07 Data analysis

run lists, key intermedaite results, mostly for real data analysis

01 run QA

Run QA of pp500 2012 data will be W-analysis oriented. We will focus only on detectors/problems relevant for W analysis.

The simplifed list of 2-letter key-words is below. In general keys have only 4 values: good, acceptable,fatal, don't-know (not all present). Multiple keys should be separated by white space (do NOT use , )

?? - no one checked this run
???? - * and runs macred as questionable or bad by the shift leader

Ok - all seems to be reasonable for this run

Xx - *never use this run for any analysis. If runs are marked "junk","bad" or questionable by shiftleader, use your judgement as to whether these are analyzable at least for some detector and you wish to over rule them.

Tm - TPX minor problems
Tx -* TPX unusable
T? - TPX problem I do not understand

Em - ETOW some tower bad but most were working
Ex -* ETOW towers unusable, e.g. all corrupted or off

Mm - EEMC some MAPMT box was bad but mostly usable
Mx -* EEMC MAPMTs not usuble this run
M? - EEMC MAPMT problem I do not understand

Bx -* BEMC towers unusable, e.g. all corrupted or off
Bm - BEMC some towers have problem but mostly usable
B? - BTOW MAPMT problem I do not understand

Pc - data taken during polarization measurement

Qx - bunch crossing or spin pattern problems
Tr - trigger rate anomalies for small part of run as seen in the rate plots

Tb -* L2WB low yield

Te -* L2WE low yield

Oe - TOF problem at the end of run

Om - TOF errors on JPLOT figure

Ox -* TOF not usable
O? - TOF problem you do not understand
Sm - BSMD minor problem
Sx - * BSMD not usable
S? - BSMD problem you do not understand

Gx - * FGT not usable
Ux - * missing muDst or run not produced , typically for 2011 data
Zm - ZDC scaler attenuator was inserted, different reference rate, corrected in the table using Gene's formula: (0.09/0.005) *ZDCX_days85,86 = ZDCX_other_days

Jx - discared based on muDst content (track, vertex,etc QA) by Justin, details are here http://drupal.star.bnl.gov/STAR/node/25066/

Obsolete: u1 - * muDst were produced by Jan+Justin during data taking and reside at MIT or IUCF disc

Comments:

FGT QA: it will not be QA'ed by humans. If at a later stage FGT will be found usefull then an automat will crawl over root files form JPlot and extract relevant info.
The main emphasis of human run QA should be on reading the shift log or inspecting the trigger rate plots. This is hard to automatize
results available 2-3 weeks after the end of the assigned week.
each person should use the ";" table prepared by Justin - it is avaliable as 'link' in the table below.
keys tagged with * can be filled by automatic run QA

The output of your QA should be 3 columns:

runNumber, keys separated by spaces, additional remarks you may have

Table 1

Run QA assignment of 2012 pp data
Dates	Lt (from Jamie)	Run ranges	Person Name	# QA runs	QA file	Completion date
Mar 17 - 23	~ 13/pb	R13077059-R13083084	Jan	144	-	May 25
Mar 24 - 30	~ 13/pb	R13084001-R13090050	Hal	130	-	May 7
Mar 31 - Apr 6	~ 20/pb	R13091001-R13097052	Hal	152	-	July 19
Apr 7 - 13	~ 22/pb	R13098026-R13104063	Hal	150	-	Aug. 9
Apr 14 - 2?	~ 18/pb	R13105003-R13109042	Justin	149	-	June 5

OLD raw tables from Justin: link1, link2, link3, link4, link5

Attachment A) ver 1 run list for whole pp510 2012 running time , CSV table (gzipped) , automatically QAed ,

rejected (Xx) production runs: 139
non-rejected production runs : 755, contain 8.1M L2WB events and 2.7M L2WE events

Attachement C) ver 1 run list for whole pp500 2011 running time , CSV table (gzipped) , automatically QAed

ppTrans data cover days 38-98, Ok'ed only automatically (no human inspection)
- Ok'runs: 924, contain events: L2WB=3.6M, L2WE=1.3M
ppLong data cover days 99-108, automatically QA'ed, still needs human inspection
- non-rejected runs: 196, contain events: L2WB=1.5M, L2WE=0.4M

