Top Cross Section

Data file locations

• Data stripped to pretags:
  /data/harv11/b/Top533Strips/data_strip_tem_jetcorr04_updated032205_533_nt5.root
• Top MC strips:
  /data/harv12/e/Top533Strips/ttopel_strip_tem_jetcorr04_updated032205_533_nt_5*.root
• Other MC strips:
  /data/harv11/b/Top533Strips/*_strip_tem_jetcorr04_updated022305_533_nt_5.root

talk btag meet 22 apr 05

Download file

Signal Events in Gen5 and New Tags

This file contains detailed information on all the tagged signal events in the Gen5 top cross section.
For now, it is based on the unoptimized cross section. It also contains information about the events/jets that were not tagged in gen4 but there were tagged in gen5.

Download file

Pretag events in Gen5

This is the PRETAG
event list
from Sal. The list was made on March 15, 2005, possibly before the lastest bug fixes (i.e. removal of loose CMX/ tight BMU events).

Plots

• Compare tags

• 
  Met, total

• 
  Met, Signal regin, htmt

• 
  Met, signal region, no htmt

• Mt, htmt

• 
  Mt, pretags
  

Met in Signal region

It looks better than in the 2-jet bin. Don't tell anyone we looked.

Total
Leptons

New event selection

Since the majority of the excess in the one and two-jet bins looked like it was a low missing energy, and since these bins are supposed to be our control sample, I've done a quick and dirty cross section calculation using the standard event selection (no Ht or mT cuts) but requiring missing Et > 30. I had to make some assumptions about HF fractions, etc., but the resulting cross section is consistent with the other measurements and actually almost as good as the optimized one. See here for the plots.

Incidentally, with this MET cut, the excess is strictly confined to the muons. This is interesting. See here. The electron plot's gorgeous, isn't it?

2-Jet Bin Kinematics

Plots

JET50 Et spectrum comparison

Download file

Updates / Questions from Preblessing

Changes from blessing:

• LF asymmetry changed from 1.1 to 1.27 (changes bkgs by 2.2 events)


• Put systematic uncertanties in weighted average (very tiny effect, see table
  below)


• Use Mt cut in regions A+B for Non-W background estimate.
  
  
  
  • Met Vs Iso (no Mt Cut): Download file
    
  
  
  • Met Vs iso (Mt cut applied everywhere): Download file
    
  
  
  • MT in regions A+B for the 1+2 jet bins (only for Met > 10 GeV, but you get the idea.)Download file
    
  
  

  Numbers:
  

  			
  
                           1-jet		2-jet		3-jet		4-jet		5-jet		Signal (>= 3 jet)
  
  QCD, no sys in avg	29.8 +- 9.9	11.5 +- 3.9	1.2 +- 0.5	0.6 +- 0.4	0.3 +- 0.2	2.1 +- 1.1
  
  QCD, sys in avg		31.9 +- 9.1	11.8 +- 3.5	1.2 +- 0.4	0.6 +- 0.3	0.3 +- 0.2	2.1 +- 1.1
  
  QCD, Mt > 20,all reg's	39.3 +- 15.2	16.2 +- 6.5	1.9 +- 0.9	1.0 +- 0.7	0.3 +- 0.3	3.1 +- 1.9
  
  

  
  These plots illustrate that there is an effect if you apply the Mt cut, although the statistics
  then become very poor. My feeling is that we should apply the Mt cut for regions A+B as well, since it only makes a difference of 1+-1 event in the signal region. This moves the xs slightly in the third decimal place.
  


• These plots also show that the effect from taking Region C as the template for the Mt cut optimization will be completely negligible.


• Fixed bug in W+HF efficiency calculation. Was only using muon events, now use both
  (where available). No change in anything outside MC stats.
  
  
  
  
  

  	1-jet	2-jet	3-jet	4-jet
  1 species	35.1	37.4	40.3	41.0
  2 species	35.0	37.9	39.4	42.6

  These numbers have errors of order 1% absolute in the 3,4,5 jet bins, so nothing changed outside of MC stats.



• Use per-jet efficiency from 4-jet bin in 5-jet bin for W+HF. There are weirdnesses associated with the 5-jet bin because we do not have W+BB+3p, so we are relying on HERWIG entirely to get the spectrum right, and that's not the right thing to do.
• Changed K-factor for applying the mistag matrix to MC from 1.00 +- 0.26 to 0.88 +- 0.26.
  This is because the number of good tracks is different. I used JET50 data and PYTHIA MC, with Et > 50 GeV for these plots. I also checked for eta < 1.0, and there is no change in the K-factor. Plot:
  Download file.
  
