ADM Nov-7-2003 G. Bernardi for CALGO CALGO TMBfix CALGO recommends to implement in the next...

39
ADM Nov-7-2003 G. Bernardi CALGO TMBfix CALGO recommends to implement in the next TMBfixing: 1) New Hot Cell Killer 2) Restore ICR region to p13 status 3) T42 in killing mode G. Bernardi for the CALGO group many thanks to: S. Beauceron, E. Busato, L. Duflot, T. Golling, L. Groer, J. Stark, S. Shary, J.R. Vlimant for the plots shown in this talk, and to all the CALGO/CAT

Transcript of ADM Nov-7-2003 G. Bernardi for CALGO CALGO TMBfix CALGO recommends to implement in the next...

ADMNov-7-2003

G. Bernardi for CALGO

CALGO TMBfixCALGO recommends to implement in the

next TMBfixing:

1) New Hot Cell Killer

2) Restore ICR region to p13 status

3) T42 in killing mode

G. Bernardi for the CALGO group

many thanks to: S. Beauceron, E. Busato, L. Duflot, T. Golling, L. Groer, J. Stark, S. Shary, J.R. Vlimant for the plots shown in this talk, and to all the CALGO/CAT members for their contributions

ADMNov-7-2003

G. Bernardi for CALGO

DQ_Calo Hot Cell Killer

Offline zero bias monitoring tool (dq_calo)

ASCII files with data quality information including bad cells list (mean energy E< -2.0 or > 2.5 GeV)

cal_corr_dst/KillBuggyCells package

Thumbnails re-reconstruction

Allows also to provide good CALO run-list

ADMNov-7-2003

G. Bernardi for CALGO

JETS RESULTS

• MHT30_3CJT5 trigger stream– L1: at least 3 calorimeter trigger towers with

ET>5 GeV. – L2: Require jet MHT>20. – L3: Vector Ht sum >30 GeV

• trigger List v12.10• runs: 178096 – 178991 • p14.03.01 • 453 471 events• JCCB jets• in average 2 – 6 bad cells killed per run and

0 – 2 of them are hot (with large occupancy).

ADMNov-7-2003

G. Bernardi for CALGO

PT distribution before and after (the jets quality criteria are not

applied)

!

ADMNov-7-2003

G. Bernardi for CALGO

Reducing influence of Noise: T42T42: reject ALL isolated cells below 4

sigmas (and ALL negative energy cells)

• Select high signal cells (+4 sigmas)

• Keep their significant neighbours (2-4 sigmas) Thresholds “2” is 2.5 at the moment

T42.5

What neighbours means

Full description is available in DØNotes

4124, 4146

Available in D0Reco in shadow mode

Not running yet in killing mode:

We wanted to have first a tmbfix w/o T42

Let’s go for it Now!

ADMNov-7-2003

G. Bernardi for CALGO

T42 on Missing and Scalar ET

WZ skim , p13.06 , W e nu selection (ET > 20 GeV, MET > 25 GeV)

Compared to Pythia MC

No T42

After T42

ADMNov-7-2003

G. Bernardi for CALGO

Run Selection with T42

Define the average shift:

MET-xy = sqrt [ (<MET-x>)2 +(<MET-y>)2] RMS-xy = sqrt [ ( RMS-x )2 + ( RMS-y )2 ]

To declare a RUN “GOOD”:

1) Require MET-xy < 6 GeV in all files of a run, and MET-xy < 4 GeV in average

2) Require RMS-xy < 18 GeV in all files of a run and RMS-xy < 16 GeV in average

3) Require scalar ET > 70 GeV in all files of a run and scalar ET > 80 GeV

Compare Run selection with and without T42 on 55 pb-1 of p14.03 data

ADMNov-7-2003

G. Bernardi for CALGO

Met-xy Shift

P14.03.02

Met-xy Shift

same data

With T42

One entry= 1 file or 20 lumi blocks

Shift in x,y of Missing ET

ADMNov-7-2003

G. Bernardi for CALGO

P14.03Jul-Aug 2003 one entry / good “file”

(about 20 lumi blocks)

Same p14.03 files

but processed with

T42

xy-shift in Missing ET, xy-rms

ADMNov-7-2003

G. Bernardi for CALGO

Scalar ET

P14.03.02

Scalar ET

With T42

T42 is “cleaner”

Scalar ET evolution with Lumi

ADMNov-7-2003

G. Bernardi for CALGO

Scalar and Missing ET (setb,metb)

P14.03Jul-Aug 2003 one entry / good “file”

(about 20 lumi blocks)

Same p14.03 files but processed with T42

SET reduced by 21 GeV MET unchanged !