TABLE 2
runQC_2012_day77_109

JAN:
+F: 16582 R13077059 ,  nRunOk= 22 , eve sum= 165488 , nFill= 0 delEve= 165488
+F: 16586 R13078024 ,  nRunOk= 42 , eve sum= 355145 , nFill= 1 delEve= 189657
+F: 16587 R13079013 ,  nRunOk= 58 , eve sum= 516386 , nFill= 2 delEve= 161241
+F: 16592 R13079069 ,  nRunOk= 69 , eve sum= 729752 , nFill= 3 delEve= 213366
+F: 16593 R13080008 ,  nRunOk= 74 , eve sum= 826984 , nFill= 4 delEve= 97232
+F: 16594 R13080078 ,  nRunOk= 92 , eve sum= 1018269 , nFill= 5 delEve= 191285
+F: 16597 R13081003 ,  nRunOk= 95 , eve sum= 1096636 , nFill= 6 delEve= 78367
+F: 16602 R13081018 ,  nRunOk= 109 , eve sum= 1320925 , nFill= 7 delEve= 224289
+F: 16618 R13083008 ,  nRunOk= 125 , eve sum= 1425379 , nFill= 8 delEve= 104454
+F: 16619
+F: 16620 R13084021 ,  nRunOk= 158 , eve sum= 1905415 , nFill= 10 delEve= 259299

HAL:
+F: 16622 R13084045 ,  nRunOk= 169 , eve sum= 2115063 , nFill= 11 delEve= 209648
+F: 16625 R13085023 ,  nRunOk= 183 , eve sum= 2327822 , nFill= 12 delEve= 212759
+F: 16626 R13085058 ,  nRunOk= 188 , eve sum= 2366182 , nFill= 13 delEve= 38360
+F: 16627 R13086064 ,  nRunOk= 206 , eve sum= 2605485 , nFill= 14 delEve= 239303
+F: 16632 R13087008 ,  nRunOk= 220 , eve sum= 2799645 , nFill= 15 delEve= 194160
+F: 16633 R13087046 ,  nRunOk= 232 , eve sum= 2939987 , nFill= 16 delEve= 140342
+F: 16643 R13089015 ,  nRunOk= 241 , eve sum= 3043785 , nFill= 17 delEve= 103798
+F: 16650 R13090003 ,  nRunOk= 259 , eve sum= 3276440 , nFill= 18 delEve= 232655
F: 16655 R13090034 ,  nRunOk= 274 , eve sum= 3441661 , nFill= 19 delEve= 165221

Hal for WILL:
+F: 16656 R13091018 ,  nRunOk= 292 , eve sum= 3727202 , nFill= 20 delEve= 285541
+F: 16657 R13091048 ,  nRunOk= 296 , eve sum= 3782790 , nFill= 21 delEve= 55588
+F: 16659 R13092004 ,  nRunOk= 300 , eve sum= 3872553 , nFill= 22 delEve= 89763
+F: 16662 R13092014 ,  nRunOk= 317 , eve sum= 4065611 , nFill= 23 delEve= 193058
+F: 16667 R13093006 ,  nRunOk= 336 , eve sum= 4304284 , nFill= 24 delEve= 238673
+F: 16668 R13093043 ,  nRunOk= 358 , eve sum= 4622506 , nFill= 25 delEve= 318222
+F: 16669 R13094033 ,  nRunOk= 367 , eve sum= 4737795 , nFill= 26 delEve= 115289
+F: 16671 R13094078 ,  nRunOk= 388 , eve sum= 5078514 , nFill= 27 delEve= 340719
+F: 16678 R13095041 ,  nRunOk= 403 , eve sum= 5265207 , nFill= 28 delEve= 186693
+F: 16685 R13096055 ,  nRunOk= 421 , eve sum= 5569174 , nFill= 29 delEve= 303967
+F: 16686 R13097011 ,  nRunOk= 440 , eve sum= 5891274 , nFill= 30 delEve= 322100

