Track Reconstruction with iTPC Sector in Run 18
Dmitri Smirnov
March 13, 2019
Updated
TPC in Run 18
- In Run 18 the inner section of TPC sector 20 trippled the number of layers - 13 TPC layers to 40 iTPC layers resulting in change from 45 to 72 total layers in the sector - It was decided not to use iTPC hits in Run 18 tracking - We would like to know how tracks are affected by missing hits in sector 20
First Observations
-
Early observations reported by Mike Lisa and Xiaoyu Liu
- The data was reconstructed by fast offline during the Run, i.e. early feedback, no fixed release
- Claimed: $\sim 75\\%$ of tracks lost at hight $P_T$
First Observations
- Asked Xiaoyu to analyze recently reconstructed data - The effect is much smaller
Our Investigation
- Look at the data reconstructed with SL18h with vetted calibrations - Model the situation in sector 20 by excluding hits in the inner sector 14 - Reconstructed same sample twice: With and without hits in sector 14 - Labeled samples as YES14 and NO14 respectively - We confirm the expected loss of track efficiency around sector 14 for global and primary tracks
Track Phi vs Eta Distribution
Track Phi vs Eta Distribution
Track Phi vs Eta Distribution
Track Phi Distribution
Track Phi Distribution
Track Phi Distribution
Track Phi Distribution
Track Phi Distribution
Track Phi Distribution
Track Phi Distribution
Track Phi Distribution
Track Phi Distribution
Track Phi Distribution
Estimated Track Efficiency
Hits on Track
Hits on Track
Hits on Track
Hits on Track
Hits on Track
Hits on Track
Hits on Track
Hits on Track
Hits on Track
Hits on Track
Impact Parameter
Impact Parameter
Impact Parameter
Impact Parameter
Impact Parameter
Impact Parameter
Impact Parameter
Impact Parameter
Impact Parameter
Impact Parameter
Summary
- We do not confirm the initially claimed $\sim 75\\%$ loss in efficiency in iTPC sector 20 - The effect is much less visible in properly calibrated data - We modeled track reconstruction without the inner hits using sector 14 - The average expected loss due to missing hits is $\sim 20\\%$ for global tracks with higher $P_T$ ($\gtrsim 0.4$ GeV) and an additional $\sim 7\\%$ loss for primary tracks - The $\sim 7\\%$ loss in primaries is consistent with the widening of the track impact parameter distribution.
As a result the tracks do not pass the 3 cm cut and are not assigned to the vertex - For lower $P_T$ ($\lesssim 0.4$ GeV) tracks we observe an increase in the number of short tracks in the adjacent sectors. The exact reason for such behavior is under investigation