Transcript MC-driven mistags (normalization)

A Limit on the Branching Ratio of
the Flavor-Changing Top quark
decay t→Zc
Alexander Paramonov, Henry Frisch, Carla Pilcher, Collin Wolfe, Dan Krop
Exotics Meeting. April 03, 2008
Internal reviewers: Beate Heinemann, Kevin Lannon, and Jason Nielsen
CDF 9101 + http://www-cdf.fnal.gov/~paramon/internal/
Outline
• Figures to bless
• Additional statistical test of the
expected result for Br(tZc) = 20%
• Understanding of the expected limits for
the null-hypothesis
• “Evolution” of the limits (requests)
• MC-driven mistags (normalization)
• Conclusions and Results
Figures to bless. Tier 1
Figures to bless. Tier 1
Figures to bless. Tier 1
Figures to bless. Tier 1
Figures to bless. Tier 1
Figures to bless. Tier 2
Figures to bless. Tier 2
Statistical test for Br(tZc) = 20%
• 50 pseudo experiments
• 100% longitudinally polarized Z’s
• MC-driven mistags
The expected limits for the null-hypothesis
• We did 50 pseudo-experiments with Data-driven
mistags two times and we got slightly different
results each time:
a. 8.3±2.5% (Gaussian Prior) & 9.2±2.8%
b. 8.7±2.8% (Gaussian Prior) & 9.6±3.1%
• The code looks fine. We did add statistical
uncertainties in backgrounds to pseudo-experiments.
• There’s a slight dependence on the number of
ensembles in numerical integration.
• We run 50 more pseudo-experiments twice to
calculate the expected limits for MC-Driven mistags:
a. 8.9±2.8% (Gaussian Prior) & 9.9±3.2%
b. 8.X±2.X% (Gaussian Prior) & 9.X±2.X%
Evolution of the limits
1. Request from the internal reviewers to
change the limit estimation procedure
to a more conventional Bayesian
approach. The limit moved from
~12.8% (this in not a Bayesian limit)
to ~9.9-11% (depending on the prior).
2. The internal reviewers asked to
improve the uncertainty of the
mistagged events by using MC
samples. This changed the limits to
8.3-9.2% (depending on the prior).
MC-driven Mistags
• We normalize MC using pretagged sample of Z+jets.
• Z+HF events are taken into
account.
• Them we run the mistagg
matrix
• We get the following numbers
of expected mistaggs:
– A. Data: 2.96±0.50 (electrons),
2.2±0.5 (muons)
– B. MC: 3.3±0.03 (electrons),
2.30±0.02 (muons)
Conclusions and Results
• Upper limits on Br(tZc) for five
longitudinal polarizations of the Zbosons.
• The results are model-independent.
• The expected limits are consistent with
observations