Hal for BERND:
+F: 16690 R13098026 ,  nRunOk= 456 , eve sum= 6112638 , nFill= 31 delEve= 221364
+F: 16691 R13098060 ,  nRunOk= 471 , eve sum= 6384325 , nFill= 32 delEve= 271687
+F: 16693 R13099025 ,  nRunOk= 488 , eve sum= 6706365 , nFill= 33 delEve= 322040
+F: 16697 R13099055 ,  nRunOk= 505 , eve sum= 7029246 , nFill= 34 delEve= 322881
+F: 16698 R13100022 ,  nRunOk= 522 , eve sum= 7309288 , nFill= 35 delEve= 280042
+F: 16699 R13100048 ,  nRunOk= 538 , eve sum= 7614720 , nFill= 36 delEve= 305432
+F: 16701 R13101011 ,  nRunOk= 548 , eve sum= 7747627 , nFill= 37 delEve= 132907
+F: 16704 R13101037 ,  nRunOk= 561 , eve sum= 8003709 , nFill= 38 delEve= 256082
+F: 16705 R13101055 ,  nRunOk= 564 , eve sum= 8053633 , nFill= 39 delEve= 49924
+F: 16710 R13103002 ,  nRunOk= 576 , eve sum= 8275155 , nFill= 40 delEve= 221522
+F: 16716 R13104001 ,  nRunOk= 585 , eve sum= 8400716 , nFill= 41 delEve= 125561
+F: 16717 R13104017 ,  nRunOk= 588 , eve sum= 8458036 , nFill= 42 delEve= 57320
+F: 16720 R13104043 ,  nRunOk= 606 , eve sum= 8780320 , nFill= 43 delEve= 322284

JUSTIN:
+F: 16722 R13105003 ,  nRunOk= 623 , eve sum= 9046831 , nFill= 44 delEve= 266511
+F: 16723 R13105037 ,  nRunOk= 642 , eve sum= 9337249 , nFill= 45 delEve= 290418
+F: 16724 R13105060 ,  nRunOk= 658 , eve sum= 9636012 , nFill= 46 delEve= 298763
+F: 16725 R13106023 ,  nRunOk= 677 , eve sum= 9877899 , nFill= 47 delEve= 241887
+F: 16726 R13106056 ,  nRunOk= 693 , eve sum= 10086950 , nFill= 48 delEve= 209051
+F: 16727 R13107012 ,  nRunOk= 712 , eve sum= 10357967 , nFill= 49 delEve= 271017
+F: 16730 R13107051 ,  nRunOk= 726 , eve sum= 10522171 , nFill= 50 delEve= 164204
+F: 16731 R13108022 ,  nRunOk= 739 , eve sum= 10697106 , nFill= 51 delEve= 174935
+F: 16732 R13108069 ,  nRunOk= 746 , eve sum= 10768736 , nFill= 52 delEve= 71630
+F16735 R13109042 ,  nRunOk= 755 , eve sum= 10886591 , nFill= 53 delEve= 117855

#runs seen = 1069 ,  nRunOk= 755 , prod_Xx= 139 , eve in OK runs L2WB= 8141240  L2WE= 2745351  sum= 10886591

TABLE 3
runQC_2011_day99-108 (only ppLong pol data)

HAL:
+F15426
+F15427
+F15431
+F15435
+F15436
+F15438
+F15443
+F15444
+F15452
+F15457
+F15464
+F15466
+F15467
+F15470
+F15472

June 12, 2012

attachement D) contains 4 CVS tables with compact, only-good run lists, for 3 years we care. Below is the content. trigIDs should allow you for controlling W algos.

Sometimes some non-essential values are missing, e.g. JamieLumi - I'll try to compute equivalent values if needed using other yields

Note, 2/5 of 2012 runs is only machine QA'ed - the next deadline for Will and Bernd is this Friday. - VOID

02 Run 12 L2W Stream QA Summary

This page summarizes the Run 12 pp510 QA for the preliminary W analysis. The presentation below describes how runs were selected and some outlier runs that were removed from the final analysis list.

More information can be found at the following links:

Manual QA (lots of spreadsheets)

Automated QA (lots more plots)

Final list of 640 runs for analysis

03 Evaluation of Run 12 L2W preview production (P12ic) TPC calibrations

Here we present some plots from the recently produced L2W stream preview production announced here. Some details of Run QA and selection can be found here.

Note: Run 12 day 85-86, have a different SC&GL calibration than the rest of the run due to changes to the ZDC which required using a different scaler for the luminosity dependence during these days. This is evaluated some below.

