There are concerns that the current L2jet code, described here, is finding jets with less energy than what would be expected based on what is being passed on from L0. To investigate this, I have developed a true 2-D search algorithm which scans every 5x5 patch to find the one with the higest energy. I have run the codes over 500k events from a filtered pythia QCD sample found here: /star/institutions/mit/corliss/l2/sample_runs/pythia500_QCD_pt10_Filter20Bemc_LT10invpb.eve.bin

I have included my comparison code here. The conditions for accepting jets, ie mono-jets, barrel-barrel jets, endcap-barrel jets, and endcap-endcap jets start around line 365. The threshold values I used are in this file. Note, these thresholds may be different on the actual L2 machine.

Here I compare my previous 5x5 algorithm with the true 2-D search algorithm.

I have also made the true 2-D search algorithm into a stand-alone code so that the speeds of the two codes can be compared.

Fig 1: This is the number of KTicks it takes the original algorithm to run.

Fig 2: This is the number of KTicks it takes the true 2-D algorithm to run.

I have also included the standard L2jet monitoring plots for the current algorithm here, and the true 2-D algorithm here.