HLT Threshold Studies

The basic goal of this procedure is to estimate the pT thresholds that trigger objects would need at higher luminosities, in order for the extrapolated HLT rate at higher luminosities to match the rate seen at 1.3e34 cm-2s-1 (about 1 kHz). See this presentation for details on how these trigger thresholds are computed.

Running over data

Inside the folder /afs/cern.ch/user/j/jjbuch/public/Trigger_Threshold_Studies there are three files: triggerScanner.py, filesInput.txt, and thresholdFinder.py. Copy these into a CMSSW area, release 8_0_19 or higher.

The first script to use is triggerScanner.py, via:

python triggerScanner.py start_file_index files_to_scan

For example, to run over the first 200 files of a data set, you'd use:

python triggerScanner.py 0 200

The files to run over are specified in filesInput.txt. The provided list represents HLTPhysics0 data taken in Run2016H. There are several thousand files in the list, and it may be advantageous to run over only a few hundred at a time with triggerScanner.py and then merge the output with hadd.

The resulting output is a collection of histograms containing the pT distribution of objects passing the triggers of interest.

Computing Thresholds

Once the output histograms from triggerScanner.py have been obtained, the next step is to run thresholdFinder.py:

python thresholdFinder.py output_from_triggerScanner.root

To analyze a single object trigger, or to just vary one object's pT threshold in a multi-object trigger while keeping the others constant, use simple_trigger().

To vary two objects in such a way that their pT thresholds remain constant, use two_object_trigger().

For the special case of a three-object trigger in which the pT thresholds of two of the objects (e.g. PFMHT and PFMET) should be kept equal, and furthermore the pT ratio of the third object vs. the other two should be kept constant - effectively reducing the three-object trigger to a two-object trigger - use twoFromThree_object_trigger().

The text output from thresholdFinder.py lists the thresholds that most closely reproduce the desired rate reduction factors for the triggers of interest. In the case of a two-object trigger, a ROOT file containing a 2D histogram of reduction factors will also be produced.