PF Jet Performances at High Level Trigger using Patatrack pixel tracks (CMS DP-2021/005)

As the start of the LHC Run III approaches, there are substantial efforts towards accelerating the Particle Flow (PF) reconstruction at the High Trigger-Level (HLT) to allow new physics searches to probe unexplored regions of parameter space at lower masses and weaker couplings. Since Run II, the quality of pixel track reconstruction has been greatly improved, which in turn allows us to consider using these optimized pixel tracks, called "Patatrack'' pixel tracks, as inputs to the PF reconstruction algorithm at HLT, leading to significant decrease of processing time. In this note, the performance of PF jets reconstructed with "Patatrack'' pixel tracks as inputs is compared to that of PF jets reconstructed using the standard Run II tracks reconstructed at the HLT. The comparisons include PF jet energy fractions, PF particle multiplicities, as well as PF jet resolution and response.

More details can be found in the CMS-DP-2021-005 note.

PF scouting reconstruction timing A comparison of the time it takes to reconstruct scouting objects for different track reconstruction algorithms used by the PF algorithm at the HLT. The time was measured on a “Skylake” Xeon Gold 6130 processor using 4 threads running on 2018 data events passing L1 seeds of interest to the scouting program. [Get pdf version] Contact: Dimitrios Karasavvas

Charged Hadron pT A comparison of the transverse momentum of charged hadrons clustered into PF jets for the different track reconstructions used by the PF algorithm at HLT. The normalized distributions for PF candidates reconstructed using Run2 tracking (black), patatrack pixel tracks with the tight selection (blue) and patatrack pixel tracks with the loose selection (red) are shown. The comparisons are performed for jets with (left) and (right), and transverse momentum 30-100 GeV. [Get pdf version left] [Get pdf version right] Contact: Dimitrios Karasavvas

Charged Hadron pT A comparison of the transverse momentum of charged hadrons clustered into PF jets for the different track reconstructions used by the PF algorithm at HLT. The normalized distributions for PF candidates reconstructed using Run2 tracking (black), patatrack pixel tracks with the tight selection (blue) and patatrack pixel tracks with the loose selection (red) are shown. The comparisons are performed for jets with (left) and (right), and transverse momentum 100-400 GeV. [Get pdf version left] [Get pdf version right] Contact: Dimitrios Karasavvas

Charged Hadron pT A comparison of the transverse momentum of charged hadrons clustered into PF jets for the different track reconstructions used by the PF algorithm at HLT. The normalized distributions for PF candidates reconstructed using Run2 tracking (black), patatrack pixel tracks with the tight selection (blue) and patatrack pixel tracks with the loose selection (red) are shown. The comparisons are performed for jets with (left) and (right), and transverse momentum 400-1000 GeV. [Get pdf version left] [Get pdf version right] Contact: Dimitrios Karasavvas

Charged Hadron Energy Fraction A comparison of the energy fraction carried by charged hadrons of a PF jet for different track reconstructions used by the PF algorithm at HLT. The normalized distributions for PF candidates reconstructed using Run2 tracking (black), patatrack pixel tracks with the tight selection (blue) and patatrack pixel tracks with the loose selection (red) are shown. The comparisons are performed for jets with (left) and (right), and transverse momentum 30-100 GeV. [Get pdf version left] [Get pdf version right] Contact: Dimitrios Karasavvas

Charged Hadron Energy Fraction A comparison of the energy fraction carried by charged hadrons of a PF jet for different track reconstructions used by the PF algorithm at HLT. The normalized distributions for PF candidates reconstructed using Run2 tracking (black), patatrack pixel tracks with the tight selection (blue) and patatrack pixel tracks with the loose selection (red) are shown. The comparisons are performed for jets with (left) and (right), and transverse momentum 100-400 GeV. [Get pdf version left] [Get pdf version right] Contact: Dimitrios Karasavvas

