Adding algorithm that writes jet-associated pixel and SCT hits to AOD

Spyros Argyropoulosrequested to merge
(removed):master-jethitassociation into master

Added JetHitAssociation algorithm that runs on ESD, taking as input a jet collection and a hits collection (Pixel & SCT), finds the hits that are within a given dR of a jet above a certain pt threshold and saves these hits into a new collection (by default these collections are called JetAssociatedPixelClusters and JetAssociatedSCTClusters).

We would want this algorithm to run by default in all ESD->AOD jobs, the size increase from the extra collections is expected to be minimal. Below are the numbers for different jet association radii and a jet pT threshold of 300 GeV that we have as a default. The numbers are taken from a private test (running on official mc16d ESDs with Athena,master,latest) with a Z' sample (800030) and the standard ttbar sample (410470).

Z' dR=0.2 Z' dR=0.4 ttbar dR=0.2 ttbar dR=0.4
5% (35kB/event) 10% (73kB/event) 0.2% (1kB/event) 0.3% (2.1kB/event)

The Z' sample has a flat pt spectrum extending to several TeV, so for most ATLAS samples we would expect a minimal increase (as in ttbar). More details can be found here

Using dR=0.4 and a pT threshold of 300 GeV (the default values) We get the following for two 2018 data runs (35770 with 55 average pile up and 364485 with 90 average pile up)

Data mu = 55 Data mu = 90 ttbar MC
0.5% (1.7kb/event 0.15% (0.8kb/event) 0.5% (1.3kB/event)

We would hope to have this in to be used for the upcoming reprocessing.

Tagging:
FTAG conveners: @vdao @cschiavi FTAG software conveners: @duperrin @sanmay Trackless FTAG: @thuffman @cpollard

To do:

  • Decide in which package the algorithm (C++ code) will live
  • Potentially move configuration file to a different location? (some adjustments might be necessary there?)
  • Make sure that the algorithm runs when no jets are present - to test this we would need a config (from discussion it seemed like jets will always be available at ESD level)

Other todos from discussion in software meeting:

  • Provide size increase for data (as a function of PU or perhaps a run with high PU (worst case scenario) would be enough?)
  • make algorithm compliant with athenaMT
  • switch algorithm off by default
  • provide timing information
Edited by Todd Brian Huffman

Merge request reports