  
  This is only used in PhiE, and it does not affect the result at all.
  


• Re-estimated PYTHIA vs HERWIG systematic. HERWIG now includes QED radiation, so we only have to compare PYTHIA default vs HERWIG default. We do, however, have to take into account the differences in BR(W-> lnu). PYTHIA uses 0.108 (measured by LEP) and
  HERWIG uses 0.111 (theoretical prediction). That is where the following reweighting comes from.
  

  
  
  Signal Region:
  
  Herwig:			8.99%
  
  Herwig, reweighted:	8.75%
  
  Pythia:			8.66%
  
  Change:			1.0%
  
  

  
  So we now use 1.0% as the systematic, which went down from 1.4% before.
• JES: This is not done yet

Questions and X-Checks

• B-tag eff vs njets in ttbar MC. Flat within statistics. Only used in PhiE, so has
  a small impact anyway: Download file
  


• Subsample cross checks:
  

  
  
  ******** NOTE: These were done using the numbers ***********
  
  ******** from the Full Status Report Optimized   ***********
  
  
  >=4 jet x-check
  
  	>=3 jets		>=4 jets
  
  Nobs	138			76
  
  Nbkg	32.6 +- 1.2		11.1 +- 1.6
  
  Den.	13.0 +- 1.4		7.6 +- 0.8
  
  XS	8.1 +- 0.9 +- 0.97	8.6 +- 1.2 +- 1.0
  
  These are about 50% correlated statistically. The systematics
  
  are 100% correlated.
  
  
  
  Ele Vs Muo
  
  	Ele			Muo
  
  Nobs	82			56
  
  Nbkg	20.8 +- 3.0		11.7 +- 1.7
  
  Den.	7.6 +- 0.9		5.6 +- 0.6
  
  XS	8.1 +- 1.2 +- 1.1	7.9 +- 1.3 +- 0.9
  
  These are 0% correlated statistically. The systematics are 
  
  about 90% correlated.
  
  
  
  ******** NOTE: These were done using the updated numbers ***********
  
  ******** in the UNOPTIMIZED measurement                  ***********
  
  Run Dependence (162 pb-1 vs 156 pb-1)
  
  
  
  	162			156
  
  Nobs	86			91
  
  Nbkg	29.7 +- 4.9		32.0 +- 5.3
  
  Den.	7.1 +- 0.8		7.0 +- 0.7
  
  XS	7.9 +- 1.3 +- 1.1	8.5 +- 1.4 +- 1.2
  
  These are 0% correlated statistically. The systematics
  
  are 100% correlated.
  
  

  
  


• New tags: Do they look like the old ones? I have a slew of plots here
  comparing various distributions. They compare the common tags in Gen4 and Gen5
  to the extra tags in Gen5 (in 162 pb-1). They also compare the distributions from ttbar.
  There are KS probabilities on the plots, one for the extra tags to the common tags, and another
  for the extra tags to the MC. No probabilities are heinous, although there are some differences.
  In essence, the old tags look the same as the new.

  Download file
  

SoftElectron Test

To test the SoftElectronTagger you can use the BTagMods/test/SoftElectronTaggerTest.cc

get the following packages from 6.1.0:
setup cdfsoft2 6.1.0
newrel -t 6.1.0 test_lset_610
cd test_lset_610
addpkg BTagMods ; addpkg BTagAlgs ; addpkg TopMods

On another directory get the code, which is in harv11's CVS repository:
export CVSROOT=/data/harv11/a/cvsroot
cvs co LSoftElectronTagger
cd LSoftElectronTagger
~gris/bin/bash/run_install.sh path_to/test_lset_610

In path_to/test_lset_610
gmake BTagMods.all
gmake BTagAlgs.all
gmake BTagMods.tbin

at this point you will have:
> ls bin/Linux2-GCC_3_4/
SecVtxTest* SoftElectronTaggerTest*

run SoftElectronTaggerTest* with:
> ./bin/Linux2-GCC_3_4/SoftElectronTaggerTest BTagMods/test/run_set_test.tcl

LSoftElectronTagger in harv11's cvs

To try it out, first:
setup cdfsoft2 6.1.0
newrel -t 6.1.0 r610
cd r610
addpkg BTagAlgs ; addpkg BTagMods ; addpkg TopMods

Now check out harv11's LSoftElectronTagger cvs package and copy the files
to the same directory structure in the 6.1.0 release area

compile:
gmake BTagAlgs.all ; gmake BTagMods.all ; gmake TopMods.bin
run:
bin/Linux2-GCC_3_4/TopFind TopMods/test/run_test_SET.tcl