SST hit use studies...
Updated on Mon, 2017-02-27 18:09. Originally created by jwebb on 2017-02-27 15:35.
A quick review of the SST hit inclusion studies done ~1 year ago.
I could not find any studies which examined QA/efficiency of tracks with only SST, or SST*IST hits.
Presentation at the collaboration meeting (1/2016) indicated that we looked at HFT tracks with as few as one SST hit. But we included other hits as they were encountered. The summary table is on slide 9. NOTE: Last column says that we looked at SST hit OR 3 HFT hits... but I don't see the plots in my archive.
https://drupal.star.bnl.gov/STAR/system/files/STAR%20Offline%20Software%20Presentation%20v10.pdf
This study indicated that all of the possible 3-layer HFT combinations gave comparable quality:
http://www.star.bnl.gov/~jwebb/2015/Sti_BaseQA/BaseQA_Sti_08_27_2015_3layer.html
Additional studies, linked below, examined different combinations including the SST and IST as sufficient to form an HFT hit. But additional hits were accumulated when they were found. So... we haven't really done a study showing that the SST or IST alone is sufficient.
Only thing I found showing IST only tracking was the initial study where we decided to require IST*PXL1*PXL2 in 2014. This showed that combinations of IST*PX1 and IST*PX2 were problematic w/respect to track quality.
http://www.star.bnl.gov/~jwebb/2015/Sti_BaseQA/BaseQA_Sti_02_18_2015.html
So bottom line is, from the studies I can track down, we do not have anything which looks at tracks with just IST * SST.
Studies done on 8/25/2015 show efficiencies for combinations where one SST hit, or one SST and one IST hit, are sufficient to form an HFT track. The remaining hits in the HFT are used if found.
http://www.star.bnl.gov/~jwebb/2015/Sti_BaseQA/BaseQA_Sti_08_25_2015.html
We evaluated 8 different combinations of hits for HFT tracking:
Columns 7 and 8 added in tracks which had hits in SST*IST and SST alone. These two classes of tracks degraded track quality in some worrisome ways:
Fig 13 -- chi2_vx for primaries shows an asymmetry vs eta
Fig 27 -- DCAz increases for small number of fit poitns (<15)
Fig 33 -- pT resolution degrades for these two classes (but this is expected and likely reasonable)
I conclude from this that the SST only and SST*IST only tracks are degrading the track quality slightly.
Efficiency plots show a few things of interest... should compare columns 6 (any 3), 7 (only SST required), and 8 (only SST and IST required):
Fig 155 -- Acceptance vs eta
Fig 158 -- Efficiency vs eta (normalized to all tracks thrown)
Fig 164 -- Efficiency vs eta (normalized to tracks in acceptance)
Any 3 HFT hits -- 10% overall efficiency, 85% normalized to acceptance
only SST and IST required -- 20% overall efficiency, ~70% normalized to acceptance
only SST required -- 30% overall efficiency, ~70% normalized to acceptance...
However, we threw MC events with wide vertex... real data trigger constrains the vertex to be w/in the pixel acceptance. So my guess is that we would stand to gain less than what is shown here.
Open question remains whether SST*IST hits can help in pp where we don't have pixel in the daq file...
Update on 9/3/2015:
I could not find any studies which examined QA/efficiency of tracks with only SST, or SST*IST hits.
Presentation at the collaboration meeting (1/2016) indicated that we looked at HFT tracks with as few as one SST hit. But we included other hits as they were encountered. The summary table is on slide 9. NOTE: Last column says that we looked at SST hit OR 3 HFT hits... but I don't see the plots in my archive.
https://drupal.star.bnl.gov/STAR/system/files/STAR%20Offline%20Software%20Presentation%20v10.pdf
This study indicated that all of the possible 3-layer HFT combinations gave comparable quality:
http://www.star.bnl.gov/~jwebb/2015/Sti_BaseQA/BaseQA_Sti_08_27_2015_3layer.html
Additional studies, linked below, examined different combinations including the SST and IST as sufficient to form an HFT hit. But additional hits were accumulated when they were found. So... we haven't really done a study showing that the SST or IST alone is sufficient.
