questions.
Lucio, this is the first simulation that allows reproducing the birth and evolution of a Milky Way-like galaxy. What are the main difficulties and challenges in achieving such results? Why couldn´t we get to this point before?
How can we be confident that this is the way things actually took place? Can we compare to real objects that we observe in the sky?
You spoke about the key-role of Dark Matter in the simulation. What is Dark Matter? Which are the evidences that it exists?
Many viewers complained about the low resolution of the movie. Wouldn´t it have been possible to produce a HD movie?
The costs of this simulation, in terms of computing systems, power, and work is very high. What´s the return of this effort and its impact on society?
What are the main challenges in astrophysics currently studied with the help of the supercomputers?
What are the perspectives for this kind of research? What´s still missing in your model? How will you improve it in the future?
You stopped your simulation after 13 billion of years. What will happen if you would continue? What will happen in the end?
hpc-ch interviewed Lucio Mayer, Professor at the Unversity of Zurich about simulating the birth of a galaxy. For this work his research group used the HPC resources at the Swiss National Supercomputing Centre (CSCS) in Manno and at NASA.
For almost 20 years astrophysicists have been trying to recreate the formation of spiral galaxies such as our Milky Way realistically. Now astrophysicists from the University of Zurich present the world´s first realistic simulation of the formation of our home galaxy together with astronomers from the University of California at Santa Cruz. The new results were partly calculated on the computer of the Swiss National Supercomputing Center (CSCS) and show, for instance, that there has to be gas on the outer edge of the Milky Way.
The aim of astrophysical simulations is to model reality in due consideration of the physical laws and processes. Astronomical sky observations and astrophysical simulations have to match up exactly. Being able to simulate a complex system like the formation of the Milky Way realistically is the ultimate proof that the underlying theories of astrophysics are correct. All previous attempts to recreate the formation of spiral galaxies like the Milky Way faltered on one of two points: Either the simulated spiral galaxies displayed too many stars at the center or the overall stellar mass was several times too big. A research group jointly run by Lucio Mayer, an astrophysicist at the University of Zurich, and Piero Madau, an astronomer at University of California at Santa Cruz, is now publishing the first realistic simulation of the formation of the Milky Way in the Astrophysical Journal. Javiera Guedes and Simone Callegari, who are PhD students at Santa Cruz and the University of Zurich respectively, performed the simulation and analyzed the data. Guedes will be working on the formation of galaxies as a postdoc in Zurich from the fall.
Removing standard matter central to formation of spiral galaxies
For their study, the scientists developed a highly complex simulation in which a spiral galaxy similar to the Milky Way develops by itself without further intervention. Named after Eris, the Greek goddess of strife and discord, because of the decades of debate surrounding the formation of spiral galaxies, the simulation offers a glimpse in time lapse into almost the entire genesis of a spiral galaxy. Its origins date back to less than a million years after the Big Bang. “Our result shows that a realistic spiral galaxy can be formed based on the basic principles of the cold dark matter paradigm and the physical laws of gravity, fluid dynamics and radiation physics,” explains Mayer.
The simulation also shows that in an entity that is supposed to develop into a spiral galaxy, the stars
form in the areas where giant cloud complexes are. In these cold giant molecular clouds, the gas exhibits extremely high densities. The star formation and distribution there does not occur uniformly, but rather in clumps and clusters. This in turn results in a considerably greater build-up of heat through local supernova explosions. Through this massive build-up of heat, visible standard matter is removed at high redshift. This prevents the formation of a bulge in the center of the galaxy. The removal of baryonic matter, as the visible standard matter is also known, also reduces the overall mass of the gas present at the center. This results in the formation of the correct stellar mass, as can be observed in the Milky Way. At the end of the simulation, a thin, curved disk results that corresponds fully to the astronomical observations of the Milky Way in terms of the mass, angular momentum and rotation velocity ratios.
On the left the simulated galaxy after 13 billion of years. The gas clouds are represented in red, the stars in blue. On the right as comparison the spiral galaxy M74 that is similar to the Milky Way (University of Zurich and NASA)
Astronomical computing power
For the calculations, the model Mayer and co. developed for the simulation of disk-shaped dwarf galaxies and published in the journal Nature in 2010 was refined. The high-resolution model simulates the formation of a galaxy with 790 billion solar masses and comprises 18.6 million particles, from which gas, dark matter and stars form. The high resolution of the numerical simulations is essential for the groundbreaking new results. For the calculations, the high-performance supercomputers Cray XT5 “Monte Rosa” at ETH Zurich´s Swiss National Supercomputing Center (CSCS) and the NASA Advanced Supercomputer Division´s Pleiades were used. A regular PC would have needed 570 years for the calculations.
Stars and gas at the outermost edge of the galaxy, hot gas at its center
The new simulation confirms the results for the formation of disk-shaped dwarf galaxies published by Mayer and demonstrates that the model – unlike all previous approaches – can recreate both small and extremely large galaxies realistically. Moreover, from the simulation it an also be deduced that protogalaxies with a large disk made of gas and stars at the center already formed a billion years after the Big Bang, and therefore long before our present galaxies.
Based on the simulation, the ratio of “cold dark matter” (CDM) and standard matter in spiral galaxies can also be adjusted. In order to obtain the correct overall stellar mass in the final stage of the galaxy – until now, one of the main difficulties – it is imperative that standard matter be removed from the center by supernova winds. On the strength of the simulation, it is highly probable that the ratio of standard matter to CDM on the outermost edge of the CDM rings of a spiral galaxy is 1:9, not 1:6 as previously assumed.
The simulation also predicts stars and gas for the outer halo of the Milky Way six hundred thousand light years away. Only the next generation of space probes and telescopes will be able to detect these extremely faint stars. Furthermore, the simulation makes predictions with regard to the radial distribution of hot gas around the galaxy´s central disk. Future telescopes that can measure X-rays, as the IXO Mission of the European Space Agency (ESA) is planning, for example, will test these predictions.
Further reading
Javiera Guedes, Simone Callegari, Piero Madau, Lucio Mayer, Forming Realistic Late-Type Spirals in A CDM Universe: The Eris Simulation, IN: The Astrophysical Journal, in press (http://xxx.lanl.gov/abs/1103.6030)
On top, an edge-on image of our simulated galaxy, as an observer would see it if he could enter in the computer and look at the light emitted by our simulated stars. He could clearly see that the galaxy is composed by a thin disk of stars, and a brighter, small bulge at the center. At the bottom, a real image of our Milky Way galaxy as you would see it in infrared, where stars emit most of their light. Here too the Milky Way is seen edge-on, as our solar system lies in the plane of the disk.
