LHCb Trigger

The design luminosity of2_⇥1032_cm 2_s 1_{(about a factor 40 less compared}

to ATLAS or CMS) results in about10M Hzrate of events with at least two charged

tracks with enough hits in the VELO and Tracking to be reconstructed into long

written to tape for offline analysis while enriching the selected data with events

of high interest (see Figure2.21). Within this,300Hzhas been allocated to charm

physics and the charm production at LHCb is factor⇠10larger thanbproductiond.

The trigger is separated into a custom-electronics hardware L0 trigger operating synchronously with collisions and a software High Level Trigger run on a processor farm. The original design also included a hardware L1 trigger which has since been absorbed into the other two.

Functionally there are three eventualities with a signal chain present (i.e. not pure combinatoric background) that can result in a trigger accepting the event:

• Trigger On Signal (TOS)

• Trigger Independent of Signal (TIS)

• Trigger On Both (TOB)

The TOS is the case where the signal chain in the event has been recognized by the trigger which fired.

In the TIS case, the signal chain is present but something else in the event caused the trigger to fire. This should result in data that is unbiased by trigger decisions provided that the cause of the trigger firing is independent of the signal chain.

The TOB case results from both the signal chain and some other part of the event causing the trigger to fire. The TOB events are notoriously difficult to analyse due to correlations and provided the number of TOB events is small, they are usually removed from the dataset and neglected.

2.9.1 L0

The L0 is an extremely fast trigger implemented on custom-built hardware aimed at fast rejection of uninteresting events. It reduces the rate of the events from

the collision rate to less than1.1M Hze. The L0 is composed of the L0 Calorimeter

trigger, the L0 Muon trigger and L0 Pileup. The Calorimeter and Muon L0 trig- gers select events with large transverse momentum and energy, which are com-

Figure 2.21: Diagram showing the trigger implementation at LHCb. Adapted from [43]

trigger combines information from all sections of the Calorimeter, the transverse

energy (ET) sum in HCAL is used to veto events with no visible interactions and

the SPD count is used to determine track multiplicity. The Muon trigger selects one

or two muon tracks with highestpT in each quadrant. The Pileup estimates the

overall number of pp interactions (the number of Primary Vertices) in the event.

The L0 Decision Unit (L0DU) collates this information and combines it into a sin- gle L0 Decision. The time between the collision and an L0 Decision being accepted

in the Front End (FE) electronics is4µsincluding time of flight of particles, cable

delay, and delays in FE electronics [43].

2.9.2 High Level Trigger

The High Level Trigger (HLT) is a C++ built application running on a filter farm of several thousand CPU nodes. Divided into two parts the HLT1 and HLT2 it

reduces the event rate from1.1M Hzto3kHz. It is completely software based, with

configuration facilitated by the use of Trigger Configuration Keys (TCK) which al- lows for modularity and for complete rerunning of the HLT offline. This is also used in a deferred trigger (implemented in 2012) which writes a portion of the events which are not immediately processed by the HLT to disk. These events are processed by the HLT during time between individual fills. The HLT is split into HLT1 which is used to confirm L0 decisions and partially reconstruct tracks and HLT2 which contains more complicated topological trigger conditions and dedi- cated settings for for individual analyses.

HLT1

The main feature of HLT1 is partial event reconstruction with full VELO track reconstruction. The VELO tracks are used to find vertices with at least 5 tracks

originating from them and such vertices within300mmaround the bunch crossing

point are considered primary vertices. Cuts are applied on the quality of VELO tracks and signal-like VELO tracks are associated with track hits. HLT1 also has a Muon trigger where VELO tracks are extrapolated to M3 and hits are searched for in a window of interest around the extrapolation. The track is then flagged as a muon if there is at least one extra hit in M2, M4 and M5. The track is then reconstructed using the VELO seed, the Tracking and potential Muon hits and the momentum and the Impact Parameter (IP) are determined. The HLT1 lines used in

this analysis are discussed in section3.3.

HLT2

With the output of HLT1 low enough to allow for offline track reconstruc- tion, the HTL2 uses a loose selection on momentum and on the IP of a track before categorising tracks as individual particles and using them to create common reso-

nances (for exampleK0

S ! ⇡+⇡ ). These are used for the reconstruction of decays

and an invariant mass cut is generally applied. The HLT2 lines used for this analy- sis are discussed in section3.3.