Only thing I found showing IST only tracking was the initial study where we decided to require IST*PXL1*PXL2 in 2014. This showed that combinations of IST*PX1 and IST*PX2 were problematic w/respect to track quality.
http://www.star.bnl.gov/~jwebb/2015/Sti_BaseQA/BaseQA_Sti_02_18_2015.html
So bottom line is, from the studies I can track down, we do not have anything which looks at tracks with just IST * SST.
Studies done on 8/25/2015 show efficiencies for combinations where one SST hit, or one SST and one IST hit, are sufficient to form an HFT track. The remaining hits in the HFT are used if found.
http://www.star.bnl.gov/~jwebb/2015/Sti_BaseQA/BaseQA_Sti_08_25_2015.html
We evaluated 8 different combinations of hits for HFT tracking:
- Column 1: old PXL1 * PXL2 * IST
- Column 2: eval PXL1 * PXL2 * IST * ( 1 + SST )
- Column 3: eval PXL1 * PXL2 * SST * ( 1 + IST )
- Column 4: eval PXL2 * IST * SST * ( 1 + PXL1 )
- Column 5: eval PXL1 * IST * SST * ( 1 + PXL2 )
- Column 6: eval Any 3 HFT layers
- Column 7: eval SST * IST * ( 1 + PX1 + PX2 )
- Column 8: eval SST * ( 1 + IST + PX1 + PX2 )
Columns 7 and 8 added in tracks which had hits in SST*IST and SST alone. These two classes of tracks degraded track quality in some worrisome ways:
Fig 13 -- chi2_vx for primaries shows an asymmetry vs eta
Fig 27 -- DCAz increases for small number of fit poitns (<15)
Fig 33 -- pT resolution degrades for these two classes (but this is expected and likely reasonable)
I conclude from this that the SST only and SST*IST only tracks are degrading the track quality slightly.
Efficiency plots show a few things of interest... should compare columns 6 (any 3), 7 (only SST required), and 8 (only SST and IST required):
Fig 155 -- Acceptance vs eta
Fig 158 -- Efficiency vs eta (normalized to all tracks thrown)
Fig 164 -- Efficiency vs eta (normalized to tracks in acceptance)
Any 3 HFT hits -- 10% overall efficiency, 85% normalized to acceptance
only SST and IST required -- 20% overall efficiency, ~70% normalized to acceptance
only SST required -- 30% overall efficiency, ~70% normalized to acceptance...
However, we threw MC events with wide vertex... real data trigger constrains the vertex to be w/in the pixel acceptance. So my guess is that we would stand to gain less than what is shown here.
Open question remains whether SST*IST hits can help in pp where we don't have pixel in the daq file...
Update on 9/3/2015:
QA of HFT tracks (any 3-layer combination) vs TPC-only tracks which did not find 4 HFT hits --
http://www.star.bnl.gov/~jwebb/?dir=2015/Sti_BaseQA/&file=BaseQA_Sti_08_27_2015_any.html
QA of HFT tracks (specific 3-layer combination) vs TPC-only tracks which did not find 4 HFT hits --
http://www.star.bnl.gov/~jwebb/?dir=2015/Sti_BaseQA/&file=BaseQA_Sti_08_27_2015_sst
http://www.star.bnl.gov/~jwebb/?dir=2015/Sti_BaseQA/&file=BaseQA_Sti_08_27_2015_ist.html
http://www.star.bnl.gov/~jwebb/?dir=2015/Sti_BaseQA/&file=BaseQA_Sti_08_27_2015_px2.html
http://www.star.bnl.gov/~jwebb/?dir=2015/Sti_BaseQA/&file=BaseQA_Sti_08_27_2015_px1.html
»
- jwebb's blog
- Login or register to post comments