LVL1 rates & H hh4b

LVL1 rates & Hhh4b

Erik Brubaker

U of Chicago

September 7, 2006 FTK Meeting 2

Outline• Almost-final word on LVL1 jet rates.

• Understanding cross-section for Hhh4b process.

• Talk descoped due to UC Tier2 going down last night…

September 7, 2006 FTK Meeting 3

Jet rates from Sherpa• Check different ME

combinations, kT,cut values to balance stats/resources, and to estimate systematics.

• LVL1 rates scatter in 40-80 Hz.

• “LVL2” rates require 4 tags, have units of mHz! Not too interesting as is.

SampleLVL1 4jet(40)

rate@1033 [Hz]

LVL2 4tag rate@1033 [10-3 Hz]

2->2+3 @ 10 38.5 ± 4.4 0.2 ± 0.2

2->2+3 @ 25 47.5 ± 1.0 3.0 ± 2.9

2->2+3 @ 50 50.7 ± 0.3 0.8 ± 0.4

2->2+3+4 @ 10

42.0 ± 22.7 0.0 ± 0.0

2->2+3+4 @ 25

65.9 ± 3.8 0.0 ± 0.0

2->2+3+4 @ 50

56.2 ± 0.6 0.4 ± 0.2

2->3 @ 10 58.4 ± 2.3 0.1 ± 0.1

2->3 @ 25 58.7 ± 0.4 0.3 ± 0.0

2->3 @ 50 66.7 ± 0.2 0.8 ± 0.1

2->3+4 @ 10 75.2 ± 32.5 0.0 ± 0.0

2->3+4 @ 25 82.8 ± 4.0 18.7 ± 18.6

2->3+4 @ 50 73.7 ± 0.6 2.3 ± 1.3

September 7, 2006 FTK Meeting 4

Cross-check: jet rates from Alpgen• Samples from

Ambreesh/SUSY group.• ALPGEN light jet

samples only.– LVL1 rates slightly

underestimated• Total rate lower than, but

on the order of, Sherpa results.

• Caveat: 4- and 5-jet samples ntupled with 0.7-cone jets—inconsistent with parametrizations…

SampleLVL1 4jet(40) rate@1033 [Hz]

3 jet 8.79 ± 0.03

4 jet 15.71 ± 0.01

5 jet (incl) 9.64 ± 0.02

Total 34.14 ± 0.04

September 7, 2006 FTK Meeting 5

Cross-check: jet rates from Pythia• Check against Pythia dijets

• Generic QCD incl heavy flavor

SampleLVL1 4jet(40) rate@1033 [Hz]

pT > 50 GeV 95.0 ± 11.9

So…

1. Consistency among models?

2. All are consistently lower than TDR rate:200 Hz for 4j@90 GeV, where 90~=60 inour calibration…

Caveat: DefaultPythia settings forUE etc. have beenunstable…

September 7, 2006 FTK Meeting 6

Systematics on LVL1 jet rates• Variations in ycut, ME configurations in

Sherpa– RMS of the 12 sample results, weighted by

reported stat unc.: 12%

• Differences among generators– Alpgen lower, Pythia higher with caveats:

30% for now

• Additional possible factors—need to check explicitly?– PDFs

– UE/MPI

– Parametrizations

RMS=13.1

RMS=7.5

September 7, 2006 FTK Meeting 7

LVL1 rates conclusion• Need to finish checking effect of UE/MPI in Sherpa,

also different pythia settings (MWT2 died on me last night).

• Run this by ATLAS trigger gurus.– Any similar recent studies?– Difference w.r.t. TDR understood?

Recommendation for FTK studies:

Use Sherpa sample with best combination of statistics,ME configuration, and kT,cut for your proposed thresholds.(Best to try more than one.)

Use my machinery to calculate rates for a range of thresholds.

Assign 35% uncertainty on absolute rate.http://hep.uchicago.edu/~brubaker/ftk

September 7, 2006 FTK Meeting 8

Example LVL1 rate scan

Require 4 jets:1st & 2nd at cuton x-axis;3rd & 4th at cutin legend.

4th jet cutdrives rate—not surprising

September 7, 2006 FTK Meeting 9

Hhhbbbb sigma*BR• I had a problem with low *BR (2 fb) reported by pythia

for Hhhbbbb w/ mh=130, mH=300.– What MSSM model was pythia assuming?

• Turns out Hhh is large only for lowish tan, and for 2mh<mH<2mt.

• To find operating points:– Use mh

max scenario to stay away from LEP direct search limits.– For a given tan, scan mA to maximize (ggH) x BR(Hhh) x

BR(hbb) x BR(hbb).

tan 2.5 2.75 3.0 3.25 3.5 4.0 4.5 5.0

mA 360 260 230 240 240 250 260 260

*BR [fb] 233 1019 1060 818 637 398 250 168

Additional slides

September 7, 2006 FTK Meeting 11

Physics Case (Details)Statement #1

FTK vs TDR menus,i.e. no tracking atLVL2—ignore timing.

Statement #2

FTK vs nominal LVL2,incl. tracking. Need:

• FTK allows b-jet ()ID w/ b, c, q, .

• Correlation w/ offline?• Depends on environment

• LVL1 trigger rate, drivenby multijet evts, estimatedusing modern generator +parametrization of fullATLAS LVL1 simulation.

Fix LVL2 outputrate.

Limit LVL1 outputrate to sth reasonable.

Optimize LVL1 cuts& LVL2 tagging reqs, maximizing signal acceptance/significance.

Same as stmt #1,but add triggertiming into the eqs.

Hard to estimate,large uncertainties.

But more realistic,stronger case.

September 7, 2006 FTK Meeting 12

CKKW vs MLM• CKKW: 2 parts

– Reweight ME to reproduce PS behavior for soft emissions.

– Reject showers that overlap higher-order ME.

• Boundary is kT,cut

or ycut kT,cut2/s.

• MLM: matching prescription– Cluster partons after shower,

compare to ME partons.

– Require each ME parton to match a unique “jet”, else reject event. ALPGEN

SHERPA

4p ME 3p ME

CKKW

MLM