ADMNov-7-2003

G. Bernardi for CALGO

Run Selection Results

Preliminary run selection performed on p14 data

With and without T42 T42 looks better

55 pb-1 recoed / 49 pb-1 good runs available in july-august i.e. 89% of the data

After T42 55 pb-1 recoed / 50 pb-1 good runs 91% of the

data

… and gain is higher if tightening the selection.

.

ADMNov-7-2003

G. Bernardi for CALGO

Top group’s alljets skim (reco version p13.06.01) : * Passes the 4JT10 trigger.

* At least 4 jets (JCCB). * HT > 100 GeV (just plain sum of uncorrected JCCB jets).

20000 events

Jets are JES corrected.No quality cuts applied.

t42 no t42

Jets properties before and after T42

Fake jets removed

p13

Jet multiplicity lowered by 0.5 jets /event :

ADMNov-7-2003

G. Bernardi for CALGO

Why do we need Jet-id ? Because there are noise/fake jets

What are the Noise/Fake jets ?1) Noise jets (partly cured by on-line/off-

line Data quality corrections)2) Fake jets can also be overcalibrated

“jets” (either by picking up noise, or in difficult regions like ICR)

What are the handles we have ?Precision readout estimators like EMF, F90 L1 information allows confirmation (in p14)Tracking also better in p14CPS in p14.05T42 (as a byproduct) in TMB-fix

More on Jet-id / Fake Jets

ADMNov-7-2003

G. Bernardi for CALGO

Jet-ID DefinitionsJID: 0.05<EMF<0.95, HOTF<10, CHF<0.4, N90>1

JID-f90: JID and ( if f90 < 0.5 or CHF < 0.15 )

L1: JID and ((CC or EC) and L1SET/pT >0.2 or ICD and L1SET/pT >0.1)

Track: JID and Sum tracks in 0.5-cone pT / jet pT > 0.05

L1/Track: L1 in CC and EC, Track in ICD

L1+Track: L1/Track + Track in CC

Good jets: JID & (L1 or Track confirmation) Bad jets : JID but no L1 and no Track Rest: which includes “p14 good” jets (JID-f90)

ADMNov-7-2003

G. Bernardi for CALGO

Number of bad/good Jets in 50k dijet sample

all erem hotf emf> emf< n90 JID f90 L1 L1&T L1/T or L1t Track coml comt

6544 6535 6533 6488 6480 6479 6347 4438 4224 3290 3839 4166 4034 3674 4052 3862

2797 2790 2789 2753 2731 2730 2725 2552 1896 1516 1789 1882 1830 1659 1860 1794

188 187 186 177 175 175 151 97 103 79 94 99 100 90 98 95 95 93 93 89 83 83 82 78 54 39 49 53 52 46 53 51

1138 1111 1110 1102 1085 1085 1063 1007 1058 953 1019 1043 1055 958 1043 1040

1296 1239 1239 1233 1222 1222 1212 1184 1203 1086 1166 1198 1202 1093 1197 1196

dijet_GRL: djT42GRL:

dijet_GRL:

diT42GRL:

dijet_GRL:

djT42GRL:

black: bad jets from high jet multiplicity

green: bad third jet in dijet event

blue: good dijet events

GRL = good run list applied

Bad Jets

Bad Jets

Good Jets

ADMNov-7-2003

G. Bernardi for CALGO

all erem hotf emf> emf< n90 JID f90 L1 L1&T L1/T or L1t Track coml comt

100 99.9 99.8 99.1 99 99 97 67.8 64.5 50.3 58.7 63.7 61.6 56.1 61.9 59 100 99.7 99.7 98.4 97.6 97.6 97.4 91.2 67.8 54.2 64 67.3 65.4 59.3 66.5 64.1

100 99.5 98.9 94.1 93.1 93.1 80.3 51.6 54.8 42 50 52.7 53.2 47.9 52.1 50.5 100 97.9 97.9 93.7 87.4 87.4 86.3 82.1 56.8 41.1 51.6 55.8 54.7 48.4 55.8 53.7

