Major mergers between galaxies are fairly rare events, but when they happen, they produce some of the most awe-inspiring images from space. Such events can also completely change the size and shape of galaxies, disrupting spiral arms and funneling stars into central bulges. Based on the relatively undisturbed structure of our own galaxy, for example, we can estimate that the Milky Way has not had any major mergers within the last ten billion years. It is believed that galaxy mergers are important for triggering episodes of star formation, and may be a cause of active phases of some galaxies where the supermassive black holes at the centers spew enormous jets of particles and radiation.
Mergers are unfortunately difficult to see observationally, because they are rare and because in the distant Universe it is hard to visually resolve a merging system. To answer questions about how often major mergers happen and when the next one might occur in a galaxy like ours, as well as to probe details of the properties and consequences of mergers, scientists often use large computer simulations which contain millions of galaxies and the halos of dark matter in which they live, in the hopes of catching a few galaxies merging together at any one time. These simulations generate tens of terabytes of data which describe the evolution of matter particles over the history of the Universe, a volume of data which is impossible to comb through by hand. Instead, scientists use sophisticated tools to search for particle clusters in this data, which correspond to where galaxies are located.
However, the current generation of analysis tools performs poorly precisely when galaxy interactions are at their most exciting: during major mergers. This is because most current analysis tools separate galaxies using only spatial information. By definition, when two galaxies are merging, they are located in the same place, which means that current tools often lump two galaxies together into one object as soon as they touch, and don't catch the uniqueness of the merging system.
To solve this problem, KIPAC graduate student Peter Behroozi along with Professor Risa Wechsler and recent KIPAC alumnus Heidi Wu developed a new computational algorithm to distinguish galaxies in simulations using both spatial and velocity information. The algorithm goes by the humble name of 'Rockstar', which stands for Robust Overdensity Calculation using K-Space Topologically Adaptive Refinement. As merging galaxies often have distinct speeds relative to each other, the Rockstar algorithm makes it possible to calculate galaxy properties well into the merger process and better understand the dynamics which lead to the disruption of pre-existing distributions of stars and matter. Scientifically, this means that questions about how common major mergers are, how long they take, and what happens to the galaxies involved during and afterwords can be answered on a much more solid basis, greatly increasing the utility of large computer simulations for studying galaxy mergers. The algorithm's increased accuracy has many benefits for other scientific projects as well, including better predictions for galaxy locations and velocities, which will aid precision determinations of the matter and energy content of the Universe.
This work is based in part on a paper submitted to The Astrophysical Journal and available from astro-ph at arXiv:1110.4372.
Tidbit author: Peter Behroozi and Jack Singal