1. Software News Takashi Hachiya RIKEN

2. Current status Status – All the updates are committed into CVS – The production macro for the full production is prepared. Update – Updating the algorithm on the SvxCentralTrackReco Reduce random association, update the chi2 calculation – SvxPrimVertexFinder Loose the track requirement if not found the prim vertex, but back to the normal cut – SvxSelectCluster Save the clusters only around the CNT track to re-do tracking if necessary – CompactCNT Convert DST to compact format. Cluster, Segment, CNT+BG is stored. Production Macro and output – Next page

3. Production Macro SvxParManager FullZero field SxvDecode includePixel(true); includeStripixel(true); setAdcOffset(24); setAdcCutoff(-24); includePixel(true); includeStripixel(true); setAdcOffset(24); setAdcCutoff(-24); SvxApplyHotDead N/A SvxReco set_UseStripThresholdDatbase(true); set_StripixelAdcSumThreshold(0); SvxPriVertexSeedFinder SvxStandaloneReco setWindowScale(3); setZerofieldFlag(true) SvxPrimVertexFinder SvxCentralTrackRecoRemove (no zero field treatment) SvxCentralTrackReco(BG)RndmAssocDchFlag(1)Remove (no zero field treatment) SvxRpSumXYReco Remove SvxSelectClusters Remove FillSvxHitssetSaveOnlySelectedHits(true)setSaveOnlySelectedHits(false)

4. QA code Commenttest scanFull FieldZero field SvxAlignment_QAQA for alignmentNoYes SvxFillHistoPixelYesNo SvxFillHistoStripYesNo SvxFillHistoBeamCenterBeamCenter checkYes SvxStripClusterQA.CCluster check (John C)NoYes SvxQAForProduction.hAll SVX checkNoYes SvxCntQACNT check in detailNo(next yes)YesNo SvxRpEventQA.hRP check (Nakagomi)NoYes SvxStripEventQA.hCellID check Write_StripQAH isto(1)  SvxStripFindHotDeadStrip hotdeadYesNo

5. OutputNode VTX_DST SvxEventInfo SvxSelectedClusterList SvxSegmentList SvxCentralTrackList SvxQAInfo SvxCentralTrackBackList BbcOut BbcRaw EventHeader Sync TrigLvl1 PreviousEvent ErtOut VtxOut SvxClusterList(Zero field only) CNT_DST RpSumXYObjectPHGlobal PHGlobal_CENTRAL SvxHit_VarArrayRpSumXYObject SvxTrack_VarArrayDchHit_VarArray SvxCentralTrack_VarArrayEmcHit_VarArray SvxCentralTrackBack_VarArrayPc1Hit_VarArray Pc2Hit_VarArray EventHeaderPc3Hit_VarArray SyncTofeHit_VarArray TrigLvl1TofwHit_VarArray PreviousEventCrkHit_VarArray ErtOutAccHit_VarArray VtxOutCglTrackHits_VarArray CglTrackBackHits_VarArray TrackProjection_VarArray TrackLineProjection_VarArray TrackPathLength_VarArray emcHitContainer

6. SvxCntQA Cut condition – |zvtx| =4 Contents – CNTTRK : SvxCentralTrack ntuple Nentry: NCNT, (10% below pT<0.5GeV/c) Content: event info(vtx, bbcq), CNT(dch, emc, rich, pc2/3), SVXCentralTrackInfo(dphi/dz, dca), siminfo(simdca, simpt, vx/y/z) – EVT: event based ntuple Nentry: Nevent Content: zvtx, bbcq, nhit at layer, nseg – Cluster: cluster based ntuple Nentry: Ncluster with early 100 event – SATrack: SATrack based ntuple Nentry: N_SATrack with early 100 event

7. Some plots from ntuple

8. Issues Check the compactCNT data if it is really working. – Read and check the compactCNT. – Then, can start Data QA. Chi2 calculation (fitting) on the SvxCentralTrackReco will be updated We need to break the tasks into people. – We have a lot of tasks DataQA/Signal Ext./Sim. And so on (written in WIKI) – Need to do the task which might not be related to your poster/talk topics. – Share same code/knowledge/analysis technique to speed up the analysis.

9. Update on the tracking algorithm Current Algorithm Most outer layer – 4mm in dphi – 3cm in dz Correlation cut – 4mm in dphi – 1cm in dz Hit requirement – More than 2hits Priority on the track selection 1.Large Nhit (4  3  2) 2.Minimum chi2/ndf Chi2 calculation – Correlation method DCA – CNT track vector at B0 Others Proposed Algorithm Most outer layer – 4mm in dphi – 10mm in dz (depending on which vtx is used) Correlation cut – 0.5 ~ 3mm in dphi (0.5 for B1-B0) – 0.7 ~ 3mm in dz (0.7 for B1-B0) Hit requirement – More than 2hit – At least 1 hit at the strip layer (B2 or B3) Priority on the track selection 1.3 or 4 hit more prefer than 2 hit 2.Minimum chi2/ndf Chi2 calculation – Fitting method DCA – track vector by fitting Others – If B3-B2-B1-B0, B2-B1-B0 is also checked 2012/5/2Analysis meeting9

10. Updating After last meeting, I checked some distributions from data and Sasha made DCA of B  e from the embedding study. – Using the updated library – Discuss with Yasuyuki and Maya. It is found that the most of the track associated with the VTX clusters are background (random association) – Large chi2 – pT distribution – Many tracks at the large DCA. Update on the tracking algorithm – Apply narrow association window – Update Chi2 calculation method. 2012/5/2Analysis meeting10