Fig 1 Signed DCA: In the run QA linked above you will find the slide shown below, which shows the signed DCA for primary tracks from highest positively rank vertices (average vs. runindex) for the L2 barrel and endcap W trigger separately and Q+ and Q- separately. Days 85-85 are circled in red.

Fig 2 Charge ratio: In the plots below you see distributions of tracks which satisfy our track QA cuts of nHits, etc. and tracks with pt>10 GeV are candidates for Ws. In the left panel you see the distributions for days 85-86 which show lower yield of Q- global tracks around pT~15 GeV, and then an excess of Q- tracks at high pT (Note: this is not reflected in the charge separation of the final selected W candidates in Fig 3). The right panel shows the results for the remainder of the runs, and these distributions match our expectation from previous productions (ie. Run 9 pp500).

Fig 3 Charge Separation: The Q/pT is shown for global and primary tracks for events witch satisfy our W requirements. For the left panel the charge sign separation looks very good, and sufficient for the W analysis. For days 85-86 the global track distribution looks strange, and the primary track distribution seems to be shifted towards higher Q/pT, with respect to our expectation from the other days. Additionally, I've made this plot of the charge separation for each TPC sector independently in the attachment below and plots of the sDCA for each sector attachment below.

Fig 4 Reconstructed Zs: As a cross check of the charge sign separation it is useful to look at the number of Z candidate events which have the same charge sign, as this would mean either one of the charge signs from a real Z was misreconstructed or it was a background QCD event. In both the panels below we see no Z candidates with the the same charge sign pairs.

From Jan:

This summary based on Justin's sDCA plots per sector. For the West TPC we do have sin-wave pattern of sDCA vs. sector phi - this may well be due to beam line x0,y0 being off the correct value . For the East sector 21 is clear outlier. If this gets corrected there is almost no sin-wave residua. Rather all East sectors have one common offset of sDCA.
Fig 4.

04 testing eta-dependent spin sorting (A)

I'll test here macro computing asymmetry .

Generate high statistic M-C sample to verify the simultaneously reco AL and ALL do not interefere. Below you see reco SSA, DSA for 100k events per STAR eta bin, pol=0.6 for both beams. The red and green lines overlayed on data points show AL and ALL used by event generator. It agrees good enough.
Fig 1.

******* W(eta) summary for charge=P  INPUT=run12toySetB *********
lumi-corrections:  0.968, 0.997, 1.009, 1.027,  applied
star-bin, sum , yield ++ +- -+ --  ,  1/sqrt(sum)
1  100000, 34711  18631  33511  13674  ,  0.003
2  100000, 40620  19145  21934  18976  ,  0.003
3  100000, 40366  22088  19301  18945  ,  0.003
4  100000, 34876  33862  18369  13606  ,  0.003
5   20000,  5776   7370   3747   3199  ,  0.007
6   20000,  3970   7333   4181   4510  ,  0.007
7   40000,  9746  14703   7928   7709  ,  0.005
8  400000, 150574  93726  93116  65202  ,  0.002
Spin results: pol1=0.60  pol2=0.60
polBeam-bin, events, *** AL ***,sig*sqrt(M)
10   40000   -0.197 +/- 0.009  nSig=22.0 ,  1.79
11   20000   -0.308 +/- 0.013  nSig=24.2 ,  1.80
12   20000   -0.087 +/- 0.012  nSig=7.3 ,  1.68
13  200000   0.099 +/- 0.004  nSig=25.3 ,  1.75
14  200000   0.310 +/- 0.005  nSig=59.4 ,  2.33
15  200000   0.403 +/- 0.006  nSig=66.7 ,  2.70
16  200000   0.602 +/- 0.008  nSig=75.0 ,  3.59
17   20000   0.514 +/- 0.014  nSig=36.3 ,  2.00
18   20000   0.218 +/- 0.012  nSig=17.8 ,  1.73
19   40000   0.366 +/- 0.010  nSig=35.9 ,  2.04
20  400000   0.353 +/- 0.006  nSig=56.2 ,  3.97
polBeam-bin, events,*** ALL *** ,sig*sqrt(M)
15  200000   0.503 +/- 0.009  nSig=58.5 ,  3.84
16  200000   -0.104 +/- 0.006  nSig=16.4 ,  2.83
17   20000   -0.296 +/- 0.020  nSig=14.7 ,  2.85
18   20000   -0.422 +/- 0.021  nSig=20.4 ,  2.93
19   40000   -0.359 +/- 0.015  nSig=23.8 ,  3.01
20  400000   0.200 +/- 0.006  nSig=36.0 ,  3.50
polBeam-bin, events, *** NULL ***  
15  200000   0.000 +/- 0.009  nSig=0.0 
16  200000   0.002 +/- 0.006  nSig=0.4 
17   20000   0.180 +/- 0.020  nSig=8.9 
18   20000   0.158 +/- 0.021  nSig=7.6 
19   40000   0.169 +/- 0.015  nSig=11.2 
20  400000   0.002 +/- 0.006  nSig=0.3 
******* end ************** charge=P  ********

