At long last it is time to start generating mcp07. I apologize for the delay
but we had to wait for the release of the cardfiles containing the up to
data Minimum Bias Model. The following message contains instructions on what
to run in mcp07. I will also explain why we are running these versions of
the code.
mcp07
Generators: p07.00.05a
Dogstar: p07.00.05a
D0sim: p07.00.05a
d0reco: p08.12.00 maxopt
recoanalyze: p08.12.00 maxopt
mc_runjob: v03-03-05
cardfiles: v00-02-07
Most of this code should already be on your farms. The new code is for
running reco and reco-analyze, mc_runjob and the cardfiles. These can be
found in the following locations:
d0mino.fnal.gov:/d0dist/dist/minitar/tarfiles
mc-p07.00.05a-Linux2-KCC_3_4.tar.gz
sim-p07.00.05a-Linux2-KCC_3_4.tar.gz
rec-p08.12.00-Linux2-KCC_3_4-maxopt.tar.gz
mc_runjob_v03-03-05.tar.gz
cardfiles-v00-02-07.tar.gz
When unpacking the code you need to check a couple of things:
1/ In the mcc-dist directory check the file SRT_SUBDIR. This must
contain
Linux2-KCC_3_4
and nothing else. It must not have Linux2-KCC_3_4-maxopt or
your mc_runjob code will not work.
2/ There is a feature/bug in d0reco that requires it to have a link
to the MC magnetic field. Hence in the directory
mcc-dist/releases/p08.12.00
you need to make the following link:
ln -sf mcc_dist/releases/p07.00.05a/d0gstar d0gstar
3/ Unfortunately you need to run the jobs in 250 event blocks due
to the memory leak in recoanalyze.
Minimum Bias Model:
We will be changing the MB model for overlaying events in d0sim. The
model uses the new cardfiles package and to run you should use a macro
essentially identical to the attached file. You will need to change the
facility and site specific parts of the macro.
Cardfiles:
It is very simple to use the cardfiles package with mc_runjob, in addition
to
the production and decay strings, you provide the following two parameters:
cfg pythia define string CardfileVersion <version here: eg v00-02-07>
cfg pythia define string CardfileDir <directory of Cardfile pkg here: eg
np,
hit,ckm etc>
See http://www-clued0.fnal.gov/mc_runjob/mainframe.html
for instructions on
how to use the cardfiles.
WARNING: Do not use carddfiles that are not part of the official release. If
some-one wants a special sample using a new cardfile they have to get it
released (a quick procedure) by contacting a physics group representative.
Once released we can release a new version of the cardfiles package to the
farms. This will be a much faster process and does not require a full
production release of the software.
If you need to run a test of a new cardfile YOU MUST ensure that it does not
go into sam. Also remove the phase mcp07 and replace with a unique string
such as test01. Files that do not have an official release pass should not
go into SAM for the time being.
Cardfiles to use:
When obtaining a request to generate MC events (which should be
submitted centrally so we can keep track of the requests) you will be given
a cardfiles version, directory, and a cardfile name. For example to generate
QCD events you would be asked to generate:
v00-02-07 hit pythia_qcd.cards
and the usual PT ranges and number of events to overlay.
Future Plans:
Code Versions: We are getting a new release of p08 (p08.13.00) which
will remove the bug in recoanalyze. We will upgrade to this once it is
tested as part of mcp07. This will allow us to run 500 event files
again.
p09 is not certified. We will consider releasing this code when it is
certified. The same goes for p10.
mc-runjob is going to be integrated into the production releases. This
should speed up the certification. Some of the farms may be used for
helping run certification samples.
Job Submission
Short Term: I will be setting up a web page that describes what has to
be included in a submission. The jobs will be submitted to myself and a
deputy and distributed to the farms.
Long Term: Sam is being upgraded to be a great deal more flexible
with the description of MC samples. When this available we will
have a GUI/web interface for submitting requests.
Iain Bertram