Last week my advisor and I were discussing a very important part of the ATLAS detector. Although you won’t see it on the colorful diagram that most often pops into your head when you think of ATLAS, it’s one of the most important parts. It’s the trigger system. A trigger makes a decision to keep an event or not. There will be multiple interactions (23 under “normal” operation) per bunch crossing, which occur every 25 ns. In a perfect world, we would be able to keep all of the interactions, but in reality we just don’t have enough disk space and computing power to do so. But getting the triggers correct is extremely important, because if a trigger decides to throw an event away, we will never see it.
The trigger decision is a multi-layered process, I’ll describe the basic idea. (This isn’t just true for ATLAS, but all HEP experiments). I’m sure you all remember how we see particles in the detector, so I won’t cover it here (but can later if you’d like).
We’ll start with an interaction, protons collide and through the power of physics new particles are created and decay in to their lightest daughter particles.
Quick! We’ve got to decide if we consider this event “interesting” or not. We define an event as being interesting if the combination of daughter particles we see in the detector is rare (low cross section). In general the level of interest goes as follows: hadrons (jets) aren’t as interesting – just because we get so many of them – but as their pT (transverse momentum) goes up, they get more interesting. Leptons are more interesting than hadrons and like jets and pretty much everything else, they get more interesting with higher pT. Multiple leptons are more interesting than single ones, and because flavor is conserved multiple flavored leptons is very interesting (flavor meaning the type: electron, muon or tau). For example, if the daughter particles were a combination of different flavor leptons we’d find the event interesting; if it were a couple of low pT jets, probably not. (note: pT is transverse momentum, meaning the momentum that occurs in the x and y direction, I can explain why we use this if anyone is curious…). This also means that if there is new physics in the low pT region of hadrons, we’d probably never see it. Don’t worry though, we have generations of previous experiments that already probed those areas and we trust that they didn’t find anything that we haven’t already seen.
The triggers get fired in order of interest as well. Back to that event… let’s start the process: multi-flavored leptons, no; multi-leptons, no; any high pT leptons, yes! Congratulations, you just passed the high pT lepton trigger, your event gets saved. But wait there’s more, even though we decide to save the event, some interactions are still so common that we employ a prescale. This is a number which decides the percentage of these particular types of events to keep. For example if a certain pT hadronic jet trigger has a prescale of 10, only one out of every 10 events would be kept. The first 9 would be thrown away. This unfortunately has to be done for the same reason we have the trigger, because we just can’t keep all those events.
I am in oversimplifying the process, but I hope I gave a general idea of how triggers work. I’ll also try to keep everyone up to date as to the status of the LHC with my fellow bloggers. We should be hearing something on the machine soon.
Interesting.
Not much talk about the fine points of the actual operation of the collider. The only really detailed talk that I have seen was by Dr. Thomas Junk, on the fine tuning of LEP from way back in 2001. Really intersting talk, gives us outsiders some insight into the complexity of a modern high energy collider. Amazing how the team squeezed every last ounce (er.. gram) of performance from the machine before it was (reluctantly) shut down for conversion to the LHC.
Definitely worth a listen, here’s the link to the talk onthe KITP site.
Searches for the Higgs Boson at LEP:
http://online.kitp.ucsb.edu/online/colloq/junk1/
I have a question about the trigger system. Will the trigger parameters be the same for the 1/2 design energy and likely 1/100th design luminosity run coming up as they would be if running at full energy and luminosity?
I imagine there will be far fewer events at the lower energy and luminosity so the computer system would have some unused capacity to record less interesting events. So would the trigger parameters be relaxed to capture more events for calibration purposes?
Also are the trigger parameters set the same above for the cosmic ray data taking or are they opened up to capture these relatively rare events?
Hi Harbles,
Excellent question. We do design triggers for specific running needs. Although I don’t know the specifics about how we will change them for the upcoming run, we will definitely maximize our computing capabilities.
I know for the cosmic runs we designed special triggers for cosmic rays. They had a much lower threshold than the triggers we’ll use for collision data and also had specific fiducial requirements since in general, cosmic rays don’t go through the interaction point.
I hope this answers your question!
Regina
Well, and what about software triggers, do you use some? As I understand it’s neccesary to have some in order to design the hardware ones.
The problem is to apply the software trigger to the MC simulators (Pythia for example), but I don’t manage to find any yet. Maybe know some?
oleksandr gituliar
Hi Git,
I assume you mean employing a “trigger” on the simulated data. That is usually just done with things like energy or momentum cuts (what is essentially what a trigger does). It would probably be best to talk to trigger people in the experiment you’re associated with, I don’t know of any specific software trigger included in say Pythia.
Regina