bfield projs06aOn the last  September our group published the first results of a numerical investigation on the origin about the magnetisation of the Cosmic Web, that we conducted producing challenging magneto-hydrodinamical simulations on Piz Daint at CSCS (F. Vazza, M. Brüggen, C. Gheller, P. Wang, 2014, MNRAS) These results were also presented in a press-release by CSCS/ETH. 
Why the magnetic field in galaxies and galaxy clusters is so "high" (a few microGauss, i.e. hundreds of thousands time smaller than the magnetic field intensity on the surface of Earth) as we observe it through telescopes is a puzzle since decades. Such fields are usually observed across gigantic scales of millions of lightyears, and display chaotic patterns as if some turbulent mechanism is responsible for their amplification. Actually, we also know that the plasma filling the nearly empty space between galaxies is in a fairly turbulent state, and galaxies (or cluster of galaxies) flying through it are continuously stirring up the medium. Therefore, it seems natural to assume that this turbulence leads to a cosmic dynamo that can amplify some extremely tiny level of early magnetisation, that should have been delivered by super-high energy phenomena soon after the big bang. The big puzzle is this picture is  the (so far hypothetical) primordial magnetic field ought to be extremely small, because we do not see it in the deepest available observation of the early Universe (from the cosmic microwave background). This set a strong constrain on the initial  "seed field" that cosmic turbulence can use to get to the magnetic fields observed around us nowadays, and only a very efficient cosmic dynamo can do the job. 
The formation of cosmic structures is a complex  physical problem that has become since a few decades extremely appealing to parallel computing, and even so is the quest for cosmic dynamos, given the highly non-linear nature of turbulence associated to that. 
Our recent work started from these ideas to simulate at the highest available resolution the formation of cosmic structure and the associated growth of turbulence and magnetic fields across cosmic epochs, running with the cosmological code ENZO on Piz Daint, thanks to an allotted CHRONOS (Computationally-Intensive, High-Impact Research On Novel Outstanding Science) project. The use of a large parallel architecture is mandatory to achieve the necessary dynamical range (the ratio between the largest scale in the system and the spatial resolution, which also gives the maximum Reynolds number that can be simulated) to simulate turbulence in a realistic fashion. In this project we simulated a 2400^3 cubic mesh sampling a cubic volume with a side of 150 millions of lightyears, where the birth of galaxies and galaxy clusters is followed in great detail, running on 1024 computing nodes of Piz Daint for a total of 2.4 million hours. The use of Tesla Graphics Processing Units was also fundamental for our work, as it warranted a speed-up of more than a factor 4 in our full calculation, thanks to the porting into CUDA of the most time-consuming parts of the algorithm (by P.Wang and T.Abel). This enabled us to get to a solution faster in time, and as well as to test our numerical parameters with smaller simulations in a much faster way.
In this first simulations we just delivered (that as far as we can tell represents the biggest magnetohydrodynamical simulation done in cosmology so far) we investigated whether this scenario can explain the magnetic field we see in galaxies and galaxy clusters, and we produced forecasts about the strength of magnetic fields in the remaining volume of the Universe, that only future big radio telescopes (such as the Square Kilometer Array) will observe. 
On one hand, we showed that cosmic turbulence can produce an efficient enough dynamo to amplify the primordial magnetic fields up to the observed values, and confirmed earlier results that several other authors already suggested. On the other hand, we showed that even with the unprecedented resolution we could reach in this project, the dynamo produced  in the very rarefied regions like the filaments of matter that connect galaxies and galaxy clusters is weak, and the final magnetic fields hardly get to the nanoGauss, i.e. one billion time smaller than the magnetic field on Earth. At first sight this is discouraging, because most probably with such low magnetic fields it will be extremely difficult to detect any emission from cosmic filaments even with the next generation of radio telescopes. But on another perspective, the existence of such inefficient cosmic dynamo can be a good thing. Indeed, having experience nearly no dynamo amplification, the magnetic fields of filaments will preserve primordial information about the strength and topology of the primordial magnetic fields released soon after the Big Bang. Whenever we will become able to  observe them in the future, they might teach us a great wealth of priceless information about the early epochs of cosmic evolution, including high-energy phenomena we can only speculate about at the moment. 
Our future simulations will try to address this issue in more detail, and to characterise by means of complex simulations the complementary role played by magnetic fields released by other astrophysical players, like exploding stars and accretion disks around black holes. "
Image: the distribution of cosmic magnetic fields for the simulated cosmic web using Piz-Daint (F. Vazza, M. Brüggen, C. Gheller, P. Wang, 2014, MNRAS, )