The parametrisation of cross sections for Dark Matter particle processes

Experiments at particle colliders provide experimental verifications of theories in particle physics, and allow to search for new particles. Computer programs are used to simulate particle collisions. Those so-called event generators can be used to prove theories and compare their results to actual collisions from the colliders. Even though those programs run on powerful computer systems, the event generation takes long and is also cost-intensive. Therefore, the aim of this thesis is to reduce this computation time in the aid of searches for new particles. One of the goals of modern particle physics experiments, such as the ATLAS and CMS detectors at the Large Hadron Collider, is to shed light on the problem of Dark Matter. The presence of matter in our universe, beyond known matter, is motivated by gravitational interaction. In this project, I simulated particle collisions producing particles that exist in theories of Dark Matter. I then parametrised the cross sections of these processes, a measure for the probability of this process occurring, depending on parameters of the Dark Matter theory studied, the mediator mass and Dark Matter mass. This parametrisation was initially studied using a simulation of many Dark Matter signal points with different mediator mass and Dark Matter mass, and then applied to a grid with fewer simulated points. Altogether, the parametrisation derived from the grid with fewer points shows cross-sections that are consistent with those of the full grid of points. This allows the generation of fewer signal points and to parametrise them instead, which then results in a much shorter computation time. These results are used for a publication on the constraints of Dark Matter searches performed at ATLAS, a particle detector, which uses data from the Large Hadron Collider (LHC).Dark Matter is a physical phenomenon, which has bothered physicists for decades. The first observations for a new invisible type of Dark Matter were made in the 1930s. However, how could one observe something, that is invisible? If one observes an arrow flying through the air straight, then suddenly sees a change in direction, it is logical that the arrow has to be influenced by something. In our case similar observations were made in space, where large objects were influenced by the gravitation of an invisible mass. This invisible, gravitational mass is called DarkMatter. The exploration of the Dark Matter’s character turned out to be very complicated and is still ongoing. Over the years, many theories were constructed and many of them were not coincident with observations. The most recent one is the introduction of a new particle, which is called weakly interacting massive particle or just WIMP. Unfortunately, this new fellow is not very easy to find and therefore, much effort is required. Large research facilities were founded and huge machines build to find a Dark Matter particle. The problem with this search is that, no one knows where to look and what to look for anyway. Unfortunately, we do not know how the WIMP looks like. It could be large or small, light or heavy or have any other properties. Therefore, we do not know where to look. This lack of information makes the search for a DarkMatter particle like playing an extremely difficult round of Battleship. The problem is thatwe cannot only hit points like A6 or F3, but also every interimvalue, like a BCC1.531. To have a chance against this well hidden ship, physicist use super computers, which look for indicators, how DarkMatter looks like. In this project I yield another advantage in the Battleships match by improving the computational search for possible candidates. This was done by only looking at simple, close-by points like F3 and F4 on our map and then assuming that the intermediate zone has to look comparable and is a mixture of those two points. It turned out that this approach is reasonable under certain conditions and can, therefore, help to find a Dark Matter particle by making computer simulations faster