Most dynamical astrophysical problems can only be investigated theoretically in detail using large computer models. We use some of the largest supercomputers in the world to model a wide selection of astronomical phenomena, ranging from generic small scale energy release process in magnetized plasmas, through relativistic collisionless shocks, to the formation of stars and planets.
One of the key objects for testing our codes and theories is the Sun. The close proximity of the Sun allows for detailed observations of dynamical phenomena taking place in its photosphere and coronal regions. Here the presence of magnetic field gives rise to a large variety of phenomena on different length and time scales. Typical time scales span from fractions of a second to months, and spatial details can be discerned down to fractions of an arcsecond (on the order of 100 Km). A wealth of space based and Earth based observations are thus available for comparison with our computational results, which are also being used by observers for calibrations and design of new types of observations.
Our computer models of the convection zone below the solar surface show how the magnetic field is carried around and concentrated by convective motions before it emerges at the surface. Emerging new magnetic flux interacts violently with existing coronal magnetic flux systems, which gives rise to energy release processes ('solar flares') where free magnetic energy is converted to high velocity jet-like motions, and also heats the plasma to many million degrees.
In the process a fraction of the electrons and protons in the corona are accelerated to relativistic energies. Some of these particles are ejected along magnetic field lines that reach far into the interplanetary space and contribute to the generation of the 'solar wind', and to 'space weather' and 'solar storms' in the Earths magnetosphere -- with nice looking auroras as a byproduct. To be able to predict solar storms is important, because of the effects they can have on satellites and astronauts in Earth orbit, on telecommunication, and on our power supply systems.
(Parts of this work is done under the EU/FP7 network Space Weather Integrated
Forecasting Framework (SWIFF: http://www.swiff.eu/ ))
Star and planet formation
In the context of star and planet formation we study the formation of 'molecular clouds', consisting of cold (~10 K) turbulent, supersonic, magnetized gas, and the formation of stars within. Stars are formed in these regions through gravitational collapse at focal points of multiple shocks formed by nearby supernovae. Because these phenomena obey generic laws of turbulence we can successfully predict statistical properties of the star formation process -- for example the distribution of initial stellar masses.
To be able to follow the collapse down to small scales requires special computer codes that are able to locally enhance the numerical resolution many times. A typical size of an star forming cloud is many light years, while the region in which a star form is collapsed to diameters of light days, in which the final star is only on a few light seconds in diameter.
Similar techniques can be used to follow the collapse of individual proto-stellar envelopes, and to follow the formation of the central stars and their planets. To model this process in detail requires a complicated treatment of radiative energy transfer in the plasma, and also requires modeling the coupled dynamics of gas and dust. Dust in the proto-planetary disk is what eventually makes the types of planets that we find in our solar system, and detailed modeling of this process will be able to show how the dust can be assembled to planetesimals and planets.