CDF TOP CROSS SECTION MEASURMENT IN THE ALL HADRONIC CHANNEL

Authors (cdf-tophad@fnal.gov)
Authors (cdf-tophad@fnal.gov)	A. Gresele, I. Lazzizzera (Univ. of Trento) A. Castro, F. Margaroli (Univ. of Bologna) P. Azzi, G. Cortiana, T. Dorigo (Univ. of Padova) J. Konigsberg, G. Lungu, A. Sukhanov (Univ. of Florida)

tt Cross Section Result (for M_top=178 GeV/c², using 311 pb^-1): 7.5 +- 1.7 (stat.) ^+3.3_-2.2 (syst.) pb

Description (Conference_Note.ps)
At the Tevatron the top decays in Wb ~100% of the time. When both the W's decay hadronically the final state is called "all hadronic" and is characterized by a large jet multiplicity. This signature is overwhelmed by a QCD background that is orders of magnitudes larger. We consider events satisfying a multijet trigger and to reduce the background and increase the S/B a kinematical selection is performed. The variables used for the kinematical selection are: total sum of the jet Et's (SumEt), total sum of the sub-leading jet Et's (SumEt-Et1-Et2), Aplanarity and Centrality. The signal region is defined for 6<=N(jet)<=8. Silicon Vertex b-Tagging (SECVTX) is then required to improve the S/B even more. The measurement uses the total number of positively tagged jets for the final cross section extraction. To estimate the background after the btagging requirement we use a 3-dim parametrization of the positive tag rate in the multijet sample in a region where the contamination of the signal is minimal. We verify that before any kinematical selection, the background estimate obtained with this method agrees with the observed number of tags as a function of the jet multiplicity., After the kinematical selection an excess of the data over the background is observed in the signal region consistent with Standard model expectation, while a good agreement is mantained in the control region (N_jet<6).

Description (Conference_Note.ps)

At the Tevatron the top decays in Wb ~100% of the time. When both the W's decay hadronically the final state is called "all hadronic" and is characterized by a large jet multiplicity. This signature is overwhelmed by a QCD background that is orders of magnitudes larger. We consider events satisfying a multijet trigger and to reduce the background and increase the S/B a kinematical selection is performed. The variables used for the kinematical selection are: total sum of the jet Et's (SumEt), total sum of the sub-leading jet Et's (SumEt-Et1-Et2), Aplanarity and Centrality. The signal region is defined for 6<=N(jet)<=8.
Silicon Vertex b-Tagging (SECVTX) is then required to improve the S/B even more. The measurement uses the total number of positively tagged jets for the final cross section extraction.
To estimate the background after the btagging requirement we use a 3-dim parametrization of the positive tag rate in the multijet sample in a region where the contamination of the signal is minimal.
We verify that before any kinematical selection, the background estimate obtained with this method agrees with the observed number of tags as a function of the jet multiplicity.,
After the kinematical selection an excess of the data over the background is observed in the signal region consistent with Standard model expectation, while a good agreement is mantained in the control region (N_jet<6).

Dataset

We are using for the data the sample collected (311pb^-1, up to August 04) with a specifically developed multijet trigger that requires 4 high Pt jets (>15 GeV) and a large total event transverse energy (>125 GeV). For the signal we have Monte Carlo samples generated with Pythia and Herwig (with a top mass of 178 GeV/c²) under several conditions. Jets are corrected for detector and cone size effects.

Pre-requisites

In order to begin with a well defined sample, we apply the follow pre-requisites:

1) number of tight lepton = 0 (to guarantee the orthogonality with lepton+jets analysis);

2) |Zjvert| < 60 cm;

3) |Zjvert - Zpvert| < 5 cm;

4) N(vert_class12) >= 1

5) MET_significance < 3 (to guarantee the orthogonality with tau+jets analysis).

Kinematical selection

We follow 2 different approaches:

1) if we consider only the statistical uncertainty then we seek the best statistical significance, i.e. the largest S/sqrt(B+S) achievable;

2) if we consider also the systematic uncertainty, the best significance of the measurement corresponds to the maximum of S/sqrt(B+S+(K_b*B)*(K_b*B)+(K_s*S)*(K_s*S)) where K_b = 5% and K_s = 25% are the expected relative systematic uncertainties on background and signal.

We'll choose the best one using the pseudoexperiments.

