Original Date: Tue, 13 Aug 2002
Updated:       Wed Aug 14, 2002
From: Marek Zielinski <marek@d0mino.fnal.gov>
To: d0caltf@fnal.gov
Subject: CTF plan for MC generation

Hi all,

We had the first Calorimeter Task Force meeting this Monday.
Part of the discussion concerned deciding the Monte Carlo data
samples to be used in subsequent studies.
We should start generating these files as soon as possible
to ensure fast progress on the studies of zero-suppression, noise
and non-linearity effects and comparisons to data.

It quickly  became very clear that for CTF mode of operation
it is important to have the ability to reprocess various stages
of the Monte Carlo generation without repeating the previous
stages. In particular, this includes re-processing the d0sim stage,
and hence requires storing the d0gstar output (which may not be the
usual practice for Monte Carlo production).

The initial request is for the following d0gstar samples:

1. empty events (eg. single neutrinos)  - 10k
   for pure noise simulation/studies
2. QCD events for different pt thresholds
   (  - pt > 10 GeV    -  10K ?)
     - pt > 20 GeV    -  100K
     - pt > 40 GeV    -  100K
     - pt > 80 GeV    -   10K
   for jet resolution/shapes, EM fakes, QCD backgrounds
3. Z->ee (60<mass<130 GeV)  - 20K
   for EM resolution, shower shapes
4. Z->tau tau (60<mass<130 GeV)  - 20k
   for tau characteristics, separation from QCD jets
5. W->enu  - 20k
   for MET studies (esp in events with jets), EM properties
6. ttbar->enu4j  - 10k
   for MET and jet studies in a complex enviroment
7. gamma+jet  - 10k  (pt cut to be specified)
   for jet response, MET, photon shower shape
8. H(bbbar)Z(nunu) (H mass = 115 GeV)  - 10k
   for MET studies
9. min bias events, if needed, to overlay on the above physics events
   (+noise) with 0.1 and 0.5 Poisson averages

All of the samples should be generated using plate calorimeter geometry,
and the same production release (hence we request the samples even if
similar previous samples exist. Since  the original d0gstar files
have likely been discarded, the steps below could not be performed
in a well controlled way.).
We envision processing the files several times trough d0sim with varying
noise and min bias overlay choices, and then through reco, with varying
zero suppression levels.
Empty-event request does not seem to require the full power of the
MC production farms, but we decided to include it with the rest to avoid
issues of using different computers or releases, and to make the
subsequent steps uniform.

We will provide the data cards upon contacting relevant experts in
the ID/physics groups, to maintain maximal consistency with
previous standard choices.

The current scheme for generating calorimeter noise in d0sim requires
several verification checks, possibly rewriting parts of the code, and
studies/decisions concerning the numerical values of the noise levels
(plus identifying personpower for longer term upkeep of the package).
Thus, it may be not practical to proceed with the d0sim and reco steps
immediately, and the cpu time would be best used for producing
the d0gstar-level files for rapid use in near future.
At this point, we have not identified the need to change d0gstar settings
or code.

Please comment on this plan before we make a request to the MC production
crews for generation of events. We would appreciate responses in time for
the discussion at the simulation meeting this Wednesday.

Thanks.
Leslie and Marek