Computational Astrophysics in Kürze:
PHOENIX is a general-purpose state-of-the-art stellar and planetary atmosphere code. It can calculate atmospheres and spectra of stars all across the HR-diagram including main sequence stars, giants, white dwarfs, stars with winds, TTauri stars, Novae, Supernovae, brown dwarfs and extrasolar giant planets.
The Hamburger Sternwarte is the 'home base' of the PHOENIX project. We develop our methods and algorithms with the goal to produce as general and realistic as possible physical simulations while still being (barely) able to run them on existing supercomputers. This allows us to tackle a very wide range of applications, from terrestrial planets to supernovae and even cosmological applications. Due to the physical complexity of the model atmosphere problem, PHOENIX has grown to more than 1.2 million lines of code, using more than 20GB of physical input data and running the code in its more advanced modes requires 2048 to more than 16000 CPUs on parallel supercomputers (our present favorites: HLRN and NERSC). Nevertheless, the different modules of PHOENIX are developed on local computers (or even a PS3) and for smaller runs we have our own parallel computers at the observatory.
Woran wir forschen:
My research is similar to global climate simulations, but for the conditions found in the atmospheres of stars and extrasolar planets. The basic differences to the Earth's atmosphere are the different thermodynamics of the stellar/planetary atmospheres and the problem of handling the complex transfer of radiation through the atmospheres. The simulations are performed with our PHOENIX code, which my group continues to develop at the Hamburger Sternwarte.
The results of the simulations are physical models for the structure of the atmospheres, e.g., temperatures, pressures and so on, as function of depths into the planet or star. Knowing the structure of the atmosphere we can calculate the spectrum that the star or the planet emits by solving the radiative transfer equation for a large number of wavelengths. By comparing these synthetic spectra to observed spectra and by adjusting the parameters of the simulation we can determine things like the abundances of the elements found in the atmosphere.
An key feature of my research is that the methods and algorithms that we use are designed as general as possible and can, therefore, be applied to a wide range of very different astronomical objects, ranging from terrestrial and giant planets to supernovae and accretion disks. The big advantage of this approach is that modules of PHOENIX that have been tested on one class of objects can be re-used on many other types of objects.
The main project I am currently working on is the extension of PHOENIX to 3D (spatial). For this, Ed Baron and I have developed a very general framework to solve radiative transfer problems in 3D spatial configurations. We have added the framework to PHOENIX, creating PHOENIX/3D (which also included PHOENIX/1D, of course). This is a long term scientific project, I expect that in 10-20 years computers will be fast enough to fully unlock the potential of PHOENIX/3D.