100 97.6 97.5 96.8 95.3 95.3 93.4 88.5 93.0 83.7 89.5 91.7 92.7 84.2 91.7 91.4

100 95.6 95.6 95.1 94.3 94.3 93.5 91.4 92.8 83.8 90 92.4 92.7 84.3 92.4 92.3

black: bad jets from high jet multiplicitygreen: bad third jet in dijet event blue: good dijet events

dijet_GRL: djT42GRL:

dijet_GRL:

diT42GRL:

dijet_GRL:

djT42GRL:

Bad Jets

Bad Jets

good Jets

Efficiencies for bad/good Jets in 50k dijet sample

ADMNov-7-2003

G. Bernardi for CALGO

Conclusions on jet-id / fake&noise jets

Some properties of the noise jets: low pT, sit mainly in the ICD and CH fake high met, large HT and high jet Multiplicity

the noise is found where we look for interesting physics

In p13 LP03 sample only 45 highest ET towers: lower efficiency to find L1 confirmed jets!

T42: Significant reduction of bad jets due to noise

Use T42 and Use L1 confirmation for Jet-id. Use Tracks to monitor jet-id efficiencies

ADMNov-7-2003

G. Bernardi for CALGO

T42 Reconstruction efficiency in MC (qcd pT20)

Efficiency lower with t42 when pTjet 8 GeV. Gain a lot of efficiency by lowering the threshold to 6 GeV

What about noise jets with this new threshold ?

Method : match particle jets to calorimeter jets with R 0.5

reco efficiency in pt bins = particle jets found in calo / all particle jets

ADMNov-7-2003

G. Bernardi for CALGO

New low pT good jets (reco efficiency higher).

Data (alljets skim)

New jets found with [t42, pTjet 6 GeV] in second bump may also be low pT good jets. Even if they are fake, situation better than in the [not42, pTjet 8 GeV] case.

T42 Reconstruction efficiency in DATA

ADMNov-7-2003

G. Bernardi for CALGO

Jet Energy loss with T42

735 MeV difference for the same jet need new JES correction

Data (alljets skim - top group)

MC (single top events) :

with noise

without noise

T42 cuts mostly noise but also a bit of real energy

ADMNov-7-2003

G. Bernardi for CALGO

Jet Energy resolution with T42 in MC (qcd pT20)

Method : match particle jets to calorimeter jets with R 0.5 E / E = (E particle jets – E cal jets ) / E particle jet distribution in pT bins

Fit with a gaussian in each pT bin.

ADMNov-7-2003

G. Bernardi for CALGO

T42 resolution better at low energy better fit with T42

T42 Results on Jet Resolution (MC)

No T42

T42

ADMNov-7-2003

G. Bernardi for CALGO

T42 on high energy electronsEstimators for EM candidates in data, when pT>13 GeV

OK !

ADMNov-7-2003

G. Bernardi for CALGO

no T42 Z ee with T42

Similar results with or w.o. T42.

Slightly better results with T42

ADMNov-7-2003

G. Bernardi for CALGO

Effect on low Energy Electrons selected with the

Road Method

• Small gain ( 0.1 %) at low energy (improvement limited by the number of isolated tracks )

Use all available p13.06.01 data.

J/ selection criteria:

• pairs of road electron candidates with opposite charge,

• || < 1.1 and pT > 4 GeV for each electron candidate

• pT asymmetry

where pT,1 and pT,2 are the transverse momenta of the two electron candidates,

• pT > 12 GeV for J/ candidate.

ADMNov-7-2003

G. Bernardi for CALGO

Effect of T42 on Soft Electrons

290 +- 17 J/Psi

286 +/- 17 J/Psi

• Slightly more J/Psi• Better Soft Electron Tag

efficiency (not shown)

TighSoftEl : 882 % no T42 922 % with

T42Is there a Fake rate issue at low Energy ?

