BJES systematics

1
B-JES systematics

• CDF/D0 Systematics Meeting
• 11/27/07
• Erik Brubaker (CDF)
• Jochen Cammin (D0)

2
b-JES Overview

• bJES systematic Uncertainty in modeling of the
  b-jet energy response w.r.t. generic jets.
• Not the total uncertainty in b-jet energy
  response, just the b-q difference.
• Not the b-q difference itself, but the
  uncertainty.
• Developed in the context of 2dmeasurements
• Derive JES from W daughter jets,but b jets carry
  most Mtopinformation.
• But nothing special about 2dthe uncertainty is
  there for all analyses.

3
bJES components

• B fragmentation uncertainties
• B decay BR uncertainties (semi-leptonics!)
• Calorimeter response uncertainties
• Anything that induces a different scale for W
  daughters vs b jets in ttbar events!

4
Semi-leptonic BR details

• Pieces we need
• Definition of one sigma uncertainty.
• Store relevant info per event.
• Dredge truth information.
• Calculate reweighting factor.
• Example from CDF 2D template analysis.

5
What is the uncertainty in SLD BRs?

• From Z-peak physics summary
• Physics Reports 427 (2006) 257-454, Table 5.10
• B(b?l) 0.1071 0.0022
• B(b?c?l) 0.0801 0.0018
• B(c?l) 0.0969 0.0031
• From PDG 2007 update (meson summary tables pg
  81)B meson/baryon admixture decays
• nanything (23.1 1.5)
• enanything (10.86 0.35)
• mnanything (10.95 0.27)
• tnanything (2.48 0.26)

Averagesof le,m
6
Uncertainty in BRs cont.

• What to use for 1s in Mtop syst?
• Need to balance
• Adequately address thepossible systematics
• Keep it straightforwardto evaluate
• Check central value insimulation.
• Is it consistent?

Proposed systematic shifts B(b?l) 23.9
0.7. Includes b?t. B(c?l) 20 0.7. Use
for all charm, incl sequential decay Vary these
independently. Here correlated ! conservative.
7
SemiLepDecayInfo.hh/cc
Keep everything you need in one int (4 bytes).

• Count hadrons from bottom, charm in the entire
  event.
• Count b, c decays to e, m, t, or none of the
  above.
• Eight numbers (32 bits) characterizes the event.
• Works for background too.
• Retrieve the int from your ntuple and call a
  function.
• Only initialization from truth info
  is CDF/D0 specific.

Then calculate an event weight according to
decays.
8
Try it out

• Used 1s shifts from pg 6.
• Here b, c shifted together for simplicity.
• Reconstructed top mass shifts as expected.
• Magnitude is small O(100 MeV).
• Need PEs for final evaluation.
• Also try with 10s.

weights
ratio
comparison
10sshift!!
9
B fragmentation (CDF)

• See later talk for D0 treatment
• Both expts should use reweighting approach like
  D0.
• CDF doesnt store the fragmentation z parameter
  from pythia.
• Can recalculate it after the fact using
• How good is this approximation?
• Probably good enoughwe just need a handle to
  apply reweighting.

10
Calorimeter response effects

• The problem
• Even effects that are not specific to b jets can
  induce difference in the b/q response ratio, due
  to different characteristics of b jets.
• Example from D0 b, q jets have different fchgd.
  Thus uncert. on hadronic E/p (due to e/h
  uncert.) affect b, q jets differently and shift
  relative calibration.

Evaluate separately ateach experiment. Uncertaint
y may still becorrelated.
11
Conclusions

• A standard lineup of b-JES systematics should
  include
• B fragmentation uncertainty
• B decay uncertainty
• Calorimeter (hadronic) response effects.
• The list of things that possibly affect the b vs
  q response ratio is infinite.
• Cant check or add a systematic for them all.
• Be vigilant for new potential effects study and
  add them to the list as needed.