|A slice of the CMS detector.|
Broadly speaking, the aim of the talk is to give the theorists in the audience an
introduction to state-of-the-art reconstruction (e.g. particle flow, techniques for dealing with high pile-up, the status of tau reconstruction) and their implications for searches. A discussion of triggering (whether focused on hadronic or more general) would also be very useful. Beyond these vague suggestions, you can define the scope of the talk however you think will do best to motivate, focus, and inform discussions about possible future analyses. The theorists in the audience will have a mix of BSM and SM expertise, and a somewhat more appetite than average for experimental details. See: Jet Reconstruction and Triggering
|A block diagram of the CMS L1 trigger|
In particle physics, a trigger is a system that uses simple criteria to rapidly decide which events in a particle detector to keep when only a small fraction of the total can be recorded. Trigger systems are necessary due to real-world limitations in data storage capacity and rates. Since experiments are typically searching for “interesting” events (such as decays of rare particles) that occur at a relatively low rate, trigger systems are used to identify the events that should be recorded for later analysis. Current accelerators have event rates greater than 1 MHz and trigger rates that can be below 10 Hz. The ratio of the trigger rate to the event rate is referred to as the selectivity of the trigger. For example, the Large Hadron Collider has an event rate of 1 GHz (109 Hz), and the Higgs boson is expected to be produced there at a rate of at least 0.01 Hz. Therefore the minimum selectivity required is 10−11.Taking A Closer Look
To have a good chance of producing a rare particle, such as a Higgs boson, a very large number of collisions are required. Most collision events in the detector are “soft” and do not produce interesting effects. The amount of raw data from each crossing is approximately 1 MB, which at the 40 MHz crossing rate would result in 40 TB of data a second, an amount that the experiment cannot hope to store or even process properly. The trigger system reduces the rate of interesting events down to a manageable 100 per second.
To accomplish this, a series of “trigger” stages are employed. All the data from each crossing is held in buffers within the detector while a small amount of key information is used to perform a fast, approximate calculation to identify features of interest such as high energy jets, muons or missing energy. This “Level 1” calculation is completed in around 1 µs, and event rate is reduced by a factor of about thousand down to 50 kHz. All these calculations are done on fast, custom hardware using reprogrammable FPGAs.
If an event is passed by the Level 1 trigger all the data still buffered in the detector is sent over fibre-optic links to the “High Level” trigger, which is software (mainly written in C++) running on ordinary computer servers. The lower event rate in the High Level trigger allows time for much more detailed analysis of the event to be done than in the Level 1 trigger. The High Level trigger reduces the event rate by a further factor of about a thousand down to around 100 events per second. These are then stored on tape for future analysis.