System Test with MOKKA input

1
System Test with MOKKA input

• currently look at Z-peak results, using the
  sample Ben generated on the csf farm
• results look considerably different from the
  ones of Marlin with SGV input, due to effects
  only
• simulated in the full MC (non-Gaussian tails
  on angular distributions from multiple
  scattering)
• does not make much sense to compare to
  Marlin(SGV input), instead compare with BRAHMS
• two BRAHMS results available, which differ due
  to different level of cheating, differences in
• track selection etc TESLA-TDR result and
  plots from LC-note LC-PHSM-2003-061
• MARLIN(Mokka input) uses same track selection
  that was used in LC-PHSM-2003-061 which was
• derived from the BRAHMS sample should be a
  good first approximation of what will be
• needed for MOKKA, though this should be
  revisited after the release
• in addition, in MARLIN we currently suppress
  K-shorts and Lambdas and switch off photon
• conversions in GEANT4
• MOKKA detector has layer thickness 0.12 X0,
  known to considerably degrade performance

2
The two BRAHMS results in comparison
b
c (b-bkgr)
open TESLA-TDR full LC-PHSM-2003-061
c

• LC-note result uses more realistic tracking and
  track selection derived from the
• sample used performance slightly worse than
  previous TESLA-TDR result
• results shown for nominal layer thickness at
  time of TESLA-TDR of 0.064 X0

3
MARLIN (MOKKA input) compared to BRAHMS
b
c (b-bkgr)
c
open BRAHMS, LC-note full MARLIN (Mokka)

• compare to result from LC-note (same track
  selection used in MARLIN)
• Marlin result falls significantly short of
  BRAHMS performance ? look at inputs

4
b
full MARLIN(MOKKA) line MARLIN(SGV input) for
some variables find excellent agreement for
example momentum of most significant track in the
jet
c
uds
5
b
significant difference seen in joint
probability (both in R-phi, see plots, and in
z) uds jets look like heavy flavour
c
uds
6
b
this is already seen at the level of the impact
parameter significances that are used to
calculate the joint probability
c
uds
7
b
c
uds
8
b
also, ZVRES seems to find secondary vertices in
light jets! (plots show number of tracks in
non-primary vertices)
c
uds
9
b
this is confirmed by the resulting
M_Pt distributions
c
uds
10
Preliminary conclusions

• confidence in the MARLIN code obtained from
  System Test with SGV input
• continues through to System Test with MOKKA
  input, e.g. some input distributions
• agree very well with MARLIN(SGV input)
• some of the degradation seen wrt SGV-input is
  expected from full MC and from the
• increased layer thickness used in the
  detector model
• however there are some effects needing further
  investigation
• ZVRES finds larger fraction of light jets with
  secondary vertices compared to
• cleaner SGV-events,
• impact parameter significances could also give
  rise to confusion of light with
• heavy flavour jets