Transcript ppt - High Energy Physics

Update on Tools
FTK Meeting
06/06/06
Erik Brubaker
U of Chicago
Physics Case (Reminder)
Statement #1
Statement #2
FTK vs TDR menus,
i.e. no tracking at
LVL2—ignore timing.
FTK vs nominal LVL2,
incl. tracking.
Fix LVL2 output
rate.
Limit LVL1 output
rate to sth
reasonable.
Optimize LVL1 cuts
& LVL2 tagging
reqs, maximizing
signal acceptance/
significance.
June 6, 2006
Need:
• FTK allows b-jet (t)
ID w/ eb, ec, eq, et.
• Correlation w/ offline?
• Depends on environment
Same as stmt #1,
but add trigger
timing into the eqs.
• LVL1 trigger rate, driven
by multijet evts, estimated
using modern generator +
parametrization of full
ATLAS LVL1 simulation.
But more realistic,
stronger case.
Hard to estimate,
large uncertainties.
This talk
FTK Meeting
2
Tools needed to estimate LVL1 rates
1. Multijet Monte Carlo with accurate treatment
of 3rd/4th jets  SHERPA.
–
Run through ATLFAST fast simulation.
2. Parametrization of LVL1 jet & tau trigger
response using full simulation.
3. Code to apply arbitrary set of trigger thresholds
to a given event, determine probability (weight)
for the event to pass.
June 6, 2006
FTK Meeting
3
Sherpa issues
• Several issues discovered with Sherpa samples:
– No b quarks in ME
• Clearly important for tagging rates. Fix next slide.
– No underlying event/mult. parton interactns (known)
•
•
•
•
Could be important, falling spectrum and all that.
Off by default in Sherpa, bugs/crashes when it’s turned on.
Improved by upgrading to last month’s Sherpa release.
Still under validation, so not in time for default samples.
– Some warnings + rare crashes fixed by upgrading to
Sherpa 1.0.8.
June 6, 2006
FTK Meeting
4
Sherpa: b quarks in ME
• No b quarks in ME for previously
produced Sherpa samples.
• Can add massive b quark by
explicitly writing out all processes:
– 2->2: 8 processes.
– 2->3: 10 processes.
– 2->4: 14 processes.
• Or declare b quark massless and it
comes for free with other quarks.
• Test samples of each: no difference
in kinematics, so use easier/quicker
massless b approach.
• But b jets are now clearly more
numerous and harder.
June 6, 2006
FTK Meeting
5
Final Sherpa samples
• Use Sherpa 1.0.8, massless b quarks in the ME,
UE off. (But in parallel work on validating a
sample with UE to determine effect.)
– 22 + 23 ME, ycut  25 GeV.
– Sample with 24 is running, but takes much longer.
• Additional samples can be created as needed.
– E.g. tried ycut  10 GeV, but found difficult to use
since effective statistics are miserable, and
understanding of low-pT jets/triggers is shaky at best.
June 6, 2006
FTK Meeting
6
Tau LVL1 trigger parametrization
• Two pieces
– Probability to pass iso cuts (2 GeV EM & HAD)
A  Landau ( pT , B, C)  D
– Energy of LVL1 cluster passing iso cuts
 2  Gaus( ETL1,  0 , 1 )  (1   2 )  Exp(  ETL1 /  3 );
 i  p2i  PTATL  p2i 1
• This characterizes the response of the LVL1 tau
trigger to jets. Would be useful to have as well
the response to real taus.
June 6, 2006
FTK Meeting
7
Tau LVL1 parametrization
• Histograms are dijet
MC samples, tau trigger
response for slices of
ATLFAST jet pT.
• Curves are from the
fitted parametrization.
• Good (enough)
agreement in both prob
to pass isolation cuts
and energy response.
• Param valid for ATL
pT>10, LVL1 t ET>10.
June 6, 2006
FTK Meeting
8
Jet LVL1 parametrization
Response in
slices of
ATL pT.
(From earlier
talk)
June 6, 2006
FTK Meeting
9
LVL1 parametrization
• Have characterization of jetjet, jettau.
– Would be useful to know tautau.
• Have code that, given the parametrization and a
list of jets for one event, returns the probability
that N of the jets pass a given trigger threshold.
– Should extend it to multiple thresholds.
June 6, 2006
FTK Meeting
10
Conclusions
• LVL1 rate tools are mostly final.
– Took a long time, but was also my introduction to
doing analysis/software work in ATLAS.
– Work on Sherpa can perhaps be used by rest of
ATLAS.
• Write up the tools as part of FTK physics case
note.
• Focus on physics channel (Hhhbbbb) for
next time.
June 6, 2006
FTK Meeting
11