11. Comparison of Chi2 distributions before and after update Chi2/ndf (old) Chi2/ndf(new) 2/3 * Chi2/ndf (new) Chi2/ndf Old version Large chi2 tail Updated version No large chi2 tail Chi2/ndf Left – Right comparison – Requiring narrow association windows Wide(left) – Narrow(right) – Large chi2 tail is disappeared Bottom-Right (both right) – Old calculation method – New calculation method (2/3* new chi2) Chi2 is improved. – new chi2 looks the exponential curve and better property. But slightly wider than the current. – new chi2 is good enough for the track reconstruction (pattern recognition) – But can be improved more. (in recalibrator) Sigma can be improved. 2012/5/2Analysis meeting11 Large ch2 tail from BG before After

12. Embedding with new SvxCentralTrackReco DCA cut of 500mkm kills ~0.5 electrons from bottom, and ~0.98 electrons from primary vertex. 12 MC HF Reconstructed HF Reconstructed from primary vertex bottom and primary electrons embedded in 0-20% central AuAu nhits≥3 && chi2/ndf<3

13. pT distribution 1.West (-0.8<phi0<1.2) 2.1+ Vtx associated (nhit>=2) 3.2+ b0 hit 4.2+Nhit>=3(up) 5.4+chi2/ndf<5 pT(GeV/c) Using the updated tracking. – After many BG are reduced by narrow cut. – Nhit>2&b0 hit requirement reduces the yield to 1/10. Compare 1,2,3 – It looks that Chi2 reduce the random background. Compare 4 and 5 2012/5/2Analysis meeting13

14. DCA (data in run 349369 100kevent) 1.All pT a.Chi2(old)/ndf<3 R=9687/61440=0.157 b.chi2(new)/ndf<4.5 R=8224/53901=0.152 2. pT>1GeV a.Chi2(old)/ndf<3 R=639/14835=0.043 b.chi2(new)/ndf<4.5 R=530/12955=0.041 3. pT>2GeV a.Chi2(old)/ndf<3 R=37/1325=0.028 b.chi2(new)/ndf<4.5 R=50/1254=0.0399 In DCA distribution DCA with new chi2 cut is narrower than that with current one and BG is reduced. – BG (|DCA|>1mm) / Real Track (|DCA|<1mm) ~ 4% – Background yield is reducing for higher pT. DCA(cm) 2012/5/2Analysis meeting14

15. Hadron: |zvtx| =3&&b0 and b1 hit and chi2/ndf 1GeV Electron Hadron + n0>2 + chi2/npe 0.7 yield |dca|<1mm Hadron : 9892 Electron : 20 Ratio(ele/had) = 2.0e-3 Expected ratio of electron to hadron : 2.4e-3 at pT=1GeV/c – PHENIX electron and pion measurement. If eID efficiency is 80%, the ratio (electron/hadron) in data (2e-3) is good agreement with the expected value – But the statistics is too small in this analysis 2012/5/2Analysis meeting15 DCA Distribution

16. Update on the chi2 calculation 2012/5/2Analysis meeting16

17. Update on the Chi2 calculation CNT track B0 B1 B2 B3 DC dphi3 dphi2 dphi1 dphi0 Actual track DCA CNT track B0 B1 B2 B3 DC dphi3 dphi2 dphi1 dphi0 Actual track DCA 2012/5/2Analysis meeting17

18. Expected sigma of each layer (by Multiple Scattering) 4% X0  3MeV/c 2% X0  2MeV/c 170mm 110mm 50mm 25mm 50  p 85mm 170  p 145mm 290  p 60mm 120  p 120mm 240  p 60mm 120  p MS by B0 MS by B1MS by B2 Total B0 25mm 50  p 85mm 210  p 145mm 420  p 85mm 120  p 145mm 300  p Total B1 2012/5/2Analysis meeting18

19. hit correlation on the fitting We checked that the correlation between DCA and the residual at each layer – Negative slope at B0, but positive at the other layer DCA in the current fitting is affected by multiple scatting – Fitting can be improved including MS effect. Now, to take the MS effect into account, the correlation at B0 is rotated to be flat. – Red circle in B0. – We can check the (expected) DCA resolution. – Sigma_DCA = 127um  72um before/after the rotation. DCA vs fitdp0 DCA vs fitdp1DCA vs fitdp3 fitdp0fitdp1fitdp3 DCA B0 B1 B2 B3 Scattered track DCA Original Track direction 2012/5/2Analysis meeting19

20. DCA with BG The background shape is reproduced by the histogram (red) from 5degree method reproduces t DCA DCA with 5dgree rotation 2012/5/2Analysis meeting20

21. Summary I updated the tracking algorithm and parameters in the SvxCentralTrackReco – Narrow association windows – Fitting for track selection and chi2 calculation –.. The updated code is working well – Large chi2 problem is fixed – Smaller background yield ( from the electron / hadron ratio) This code is committed into CVS – Ready to use for the mini-production. Current status of the track fitting. DCA & chi2 – Currently simple chi2 fitting is used. The result is good enough. – This could be refined in the recalibrator I would like to ask Sasha to do embedding study of B  e again. 2012/5/2Analysis meeting21

22. Next step Improve the chi2 calculation 2012/5/2Analysis meeting22

23. 2012/5/2Analysis meeting23

24. Multiple Scattering の効果は相関スロープの角度として現れる。 運動量毎の相関を見ると Slope が変わるのが見える。 – Multiple scattering を含めた Fitting が必要。 DCA vs fitdp0(0.5<p<1) fitdp0 DCA DCA vs fitdp0(1<p<1.5) fitdp0 DCA DCA vs fitdp0(1.5<p<2) fitdp0 DCA 2012/5/2Analysis meeting24