05 W-algo pass-A through full 2012 data and MC

Here are the EtaBin sorted histograms for the data and MC /star/u/stevens4/forJan/8.28.12/

For data day 84-86 are rejected.

jba310 = W+
jba311 = W-
jba322 = Z/gamma*
jbb330 = filtered QCD

Also, standard movies for each sample and eta slice are linked from
http://drupal.star.bnl.gov/STAR/pwg/analysisstatushtml/w2012/preliminary-result-documentation

Attachements A-G,I-K show examples of critical plots for W-Algo, sorted by 6 eta bins and for 4 types of events: 2012 real data + MC: W+, W-, fileterd QCD

The plotting macro is in attachement H. The following relative normalization was used:

dataNameA[mxF]={"STAR data 2012", "Pythia W+", "filter Pythia QCD" , "Pythia W-"};

lumScale[mxF]={80., 192/0.65, 27*2.2, 198/0.84};

It was choosen to reproduce W+,W- eta-integrated yield.

06 2nd Endcap Bckg vs. EtaBin

To estimate QCD events passing W-algo with lepton limited to a narrow eta-range due to missing East Endcap we will use amount of events passing W-algo for the mirror eta-bin while West Endcap was disabled on purpose in the data pass.

07 Preview of W AL(eta) from 2012 data for spin PWG

Slides and writeups consistent with final eta-bin selection for 2012 W-data

08 Q/PT charge separation

Transformation Q/PT --> Q * ET/PT is applied to remove hyperbolic correlation between reco electron charge reco in TPC vs. reco ET from EMC.

The 5 eta bins cover electron rapidity ranges:

[-1,-0.5]
[-0.5,0]
[0,0.5]
[0.5,1]
[1,1.5] (all endcap)

Fig 1. The 2D distributions for pp500 2012 data are shown in the top row.

Bottom row shows projection of 'gold Ws' with ET [25,50] GeV fitted with gaussian, the mean and sigma of gauss are given on each plot.

Fig 2. Similar plots for Pythia W+ run thrugh BFC and best avaliable TpcRS simulation params. Clear difference in TPC reco accuracy is visible.

09 Alternative rel lumi monitors

Alternative methods of monitoring relative spin dependent luminosities for W AL analysis.

Alternative low Pt QCD events recorded in the W-stream were spin sorted. The rato of the yields for those alternatives differ by less than 0.5% from the one obtained using the default rel lumi monitor events used in W AL.

Since the magnitude of W AL is of 0.1, the stability of lumi monitor of 0.005 is sufficient to be neglacted in the error propagation.

/* mapping of spin4-index to helicity at STAR */

ka=10, /* STAR pol B+ Y + */

kb=9, /* STAR pol B+ Y - */

kc=6, /* STAR pol B- Y + */

kd=5, /* STAR pol B- Y - */

Fig.1 Top row: left - defaul rel lumi monitor: events with flipped isolation cut and ET<20 GeV.

Middle - event count stored in the ttree. Right - for events with reco prim vertex and at least 1 TPC track w/ pT >10 GeV/c I pick the highest 2x2 ET matched to such 10+ GeV/c track.

Bottom row - 3 subset of events: Left - all with a 10+ GeVc track, middle - additionally 2x2ET=[5,10] GeV, right - as before but 2x2 ET=[10,14] GeV.

Plots show ratio of yileds to the top left plot (the default) - deviatins from a constant are on the level of 0.5%.