ADMNov-7-2003

G. Bernardi for CALGO

T42 on EM objects at low energy estimators pT<13 GeV

Fake rate increases below 6 GeV: Tuning of estimator cuts needed

If track-match required

No big diff. for Ee > 6 GeV

ADMNov-7-2003

G. Bernardi for CALGO

Fake Rate at 20 and 15 GeV from MW

20 GeV

15 GeV

T42 vs. no-T42

fake

ADMNov-7-2003

G. Bernardi for CALGO

T42+“tmb-fixed” W e nu sample

P13

P13 with TMB fix and T42

Em-id:• Em Fraction > 0.90• Isolation < 0.15• Em HM8 < 20• || < 1.1, In Fiducial vol.

• pTe > 20 GeV

•MET > 25 GeV

Standard Jets Criteria:• Em Fraction < 0.95• HOTF < 10• f90 < 0.5 or CHf<0.15 •|| < 2.5• PT > 15 GeV

(left: loose elec.sel.right: tight elec. sel.)

ADMNov-7-2003

G. Bernardi for CALGO

Missing ET , W Transverse Mass

P13

P13 with TMB fix and T42

ADMNov-7-2003

G. Bernardi for CALGO

Eta of Leading Jet

P13 (left: loose elec. sel.right: tight elec. sel.)

huge change in the distribution. Coming from the change of ICD Cal_Weight & MG bug

P13 with TMB fix and T42

Preamps used in readout 10 pF feedback capacitor rather than the usual 5 pF found at these layers. Increase weight by ~ 10.5/5.5 (actually 1.95)

EC MH layers

CC MG layer

ICD

ICD response highered by factor 3.8

CCMG and EC FH were highered twice BUG

ADMNov-7-2003

G. Bernardi for CALGO

Leading Jet Eta (e.m.+MET)

MG bug ICR x 3.8 No MG

bug ICR x 3.8

No MG bug ICR x 1.0

Horns at Eta +/- 1.2 MG bug +ICR boost

Smaller (and more central) w/o MG bug

No Horns in p13 situation.

ADMNov-7-2003

G. Bernardi for CALGO

ICD-MG Expert Comitee Composition:

Andy White (chair), Leslie Groer,Bob Kehoe,Ia Iashvili, Vivian O'Dell, Lee Sawyer, Dean Schamberger, Jan Stark, Alan Stone ,Vishnu Zutshi, Gregorio Bernardi (ex officio), Robert Zitoun (ex officio)

Charge of the comittee:

A) Review the harware motivations for having scaled up by a factor 3.8

the ICD response, and by a factor of 2 some of the MG, and report it in a

short document to the CALGO steering group (CAT) and algorithms chair.

The Jet energy scale and resolution issues are studied separately, inside

JES&CALGO.

B) Propose a complete calibration strategy (on-line,off-line) for the ICD,

including its phi intercalibration.

Be in charge of providing updates of the ICD/MG numbers if necessary.

ADMNov-7-2003

G. Bernardi for CALGO

ICD-MG Comitee Recomendations

 

1)      The ICD ADC to GeV conversion factor should stay at its present value: 0.0000945 GeV/ADC count.

2)      The factor of 3.8 should not be used. For the Winter conferences we should adopt the conservative approach of using the ICD signals without a boost factor.

3)      More studies of the required ICD boost factor should be carried out as soon as possible and the best factor determined and then used in energy correction.

4) The channel to channel relative calibration for the ICD should be tested and implemented as soon as possible.

ADMNov-7-2003

G. Bernardi for CALGO

P13 vs P13-fix/after MG-bug corrected

T42 MC: T42 vs. no-T42

Ready to look at finer details

ADMNov-7-2003

G. Bernardi for CALGO

Summary

CALGO group would like to have a full new round of TMBfixing, for having:

1) more efficient HC killing

2) ICR restored to the p13 status

3) T42 which will bring an overall improvement of calorimetric response.

Detecting muons in the calo with T42 implemented can be solved simply.

Other low energy issues can be tested in the next days if needed.

ADMNov-7-2003

G. Bernardi for CALGO

Comparison on a subset of All-jets skim <pTjet > = 28 GeV, (same events)

Rejected by the p13 cut

The number of jets rejected is ~ the same with both cuts within 5% About 10% of the jets are not classified as good jets with this cut in dataWith the new cut the rejection is stronger in the (eta= –1) ICR region, where noise is the most important, and it’s less important in the central region. Some good ICR jets are rejected by this cut (but also by the old cut)

Rejected by the p14 cut

chf > 0.15 && f90 > 0.5