What Are Pileups? Explanation & Analysis

[Silviu]

Hello! Can someone explain to me what pileups are? I read some stuff but I am not sure I understand. So, if you collide a bunch of protons looking for something, from what I understood pileups are the other proton-proton collisions that don't give you the interaction of interest? However, I understand that in particle accelerators they try to get rid of pileups, but how can you know before doing any analysis on the data if a certain interaction is a pileup or the interaction you are looking for?
Thank you.

 

[ChrisVer]

You are correct. Pileup comes when the detector measures objects that do not come from the Primary Vertex (PV), where the hard scattering occurs. There are two types of pileup: in-time and out-of-time.
The first (intime) comes from the same proton bunch crossing. You can determine the interaction points by identifying vertices, and comparing their distances to the PV.
The second occurs when other bunches cross and your detector has not yet recorded the signal completely (eg due to deadtime). The silicon detectors are having a good time resolution of ~25ns (as much as the time that bunches cross), and so the out-of-time pileup is determined by the silicon detectors (in pseudorapidity region $\eta < 2$).
Also in general, pileup events are low energetic. So, by applying high-pT cuts (eg pT>10GeV etc), you throw away most of it.
Pileup is not known before you look at data- you can estimate it by simulations (MC), but in the end you will have to apply correction weights to your MC to bring it to agree with the observed data's interactions.

 

[mfb]

Interesting collisions are rare, so the experiments usually try to get as many as possible. You can't change the bunch spacing (one bunch crossing every 25 ns at the LHC), so you need many collisions per bunch crossing.

The definition which collision is the one you are looking at comes from your analysis. If you search for two electrons (random example), and find two in an event, then the collision producing those two electrons is the one you are interested in, and all other collisions are called pileup: It is not what you are interested in, but it can disturb the measurement. You try to remove the impact from these pileup collisions in your analysis.
It can also happen that one electron comes from one collision and the other one from a different collision - that is also an effect of pileup you have to take into account (although it is very rare, especially with electrons).

 

[Vanadium 50]

The example of out-of-time pileup (the second kind) is not the best. It takes time for your signal to form in your detector. For a scintillator-calorimeter, it is around 60ns, so 2-1/2 bunch crossings: so when you measure the energy, you may have leakage from past events. For other detectors, even more - if you use drift tubes to measure muons, they can have drift times as long as a microsecond or even more: so you are integrating over 50 or more crossings.

 

[ChrisVer]

Well, for the calorimeter here : http://arxiv.org/abs/1609.09324 (last paragraph of Section 3.1) they say that the calorimeter parts can take up to 450ns for the charge collection time. Also the same paragraph gives a short answer to the OP ..