Charged Hadron Energy Fraction A comparison of the energy fraction carried by charged hadrons of a PF jet for different track reconstructions used by the PF algorithm at HLT. The normalized distributions for PF candidates reconstructed using Run2 tracking (black), patatrack pixel tracks with the tight selection (blue) and patatrack pixel tracks with the loose selection (red) are shown. The comparisons are performed for jets with (left) and (right), and transverse momentum 400-1000 GeV. [Get pdf version left] [Get pdf version right] Contact: Dimitrios Karasavvas

Charged Multiplicity A comparison of the multiplicity of charged particles comprising the PF jets for different track reconstructions used by the PF algorithm at HLT. The normalized distributions for PF candidates reconstructed using Run2 tracking (black), patatrack pixel tracks with the tight selection (blue) and patatrack pixel tracks with the loose selection (red) are shown. The comparisons are performed for jets with (left) and (right), and transverse momentum 30-100 GeV. [Get pdf version left] [Get pdf version right] Contact: Dimitrios Karasavvas

Charged Multiplicity A comparison of the multiplicity of charged particles comprising the PF jets for different track reconstructions used by the PF algorithm at HLT. The normalized distributions for PF candidates reconstructed using Run2 tracking (black), patatrack pixel tracks with the tight selection (blue) and patatrack pixel tracks with the loose selection (red) are shown. The comparisons are performed for jets with (left) and (right), and transverse momentum 100-400 GeV. [Get pdf version left] [Get pdf version right] Contact: Dimitrios Karasavvas

Charged Multiplicity A comparison of the multiplicity of charged particles comprising the PF jets for different track reconstructions used by the PF algorithm at HLT. The normalized distributions for PF candidates reconstructed using Run2 tracking (black), patatrack pixel tracks with the tight selection (blue) and patatrack pixel tracks with the loose selection (red) are shown. The comparisons are performed for jets with (left) and (right), and transverse momentum 400-1000 GeV. [Get pdf version left] [Get pdf version right] Contact: Dimitrios Karasavvas

Missing transverse energy A comparison of missing energy in events for the different track reconstructions used by the PF algorithm at HLT. The normalized distributions for events whose PF candidates were reconstructed using Run2 tracking (black), patatrack pixel tracks with the tight selection (blue) and patatrack pixel tracks with the loose selection (red) are shown. [Get pdf version] Contact: Dimitrios Karasavvas

Missing transverse energy / Scalar sum of jet pT A comparison of missing energy divided by the scalar sum of all the jet transverse momenta in event for the different track reconstructions used by the PF algorithm at HLT. The normalized distributions for events whose PF candidates were reconstructed using Run2 tracking (black), patatrack pixel tracks with the tight selection (blue) and patatrack pixel tracks with the loose selection (red) are shown. [Get pdf version] Contact: Dimitrios Karasavvas

Preliminary jet pT response The jet response is the mean of a Gaussian fit in the core of the raw distribution of PF jets in several and bins. A comparison of the jet response of PF jets with (left) and (right) for the different track reconstructions used by the PF algorithm at HLT. The normalized distributions for PF candidates reconstructed using Run2 tracking (black), patatrack pixel tracks with the tight selection (blue) and patatrack pixel tracks with the loose selection (red) are shown. [Get pdf version left] [Get pdf version right] Contact: Dimitrios Karasavvas

Preliminary jet pT resolution The jet resolution is the sigma of a Gaussian fit in the core of the raw distribution of PF jets in several and bins, after correcting the reconstructed jet by the response. A comparison of the jet resolution of PF jets with (left) and (right) for the different track reconstructions used by the PF algorithm at HLT. The normalized distributions for PF candidates reconstructed using Run2 tracking (black), patatrack pixel tracks with the tight selection (blue) and patatrack pixel tracks with the loose selection (red) are shown. [Get pdf version left] [Get pdf version right] Contact: Dimitrios Karasavvas

-- LisaBenato - 2021-04-07