Heavy Ion DiMuon? Trigger Study 2

L3 HLT reconstruction improvement

The problems

• 1) Too many warning message printing at L3 muon reconstruction
  • CAIR error at online processing, which is big problem real data taking HI events on the HI HLT team.

         %MSG-w TwoTrackMinimumDistance:  TestMuL1L2Filter:hltHIMML3Filter  07-Dec-2009 14:25:45 CET Run: 1 Event: 54
         Computation HelixHelix::CAIR failed.

• 2) Efficiency drop fraction is about 50 %
  • Looking at Trigger efficiency of Upsilon 50k sample here
  • although there is L2 muons at STA muons, it can not find theFtsFound correctly well.

Improvement

• 1) the Item "1) of the problems" is ongoing to study with Olga
  • Idea : to find where is message come up and give it some cut to avoid it.
    • the place to show this message is while vertex reconstruction with RecoVertex?.fitter(TwoTracks?)
    • Tracing Vertex fitting information
      • this fitter is Kalman vertex fitter and the specific name is SequentialVertexFitter? here
      • During Sequential vertex fitting, it uses CrossingPtBasedLinearizationPointFinder? here
      • And then it call the TwoTrackMinimumDistance?.cc here, which investigates the status of two tracks such as concentric, external or internal circle of each tracks in the CMS magnetic field.
      • this investigation for the status of two tracks gives the CAIR error message
      • Confirmed this CAIR error message is correct and we removed this case is not applied to vertex fit.

• 2) the Item "2) of the problems"
  • The main reason to be dropped at L3 from L2 is to fail to generate the seed on the tracker.
  • Because of the wrong pT measurement at L2(STA), when we predict the position on the tracker with LUT, predicted position is not enough to find the existence of hits to be supposed seed.
    • remake the LUT on the seed generator
    • to widen the rphicut from 0.5 to ??

20100210 To do list

• Fitting with Profiling
  • PtGen? - PtL2? = a + b*PtGen, PtGen? = (a+PtL2)/(1-b), a : gradient, b : y intersection. => modified PtGen?.
  • deltaPhi vs PtGenModified?.

Study with 30k Upsilons

• Data Sample : 30k Upsilon (flat 0~30 GeV?/c pT, |rapidity|<2.4) at CMSSW_3_4_0_pre2
• separate region : |eta|<2.4, |eta|<0.7, 0.7<|eta|<1.2, 1.2<|eta|<1.6, 1.6<|eta|<1.9, 1.9<|eta|<2.4

1) Seed Efficiency (Definition of efficiency : L2/L1, Seed/L1, L3/L1)

Triggered events for [eta] < 2.4

Triggered events for [eta] < 0.7

[eta]<2.4	[eta]<0.7	0.7<[eta]<1.2
1.2<[eta]<1.6	1.6<[eta]<1.9	1.9<[eta]<2.4

2) the difference between Gen Muon pT and L2 Muon pT

• data location : rfio:/castor/cern.ch/user/d/dmoon/cms340pre2/HLT_Sig/HLT_HIDoubleMu_Upsi_Sig_e100k.root

Second try to look into the difference (Gen Pt - inner L2 Pt)with 100k Upsilon events(Updated 2010 03 03)

[eta]<2.4	[eta]<2.4 profiling Y	[eta]<0.7	[eta]<0.7 profiling Y
0.7<[eta]<1.2	0.7<[eta]<1.2 profiling Y	1.2<[eta]<2.4	1.2<[eta]<2.4 profiling Y

Second try to look into the difference with 100k Upsilon events(Updated 2010 03 03)

[eta]<2.4	[eta]<2.4 profiling Y	[eta]<0.7	[eta]<0.7 profiling Y
0.7<[eta]<1.2	0.7<[eta]<1.2 profiling Y	1.2<[eta]<2.4	1.2<[eta]<2.4 profiling Y

This is the first looking at their difference but it has the problem to get L2 information(through HICFTSfromL1L2?). just referencing.

[eta]<2.4	[eta]<2.4 profiling Y	[eta]<0.7	[eta]<0.7 profiling Y
0.7<[eta]<1.2	0.7<[eta]<1.2 profiling Y	1.2<[eta]<1.6	1.2<[eta]<1.6 profiling Y
1.6<[eta]<1.9	1.6<[eta]<1.9 profiling Y	1.9<[eta]<2.4	1.9<[eta]<2.4 profiling Y

-- DongHoMoon - 07 Nov 2009