As long as the LHC is not running, the ATLAS experiment is taking data runs to measure cosmic muons. Muons are the heavier brothers of the electron. Every second 3000 of these muons pass through ATLAS leaving at least one hit in the detector. Because a muon is so heavy, it can penetrate though the 100 meters of rock above ATLAS, other particles are absorbed and therefore almost never seen in the detector. We know the paths of the muons are straight. We can use this knowledge to calibrate the detector. Very useful indeed.

On the left, you see a muon flying though the middle part of the SCT. It obviously leaves more that one YES (yellow dot) in the detector. On the right side, you see the same track, but through the disks. It only leaves one or two hits.
The part of the ATLAS detector I will be working at is the Semi Conductor Detector (SCT). This detector is built up from modules with small strips in them. These strips give a signal when a charged particle (such as a muon) passes by. This is a binary signal: YES (a particle) or NO. In order to measure as many of the particles produced in the LHC collisions later on, the detector has two regions: the middle part (barrel, cylindrical shaped) and circular disks on each side. As everything is installed by now, it is not my job to design or construct the detector, but to calibrate it. Here the muons come into play.
When a muon passes through the barrel part of the detector, the change that it leaves a YES in only one point of the detector is rather small. (See the figure.) When looking at the disks, the change of leaving only one YES is quite big.The thing is, that we have very complicated software that searches for tracks in the jungle of many hits that may come from a single event. As you can basically draw any straight line through a single point, this is not very informative. You need at least two points to determine a straight line (a line is therefore a 2D object). So, with only one hit, the software cannot find a track, with two hits it is stillambiguous and with three hits it starts to work.
Because I have been interested in the disks for the last few weeks, I have gone through many many files looking for tracks that left enough hits to find straight paths and analyse these. The good thing to report here is that some analysis can be done and it is good fun to be working with real data. Because statistics are low, I expect errors to be big. I will keep you informed.