Hit by hit and track by track analyses for AuAu200_2016 sample processed with current dev and SL24y STAR library versions.
Legend
- Samples:
- "old" :Sep 15 02:40 /star/rcf/test/dev/daq_sl302.stica/Sat/year_2016/AuAu200_production_2016_64bit/st_physics_17078006_raw_1000030.event.root
- "new" Sep 12 14:08 /star/rcf/test/gitdev/daq_sl302.stica/Wed/year_2016/AuAu200_production_2016_64bit/st_physics_17078006_raw_1000030.event.root
- Goals of this study is to see any effects of code changes (libraries dev => Sl24y) due to new alignment schema (Alignmnet2024) and Sti package (new hit errors and sharing hits between tracks) for the dataset, which is processed with old bfc chain option.
To see :
- difference in hit reconstruction,
- difference in track and vertex reconstruction :
- What is the track parameters and their errors dependence on the track kinematics (φ, η, 1/pT) ?
- What is difference in the primary vertex positions?
- What global and primaries track relative reconstruction efficiency of "new" production with respect to "old" one?
- Notations:
- Filled marker for "new" production, opened marker for "old" one.
- Black marker for positive tracks, red markers for negative tracks.
- Matching:
- The hits from old and new production are considered as matched if they have Δpad2 + ΔtimeBucket2 < 4.
- The tracks are considered as matched if the number of matched hits from the tracks is > 5 and > 0.5 × Min(no. of fit points on trackOld or trackNew).
- Efficiency, lost and clone rates:
- The efficiency is calculated for the matched track pairs, i.e. tracks with one to one match or if there is a multiple matches than it is taken the first matched track only.
- The clones are tracks with more than 1 matched track.
- The lost tracks are ones without match.
Track by track analysis
Hit by hit analysis
The conclusions from hit by hit analysis:
- Practically no difference in X and Y (Fig. 1-2, σX and σY < 2 μm). The X and Y coordinates are mainly affected by code changes for new alignment. This changes do not affect these coordinates.
- There is a difference in dZ (Fig. 3) with ~3% of hits with |dZ| > 100 μm. For both inner (Fig. 4-6) and outer (Fig. 7-9) sectors the difference is associated with hits
produced between the Gating Grid and anode wires (prompt hits). The prompt hits is a tricky business.
These hits are sitting on anode wire and "drift hit" approach does not work. This difference will be address in the next release.
Track by track analysis
The conclusions from track by track analysis:
- The new (SL24y) production has in average ~0.7 more hits on track than the old (dev) one (fig.1). This is due to that in new production is allowed to share hits between tracks which was forbidden in old one.
- There is no difference in momentum reconstructed for both global (Fig. 2) and primary (Fig. 3) tracks.
- Practically there is no difference in reconstructed primary vertex position (Fig. 4-6).
- The track reconstruction efficiency (versus pT, η, φ) for global (Fig. 8-10) and primary (Fig. 11-13) shows increase by ~4%. I attribute this to hit sharing which increases reconstruction efficiency and reduce its dependence on multiplicity.
- The clone rate is on the level (global tracks Fig. 13-15, and primary tracks Fig. 16-19) is on the level on ~0.1%.
- The overall primary efficiency on the level ~90% is pointing to reassigning the global tracks to the primary vertex due to hit sharing.
- The rate of the lost tracks (Fig. 20-22 and Fig. 23-25) has confirm the conclusion for the track reconstruction efficiency.
- The difference in the track reconstruction efficiency (~4%) has explained observed increase of K0s yield by 10% observed in the KFParaticle analysis (https://www.star.bnl.gov/~fisyak/star/Tpc/Alignment/kfp/) of P24ia and TFG24c/d productions.
Conclusion
- The codes changes due to new alignment schema (Alignment2024) with "old" alignment do not affect the drifting hits. There is an effect on the prompt hits which should be addressed later.
- No changes in the track parameters have been observed.
- The hit sharing between tracks has been lost during the last merging with TFG (~2014). The hit sharing adds ~4% to track reconstruction efficiency and this is main source of ~10% increase of yield in the reconstruction of strange particle decays observed in comparison P24ia and TFG24c/d productions.
- The rest 2% of yield are coming from upgrade TPCCATracker and related changes PR#25 made by Hongwei Ke on June 9, 2021.