1st Kinematical selection

Using only the statistical significance, we find:

1) 6 <= Njet <= 8;

2) Centrality >= 0.78;

3) SumEt >= 280 GeV;

4) (Aplanarity + 0.005 * SumEt3) >= 0.96

Variables used in the 1st kinematical selections. (gif), (eps)

2nd Kinematical selection

Using only the total significance (statistical and systematic), we find:

1) 6 <= Njet <= 8;

2) Centrality >= 0.8;

3) SumEt >= 280 GeV;

4) (Aplanarity + 0.003 * SumEt3) >= 0.76

Table 1, Systematics
Source	Systematic (%) on 1st kin. sel.	Systematic (%) on 2nd kin. sel.
Generator	1.7	2.2
JES	19.4	24.6
ISR/FSR modeling	3.9	5.9
PDF	2.6	3.9
TOTAL	20.0	24.6

Table 2, kinematical selections table summary
	1st kinematical selection	2nd kinematical selection
Signal Inclusive Efficiency (%)	6.7 +- 1.4	2.2 +- 0.6

Average number of tagged jets

Following the cominatorial method, we estimate the average number of tagged jets, in particular:

1) for 1st kin. sel., N_ave_jet = 0.845 +- 0.073;

2) for 2nd kin. sel., N_ave_jet = 0.862 +- 0.076.

3 dim Matrix

This matrix is made starting from a sample with only 4 jets and using:

1) 4 bins in Nv12,

2) 13 bins in Ntrk,

3) 5 bins in Et

The agreement is good when we apply this matrix on the sample before the kinematical selection.

If we apply the matrix after the kinematical selection, we can estimate the background in our data sample. But the presence of the tt events in the pre-tag sample leads to an overestimate of the background so we can account for it with an iterative procedure to rescale the background. We summarize all the number in the follow table:

Table 3, background estimation
	1st kin. sel.	2nd kin. sel.
Data	816	171
Background corrected	683.7 +- 36.7	127.6 +- 7.5
Data-Back(corr)	132.8	43.4


Number of tagged jets vs jet multiplicity. Data(points), expected background (iteratively corrected) and ttbar expectation (assuming xsec=6.1pb) after the 1st kinematical selection. (gif), (eps)	Number of tagged jets vs jet multiplicity. Data(points), expected background (iteratively corrected) and ttbar expectation (assuming xsec=6.1pb) after the 2nd kinematical selection. (gif) (eps)

Table 4, Tagging Related Systematics
Variable	Systematic (%) on 1st kin. sel.	Systematic (%) on 2nd kin. sel.
Apla+K Sumet3	2.6	2.1
Centrality	0.08	0.2
SumEt	0.37	0.14
Common	3.2	3.8
Luminosity	<< 1	<< 1
Run Number	<< 1	<< 1
TOTAL	4.1	4.3

Pseudo-experiments result

We have 2 possible kinematical selections and we want to choose the one which provides the best expected cross section and the smallest expected uncertainties. To do so:

1) we perform 2 sets of pseudoexperiment where we use integrated luminosity, kinematical efficiency, average number of tags (to obtain the expected signal) and the expected background;

2) signal and background are fluctuated according to the corresponding uncertainties, while the number of expected candidates n=S+B is fluctuated with a sigma = sqrt(S+B)

The spread in expected cross section is

1) sigma_tt = 6.1 +3.8 -3.2 pb for the 1st kin. sel.;

2) sigma_tt = 6.1 +4.1 -3.2 pb for the 2nd kin. sel.;

Because the spread in the 1st kinematical selection is smaller, we choose this selection to calculate the final cross section.

Expected cross section for the different kinematical selections. (gif), (eps)

Ingredients

The ingredients for the cross section are:

Table 5
Ingredient	Value
Luminosity	(311 +- 18) pb-1
Kinematical efficiency	(6.7 +- 1.4) %
Average number of tags	(0.845 +- 0.073)
Pre-tag Events (nJ>=6)	3315
Tagged Events (nJ>=6)	816
Background (nJ>=6)	(717.2 +- 29.4)
Background corrected(nJ>=6)	(683.7 +- 37.5)

The final cross section measurement

The final ttbar cross section measured in the all hadronic channel is:

sigma_tt = 7.5 +- 1.7 (stat) +3.3 - 2.2 (syst) = 7.5 +3.7 -2.8 pb